ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1001 - (hide annotations) (download)
Wed Aug 8 10:18:25 2012 UTC (2 years, 8 months ago) by ph10
File size: 121898 byte(s)
Improve documentation of \c in EBCDIC mode.

1 ph10 964 .TH PCREPATTERN 3 "04 May 2012" "PCRE 8.31"
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 859 there is now also support for UTF-8 strings in the original library, and a
25     second library that supports 16-bit and UTF-16 character strings. To use these
26     features, PCRE must be built to include appropriate support. When using UTF
27     strings you must either call the compiling function with the PCRE_UTF8 or
28 ph10 903 PCRE_UTF16 option, or the pattern must start with one of these special
29 ph10 859 sequences:
30 ph10 412 .sp
31     (*UTF8)
32 ph10 903 (*UTF16)
33 ph10 416 .sp
34 ph10 859 Starting a pattern with such a sequence is equivalent to setting the relevant
35     option. This feature is not Perl-compatible. How setting a UTF mode affects
36 ph10 412 pattern matching is mentioned in several places below. There is also a summary
37 ph10 859 of features in the
38 nigel 63 .\" HREF
39 ph10 678 \fBpcreunicode\fP
40 nigel 63 .\"
41     page.
42 nigel 75 .P
43 ph10 535 Another special sequence that may appear at the start of a pattern or in
44 ph10 859 combination with (*UTF8) or (*UTF16) is:
45 ph10 518 .sp
46     (*UCP)
47     .sp
48 ph10 535 This has the same effect as setting the PCRE_UCP option: it causes sequences
49     such as \ed and \ew to use Unicode properties to determine character types,
50     instead of recognizing only characters with codes less than 128 via a lookup
51 ph10 518 table.
52     .P
53 ph10 576 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
54 ph10 579 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
55     also some more of these special sequences that are concerned with the handling
56 ph10 576 of newlines; they are described below.
57     .P
58 nigel 77 The remainder of this document discusses the patterns that are supported by
59 ph10 859 PCRE when one its main matching functions, \fBpcre_exec()\fP (8-bit) or
60     \fBpcre16_exec()\fP (16-bit), is used. PCRE also has alternative matching
61     functions, \fBpcre_dfa_exec()\fP and \fBpcre16_dfa_exec()\fP, which match using
62     a different algorithm that is not Perl-compatible. Some of the features
63     discussed below are not available when DFA matching is used. The advantages and
64     disadvantages of the alternative functions, and how they differ from the normal
65     functions, are discussed in the
66 nigel 77 .\" HREF
67     \fBpcrematching\fP
68     .\"
69     page.
70 nigel 93 .
71     .
72 ph10 556 .\" HTML <a name="newlines"></a>
74     .rs
75     .sp
76     PCRE supports five different conventions for indicating line breaks in
77     strings: a single CR (carriage return) character, a single LF (linefeed)
78     character, the two-character sequence CRLF, any of the three preceding, or any
79     Unicode newline sequence. The
80     .\" HREF
81     \fBpcreapi\fP
82     .\"
83     page has
84     .\" HTML <a href="pcreapi.html#newlines">
85     .\" </a>
86     further discussion
87     .\"
88     about newlines, and shows how to set the newline convention in the
89     \fIoptions\fP arguments for the compiling and matching functions.
90     .P
91     It is also possible to specify a newline convention by starting a pattern
92     string with one of the following five sequences:
93     .sp
94     (*CR) carriage return
95     (*LF) linefeed
96     (*CRLF) carriage return, followed by linefeed
97     (*ANYCRLF) any of the three above
98     (*ANY) all Unicode newline sequences
99     .sp
100 ph10 859 These override the default and the options given to the compiling function. For
101     example, on a Unix system where LF is the default newline sequence, the pattern
102 ph10 227 .sp
103     (*CR)a.b
104     .sp
105     changes the convention to CR. That pattern matches "a\enb" because LF is no
106     longer a newline. Note that these special settings, which are not
107     Perl-compatible, are recognized only at the very start of a pattern, and that
108 ph10 231 they must be in upper case. If more than one of them is present, the last one
109     is used.
110     .P
111 ph10 514 The newline convention affects the interpretation of the dot metacharacter when
112     PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
113     affect what the \eR escape sequence matches. By default, this is any Unicode
114     newline sequence, for Perl compatibility. However, this can be changed; see the
115     description of \eR in the section entitled
116 ph10 231 .\" HTML <a href="#newlineseq">
117     .\" </a>
118     "Newline sequences"
119     .\"
120 ph10 247 below. A change of \eR setting can be combined with a change of newline
121 ph10 246 convention.
122 ph10 227 .
123     .
125     .rs
126     .sp
127 nigel 63 A regular expression is a pattern that is matched against a subject string from
128     left to right. Most characters stand for themselves in a pattern, and match the
129     corresponding characters in the subject. As a trivial example, the pattern
130 nigel 75 .sp
131 nigel 63 The quick brown fox
132 nigel 75 .sp
133 nigel 77 matches a portion of a subject string that is identical to itself. When
134     caseless matching is specified (the PCRE_CASELESS option), letters are matched
135 ph10 859 independently of case. In a UTF mode, PCRE always understands the concept of
136 nigel 77 case for characters whose values are less than 128, so caseless matching is
137     always possible. For characters with higher values, the concept of case is
138     supported if PCRE is compiled with Unicode property support, but not otherwise.
139     If you want to use caseless matching for characters 128 and above, you must
140     ensure that PCRE is compiled with Unicode property support as well as with
141 ph10 859 UTF support.
142 nigel 77 .P
143     The power of regular expressions comes from the ability to include alternatives
144     and repetitions in the pattern. These are encoded in the pattern by the use of
145 nigel 75 \fImetacharacters\fP, which do not stand for themselves but instead are
146 nigel 63 interpreted in some special way.
147 nigel 75 .P
148     There are two different sets of metacharacters: those that are recognized
149 nigel 63 anywhere in the pattern except within square brackets, and those that are
150 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
151     are as follows:
152 nigel 75 .sp
153     \e general escape character with several uses
154 nigel 63 ^ assert start of string (or line, in multiline mode)
155     $ assert end of string (or line, in multiline mode)
156     . match any character except newline (by default)
157     [ start character class definition
158     | start of alternative branch
159     ( start subpattern
160     ) end subpattern
161     ? extends the meaning of (
162     also 0 or 1 quantifier
163     also quantifier minimizer
164     * 0 or more quantifier
165     + 1 or more quantifier
166     also "possessive quantifier"
167     { start min/max quantifier
168 nigel 75 .sp
169 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
170 nigel 75 a character class the only metacharacters are:
171     .sp
172     \e general escape character
173 nigel 63 ^ negate the class, but only if the first character
174     - indicates character range
175 nigel 75 .\" JOIN
176 nigel 63 [ POSIX character class (only if followed by POSIX
177     syntax)
178     ] terminates the character class
179 nigel 75 .sp
180     The following sections describe the use of each of the metacharacters.
181     .
182 nigel 93 .
183 nigel 63 .SH BACKSLASH
184     .rs
185     .sp
186     The backslash character has several uses. Firstly, if it is followed by a
187 ph10 574 character that is not a number or a letter, it takes away any special meaning
188     that character may have. This use of backslash as an escape character applies
189 ph10 579 both inside and outside character classes.
190 nigel 75 .P
191     For example, if you want to match a * character, you write \e* in the pattern.
192 nigel 63 This escaping action applies whether or not the following character would
193 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
194     non-alphanumeric with backslash to specify that it stands for itself. In
195     particular, if you want to match a backslash, you write \e\e.
196     .P
197 ph10 859 In a UTF mode, only ASCII numbers and letters have any special meaning after a
198 ph10 579 backslash. All other characters (in particular, those whose codepoints are
199 ph10 574 greater than 127) are treated as literals.
200     .P
201 ph10 968 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
202 nigel 63 pattern (other than in a character class) and characters between a # outside
203 nigel 91 a character class and the next newline are ignored. An escaping backslash can
204 ph10 968 be used to include a white space or # character as part of the pattern.
205 nigel 75 .P
206 nigel 63 If you want to remove the special meaning from a sequence of characters, you
207 nigel 75 can do so by putting them between \eQ and \eE. This is different from Perl in
208     that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
209 nigel 63 Perl, $ and @ cause variable interpolation. Note the following examples:
210 nigel 75 .sp
211 nigel 63 Pattern PCRE matches Perl matches
212 nigel 75 .sp
213     .\" JOIN
214     \eQabc$xyz\eE abc$xyz abc followed by the
215 nigel 63 contents of $xyz
216 nigel 75 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
217     \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
218     .sp
219     The \eQ...\eE sequence is recognized both inside and outside character classes.
220 ph10 654 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
221     by \eE later in the pattern, the literal interpretation continues to the end of
222 ph10 607 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
223     a character class, this causes an error, because the character class is not
224     terminated.
225 nigel 75 .
226     .
227     .\" HTML <a name="digitsafterbackslash"></a>
228     .SS "Non-printing characters"
229     .rs
230     .sp
231 nigel 63 A second use of backslash provides a way of encoding non-printing characters
232     in patterns in a visible manner. There is no restriction on the appearance of
233     non-printing characters, apart from the binary zero that terminates a pattern,
234 ph10 456 but when a pattern is being prepared by text editing, it is often easier to use
235     one of the following escape sequences than the binary character it represents:
236 nigel 75 .sp
237     \ea alarm, that is, the BEL character (hex 07)
238 ph10 574 \ecx "control-x", where x is any ASCII character
239 nigel 75 \ee escape (hex 1B)
240 ph10 968 \ef form feed (hex 0C)
241 ph10 227 \en linefeed (hex 0A)
242 nigel 75 \er carriage return (hex 0D)
243     \et tab (hex 09)
244 ph10 488 \eddd character with octal code ddd, or back reference
245 nigel 75 \exhh character with hex code hh
246 ph10 745 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
247 ph10 836 \euhhhh character with hex code hhhh (JavaScript mode only)
248 nigel 75 .sp
249 ph10 1001 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
250     case letter, it is converted to upper case. Then bit 6 of the character (hex
251     40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
252     but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
253     data item (byte or 16-bit value) following \ec has a value greater than 127, a
254     compile-time error occurs. This locks out non-ASCII characters in all modes.
255 nigel 75 .P
256 ph10 1001 The \ec facility was designed for use with ASCII characters, but with the
257     extension to Unicode it is even less useful than it once was. It is, however,
258     recognized when PCRE is compiled in EBCDIC mode, where data items are always
259     bytes. In this mode, all values are valid after \ec. If the next character is a
260     lower case letter, it is converted to upper case. Then the 0xc0 bits of the
261     byte are inverted. Thus \ecA becomes hex 01, as in ASCII (A is C1), but because
262     the EBCDIC letters are disjoint, \ecZ becomes hex 29 (Z is E9), and other
263     characters also generate different values.
264     .P
265 ph10 745 By default, after \ex, from zero to two hexadecimal digits are read (letters
266     can be in upper or lower case). Any number of hexadecimal digits may appear
267 ph10 859 between \ex{ and }, but the character code is constrained as follows:
268     .sp
269     8-bit non-UTF mode less than 0x100
270     8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
271     16-bit non-UTF mode less than 0x10000
272     16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
273 ph10 903 .sp
274     Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
275 ph10 859 "surrogate" codepoints).
276 nigel 75 .P
277 ph10 211 If characters other than hexadecimal digits appear between \ex{ and }, or if
278     there is no terminating }, this form of escape is not recognized. Instead, the
279     initial \ex will be interpreted as a basic hexadecimal escape, with no
280     following digits, giving a character whose value is zero.
281     .P
282 ph10 836 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
283     as just described only when it is followed by two hexadecimal digits.
284 ph10 745 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
285 ph10 836 code points greater than 256 is provided by \eu, which must be followed by
286 ph10 745 four hexadecimal digits; otherwise it matches a literal "u" character.
287 ph10 982 Character codes specified by \eu in JavaScript mode are constrained in the same
288 ph10 978 was as those specified by \ex in non-JavaScript mode.
289 ph10 745 .P
290 nigel 63 Characters whose value is less than 256 can be defined by either of the two
291 ph10 745 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
292 ph10 836 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
293 ph10 745 \eu00dc in JavaScript mode).
294 nigel 75 .P
295 nigel 91 After \e0 up to two further octal digits are read. If there are fewer than two
296     digits, just those that are present are used. Thus the sequence \e0\ex\e07
297     specifies two binary zeros followed by a BEL character (code value 7). Make
298     sure you supply two digits after the initial zero if the pattern character that
299     follows is itself an octal digit.
300 nigel 75 .P
301 nigel 63 The handling of a backslash followed by a digit other than 0 is complicated.
302     Outside a character class, PCRE reads it and any following digits as a decimal
303     number. If the number is less than 10, or if there have been at least that many
304     previous capturing left parentheses in the expression, the entire sequence is
305 nigel 75 taken as a \fIback reference\fP. A description of how this works is given
306     .\" HTML <a href="#backreferences">
307     .\" </a>
308     later,
309     .\"
310     following the discussion of
311     .\" HTML <a href="#subpattern">
312     .\" </a>
313     parenthesized subpatterns.
314     .\"
315     .P
316 nigel 63 Inside a character class, or if the decimal number is greater than 9 and there
317     have not been that many capturing subpatterns, PCRE re-reads up to three octal
318 nigel 93 digits following the backslash, and uses them to generate a data character. Any
319 ph10 903 subsequent digits stand for themselves. The value of the character is
320 ph10 859 constrained in the same way as characters specified in hexadecimal.
321     For example:
322 nigel 75 .sp
323     \e040 is another way of writing a space
324     .\" JOIN
325     \e40 is the same, provided there are fewer than 40
326 nigel 63 previous capturing subpatterns
327 nigel 75 \e7 is always a back reference
328     .\" JOIN
329     \e11 might be a back reference, or another way of
330 nigel 63 writing a tab
331 nigel 75 \e011 is always a tab
332     \e0113 is a tab followed by the character "3"
333     .\" JOIN
334     \e113 might be a back reference, otherwise the
335 nigel 63 character with octal code 113
336 nigel 75 .\" JOIN
337     \e377 might be a back reference, otherwise
338 ph10 859 the value 255 (decimal)
339 nigel 75 .\" JOIN
340     \e81 is either a back reference, or a binary zero
341 nigel 63 followed by the two characters "8" and "1"
342 nigel 75 .sp
343 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
344     zero, because no more than three octal digits are ever read.
345 nigel 75 .P
346 nigel 91 All the sequences that define a single character value can be used both inside
347 ph10 836 and outside character classes. In addition, inside a character class, \eb is
348     interpreted as the backspace character (hex 08).
349     .P
350     \eN is not allowed in a character class. \eB, \eR, and \eX are not special
351     inside a character class. Like other unrecognized escape sequences, they are
352     treated as the literal characters "B", "R", and "X" by default, but cause an
353     error if the PCRE_EXTRA option is set. Outside a character class, these
354     sequences have different meanings.
355 nigel 75 .
356     .
357 ph10 745 .SS "Unsupported escape sequences"
358     .rs
359     .sp
360     In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
361     handler and used to modify the case of following characters. By default, PCRE
362     does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
363     option is set, \eU matches a "U" character, and \eu can be used to define a
364     character by code point, as described in the previous section.
365     .
366     .
367 nigel 93 .SS "Absolute and relative back references"
368     .rs
369     .sp
370 ph10 208 The sequence \eg followed by an unsigned or a negative number, optionally
371     enclosed in braces, is an absolute or relative back reference. A named back
372     reference can be coded as \eg{name}. Back references are discussed
373 nigel 93 .\" HTML <a href="#backreferences">
374     .\" </a>
375     later,
376     .\"
377     following the discussion of
378     .\" HTML <a href="#subpattern">
379     .\" </a>
380     parenthesized subpatterns.
381     .\"
382     .
383     .
384 ph10 333 .SS "Absolute and relative subroutine calls"
385     .rs
386     .sp
387 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
388     a number enclosed either in angle brackets or single quotes, is an alternative
389     syntax for referencing a subpattern as a "subroutine". Details are discussed
390 ph10 333 .\" HTML <a href="#onigurumasubroutines">
391     .\" </a>
392     later.
393     .\"
394 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
395 ph10 461 synonymous. The former is a back reference; the latter is a
396 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
397     .\" </a>
398     subroutine
399     .\"
400     call.
401 ph10 333 .
402     .
403 ph10 518 .\" HTML <a name="genericchartypes"></a>
404 nigel 75 .SS "Generic character types"
405     .rs
406     .sp
407 ph10 514 Another use of backslash is for specifying generic character types:
408 nigel 75 .sp
409 ph10 182 \ed any decimal digit
410 nigel 75 \eD any character that is not a decimal digit
411 ph10 968 \eh any horizontal white space character
412     \eH any character that is not a horizontal white space character
413     \es any white space character
414     \eS any character that is not a white space character
415     \ev any vertical white space character
416     \eV any character that is not a vertical white space character
417 nigel 75 \ew any "word" character
418     \eW any "non-word" character
419     .sp
420 ph10 535 There is also the single sequence \eN, which matches a non-newline character.
421     This is the same as
422 ph10 514 .\" HTML <a href="#fullstopdot">
423     .\" </a>
424 ph10 535 the "." metacharacter
425 ph10 514 .\"
426 ph10 836 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
427 ph10 745 PCRE does not support this.
428 nigel 75 .P
429 ph10 514 Each pair of lower and upper case escape sequences partitions the complete set
430     of characters into two disjoint sets. Any given character matches one, and only
431 ph10 518 one, of each pair. The sequences can appear both inside and outside character
432 nigel 75 classes. They each match one character of the appropriate type. If the current
433 ph10 518 matching point is at the end of the subject string, all of them fail, because
434 nigel 75 there is no character to match.
435     .P
436     For compatibility with Perl, \es does not match the VT character (code 11).
437     This makes it different from the the POSIX "space" class. The \es characters
438 ph10 178 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
439 nigel 91 included in a Perl script, \es may match the VT character. In PCRE, it never
440 ph10 178 does.
441 nigel 75 .P
442 ph10 518 A "word" character is an underscore or any character that is a letter or digit.
443     By default, the definition of letters and digits is controlled by PCRE's
444     low-valued character tables, and may vary if locale-specific matching is taking
445     place (see
446     .\" HTML <a href="pcreapi.html#localesupport">
447     .\" </a>
448     "Locale support"
449     .\"
450     in the
451     .\" HREF
452     \fBpcreapi\fP
453     .\"
454     page). For example, in a French locale such as "fr_FR" in Unix-like systems,
455     or "french" in Windows, some character codes greater than 128 are used for
456     accented letters, and these are then matched by \ew. The use of locales with
457     Unicode is discouraged.
458 ph10 178 .P
459 ph10 859 By default, in a UTF mode, characters with values greater than 128 never match
460 ph10 518 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
461 ph10 859 their original meanings from before UTF support was available, mainly for
462 ph10 535 efficiency reasons. However, if PCRE is compiled with Unicode property support,
463 ph10 518 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
464     properties are used to determine character types, as follows:
465     .sp
466     \ed any character that \ep{Nd} matches (decimal digit)
467     \es any character that \ep{Z} matches, plus HT, LF, FF, CR
468     \ew any character that \ep{L} or \ep{N} matches, plus underscore
469     .sp
470     The upper case escapes match the inverse sets of characters. Note that \ed
471 ph10 535 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
472 ph10 518 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
473     \eB because they are defined in terms of \ew and \eW. Matching these sequences
474     is noticeably slower when PCRE_UCP is set.
475     .P
476 ph10 579 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
477 ph10 572 release 5.10. In contrast to the other sequences, which match only ASCII
478 ph10 859 characters by default, these always match certain high-valued codepoints,
479     whether or not PCRE_UCP is set. The horizontal space characters are:
480 ph10 178 .sp
481     U+0009 Horizontal tab
482     U+0020 Space
483     U+00A0 Non-break space
484     U+1680 Ogham space mark
485     U+180E Mongolian vowel separator
486     U+2000 En quad
487     U+2001 Em quad
488     U+2002 En space
489     U+2003 Em space
490     U+2004 Three-per-em space
491     U+2005 Four-per-em space
492     U+2006 Six-per-em space
493     U+2007 Figure space
494     U+2008 Punctuation space
495     U+2009 Thin space
496     U+200A Hair space
497     U+202F Narrow no-break space
498     U+205F Medium mathematical space
499     U+3000 Ideographic space
500     .sp
501     The vertical space characters are:
502     .sp
503     U+000A Linefeed
504     U+000B Vertical tab
505 ph10 968 U+000C Form feed
506 ph10 178 U+000D Carriage return
507     U+0085 Next line
508     U+2028 Line separator
509     U+2029 Paragraph separator
510 ph10 859 .sp
511 ph10 903 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
512     relevant.
513 nigel 75 .
514     .
515 ph10 231 .\" HTML <a name="newlineseq"></a>
516 nigel 93 .SS "Newline sequences"
517     .rs
518     .sp
519 ph10 231 Outside a character class, by default, the escape sequence \eR matches any
520 ph10 859 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
521     following:
522 nigel 93 .sp
523     (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
524     .sp
525     This is an example of an "atomic group", details of which are given
526     .\" HTML <a href="#atomicgroup">
527     .\" </a>
528     below.
529     .\"
530     This particular group matches either the two-character sequence CR followed by
531     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
532 ph10 968 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
533 nigel 93 line, U+0085). The two-character sequence is treated as a single unit that
534     cannot be split.
535     .P
536 ph10 859 In other modes, two additional characters whose codepoints are greater than 255
537 nigel 93 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
538     Unicode character property support is not needed for these characters to be
539     recognized.
540     .P
541 ph10 231 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
542     complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
543 ph10 247 either at compile time or when the pattern is matched. (BSR is an abbrevation
544 ph10 246 for "backslash R".) This can be made the default when PCRE is built; if this is
545     the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
546     It is also possible to specify these settings by starting a pattern string with
547     one of the following sequences:
548 ph10 231 .sp
549     (*BSR_ANYCRLF) CR, LF, or CRLF only
550     (*BSR_UNICODE) any Unicode newline sequence
551     .sp
552 ph10 859 These override the default and the options given to the compiling function, but
553     they can themselves be overridden by options given to a matching function. Note
554     that these special settings, which are not Perl-compatible, are recognized only
555     at the very start of a pattern, and that they must be in upper case. If more
556     than one of them is present, the last one is used. They can be combined with a
557     change of newline convention; for example, a pattern can start with:
558 ph10 246 .sp
559     (*ANY)(*BSR_ANYCRLF)
560     .sp
561 ph10 859 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
562     sequences. Inside a character class, \eR is treated as an unrecognized escape
563     sequence, and so matches the letter "R" by default, but causes an error if
564     PCRE_EXTRA is set.
565 nigel 93 .
566     .
567 nigel 75 .\" HTML <a name="uniextseq"></a>
568     .SS Unicode character properties
569     .rs
570     .sp
571     When PCRE is built with Unicode character property support, three additional
572 ph10 184 escape sequences that match characters with specific properties are available.
573 ph10 859 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
574 ph10 184 characters whose codepoints are less than 256, but they do work in this mode.
575     The extra escape sequences are:
576 nigel 75 .sp
577 nigel 87 \ep{\fIxx\fP} a character with the \fIxx\fP property
578     \eP{\fIxx\fP} a character without the \fIxx\fP property
579     \eX an extended Unicode sequence
580 nigel 75 .sp
581 nigel 87 The property names represented by \fIxx\fP above are limited to the Unicode
582 ph10 517 script names, the general category properties, "Any", which matches any
583     character (including newline), and some special PCRE properties (described
584 ph10 535 in the
585 ph10 517 .\" HTML <a href="#extraprops">
586     .\" </a>
587 ph10 535 next section).
588 ph10 517 .\"
589     Other Perl properties such as "InMusicalSymbols" are not currently supported by
590     PCRE. Note that \eP{Any} does not match any characters, so always causes a
591     match failure.
592 nigel 75 .P
593 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
594     character from one of these sets can be matched using a script name. For
595     example:
596 nigel 75 .sp
597 nigel 87 \ep{Greek}
598     \eP{Han}
599     .sp
600     Those that are not part of an identified script are lumped together as
601     "Common". The current list of scripts is:
602     .P
603     Arabic,
604     Armenian,
605 ph10 491 Avestan,
606 nigel 93 Balinese,
607 ph10 491 Bamum,
608 ph10 942 Batak,
609 nigel 87 Bengali,
610     Bopomofo,
611 ph10 942 Brahmi,
612 nigel 87 Braille,
613     Buginese,
614     Buhid,
615     Canadian_Aboriginal,
616 ph10 491 Carian,
617 ph10 942 Chakma,
618 ph10 491 Cham,
619 nigel 87 Cherokee,
620     Common,
621     Coptic,
622 nigel 93 Cuneiform,
623 nigel 87 Cypriot,
624     Cyrillic,
625     Deseret,
626     Devanagari,
627 ph10 491 Egyptian_Hieroglyphs,
628 nigel 87 Ethiopic,
629     Georgian,
630     Glagolitic,
631     Gothic,
632     Greek,
633     Gujarati,
634     Gurmukhi,
635     Han,
636     Hangul,
637     Hanunoo,
638     Hebrew,
639     Hiragana,
640 ph10 491 Imperial_Aramaic,
641 nigel 87 Inherited,
642 ph10 491 Inscriptional_Pahlavi,
643     Inscriptional_Parthian,
644     Javanese,
645     Kaithi,
646 nigel 87 Kannada,
647     Katakana,
648 ph10 491 Kayah_Li,
649 nigel 87 Kharoshthi,
650     Khmer,
651     Lao,
652     Latin,
653 ph10 491 Lepcha,
654 nigel 87 Limbu,
655     Linear_B,
656 ph10 491 Lisu,
657     Lycian,
658     Lydian,
659 nigel 87 Malayalam,
660 ph10 942 Mandaic,
661 ph10 491 Meetei_Mayek,
662 ph10 942 Meroitic_Cursive,
663     Meroitic_Hieroglyphs,
664     Miao,
665 nigel 87 Mongolian,
666     Myanmar,
667     New_Tai_Lue,
668 nigel 93 Nko,
669 nigel 87 Ogham,
670     Old_Italic,
671     Old_Persian,
672 ph10 491 Old_South_Arabian,
673     Old_Turkic,
674     Ol_Chiki,
675 nigel 87 Oriya,
676     Osmanya,
677 nigel 93 Phags_Pa,
678     Phoenician,
679 ph10 491 Rejang,
680 nigel 87 Runic,
681 ph10 491 Samaritan,
682     Saurashtra,
683 ph10 942 Sharada,
684 nigel 87 Shavian,
685     Sinhala,
686 ph10 942 Sora_Sompeng,
687 ph10 491 Sundanese,
688 nigel 87 Syloti_Nagri,
689     Syriac,
690     Tagalog,
691     Tagbanwa,
692     Tai_Le,
693 ph10 491 Tai_Tham,
694     Tai_Viet,
695 ph10 942 Takri,
696 nigel 87 Tamil,
697     Telugu,
698     Thaana,
699     Thai,
700     Tibetan,
701     Tifinagh,
702     Ugaritic,
703 ph10 491 Vai,
704 nigel 87 Yi.
705     .P
706 ph10 517 Each character has exactly one Unicode general category property, specified by
707     a two-letter abbreviation. For compatibility with Perl, negation can be
708     specified by including a circumflex between the opening brace and the property
709     name. For example, \ep{^Lu} is the same as \eP{Lu}.
710 nigel 87 .P
711     If only one letter is specified with \ep or \eP, it includes all the general
712     category properties that start with that letter. In this case, in the absence
713     of negation, the curly brackets in the escape sequence are optional; these two
714     examples have the same effect:
715     .sp
716 nigel 75 \ep{L}
717     \epL
718     .sp
719 nigel 87 The following general category property codes are supported:
720 nigel 75 .sp
721     C Other
722     Cc Control
723     Cf Format
724     Cn Unassigned
725     Co Private use
726     Cs Surrogate
727     .sp
728     L Letter
729     Ll Lower case letter
730     Lm Modifier letter
731     Lo Other letter
732     Lt Title case letter
733     Lu Upper case letter
734     .sp
735     M Mark
736     Mc Spacing mark
737     Me Enclosing mark
738     Mn Non-spacing mark
739     .sp
740     N Number
741     Nd Decimal number
742     Nl Letter number
743     No Other number
744     .sp
745     P Punctuation
746     Pc Connector punctuation
747     Pd Dash punctuation
748     Pe Close punctuation
749     Pf Final punctuation
750     Pi Initial punctuation
751     Po Other punctuation
752     Ps Open punctuation
753     .sp
754     S Symbol
755     Sc Currency symbol
756     Sk Modifier symbol
757     Sm Mathematical symbol
758     So Other symbol
759     .sp
760     Z Separator
761     Zl Line separator
762     Zp Paragraph separator
763     Zs Space separator
764     .sp
765 nigel 87 The special property L& is also supported: it matches a character that has
766     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
767     a modifier or "other".
768 nigel 75 .P
769 ph10 211 The Cs (Surrogate) property applies only to characters in the range U+D800 to
770 ph10 859 U+DFFF. Such characters are not valid in Unicode strings and so
771     cannot be tested by PCRE, unless UTF validity checking has been turned off
772     (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
773 ph10 211 .\" HREF
774     \fBpcreapi\fP
775     .\"
776 ph10 451 page). Perl does not support the Cs property.
777 ph10 211 .P
778 ph10 451 The long synonyms for property names that Perl supports (such as \ep{Letter})
779 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
780 nigel 87 properties with "Is".
781     .P
782     No character that is in the Unicode table has the Cn (unassigned) property.
783     Instead, this property is assumed for any code point that is not in the
784     Unicode table.
785     .P
786 nigel 75 Specifying caseless matching does not affect these escape sequences. For
787     example, \ep{Lu} always matches only upper case letters.
788     .P
789     The \eX escape matches any number of Unicode characters that form an extended
790     Unicode sequence. \eX is equivalent to
791     .sp
792     (?>\ePM\epM*)
793     .sp
794     That is, it matches a character without the "mark" property, followed by zero
795     or more characters with the "mark" property, and treats the sequence as an
796     atomic group
797     .\" HTML <a href="#atomicgroup">
798     .\" </a>
799     (see below).
800     .\"
801     Characters with the "mark" property are typically accents that affect the
802 ph10 185 preceding character. None of them have codepoints less than 256, so in
803 ph10 859 8-bit non-UTF-8 mode \eX matches any one character.
804 nigel 75 .P
805 ph10 654 Note that recent versions of Perl have changed \eX to match what Unicode calls
806 ph10 628 an "extended grapheme cluster", which has a more complicated definition.
807     .P
808 nigel 75 Matching characters by Unicode property is not fast, because PCRE has to search
809     a structure that contains data for over fifteen thousand characters. That is
810     why the traditional escape sequences such as \ed and \ew do not use Unicode
811 ph10 535 properties in PCRE by default, though you can make them do so by setting the
812 ph10 859 PCRE_UCP option or by starting the pattern with (*UCP).
813 nigel 75 .
814     .
815 ph10 517 .\" HTML <a name="extraprops"></a>
816     .SS PCRE's additional properties
817     .rs
818     .sp
819 ph10 535 As well as the standard Unicode properties described in the previous
820     section, PCRE supports four more that make it possible to convert traditional
821 ph10 517 escape sequences such as \ew and \es and POSIX character classes to use Unicode
822 ph10 518 properties. PCRE uses these non-standard, non-Perl properties internally when
823     PCRE_UCP is set. They are:
824 ph10 517 .sp
825     Xan Any alphanumeric character
826     Xps Any POSIX space character
827     Xsp Any Perl space character
828     Xwd Any Perl "word" character
829     .sp
830 ph10 535 Xan matches characters that have either the L (letter) or the N (number)
831 ph10 968 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
832 ph10 517 carriage return, and any other character that has the Z (separator) property.
833 ph10 535 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
834 ph10 517 same characters as Xan, plus underscore.
835     .
836     .
837 ph10 168 .\" HTML <a name="resetmatchstart"></a>
838     .SS "Resetting the match start"
839     .rs
840     .sp
841 ph10 572 The escape sequence \eK causes any previously matched characters not to be
842     included in the final matched sequence. For example, the pattern:
843 ph10 168 .sp
844     foo\eKbar
845     .sp
846 ph10 172 matches "foobar", but reports that it has matched "bar". This feature is
847 ph10 168 similar to a lookbehind assertion
848     .\" HTML <a href="#lookbehind">
849     .\" </a>
850     (described below).
851     .\"
852 ph10 172 However, in this case, the part of the subject before the real match does not
853     have to be of fixed length, as lookbehind assertions do. The use of \eK does
854 ph10 168 not interfere with the setting of
855     .\" HTML <a href="#subpattern">
856     .\" </a>
857     captured substrings.
858 ph10 172 .\"
859 ph10 168 For example, when the pattern
860     .sp
861     (foo)\eKbar
862     .sp
863 ph10 172 matches "foobar", the first substring is still set to "foo".
864 ph10 500 .P
865 ph10 507 Perl documents that the use of \eK within assertions is "not well defined". In
866     PCRE, \eK is acted upon when it occurs inside positive assertions, but is
867 ph10 500 ignored in negative assertions.
868 ph10 168 .
869     .
870 nigel 75 .\" HTML <a name="smallassertions"></a>
871     .SS "Simple assertions"
872     .rs
873     .sp
874 nigel 93 The final use of backslash is for certain simple assertions. An assertion
875 nigel 63 specifies a condition that has to be met at a particular point in a match,
876     without consuming any characters from the subject string. The use of
877 nigel 75 subpatterns for more complicated assertions is described
878     .\" HTML <a href="#bigassertions">
879     .\" </a>
880     below.
881     .\"
882 nigel 91 The backslashed assertions are:
883 nigel 75 .sp
884     \eb matches at a word boundary
885     \eB matches when not at a word boundary
886 nigel 93 \eA matches at the start of the subject
887     \eZ matches at the end of the subject
888     also matches before a newline at the end of the subject
889     \ez matches only at the end of the subject
890     \eG matches at the first matching position in the subject
891 nigel 75 .sp
892 ph10 513 Inside a character class, \eb has a different meaning; it matches the backspace
893 ph10 535 character. If any other of these assertions appears in a character class, by
894 ph10 513 default it matches the corresponding literal character (for example, \eB
895     matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
896     escape sequence" error is generated instead.
897 nigel 75 .P
898 nigel 63 A word boundary is a position in the subject string where the current character
899 nigel 75 and the previous character do not both match \ew or \eW (i.e. one matches
900     \ew and the other matches \eW), or the start or end of the string if the
901 ph10 859 first or last character matches \ew, respectively. In a UTF mode, the meanings
902 ph10 518 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
903     done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
904     of word" or "end of word" metasequence. However, whatever follows \eb normally
905     determines which it is. For example, the fragment \eba matches "a" at the start
906     of a word.
907 nigel 75 .P
908     The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
909     dollar (described in the next section) in that they only ever match at the very
910     start and end of the subject string, whatever options are set. Thus, they are
911     independent of multiline mode. These three assertions are not affected by the
912     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
913     circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
914     argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
915     at a point other than the beginning of the subject, \eA can never match. The
916 nigel 91 difference between \eZ and \ez is that \eZ matches before a newline at the end
917     of the string as well as at the very end, whereas \ez matches only at the end.
918 nigel 75 .P
919     The \eG assertion is true only when the current matching position is at the
920     start point of the match, as specified by the \fIstartoffset\fP argument of
921     \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
922     non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
923 nigel 63 arguments, you can mimic Perl's /g option, and it is in this kind of
924 nigel 75 implementation where \eG can be useful.
925     .P
926     Note, however, that PCRE's interpretation of \eG, as the start of the current
927 nigel 63 match, is subtly different from Perl's, which defines it as the end of the
928     previous match. In Perl, these can be different when the previously matched
929     string was empty. Because PCRE does just one match at a time, it cannot
930     reproduce this behaviour.
931 nigel 75 .P
932     If all the alternatives of a pattern begin with \eG, the expression is anchored
933 nigel 63 to the starting match position, and the "anchored" flag is set in the compiled
934     regular expression.
935 nigel 75 .
936     .
938 nigel 63 .rs
939     .sp
940     Outside a character class, in the default matching mode, the circumflex
941 nigel 75 character is an assertion that is true only if the current matching point is
942     at the start of the subject string. If the \fIstartoffset\fP argument of
943     \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
944 nigel 63 option is unset. Inside a character class, circumflex has an entirely different
945 nigel 75 meaning
946     .\" HTML <a href="#characterclass">
947     .\" </a>
948     (see below).
949     .\"
950     .P
951 nigel 63 Circumflex need not be the first character of the pattern if a number of
952     alternatives are involved, but it should be the first thing in each alternative
953     in which it appears if the pattern is ever to match that branch. If all
954     possible alternatives start with a circumflex, that is, if the pattern is
955     constrained to match only at the start of the subject, it is said to be an
956     "anchored" pattern. (There are also other constructs that can cause a pattern
957     to be anchored.)
958 nigel 75 .P
959     A dollar character is an assertion that is true only if the current matching
960 nigel 63 point is at the end of the subject string, or immediately before a newline
961 nigel 91 at the end of the string (by default). Dollar need not be the last character of
962     the pattern if a number of alternatives are involved, but it should be the last
963     item in any branch in which it appears. Dollar has no special meaning in a
964     character class.
965 nigel 75 .P
966 nigel 63 The meaning of dollar can be changed so that it matches only at the very end of
967     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
968 nigel 75 does not affect the \eZ assertion.
969     .P
970 nigel 63 The meanings of the circumflex and dollar characters are changed if the
971 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
972     immediately after internal newlines as well as at the start of the subject
973     string. It does not match after a newline that ends the string. A dollar
974     matches before any newlines in the string, as well as at the very end, when
975     PCRE_MULTILINE is set. When newline is specified as the two-character
976     sequence CRLF, isolated CR and LF characters do not indicate newlines.
977 nigel 75 .P
978 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
979     \en represents a newline) in multiline mode, but not otherwise. Consequently,
980     patterns that are anchored in single line mode because all branches start with
981     ^ are not anchored in multiline mode, and a match for circumflex is possible
982     when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
983     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
984     .P
985 nigel 75 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
986 nigel 63 end of the subject in both modes, and if all branches of a pattern start with
987 nigel 91 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
988 nigel 75 .
989     .
990 ph10 514 .\" HTML <a name="fullstopdot"></a>
992 nigel 63 .rs
993     .sp
994     Outside a character class, a dot in the pattern matches any one character in
995 nigel 91 the subject string except (by default) a character that signifies the end of a
996 ph10 903 line.
997 nigel 91 .P
998 nigel 93 When a line ending is defined as a single character, dot never matches that
999     character; when the two-character sequence CRLF is used, dot does not match CR
1000     if it is immediately followed by LF, but otherwise it matches all characters
1001     (including isolated CRs and LFs). When any Unicode line endings are being
1002     recognized, dot does not match CR or LF or any of the other line ending
1003     characters.
1004     .P
1005 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1006 nigel 93 option is set, a dot matches any one character, without exception. If the
1007     two-character sequence CRLF is present in the subject string, it takes two dots
1008     to match it.
1009 nigel 91 .P
1010     The handling of dot is entirely independent of the handling of circumflex and
1011     dollar, the only relationship being that they both involve newlines. Dot has no
1012     special meaning in a character class.
1013 ph10 514 .P
1014 ph10 579 The escape sequence \eN behaves like a dot, except that it is not affected by
1015 ph10 572 the PCRE_DOTALL option. In other words, it matches any character except one
1016 ph10 745 that signifies the end of a line. Perl also uses \eN to match characters by
1017     name; PCRE does not support this.
1018 nigel 75 .
1019     .
1021 nigel 63 .rs
1022     .sp
1023 ph10 859 Outside a character class, the escape sequence \eC matches any one data unit,
1024     whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1025     byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \eC always
1026     matches line-ending characters. The feature is provided in Perl in order to
1027     match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1028     used. Because \eC breaks up characters into individual data units, matching one
1029     unit with \eC in a UTF mode means that the rest of the string may start with a
1030     malformed UTF character. This has undefined results, because PCRE assumes that
1031     it is dealing with valid UTF strings (and by default it checks this at the
1032 ph10 959 start of processing unless the PCRE_NO_UTF8_CHECK or PCRE_NO_UTF16_CHECK option
1033     is used).
1034 nigel 75 .P
1035     PCRE does not allow \eC to appear in lookbehind assertions
1036     .\" HTML <a href="#lookbehind">
1037     .\" </a>
1038 ph10 754 (described below)
1039 nigel 75 .\"
1040 ph10 859 in a UTF mode, because this would make it impossible to calculate the length of
1041 nigel 75 the lookbehind.
1042 ph10 737 .P
1043 ph10 859 In general, the \eC escape sequence is best avoided. However, one
1044     way of using it that avoids the problem of malformed UTF characters is to use a
1045     lookahead to check the length of the next character, as in this pattern, which
1046     could be used with a UTF-8 string (ignore white space and line breaks):
1047 ph10 737 .sp
1048     (?| (?=[\ex00-\ex7f])(\eC) |
1049 ph10 738 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1050     (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1051 ph10 737 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1052     .sp
1053 ph10 738 A group that starts with (?| resets the capturing parentheses numbers in each
1054     alternative (see
1055 ph10 737 .\" HTML <a href="#dupsubpatternnumber">
1056     .\" </a>
1057     "Duplicate Subpattern Numbers"
1058     .\"
1059 ph10 738 below). The assertions at the start of each branch check the next UTF-8
1060     character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1061 ph10 737 character's individual bytes are then captured by the appropriate number of
1062     groups.
1063 nigel 75 .
1064     .
1065     .\" HTML <a name="characterclass"></a>
1067 nigel 63 .rs
1068     .sp
1069     An opening square bracket introduces a character class, terminated by a closing
1070 ph10 461 square bracket. A closing square bracket on its own is not special by default.
1071     However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1072 ph10 456 bracket causes a compile-time error. If a closing square bracket is required as
1073     a member of the class, it should be the first data character in the class
1074     (after an initial circumflex, if present) or escaped with a backslash.
1075 nigel 75 .P
1076 ph10 859 A character class matches a single character in the subject. In a UTF mode, the
1077     character may be more than one data unit long. A matched character must be in
1078     the set of characters defined by the class, unless the first character in the
1079     class definition is a circumflex, in which case the subject character must not
1080     be in the set defined by the class. If a circumflex is actually required as a
1081     member of the class, ensure it is not the first character, or escape it with a
1082 nigel 63 backslash.
1083 nigel 75 .P
1084 nigel 63 For example, the character class [aeiou] matches any lower case vowel, while
1085     [^aeiou] matches any character that is not a lower case vowel. Note that a
1086 nigel 75 circumflex is just a convenient notation for specifying the characters that
1087     are in the class by enumerating those that are not. A class that starts with a
1088 ph10 456 circumflex is not an assertion; it still consumes a character from the subject
1089 nigel 75 string, and therefore it fails if the current pointer is at the end of the
1090     string.
1091     .P
1092 ph10 859 In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
1093     included in a class as a literal string of data units, or by using the \ex{
1094     escaping mechanism.
1095 nigel 75 .P
1096 nigel 63 When caseless matching is set, any letters in a class represent both their
1097     upper case and lower case versions, so for example, a caseless [aeiou] matches
1098     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1099 ph10 859 caseful version would. In a UTF mode, PCRE always understands the concept of
1100 nigel 77 case for characters whose values are less than 128, so caseless matching is
1101     always possible. For characters with higher values, the concept of case is
1102     supported if PCRE is compiled with Unicode property support, but not otherwise.
1103 ph10 859 If you want to use caseless matching in a UTF mode for characters 128 and
1104     above, you must ensure that PCRE is compiled with Unicode property support as
1105     well as with UTF support.
1106 nigel 75 .P
1107 nigel 93 Characters that might indicate line breaks are never treated in any special way
1108     when matching character classes, whatever line-ending sequence is in use, and
1109     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1110     such as [^a] always matches one of these characters.
1111 nigel 75 .P
1112 nigel 63 The minus (hyphen) character can be used to specify a range of characters in a
1113     character class. For example, [d-m] matches any letter between d and m,
1114     inclusive. If a minus character is required in a class, it must be escaped with
1115     a backslash or appear in a position where it cannot be interpreted as
1116     indicating a range, typically as the first or last character in the class.
1117 nigel 75 .P
1118 nigel 63 It is not possible to have the literal character "]" as the end character of a
1119     range. A pattern such as [W-]46] is interpreted as a class of two characters
1120     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1121     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1122 nigel 75 the end of range, so [W-\e]46] is interpreted as a class containing a range
1123     followed by two other characters. The octal or hexadecimal representation of
1124     "]" can also be used to end a range.
1125     .P
1126 nigel 63 Ranges operate in the collating sequence of character values. They can also be
1127 ph10 903 used for characters specified numerically, for example [\e000-\e037]. Ranges
1128 ph10 859 can include any characters that are valid for the current mode.
1129 nigel 75 .P
1130 nigel 63 If a range that includes letters is used when caseless matching is set, it
1131     matches the letters in either case. For example, [W-c] is equivalent to
1132 ph10 859 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1133 ph10 139 tables for a French locale are in use, [\exc8-\excb] matches accented E
1134 ph10 859 characters in both cases. In UTF modes, PCRE supports the concept of case for
1135 nigel 75 characters with values greater than 128 only when it is compiled with Unicode
1136     property support.
1137     .P
1138 ph10 575 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1139     \eV, \ew, and \eW may appear in a character class, and add the characters that
1140     they match to the class. For example, [\edABCDEF] matches any hexadecimal
1141 ph10 859 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1142 ph10 575 and their upper case partners, just as it does when they appear outside a
1143     character class, as described in the section entitled
1144     .\" HTML <a href="#genericchartypes">
1145     .\" </a>
1146     "Generic character types"
1147     .\"
1148     above. The escape sequence \eb has a different meaning inside a character
1149     class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1150     are not special inside a character class. Like any other unrecognized escape
1151     sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1152     default, but cause an error if the PCRE_EXTRA option is set.
1153     .P
1154     A circumflex can conveniently be used with the upper case character types to
1155 ph10 518 specify a more restricted set of characters than the matching lower case type.
1156 ph10 575 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1157 ph10 579 whereas [\ew] includes underscore. A positive character class should be read as
1158     "something OR something OR ..." and a negative class as "NOT something AND NOT
1159 ph10 575 something AND NOT ...".
1160 nigel 75 .P
1161     The only metacharacters that are recognized in character classes are backslash,
1162     hyphen (only where it can be interpreted as specifying a range), circumflex
1163     (only at the start), opening square bracket (only when it can be interpreted as
1164     introducing a POSIX class name - see the next section), and the terminating
1165     closing square bracket. However, escaping other non-alphanumeric characters
1166     does no harm.
1167     .
1168     .
1170 nigel 63 .rs
1171     .sp
1172 nigel 75 Perl supports the POSIX notation for character classes. This uses names
1173 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1174     this notation. For example,
1175 nigel 75 .sp
1176 nigel 63 [01[:alpha:]%]
1177 nigel 75 .sp
1178 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
1179 ph10 518 are:
1180 nigel 75 .sp
1181 nigel 63 alnum letters and digits
1182     alpha letters
1183     ascii character codes 0 - 127
1184     blank space or tab only
1185     cntrl control characters
1186 nigel 75 digit decimal digits (same as \ed)
1187 nigel 63 graph printing characters, excluding space
1188     lower lower case letters
1189     print printing characters, including space
1190 ph10 518 punct printing characters, excluding letters and digits and space
1191 nigel 75 space white space (not quite the same as \es)
1192 nigel 63 upper upper case letters
1193 nigel 75 word "word" characters (same as \ew)
1194 nigel 63 xdigit hexadecimal digits
1195 nigel 75 .sp
1196 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1197     space (32). Notice that this list includes the VT character (code 11). This
1198 nigel 75 makes "space" different to \es, which does not include VT (for Perl
1199 nigel 63 compatibility).
1200 nigel 75 .P
1201 nigel 63 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1202     5.8. Another Perl extension is negation, which is indicated by a ^ character
1203     after the colon. For example,
1204 nigel 75 .sp
1205 nigel 63 [12[:^digit:]]
1206 nigel 75 .sp
1207 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1208     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1209     supported, and an error is given if they are encountered.
1210 nigel 75 .P
1211 ph10 859 By default, in UTF modes, characters with values greater than 128 do not match
1212 ph10 518 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1213 ph10 535 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1214     character properties are used. This is achieved by replacing the POSIX classes
1215 ph10 518 by other sequences, as follows:
1216     .sp
1217     [:alnum:] becomes \ep{Xan}
1218     [:alpha:] becomes \ep{L}
1219 ph10 535 [:blank:] becomes \eh
1220 ph10 518 [:digit:] becomes \ep{Nd}
1221     [:lower:] becomes \ep{Ll}
1222 ph10 535 [:space:] becomes \ep{Xps}
1223 ph10 518 [:upper:] becomes \ep{Lu}
1224     [:word:] becomes \ep{Xwd}
1225     .sp
1226     Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1227     classes are unchanged, and match only characters with code points less than
1228     128.
1229 nigel 75 .
1230     .
1231     .SH "VERTICAL BAR"
1232 nigel 63 .rs
1233     .sp
1234     Vertical bar characters are used to separate alternative patterns. For example,
1235     the pattern
1236 nigel 75 .sp
1237 nigel 63 gilbert|sullivan
1238 nigel 75 .sp
1239 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1240 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
1241     process tries each alternative in turn, from left to right, and the first one
1242     that succeeds is used. If the alternatives are within a subpattern
1243 nigel 75 .\" HTML <a href="#subpattern">
1244     .\" </a>
1245     (defined below),
1246     .\"
1247     "succeeds" means matching the rest of the main pattern as well as the
1248     alternative in the subpattern.
1249     .
1250     .
1252 nigel 63 .rs
1253     .sp
1254     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1255 ph10 231 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1256     the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1257     The option letters are
1258 nigel 75 .sp
1259 nigel 63 i for PCRE_CASELESS
1260     m for PCRE_MULTILINE
1261     s for PCRE_DOTALL
1262     x for PCRE_EXTENDED
1263 nigel 75 .sp
1264 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
1265     unset these options by preceding the letter with a hyphen, and a combined
1266     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1267     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1268     permitted. If a letter appears both before and after the hyphen, the option is
1269     unset.
1270 nigel 75 .P
1271 ph10 231 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1272     changed in the same way as the Perl-compatible options by using the characters
1273     J, U and X respectively.
1274     .P
1275 ph10 412 When one of these option changes occurs at top level (that is, not inside
1276     subpattern parentheses), the change applies to the remainder of the pattern
1277     that follows. If the change is placed right at the start of a pattern, PCRE
1278     extracts it into the global options (and it will therefore show up in data
1279     extracted by the \fBpcre_fullinfo()\fP function).
1280 nigel 75 .P
1281 nigel 93 An option change within a subpattern (see below for a description of
1282 ph10 572 subpatterns) affects only that part of the subpattern that follows it, so
1283 nigel 75 .sp
1284 nigel 63 (a(?i)b)c
1285 nigel 75 .sp
1286 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1287     By this means, options can be made to have different settings in different
1288     parts of the pattern. Any changes made in one alternative do carry on
1289     into subsequent branches within the same subpattern. For example,
1290 nigel 75 .sp
1291 nigel 63 (a(?i)b|c)
1292 nigel 75 .sp
1293 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1294     branch is abandoned before the option setting. This is because the effects of
1295     option settings happen at compile time. There would be some very weird
1296     behaviour otherwise.
1297 ph10 251 .P
1298     \fBNote:\fP There are other PCRE-specific options that can be set by the
1299 ph10 859 application when the compiling or matching functions are called. In some cases
1300     the pattern can contain special leading sequences such as (*CRLF) to override
1301     what the application has set or what has been defaulted. Details are given in
1302     the section entitled
1303 ph10 251 .\" HTML <a href="#newlineseq">
1304     .\" </a>
1305     "Newline sequences"
1306     .\"
1307 ph10 859 above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
1308     can be used to set UTF and Unicode property modes; they are equivalent to
1309     setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1310 nigel 75 .
1311     .
1312     .\" HTML <a name="subpattern"></a>
1313 nigel 63 .SH SUBPATTERNS
1314     .rs
1315     .sp
1316     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1317 nigel 75 Turning part of a pattern into a subpattern does two things:
1318     .sp
1319 nigel 63 1. It localizes a set of alternatives. For example, the pattern
1320 nigel 75 .sp
1321 nigel 63 cat(aract|erpillar|)
1322 nigel 75 .sp
1323 ph10 572 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1324     match "cataract", "erpillar" or an empty string.
1325 nigel 75 .sp
1326     2. It sets up the subpattern as a capturing subpattern. This means that, when
1327     the whole pattern matches, that portion of the subject string that matched the
1328 ph10 903 subpattern is passed back to the caller via the \fIovector\fP argument of the
1329     matching function. (This applies only to the traditional matching functions;
1330 ph10 859 the DFA matching functions do not support capturing.)
1331     .P
1332     Opening parentheses are counted from left to right (starting from 1) to obtain
1333     numbers for the capturing subpatterns. For example, if the string "the red
1334     king" is matched against the pattern
1335 nigel 75 .sp
1336 nigel 63 the ((red|white) (king|queen))
1337 nigel 75 .sp
1338 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
1339     2, and 3, respectively.
1340 nigel 75 .P
1341 nigel 63 The fact that plain parentheses fulfil two functions is not always helpful.
1342     There are often times when a grouping subpattern is required without a
1343     capturing requirement. If an opening parenthesis is followed by a question mark
1344     and a colon, the subpattern does not do any capturing, and is not counted when
1345     computing the number of any subsequent capturing subpatterns. For example, if
1346     the string "the white queen" is matched against the pattern
1347 nigel 75 .sp
1348 nigel 63 the ((?:red|white) (king|queen))
1349 nigel 75 .sp
1350 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
1351 nigel 93 2. The maximum number of capturing subpatterns is 65535.
1352 nigel 75 .P
1353 nigel 63 As a convenient shorthand, if any option settings are required at the start of
1354     a non-capturing subpattern, the option letters may appear between the "?" and
1355     the ":". Thus the two patterns
1356 nigel 75 .sp
1357 nigel 63 (?i:saturday|sunday)
1358     (?:(?i)saturday|sunday)
1359 nigel 75 .sp
1360 nigel 63 match exactly the same set of strings. Because alternative branches are tried
1361     from left to right, and options are not reset until the end of the subpattern
1362     is reached, an option setting in one branch does affect subsequent branches, so
1363     the above patterns match "SUNDAY" as well as "Saturday".
1364 nigel 75 .
1365     .
1366 ph10 456 .\" HTML <a name="dupsubpatternnumber"></a>
1368     .rs
1369     .sp
1370 ph10 182 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1371     the same numbers for its capturing parentheses. Such a subpattern starts with
1372     (?| and is itself a non-capturing subpattern. For example, consider this
1373 ph10 175 pattern:
1374     .sp
1375     (?|(Sat)ur|(Sun))day
1376 ph10 182 .sp
1377     Because the two alternatives are inside a (?| group, both sets of capturing
1378     parentheses are numbered one. Thus, when the pattern matches, you can look
1379     at captured substring number one, whichever alternative matched. This construct
1380     is useful when you want to capture part, but not all, of one of a number of
1381     alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1382 ph10 175 number is reset at the start of each branch. The numbers of any capturing
1383 ph10 572 parentheses that follow the subpattern start after the highest number used in
1384     any branch. The following example is taken from the Perl documentation. The
1385     numbers underneath show in which buffer the captured content will be stored.
1386 ph10 175 .sp
1387     # before ---------------branch-reset----------- after
1388     / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1389     # 1 2 2 3 2 3 4
1390 ph10 182 .sp
1391 ph10 488 A back reference to a numbered subpattern uses the most recent value that is
1392     set for that number by any subpattern. The following pattern matches "abcabc"
1393     or "defdef":
1394 ph10 456 .sp
1395 ph10 461 /(?|(abc)|(def))\e1/
1396 ph10 456 .sp
1397 ph10 716 In contrast, a subroutine call to a numbered subpattern always refers to the
1398     first one in the pattern with the given number. The following pattern matches
1399     "abcabc" or "defabc":
1400 ph10 456 .sp
1401     /(?|(abc)|(def))(?1)/
1402     .sp
1403 ph10 459 If a
1404     .\" HTML <a href="#conditions">
1405     .\" </a>
1406     condition test
1407     .\"
1408     for a subpattern's having matched refers to a non-unique number, the test is
1409     true if any of the subpatterns of that number have matched.
1410     .P
1411     An alternative approach to using this "branch reset" feature is to use
1412 ph10 175 duplicate named subpatterns, as described in the next section.
1413     .
1414     .
1415 nigel 75 .SH "NAMED SUBPATTERNS"
1416 nigel 63 .rs
1417     .sp
1418     Identifying capturing parentheses by number is simple, but it can be very hard
1419     to keep track of the numbers in complicated regular expressions. Furthermore,
1420 nigel 75 if an expression is modified, the numbers may change. To help with this
1421 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
1422     added to Perl until release 5.10. Python had the feature earlier, and PCRE
1423     introduced it at release 4.0, using the Python syntax. PCRE now supports both
1424 ph10 459 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1425     have different names, but PCRE does not.
1426 nigel 93 .P
1427     In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1428     (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1429 nigel 91 parentheses from other parts of the pattern, such as
1430     .\" HTML <a href="#backreferences">
1431     .\" </a>
1432 ph10 488 back references,
1433 nigel 91 .\"
1434     .\" HTML <a href="#recursion">
1435     .\" </a>
1436     recursion,
1437     .\"
1438     and
1439     .\" HTML <a href="#conditions">
1440     .\" </a>
1441     conditions,
1442     .\"
1443     can be made by name as well as by number.
1444 nigel 75 .P
1445 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
1446 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
1447     if the names were not present. The PCRE API provides function calls for
1448     extracting the name-to-number translation table from a compiled pattern. There
1449     is also a convenience function for extracting a captured substring by name.
1450 nigel 91 .P
1451     By default, a name must be unique within a pattern, but it is possible to relax
1452 ph10 457 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1453 ph10 461 names are also always permitted for subpatterns with the same number, set up as
1454 ph10 457 described in the previous section.) Duplicate names can be useful for patterns
1455     where only one instance of the named parentheses can match. Suppose you want to
1456     match the name of a weekday, either as a 3-letter abbreviation or as the full
1457     name, and in both cases you want to extract the abbreviation. This pattern
1458     (ignoring the line breaks) does the job:
1459 nigel 91 .sp
1460 nigel 93 (?<DN>Mon|Fri|Sun)(?:day)?|
1461     (?<DN>Tue)(?:sday)?|
1462     (?<DN>Wed)(?:nesday)?|
1463     (?<DN>Thu)(?:rsday)?|
1464     (?<DN>Sat)(?:urday)?
1465 nigel 91 .sp
1466     There are five capturing substrings, but only one is ever set after a match.
1467 ph10 182 (An alternative way of solving this problem is to use a "branch reset"
1468 ph10 175 subpattern, as described in the previous section.)
1469     .P
1470 nigel 91 The convenience function for extracting the data by name returns the substring
1471 nigel 93 for the first (and in this example, the only) subpattern of that name that
1472 ph10 461 matched. This saves searching to find which numbered subpattern it was.
1473 ph10 459 .P
1474 ph10 488 If you make a back reference to a non-unique named subpattern from elsewhere in
1475 ph10 459 the pattern, the one that corresponds to the first occurrence of the name is
1476     used. In the absence of duplicate numbers (see the previous section) this is
1477     the one with the lowest number. If you use a named reference in a condition
1478     test (see the
1479     .\"
1480     .\" HTML <a href="#conditions">
1481     .\" </a>
1482     section about conditions
1483     .\"
1484 ph10 461 below), either to check whether a subpattern has matched, or to check for
1485 ph10 459 recursion, all subpatterns with the same name are tested. If the condition is
1486     true for any one of them, the overall condition is true. This is the same
1487     behaviour as testing by number. For further details of the interfaces for
1488     handling named subpatterns, see the
1489 nigel 63 .\" HREF
1490 nigel 75 \fBpcreapi\fP
1491 nigel 63 .\"
1492     documentation.
1493 ph10 385 .P
1494     \fBWarning:\fP You cannot use different names to distinguish between two
1495 ph10 457 subpatterns with the same number because PCRE uses only the numbers when
1496     matching. For this reason, an error is given at compile time if different names
1497     are given to subpatterns with the same number. However, you can give the same
1498     name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1499 nigel 75 .
1500     .
1501 nigel 63 .SH REPETITION
1502     .rs
1503     .sp
1504     Repetition is specified by quantifiers, which can follow any of the following
1505     items:
1506 nigel 75 .sp
1507 nigel 63 a literal data character
1508 nigel 93 the dot metacharacter
1509 nigel 75 the \eC escape sequence
1510 ph10 859 the \eX escape sequence
1511 nigel 93 the \eR escape sequence
1512 ph10 572 an escape such as \ed or \epL that matches a single character
1513 nigel 63 a character class
1514     a back reference (see next section)
1515 ph10 637 a parenthesized subpattern (including assertions)
1516 ph10 716 a subroutine call to a subpattern (recursive or otherwise)
1517 nigel 75 .sp
1518 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1519     permitted matches, by giving the two numbers in curly brackets (braces),
1520     separated by a comma. The numbers must be less than 65536, and the first must
1521     be less than or equal to the second. For example:
1522 nigel 75 .sp
1523 nigel 63 z{2,4}
1524 nigel 75 .sp
1525 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1526     character. If the second number is omitted, but the comma is present, there is
1527     no upper limit; if the second number and the comma are both omitted, the
1528     quantifier specifies an exact number of required matches. Thus
1529 nigel 75 .sp
1530 nigel 63 [aeiou]{3,}
1531 nigel 75 .sp
1532 nigel 63 matches at least 3 successive vowels, but may match many more, while
1533 nigel 75 .sp
1534     \ed{8}
1535     .sp
1536 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1537     where a quantifier is not allowed, or one that does not match the syntax of a
1538     quantifier, is taken as a literal character. For example, {,6} is not a
1539     quantifier, but a literal string of four characters.
1540 nigel 75 .P
1541 ph10 859 In UTF modes, quantifiers apply to characters rather than to individual data
1542     units. Thus, for example, \ex{100}{2} matches two characters, each of
1543     which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1544     \eX{3} matches three Unicode extended sequences, each of which may be several
1545     data units long (and they may be of different lengths).
1546 nigel 75 .P
1547 nigel 63 The quantifier {0} is permitted, causing the expression to behave as if the
1548 ph10 345 previous item and the quantifier were not present. This may be useful for
1549     subpatterns that are referenced as
1550 ph10 335 .\" HTML <a href="#subpatternsassubroutines">
1551     .\" </a>
1552     subroutines
1553     .\"
1554 ph10 572 from elsewhere in the pattern (but see also the section entitled
1555     .\" HTML <a href="#subdefine">
1556     .\" </a>
1557     "Defining subpatterns for use by reference only"
1558     .\"
1559     below). Items other than subpatterns that have a {0} quantifier are omitted
1560     from the compiled pattern.
1561 nigel 75 .P
1562 nigel 93 For convenience, the three most common quantifiers have single-character
1563     abbreviations:
1564 nigel 75 .sp
1565 nigel 63 * is equivalent to {0,}
1566     + is equivalent to {1,}
1567     ? is equivalent to {0,1}
1568 nigel 75 .sp
1569 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1570     match no characters with a quantifier that has no upper limit, for example:
1571 nigel 75 .sp
1572 nigel 63 (a?)*
1573 nigel 75 .sp
1574 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1575     such patterns. However, because there are cases where this can be useful, such
1576     patterns are now accepted, but if any repetition of the subpattern does in fact
1577     match no characters, the loop is forcibly broken.
1578 nigel 75 .P
1579 nigel 63 By default, the quantifiers are "greedy", that is, they match as much as
1580     possible (up to the maximum number of permitted times), without causing the
1581     rest of the pattern to fail. The classic example of where this gives problems
1582 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1583     and within the comment, individual * and / characters may appear. An attempt to
1584     match C comments by applying the pattern
1585     .sp
1586     /\e*.*\e*/
1587     .sp
1588 nigel 63 to the string
1589 nigel 75 .sp
1590     /* first comment */ not comment /* second comment */
1591     .sp
1592 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1593     item.
1594 nigel 75 .P
1595 nigel 63 However, if a quantifier is followed by a question mark, it ceases to be
1596     greedy, and instead matches the minimum number of times possible, so the
1597     pattern
1598 nigel 75 .sp
1599     /\e*.*?\e*/
1600     .sp
1601 nigel 63 does the right thing with the C comments. The meaning of the various
1602     quantifiers is not otherwise changed, just the preferred number of matches.
1603     Do not confuse this use of question mark with its use as a quantifier in its
1604     own right. Because it has two uses, it can sometimes appear doubled, as in
1605 nigel 75 .sp
1606     \ed??\ed
1607     .sp
1608 nigel 63 which matches one digit by preference, but can match two if that is the only
1609     way the rest of the pattern matches.
1610 nigel 75 .P
1611 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1612 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1613     greedy by following them with a question mark. In other words, it inverts the
1614     default behaviour.
1615 nigel 75 .P
1616 nigel 63 When a parenthesized subpattern is quantified with a minimum repeat count that
1617 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1618 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1619 nigel 75 .P
1620 nigel 63 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1621 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1622 nigel 63 implicitly anchored, because whatever follows will be tried against every
1623     character position in the subject string, so there is no point in retrying the
1624     overall match at any position after the first. PCRE normally treats such a
1625 nigel 75 pattern as though it were preceded by \eA.
1626     .P
1627 nigel 63 In cases where it is known that the subject string contains no newlines, it is
1628     worth setting PCRE_DOTALL in order to obtain this optimization, or
1629     alternatively using ^ to indicate anchoring explicitly.
1630 nigel 75 .P
1631 ph10 994 However, there are some cases where the optimization cannot be used. When .*
1632 ph10 488 is inside capturing parentheses that are the subject of a back reference
1633 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1634     succeeds. Consider, for example:
1635 nigel 75 .sp
1636     (.*)abc\e1
1637     .sp
1638 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1639     this reason, such a pattern is not implicitly anchored.
1640 nigel 75 .P
1641 ph10 994 Another case where implicit anchoring is not applied is when the leading .* is
1642     inside an atomic group. Once again, a match at the start may fail where a later
1643     one succeeds. Consider this pattern:
1644     .sp
1645     (?>.*?a)b
1646     .sp
1647     It matches "ab" in the subject "aab". The use of the backtracking control verbs
1648     (*PRUNE) and (*SKIP) also disable this optimization.
1649     .P
1650 nigel 63 When a capturing subpattern is repeated, the value captured is the substring
1651     that matched the final iteration. For example, after
1652 nigel 75 .sp
1653     (tweedle[dume]{3}\es*)+
1654     .sp
1655 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1656     "tweedledee". However, if there are nested capturing subpatterns, the
1657     corresponding captured values may have been set in previous iterations. For
1658     example, after
1659 nigel 75 .sp
1660 nigel 63 /(a|(b))+/
1661 nigel 75 .sp
1662 nigel 63 matches "aba" the value of the second captured substring is "b".
1663 nigel 75 .
1664     .
1665     .\" HTML <a name="atomicgroup"></a>
1667 nigel 63 .rs
1668     .sp
1669 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1670     repetition, failure of what follows normally causes the repeated item to be
1671     re-evaluated to see if a different number of repeats allows the rest of the
1672     pattern to match. Sometimes it is useful to prevent this, either to change the
1673     nature of the match, or to cause it fail earlier than it otherwise might, when
1674     the author of the pattern knows there is no point in carrying on.
1675 nigel 75 .P
1676     Consider, for example, the pattern \ed+foo when applied to the subject line
1677     .sp
1678 nigel 63 123456bar
1679 nigel 75 .sp
1680 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1681 nigel 75 action of the matcher is to try again with only 5 digits matching the \ed+
1682 nigel 63 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1683     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1684     that once a subpattern has matched, it is not to be re-evaluated in this way.
1685 nigel 75 .P
1686 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1687 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1688     special parenthesis, starting with (?> as in this example:
1689 nigel 75 .sp
1690     (?>\ed+)foo
1691     .sp
1692 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1693     it has matched, and a failure further into the pattern is prevented from
1694     backtracking into it. Backtracking past it to previous items, however, works as
1695     normal.
1696 nigel 75 .P
1697 nigel 63 An alternative description is that a subpattern of this type matches the string
1698     of characters that an identical standalone pattern would match, if anchored at
1699     the current point in the subject string.
1700 nigel 75 .P
1701 nigel 63 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1702     the above example can be thought of as a maximizing repeat that must swallow
1703 nigel 75 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1704 nigel 63 number of digits they match in order to make the rest of the pattern match,
1705 nigel 75 (?>\ed+) can only match an entire sequence of digits.
1706     .P
1707 nigel 63 Atomic groups in general can of course contain arbitrarily complicated
1708     subpatterns, and can be nested. However, when the subpattern for an atomic
1709     group is just a single repeated item, as in the example above, a simpler
1710     notation, called a "possessive quantifier" can be used. This consists of an
1711     additional + character following a quantifier. Using this notation, the
1712     previous example can be rewritten as
1713 nigel 75 .sp
1714     \ed++foo
1715     .sp
1716 ph10 208 Note that a possessive quantifier can be used with an entire group, for
1717     example:
1718     .sp
1719     (abc|xyz){2,3}+
1720     .sp
1721 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1722     option is ignored. They are a convenient notation for the simpler forms of
1723 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1724     quantifier and the equivalent atomic group, though there may be a performance
1725     difference; possessive quantifiers should be slightly faster.
1726 nigel 75 .P
1727 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1728     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1729     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1730     package, and PCRE copied it from there. It ultimately found its way into Perl
1731     at release 5.10.
1732 nigel 75 .P
1733 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1734     pattern constructs. For example, the sequence A+B is treated as A++B because
1735     there is no point in backtracking into a sequence of A's when B must follow.
1736     .P
1737 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1738     be repeated an unlimited number of times, the use of an atomic group is the
1739     only way to avoid some failing matches taking a very long time indeed. The
1740     pattern
1741 nigel 75 .sp
1742     (\eD+|<\ed+>)*[!?]
1743     .sp
1744 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1745     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1746     quickly. However, if it is applied to
1747 nigel 75 .sp
1748 nigel 63 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1749 nigel 75 .sp
1750 nigel 63 it takes a long time before reporting failure. This is because the string can
1751 nigel 75 be divided between the internal \eD+ repeat and the external * repeat in a
1752     large number of ways, and all have to be tried. (The example uses [!?] rather
1753     than a single character at the end, because both PCRE and Perl have an
1754     optimization that allows for fast failure when a single character is used. They
1755     remember the last single character that is required for a match, and fail early
1756     if it is not present in the string.) If the pattern is changed so that it uses
1757     an atomic group, like this:
1758     .sp
1759     ((?>\eD+)|<\ed+>)*[!?]
1760     .sp
1761 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1762 nigel 75 .
1763     .
1764     .\" HTML <a name="backreferences"></a>
1766 nigel 63 .rs
1767     .sp
1768     Outside a character class, a backslash followed by a digit greater than 0 (and
1769     possibly further digits) is a back reference to a capturing subpattern earlier
1770     (that is, to its left) in the pattern, provided there have been that many
1771     previous capturing left parentheses.
1772 nigel 75 .P
1773 nigel 63 However, if the decimal number following the backslash is less than 10, it is
1774     always taken as a back reference, and causes an error only if there are not
1775     that many capturing left parentheses in the entire pattern. In other words, the
1776     parentheses that are referenced need not be to the left of the reference for
1777 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1778     when a repetition is involved and the subpattern to the right has participated
1779     in an earlier iteration.
1780     .P
1781 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1782     whose number is 10 or more using this syntax because a sequence such as \e50 is
1783     interpreted as a character defined in octal. See the subsection entitled
1784 nigel 91 "Non-printing characters"
1785 nigel 75 .\" HTML <a href="#digitsafterbackslash">
1786     .\" </a>
1787     above
1788     .\"
1789 nigel 93 for further details of the handling of digits following a backslash. There is
1790     no such problem when named parentheses are used. A back reference to any
1791     subpattern is possible using named parentheses (see below).
1792 nigel 75 .P
1793 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1794 ph10 572 backslash is to use the \eg escape sequence. This escape must be followed by an
1795     unsigned number or a negative number, optionally enclosed in braces. These
1796     examples are all identical:
1797 nigel 93 .sp
1798     (ring), \e1
1799     (ring), \eg1
1800     (ring), \eg{1}
1801     .sp
1802 ph10 208 An unsigned number specifies an absolute reference without the ambiguity that
1803     is present in the older syntax. It is also useful when literal digits follow
1804     the reference. A negative number is a relative reference. Consider this
1805     example:
1806 nigel 93 .sp
1807     (abc(def)ghi)\eg{-1}
1808     .sp
1809     The sequence \eg{-1} is a reference to the most recently started capturing
1810 ph10 572 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1811     Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1812     can be helpful in long patterns, and also in patterns that are created by
1813     joining together fragments that contain references within themselves.
1814 nigel 93 .P
1815 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1816     the current subject string, rather than anything matching the subpattern
1817     itself (see
1818     .\" HTML <a href="#subpatternsassubroutines">
1819     .\" </a>
1820     "Subpatterns as subroutines"
1821     .\"
1822     below for a way of doing that). So the pattern
1823 nigel 75 .sp
1824     (sens|respons)e and \e1ibility
1825     .sp
1826 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1827     "sense and responsibility". If caseful matching is in force at the time of the
1828     back reference, the case of letters is relevant. For example,
1829 nigel 75 .sp
1830     ((?i)rah)\es+\e1
1831     .sp
1832 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1833     capturing subpattern is matched caselessly.
1834 nigel 75 .P
1835 ph10 171 There are several different ways of writing back references to named
1836     subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1837     \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1838     back reference syntax, in which \eg can be used for both numeric and named
1839     references, is also supported. We could rewrite the above example in any of
1840 nigel 93 the following ways:
1841 nigel 75 .sp
1842 nigel 93 (?<p1>(?i)rah)\es+\ek<p1>
1843 ph10 171 (?'p1'(?i)rah)\es+\ek{p1}
1844 nigel 91 (?P<p1>(?i)rah)\es+(?P=p1)
1845 ph10 171 (?<p1>(?i)rah)\es+\eg{p1}
1846 nigel 75 .sp
1847 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1848     after the reference.
1849     .P
1850 nigel 63 There may be more than one back reference to the same subpattern. If a
1851     subpattern has not actually been used in a particular match, any back
1852 ph10 456 references to it always fail by default. For example, the pattern
1853 nigel 75 .sp
1854     (a|(bc))\e2
1855     .sp
1856 ph10 461 always fails if it starts to match "a" rather than "bc". However, if the
1857     PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1858 ph10 456 unset value matches an empty string.
1859     .P
1860     Because there may be many capturing parentheses in a pattern, all digits
1861     following a backslash are taken as part of a potential back reference number.
1862     If the pattern continues with a digit character, some delimiter must be used to
1863     terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1864 ph10 968 white space. Otherwise, the \eg{ syntax or an empty comment (see
1865 nigel 75 .\" HTML <a href="#comments">
1866     .\" </a>
1867     "Comments"
1868     .\"
1869     below) can be used.
1870 ph10 488 .
1871     .SS "Recursive back references"
1872     .rs
1873     .sp
1874 nigel 63 A back reference that occurs inside the parentheses to which it refers fails
1875 nigel 75 when the subpattern is first used, so, for example, (a\e1) never matches.
1876 nigel 63 However, such references can be useful inside repeated subpatterns. For
1877     example, the pattern
1878 nigel 75 .sp
1879     (a|b\e1)+
1880     .sp
1881 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1882     the subpattern, the back reference matches the character string corresponding
1883     to the previous iteration. In order for this to work, the pattern must be such
1884     that the first iteration does not need to match the back reference. This can be
1885     done using alternation, as in the example above, or by a quantifier with a
1886     minimum of zero.
1887 ph10 488 .P
1888     Back references of this type cause the group that they reference to be treated
1889     as an
1890     .\" HTML <a href="#atomicgroup">
1891     .\" </a>
1892     atomic group.
1893     .\"
1894     Once the whole group has been matched, a subsequent matching failure cannot
1895     cause backtracking into the middle of the group.
1896 nigel 75 .
1897     .
1898     .\" HTML <a name="bigassertions"></a>
1899 nigel 63 .SH ASSERTIONS
1900     .rs
1901     .sp
1902     An assertion is a test on the characters following or preceding the current
1903     matching point that does not actually consume any characters. The simple
1904 nigel 75 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1905     .\" HTML <a href="#smallassertions">
1906     .\" </a>
1907     above.
1908     .\"
1909     .P
1910 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1911     those that look ahead of the current position in the subject string, and those
1912 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1913     except that it does not cause the current matching position to be changed.
1914     .P
1915 ph10 637 Assertion subpatterns are not capturing subpatterns. If such an assertion
1916     contains capturing subpatterns within it, these are counted for the purposes of
1917     numbering the capturing subpatterns in the whole pattern. However, substring
1918     capturing is carried out only for positive assertions, because it does not make
1919     sense for negative assertions.
1920     .P
1921 ph10 643 For compatibility with Perl, assertion subpatterns may be repeated; though
1922 ph10 654 it makes no sense to assert the same thing several times, the side effect of
1923 ph10 643 capturing parentheses may occasionally be useful. In practice, there only three
1924     cases:
1925 ph10 637 .sp
1926 ph10 654 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1927     However, it may contain internal capturing parenthesized groups that are called
1928 ph10 637 from elsewhere via the
1929     .\" HTML <a href="#subpatternsassubroutines">
1930     .\" </a>
1931     subroutine mechanism.
1932     .\"
1933     .sp
1934 ph10 654 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1935     were {0,1}. At run time, the rest of the pattern match is tried with and
1936 ph10 637 without the assertion, the order depending on the greediness of the quantifier.
1937     .sp
1938 ph10 654 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1939 ph10 637 The assertion is obeyed just once when encountered during matching.
1940 nigel 75 .
1941     .
1942     .SS "Lookahead assertions"
1943     .rs
1944     .sp
1945 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1946     negative assertions. For example,
1947 nigel 75 .sp
1948     \ew+(?=;)
1949     .sp
1950 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1951     the match, and
1952 nigel 75 .sp
1953 nigel 63 foo(?!bar)
1954 nigel 75 .sp
1955 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1956     apparently similar pattern
1957 nigel 75 .sp
1958 nigel 63 (?!foo)bar
1959 nigel 75 .sp
1960 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1961     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1962     (?!foo) is always true when the next three characters are "bar". A
1963 nigel 75 lookbehind assertion is needed to achieve the other effect.
1964     .P
1965 nigel 63 If you want to force a matching failure at some point in a pattern, the most
1966     convenient way to do it is with (?!) because an empty string always matches, so
1967     an assertion that requires there not to be an empty string must always fail.
1968 ph10 572 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1969 nigel 75 .
1970     .
1971     .\" HTML <a name="lookbehind"></a>
1972     .SS "Lookbehind assertions"
1973     .rs
1974     .sp
1975 nigel 63 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1976     negative assertions. For example,
1977 nigel 75 .sp
1978 nigel 63 (?<!foo)bar
1979 nigel 75 .sp
1980 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1981     a lookbehind assertion are restricted such that all the strings it matches must
1982 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1983     do not all have to have the same fixed length. Thus
1984 nigel 75 .sp
1985 nigel 63 (?<=bullock|donkey)
1986 nigel 75 .sp
1987 nigel 63 is permitted, but
1988 nigel 75 .sp
1989 nigel 63 (?<!dogs?|cats?)
1990 nigel 75 .sp
1991 nigel 63 causes an error at compile time. Branches that match different length strings
1992     are permitted only at the top level of a lookbehind assertion. This is an
1993 ph10 572 extension compared with Perl, which requires all branches to match the same
1994     length of string. An assertion such as
1995 nigel 75 .sp
1996 nigel 63 (?<=ab(c|de))
1997 nigel 75 .sp
1998 nigel 63 is not permitted, because its single top-level branch can match two different
1999 ph10 454 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2000     branches:
2001 nigel 75 .sp
2002 nigel 63 (?<=abc|abde)
2003 nigel 75 .sp
2004 ph10 572 In some cases, the escape sequence \eK
2005 ph10 168 .\" HTML <a href="#resetmatchstart">
2006     .\" </a>
2007     (see above)
2008     .\"
2009 ph10 461 can be used instead of a lookbehind assertion to get round the fixed-length
2010 ph10 454 restriction.
2011 ph10 168 .P
2012 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
2013 nigel 93 temporarily move the current position back by the fixed length and then try to
2014 nigel 63 match. If there are insufficient characters before the current position, the
2015 nigel 93 assertion fails.
2016 nigel 75 .P
2017 ph10 859 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2018     unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2019     it impossible to calculate the length of the lookbehind. The \eX and \eR
2020     escapes, which can match different numbers of data units, are also not
2021     permitted.
2022 nigel 75 .P
2023 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2024     .\" </a>
2025     "Subroutine"
2026     .\"
2027     calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2028 ph10 461 as the subpattern matches a fixed-length string.
2029 ph10 454 .\" HTML <a href="#recursion">
2030     .\" </a>
2031     Recursion,
2032     .\"
2033     however, is not supported.
2034     .P
2035 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2036 ph10 456 specify efficient matching of fixed-length strings at the end of subject
2037     strings. Consider a simple pattern such as
2038 nigel 75 .sp
2039 nigel 63 abcd$
2040 nigel 75 .sp
2041 nigel 63 when applied to a long string that does not match. Because matching proceeds
2042     from left to right, PCRE will look for each "a" in the subject and then see if
2043     what follows matches the rest of the pattern. If the pattern is specified as
2044 nigel 75 .sp
2045 nigel 63 ^.*abcd$
2046 nigel 75 .sp
2047 nigel 63 the initial .* matches the entire string at first, but when this fails (because
2048     there is no following "a"), it backtracks to match all but the last character,
2049     then all but the last two characters, and so on. Once again the search for "a"
2050     covers the entire string, from right to left, so we are no better off. However,
2051     if the pattern is written as
2052 nigel 75 .sp
2053 nigel 63 ^.*+(?<=abcd)
2054 nigel 75 .sp
2055 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
2056 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
2057     characters. If it fails, the match fails immediately. For long strings, this
2058     approach makes a significant difference to the processing time.
2059 nigel 75 .
2060     .
2061     .SS "Using multiple assertions"
2062     .rs
2063     .sp
2064 nigel 63 Several assertions (of any sort) may occur in succession. For example,
2065 nigel 75 .sp
2066     (?<=\ed{3})(?<!999)foo
2067     .sp
2068 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
2069     the assertions is applied independently at the same point in the subject
2070     string. First there is a check that the previous three characters are all
2071     digits, and then there is a check that the same three characters are not "999".
2072 nigel 75 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2073 nigel 63 of which are digits and the last three of which are not "999". For example, it
2074     doesn't match "123abcfoo". A pattern to do that is
2075 nigel 75 .sp
2076     (?<=\ed{3}...)(?<!999)foo
2077     .sp
2078 nigel 63 This time the first assertion looks at the preceding six characters, checking
2079     that the first three are digits, and then the second assertion checks that the
2080     preceding three characters are not "999".
2081 nigel 75 .P
2082 nigel 63 Assertions can be nested in any combination. For example,
2083 nigel 75 .sp
2084 nigel 63 (?<=(?<!foo)bar)baz
2085 nigel 75 .sp
2086 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2087     preceded by "foo", while
2088 nigel 75 .sp
2089     (?<=\ed{3}(?!999)...)foo
2090     .sp
2091     is another pattern that matches "foo" preceded by three digits and any three
2092 nigel 63 characters that are not "999".
2093 nigel 75 .
2094     .
2095 nigel 91 .\" HTML <a name="conditions"></a>
2097 nigel 63 .rs
2098     .sp
2099     It is possible to cause the matching process to obey a subpattern
2100     conditionally or to choose between two alternative subpatterns, depending on
2101 ph10 461 the result of an assertion, or whether a specific capturing subpattern has
2102 ph10 456 already been matched. The two possible forms of conditional subpattern are:
2103 nigel 75 .sp
2104 nigel 63 (?(condition)yes-pattern)
2105     (?(condition)yes-pattern|no-pattern)
2106 nigel 75 .sp
2107 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
2108     no-pattern (if present) is used. If there are more than two alternatives in the
2109 ph10 557 subpattern, a compile-time error occurs. Each of the two alternatives may
2110 ph10 579 itself contain nested subpatterns of any form, including conditional
2111 ph10 557 subpatterns; the restriction to two alternatives applies only at the level of
2112 ph10 579 the condition. This pattern fragment is an example where the alternatives are
2113 ph10 557 complex:
2114     .sp
2115     (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2116     .sp
2117 nigel 75 .P
2118 nigel 93 There are four kinds of condition: references to subpatterns, references to
2119     recursion, a pseudo-condition called DEFINE, and assertions.
2120     .
2121     .SS "Checking for a used subpattern by number"
2122     .rs
2123     .sp
2124     If the text between the parentheses consists of a sequence of digits, the
2125 ph10 456 condition is true if a capturing subpattern of that number has previously
2126 ph10 461 matched. If there is more than one capturing subpattern with the same number
2127     (see the earlier
2128 ph10 456 .\"
2129     .\" HTML <a href="#recursion">
2130     .\" </a>
2131     section about duplicate subpattern numbers),
2132     .\"
2133 ph10 572 the condition is true if any of them have matched. An alternative notation is
2134 ph10 456 to precede the digits with a plus or minus sign. In this case, the subpattern
2135     number is relative rather than absolute. The most recently opened parentheses
2136 ph10 572 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2137 ph10 579 loops it can also make sense to refer to subsequent groups. The next
2138 ph10 572 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2139     zero in any of these forms is not used; it provokes a compile-time error.)
2140 nigel 91 .P
2141     Consider the following pattern, which contains non-significant white space to
2142     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2143     three parts for ease of discussion:
2144 nigel 75 .sp
2145     ( \e( )? [^()]+ (?(1) \e) )
2146     .sp
2147 nigel 63 The first part matches an optional opening parenthesis, and if that
2148     character is present, sets it as the first captured substring. The second part
2149     matches one or more characters that are not parentheses. The third part is a
2150 ph10 572 conditional subpattern that tests whether or not the first set of parentheses
2151     matched. If they did, that is, if subject started with an opening parenthesis,
2152 nigel 63 the condition is true, and so the yes-pattern is executed and a closing
2153     parenthesis is required. Otherwise, since no-pattern is not present, the
2154     subpattern matches nothing. In other words, this pattern matches a sequence of
2155 nigel 93 non-parentheses, optionally enclosed in parentheses.
2156 ph10 167 .P
2157 ph10 172 If you were embedding this pattern in a larger one, you could use a relative
2158 ph10 167 reference:
2159     .sp
2160     ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2161     .sp
2162     This makes the fragment independent of the parentheses in the larger pattern.
2163 nigel 93 .
2164     .SS "Checking for a used subpattern by name"
2165     .rs
2166 nigel 91 .sp
2167 nigel 93 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2168     subpattern by name. For compatibility with earlier versions of PCRE, which had
2169     this facility before Perl, the syntax (?(name)...) is also recognized. However,
2170     there is a possible ambiguity with this syntax, because subpattern names may
2171     consist entirely of digits. PCRE looks first for a named subpattern; if it
2172     cannot find one and the name consists entirely of digits, PCRE looks for a
2173     subpattern of that number, which must be greater than zero. Using subpattern
2174     names that consist entirely of digits is not recommended.
2175     .P
2176     Rewriting the above example to use a named subpattern gives this:
2177 nigel 91 .sp
2178 nigel 93 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2179     .sp
2180 ph10 461 If the name used in a condition of this kind is a duplicate, the test is
2181     applied to all subpatterns of the same name, and is true if any one of them has
2182 ph10 459 matched.
2183 nigel 93 .
2184     .SS "Checking for pattern recursion"
2185     .rs
2186     .sp
2187 nigel 91 If the condition is the string (R), and there is no subpattern with the name R,
2188 nigel 93 the condition is true if a recursive call to the whole pattern or any
2189     subpattern has been made. If digits or a name preceded by ampersand follow the
2190     letter R, for example:
2191     .sp
2192     (?(R3)...) or (?(R&name)...)
2193     .sp
2194 ph10 456 the condition is true if the most recent recursion is into a subpattern whose
2195 nigel 93 number or name is given. This condition does not check the entire recursion
2196 ph10 461 stack. If the name used in a condition of this kind is a duplicate, the test is
2197     applied to all subpatterns of the same name, and is true if any one of them is
2198     the most recent recursion.
2199 nigel 75 .P
2200 ph10 461 At "top level", all these recursion test conditions are false.
2201 ph10 454 .\" HTML <a href="#recursion">
2202     .\" </a>
2203 ph10 459 The syntax for recursive patterns
2204 ph10 454 .\"
2205 ph10 459 is described below.
2206 nigel 93 .
2207 ph10 572 .\" HTML <a name="subdefine"></a>
2208 nigel 93 .SS "Defining subpatterns for use by reference only"
2209     .rs
2210     .sp
2211     If the condition is the string (DEFINE), and there is no subpattern with the
2212     name DEFINE, the condition is always false. In this case, there may be only one
2213     alternative in the subpattern. It is always skipped if control reaches this
2214     point in the pattern; the idea of DEFINE is that it can be used to define
2215 ph10 716 subroutines that can be referenced from elsewhere. (The use of
2216 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2217     .\" </a>
2218 ph10 716 subroutines
2219 ph10 454 .\"
2220 ph10 572 is described below.) For example, a pattern to match an IPv4 address such as
2221 ph10 968 "" could be written like this (ignore white space and line
2222 ph10 572 breaks):
2223 nigel 93 .sp
2224     (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2225     \eb (?&byte) (\e.(?&byte)){3} \eb
2226     .sp
2227     The first part of the pattern is a DEFINE group inside which a another group
2228     named "byte" is defined. This matches an individual component of an IPv4
2229     address (a number less than 256). When matching takes place, this part of the
2230 ph10 456 pattern is skipped because DEFINE acts like a false condition. The rest of the
2231     pattern uses references to the named group to match the four dot-separated
2232     components of an IPv4 address, insisting on a word boundary at each end.
2233 nigel 93 .
2234     .SS "Assertion conditions"
2235     .rs
2236     .sp
2237     If the condition is not in any of the above formats, it must be an assertion.
2238 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
2239     this pattern, again containing non-significant white space, and with the two
2240     alternatives on the second line:
2241 nigel 75 .sp
2242 nigel 63 (?(?=[^a-z]*[a-z])
2243 nigel 75 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2244     .sp
2245 nigel 63 The condition is a positive lookahead assertion that matches an optional
2246     sequence of non-letters followed by a letter. In other words, it tests for the
2247     presence of at least one letter in the subject. If a letter is found, the
2248     subject is matched against the first alternative; otherwise it is matched
2249     against the second. This pattern matches strings in one of the two forms
2250     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2251 nigel 75 .
2252     .
2253     .\" HTML <a name="comments"></a>
2254 nigel 63 .SH COMMENTS
2255     .rs
2256     .sp
2257 ph10 579 There are two ways of including comments in patterns that are processed by
2258 ph10 562 PCRE. In both cases, the start of the comment must not be in a character class,
2259     nor in the middle of any other sequence of related characters such as (?: or a
2260     subpattern name or number. The characters that make up a comment play no part
2261     in the pattern matching.
2262     .P
2263 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
2264 ph10 562 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2265     option is set, an unescaped # character also introduces a comment, which in
2266     this case continues to immediately after the next newline character or
2267     character sequence in the pattern. Which characters are interpreted as newlines
2268 ph10 859 is controlled by the options passed to a compiling function or by a special
2269 ph10 562 sequence at the start of the pattern, as described in the section entitled
2270 ph10 572 .\" HTML <a href="#newlines">
2271 ph10 556 .\" </a>
2272     "Newline conventions"
2273     .\"
2274 ph10 572 above. Note that the end of this type of comment is a literal newline sequence
2275     in the pattern; escape sequences that happen to represent a newline do not
2276     count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2277     default newline convention is in force:
2278 ph10 556 .sp
2279     abc #comment \en still comment
2280     .sp
2281 ph10 579 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2282 ph10 556 a newline in the pattern. The sequence \en is still literal at this stage, so
2283     it does not terminate the comment. Only an actual character with the code value
2284 ph10 562 0x0a (the default newline) does so.
2285 nigel 75 .
2286     .
2287 nigel 91 .\" HTML <a name="recursion"></a>
2288 nigel 75 .SH "RECURSIVE PATTERNS"
2289 nigel 63 .rs
2290     .sp
2291     Consider the problem of matching a string in parentheses, allowing for
2292     unlimited nested parentheses. Without the use of recursion, the best that can
2293     be done is to use a pattern that matches up to some fixed depth of nesting. It
2294 nigel 93 is not possible to handle an arbitrary nesting depth.
2295     .P
2296     For some time, Perl has provided a facility that allows regular expressions to
2297     recurse (amongst other things). It does this by interpolating Perl code in the
2298     expression at run time, and the code can refer to the expression itself. A Perl
2299     pattern using code interpolation to solve the parentheses problem can be
2300     created like this:
2301 nigel 75 .sp
2302     $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2303     .sp
2304 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2305 nigel 93 recursively to the pattern in which it appears.
2306 nigel 75 .P
2307 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2308     supports special syntax for recursion of the entire pattern, and also for
2309     individual subpattern recursion. After its introduction in PCRE and Python,
2310 ph10 453 this kind of recursion was subsequently introduced into Perl at release 5.10.
2311 nigel 75 .P
2312 nigel 93 A special item that consists of (? followed by a number greater than zero and a
2313 ph10 716 closing parenthesis is a recursive subroutine call of the subpattern of the
2314     given number, provided that it occurs inside that subpattern. (If not, it is a
2315 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2316     .\" </a>
2317 ph10 716 non-recursive subroutine
2318 ph10 454 .\"
2319 nigel 93 call, which is described in the next section.) The special item (?R) or (?0) is
2320     a recursive call of the entire regular expression.
2321 nigel 87 .P
2322     This PCRE pattern solves the nested parentheses problem (assume the
2323     PCRE_EXTENDED option is set so that white space is ignored):
2324 nigel 75 .sp
2325 ph10 456 \e( ( [^()]++ | (?R) )* \e)
2326 nigel 75 .sp
2327 nigel 63 First it matches an opening parenthesis. Then it matches any number of
2328     substrings which can either be a sequence of non-parentheses, or a recursive
2329 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
2330 ph10 461 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2331 ph10 456 to avoid backtracking into sequences of non-parentheses.
2332 nigel 75 .P
2333 nigel 63 If this were part of a larger pattern, you would not want to recurse the entire
2334     pattern, so instead you could use this:
2335 nigel 75 .sp
2336 ph10 456 ( \e( ( [^()]++ | (?1) )* \e) )
2337 nigel 75 .sp
2338 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
2339 ph10 172 them instead of the whole pattern.
2340 ph10 166 .P
2341     In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2342 ph10 572 is made easier by the use of relative references. Instead of (?1) in the
2343     pattern above you can write (?-2) to refer to the second most recently opened
2344     parentheses preceding the recursion. In other words, a negative number counts
2345     capturing parentheses leftwards from the point at which it is encountered.
2346 ph10 166 .P
2347     It is also possible to refer to subsequently opened parentheses, by writing
2348     references such as (?+2). However, these cannot be recursive because the
2349     reference is not inside the parentheses that are referenced. They are always
2350 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2351     .\" </a>
2352 ph10 716 non-recursive subroutine
2353 ph10 454 .\"
2354     calls, as described in the next section.
2355 ph10 166 .P
2356     An alternative approach is to use named parentheses instead. The Perl syntax
2357     for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2358     could rewrite the above example as follows:
2359 nigel 75 .sp
2360 ph10 456 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2361 nigel 75 .sp
2362 nigel 93 If there is more than one subpattern with the same name, the earliest one is
2363 ph10 172 used.
2364 ph10 166 .P
2365     This particular example pattern that we have been looking at contains nested
2366 ph10 456 unlimited repeats, and so the use of a possessive quantifier for matching
2367     strings of non-parentheses is important when applying the pattern to strings
2368     that do not match. For example, when this pattern is applied to
2369 nigel 75 .sp
2370 nigel 63 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2371 nigel 75 .sp
2372 ph10 456 it yields "no match" quickly. However, if a possessive quantifier is not used,
2373 nigel 63 the match runs for a very long time indeed because there are so many different
2374     ways the + and * repeats can carve up the subject, and all have to be tested
2375     before failure can be reported.
2376 nigel 75 .P
2377 ph10 464 At the end of a match, the values of capturing parentheses are those from
2378     the outermost level. If you want to obtain intermediate values, a callout
2379     function can be used (see below and the
2380 nigel 63 .\" HREF
2381 nigel 75 \fBpcrecallout\fP
2382 nigel 63 .\"
2383     documentation). If the pattern above is matched against
2384 nigel 75 .sp
2385 nigel 63 (ab(cd)ef)
2386 nigel 75 .sp
2387 ph10 464 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2388     the last value taken on at the top level. If a capturing subpattern is not
2389 ph10 724 matched at the top level, its final captured value is unset, even if it was
2390     (temporarily) set at a deeper level during the matching process.
2391 nigel 75 .P
2392 ph10 464 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2393     obtain extra memory to store data during a recursion, which it does by using
2394     \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2395     be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2396     .P
2397 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2398     Consider this pattern, which matches text in angle brackets, allowing for
2399     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2400     recursing), whereas any characters are permitted at the outer level.
2401 nigel 75 .sp
2402     < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2403     .sp
2404 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
2405     different alternatives for the recursive and non-recursive cases. The (?R) item
2406     is the actual recursive call.
2407 nigel 75 .
2408     .
2409 ph10 453 .\" HTML <a name="recursiondifference"></a>
2410 ph10 724 .SS "Differences in recursion processing between PCRE and Perl"
2411 ph10 453 .rs
2412     .sp
2413 ph10 724 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2414     (like Python, but unlike Perl), a recursive subpattern call is always treated
2415     as an atomic group. That is, once it has matched some of the subject string, it
2416     is never re-entered, even if it contains untried alternatives and there is a
2417     subsequent matching failure. This can be illustrated by the following pattern,
2418     which purports to match a palindromic string that contains an odd number of
2419     characters (for example, "a", "aba", "abcba", "abcdcba"):
2420 ph10 453 .sp
2421     ^(.|(.)(?1)\e2)$
2422     .sp
2423 ph10 461 The idea is that it either matches a single character, or two identical
2424     characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2425 ph10 453 it does not if the pattern is longer than three characters. Consider the
2426     subject string "abcba":
2427     .P
2428 ph10 461 At the top level, the first character is matched, but as it is not at the end
2429 ph10 453 of the string, the first alternative fails; the second alternative is taken
2430     and the recursion kicks in. The recursive call to subpattern 1 successfully
2431     matches the next character ("b"). (Note that the beginning and end of line
2432     tests are not part of the recursion).
2433     .P
2434     Back at the top level, the next character ("c") is compared with what
2435 ph10 461 subpattern 2 matched, which was "a". This fails. Because the recursion is
2436 ph10 453 treated as an atomic group, there are now no backtracking points, and so the
2437     entire match fails. (Perl is able, at this point, to re-enter the recursion and
2438     try the second alternative.) However, if the pattern is written with the
2439     alternatives in the other order, things are different:
2440     .sp
2441     ^((.)(?1)\e2|.)$
2442     .sp
2443 ph10 461 This time, the recursing alternative is tried first, and continues to recurse
2444     until it runs out of characters, at which point the recursion fails. But this
2445     time we do have another alternative to try at the higher level. That is the big
2446 ph10 453 difference: in the previous case the remaining alternative is at a deeper
2447     recursion level, which PCRE cannot use.
2448     .P
2449 ph10 572 To change the pattern so that it matches all palindromic strings, not just
2450     those with an odd number of characters, it is tempting to change the pattern to
2451     this:
2452 ph10 453 .sp
2453     ^((.)(?1)\e2|.?)$
2454     .sp
2455 ph10 461 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2456     deeper recursion has matched a single character, it cannot be entered again in
2457     order to match an empty string. The solution is to separate the two cases, and
2458 ph10 453 write out the odd and even cases as alternatives at the higher level:
2459     .sp
2460     ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2461 ph10 461 .sp
2462     If you want to match typical palindromic phrases, the pattern has to ignore all
2463 ph10 453 non-word characters, which can be done like this:
2464     .sp
2465 ph10 461 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2466 ph10 453 .sp
2467 ph10 461 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2468     man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2469     the use of the possessive quantifier *+ to avoid backtracking into sequences of
2470 ph10 453 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2471     more) to match typical phrases, and Perl takes so long that you think it has
2472     gone into a loop.
2473 ph10 456 .P
2474     \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2475     string does not start with a palindrome that is shorter than the entire string.
2476     For example, although "abcba" is correctly matched, if the subject is "ababa",
2477     PCRE finds the palindrome "aba" at the start, then fails at top level because
2478     the end of the string does not follow. Once again, it cannot jump back into the
2479     recursion to try other alternatives, so the entire match fails.
2480 ph10 724 .P
2481 ph10 733 The second way in which PCRE and Perl differ in their recursion processing is
2482     in the handling of captured values. In Perl, when a subpattern is called
2483     recursively or as a subpattern (see the next section), it has no access to any
2484     values that were captured outside the recursion, whereas in PCRE these values
2485 ph10 724 can be referenced. Consider this pattern:
2486     .sp
2487     ^(.)(\e1|a(?2))
2488     .sp
2489 ph10 733 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2490 ph10 724 then in the second group, when the back reference \e1 fails to match "b", the
2491     second alternative matches "a" and then recurses. In the recursion, \e1 does
2492     now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2493     match because inside the recursive call \e1 cannot access the externally set
2494     value.
2495 ph10 453 .
2496     .
2497 nigel 63 .\" HTML <a name="subpatternsassubroutines"></a>
2499 nigel 63 .rs
2500     .sp
2501 ph10 716 If the syntax for a recursive subpattern call (either by number or by
2502 nigel 63 name) is used outside the parentheses to which it refers, it operates like a
2503 ph10 716 subroutine in a programming language. The called subpattern may be defined
2504 ph10 166 before or after the reference. A numbered reference can be absolute or
2505     relative, as in these examples:
2506 nigel 75 .sp
2507 ph10 166 (...(absolute)...)...(?2)...
2508     (...(relative)...)...(?-1)...
2509 ph10 172 (...(?+1)...(relative)...
2510 ph10 166 .sp
2511     An earlier example pointed out that the pattern
2512     .sp
2513 nigel 75 (sens|respons)e and \e1ibility
2514     .sp
2515 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
2516     "sense and responsibility". If instead the pattern
2517 nigel 75 .sp
2518 nigel 63 (sens|respons)e and (?1)ibility
2519 nigel 75 .sp
2520 nigel 63 is used, it does match "sense and responsibility" as well as the other two
2521 nigel 93 strings. Another example is given in the discussion of DEFINE above.
2522 nigel 87 .P
2523 ph10 716 All subroutine calls, whether recursive or not, are always treated as atomic
2524     groups. That is, once a subroutine has matched some of the subject string, it
2525     is never re-entered, even if it contains untried alternatives and there is a
2526     subsequent matching failure. Any capturing parentheses that are set during the
2527     subroutine call revert to their previous values afterwards.
2528 nigel 93 .P
2529 ph10 716 Processing options such as case-independence are fixed when a subpattern is
2530     defined, so if it is used as a subroutine, such options cannot be changed for
2531     different calls. For example, consider this pattern:
2532 nigel 93 .sp
2533 ph10 166 (abc)(?i:(?-1))
2534 nigel 93 .sp
2535     It matches "abcabc". It does not match "abcABC" because the change of
2536     processing option does not affect the called subpattern.
2537 nigel 75 .
2538     .
2539 ph10 333 .\" HTML <a name="onigurumasubroutines"></a>
2541     .rs
2542     .sp
2543 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2544     a number enclosed either in angle brackets or single quotes, is an alternative
2545     syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2546 ph10 333 are two of the examples used above, rewritten using this syntax:
2547     .sp
2548     (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2549     (sens|respons)e and \eg'1'ibility
2550     .sp
2551 ph10 345 PCRE supports an extension to Oniguruma: if a number is preceded by a
2552 ph10 333 plus or a minus sign it is taken as a relative reference. For example:
2553     .sp
2554     (abc)(?i:\eg<-1>)
2555     .sp
2556 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2557 ph10 333 synonymous. The former is a back reference; the latter is a subroutine call.
2558     .
2559     .
2560 nigel 63 .SH CALLOUTS
2561     .rs
2562     .sp
2563     Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2564     code to be obeyed in the middle of matching a regular expression. This makes it
2565     possible, amongst other things, to extract different substrings that match the
2566     same pair of parentheses when there is a repetition.
2567 nigel 75 .P
2568 nigel 63 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2569     code. The feature is called "callout". The caller of PCRE provides an external
2570 ph10 903 function by putting its entry point in the global variable \fIpcre_callout\fP
2571 ph10 859 (8-bit library) or \fIpcre16_callout\fP (16-bit library). By default, this
2572     variable contains NULL, which disables all calling out.
2573 nigel 75 .P
2574 nigel 63 Within a regular expression, (?C) indicates the points at which the external
2575     function is to be called. If you want to identify different callout points, you
2576     can put a number less than 256 after the letter C. The default value is zero.
2577     For example, this pattern has two callout points:
2578 nigel 75 .sp
2579 ph10 155 (?C1)abc(?C2)def
2580 nigel 75 .sp
2581 ph10 859 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2582 nigel 75 automatically installed before each item in the pattern. They are all numbered
2583     255.
2584     .P
2585 ph10 859 During matching, when PCRE reaches a callout point, the external function is
2586     called. It is provided with the number of the callout, the position in the
2587     pattern, and, optionally, one item of data originally supplied by the caller of
2588     the matching function. The callout function may cause matching to proceed, to
2589     backtrack, or to fail altogether. A complete description of the interface to
2590     the callout function is given in the
2591 nigel 63 .\" HREF
2592 nigel 75 \fBpcrecallout\fP
2593 nigel 63 .\"
2594     documentation.
2595 nigel 93 .
2596     .
2597 ph10 510 .\" HTML <a name="backtrackcontrol"></a>
2599 ph10 210 .rs
2600     .sp
2601 ph10 211 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2602 ph10 210 are described in the Perl documentation as "experimental and subject to change
2603 ph10 211 or removal in a future version of Perl". It goes on to say: "Their usage in
2604     production code should be noted to avoid problems during upgrades." The same
2605 ph10 210 remarks apply to the PCRE features described in this section.
2606     .P
2607 ph10 341 Since these verbs are specifically related to backtracking, most of them can be
2608 ph10 859 used only when the pattern is to be matched using one of the traditional
2609     matching functions, which use a backtracking algorithm. With the exception of
2610     (*FAIL), which behaves like a failing negative assertion, they cause an error
2611     if encountered by a DFA matching function.
2612 ph10 210 .P
2613 ph10 716 If any of these verbs are used in an assertion or in a subpattern that is
2614     called as a subroutine (whether or not recursively), their effect is confined
2615     to that subpattern; it does not extend to the surrounding pattern, with one
2616 ph10 836 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2617     a successful positive assertion \fIis\fP passed back when a match succeeds
2618     (compare capturing parentheses in assertions). Note that such subpatterns are
2619     processed as anchored at the point where they are tested. Note also that Perl's
2620 ph10 972 treatment of subroutines and assertions is different in some cases.
2621 ph10 445 .P
2622 ph10 211 The new verbs make use of what was previously invalid syntax: an opening
2623 ph10 510 parenthesis followed by an asterisk. They are generally of the form
2624 ph10 512 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2625 ph10 716 depending on whether or not an argument is present. A name is any sequence of
2626 ph10 964 characters that does not include a closing parenthesis. The maximum length of
2627     name is 255 in the 8-bit library and 65535 in the 16-bit library. If the name
2628     is empty, that is, if the closing parenthesis immediately follows the colon,
2629     the effect is as if the colon were not there. Any number of these verbs may
2630     occur in a pattern.
2631 ph10 930 .
2632     .
2633     .\" HTML <a name="nooptimize"></a>
2634     .SS "Optimizations that affect backtracking verbs"
2635     .rs
2636     .sp
2637 ph10 512 PCRE contains some optimizations that are used to speed up matching by running
2638     some checks at the start of each match attempt. For example, it may know the
2639     minimum length of matching subject, or that a particular character must be
2640     present. When one of these optimizations suppresses the running of a match, any
2641     included backtracking verbs will not, of course, be processed. You can suppress
2642     the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2643 ph10 577 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2644 ph10 930 pattern with (*NO_START_OPT). There is more discussion of this option in the
2645     section entitled
2646     .\" HTML <a href="pcreapi.html#execoptions">
2647     .\" </a>
2648     "Option bits for \fBpcre_exec()\fP"
2649     .\"
2650     in the
2651     .\" HREF
2652     \fBpcreapi\fP
2653     .\"
2654 ph10 975 documentation.
2655 ph10 836 .P
2656     Experiments with Perl suggest that it too has similar optimizations, sometimes
2657     leading to anomalous results.
2658 ph10 210 .
2659 ph10 510 .
2660 ph10 210 .SS "Verbs that act immediately"
2661     .rs
2662     .sp
2663 ph10 512 The following verbs act as soon as they are encountered. They may not be
2664 ph10 510 followed by a name.
2665 ph10 210 .sp
2666     (*ACCEPT)
2667     .sp
2668     This verb causes the match to end successfully, skipping the remainder of the
2669 ph10 716 pattern. However, when it is inside a subpattern that is called as a
2670     subroutine, only that subpattern is ended successfully. Matching then continues
2671     at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2672     far is captured. For example:
2673 ph10 210 .sp
2674 ph10 447 A((?:A|B(*ACCEPT)|C)D)
2675 ph10 210 .sp
2676 ph10 461 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2677 ph10 447 the outer parentheses.
2678 ph10 210 .sp
2679     (*FAIL) or (*F)
2680     .sp
2681 ph10 716 This verb causes a matching failure, forcing backtracking to occur. It is
2682 ph10 210 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2683     probably useful only when combined with (?{}) or (??{}). Those are, of course,
2684     Perl features that are not present in PCRE. The nearest equivalent is the
2685     callout feature, as for example in this pattern:
2686     .sp
2687     a+(?C)(*FAIL)
2688     .sp
2689 ph10 211 A match with the string "aaaa" always fails, but the callout is taken before
2690     each backtrack happens (in this example, 10 times).
2691 ph10 210 .
2692 ph10 510 .
2693     .SS "Recording which path was taken"
2694     .rs
2695     .sp
2696 ph10 512 There is one verb whose main purpose is to track how a match was arrived at,
2697     though it also has a secondary use in conjunction with advancing the match
2698 ph10 510 starting point (see (*SKIP) below).
2699     .sp
2700     (*MARK:NAME) or (*:NAME)
2701     .sp
2702     A name is always required with this verb. There may be as many instances of
2703     (*MARK) as you like in a pattern, and their names do not have to be unique.
2704     .P
2705 ph10 836 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2706 ph10 859 path is passed back to the caller as described in the section entitled
2707 ph10 510 .\" HTML <a href="pcreapi.html#extradata">
2708     .\" </a>
2709 ph10 859 "Extra data for \fBpcre_exec()\fP"
2710 ph10 510 .\"
2711 ph10 512 in the
2712 ph10 510 .\" HREF
2713     \fBpcreapi\fP
2714     .\"
2715 ph10 836 documentation. Here is an example of \fBpcretest\fP output, where the /K
2716     modifier requests the retrieval and outputting of (*MARK) data:
2717 ph10 510 .sp
2718 ph10 836 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2719     data> XY
2720 ph10 510 0: XY
2721     MK: A
2722     XZ
2723     0: XZ
2724     MK: B
2725     .sp
2726 ph10 512 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2727     indicates which of the two alternatives matched. This is a more efficient way
2728 ph10 510 of obtaining this information than putting each alternative in its own
2729     capturing parentheses.
2730     .P
2731 ph10 654 If (*MARK) is encountered in a positive assertion, its name is recorded and
2732     passed back if it is the last-encountered. This does not happen for negative
2733 ph10 716 assertions.
2734 ph10 630 .P
2735 ph10 836 After a partial match or a failed match, the name of the last encountered
2736     (*MARK) in the entire match process is returned. For example:
2737 ph10 510 .sp
2738 ph10 836 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2739     data> XP
2740 ph10 510 No match, mark = B
2741     .sp
2742 ph10 836 Note that in this unanchored example the mark is retained from the match
2743 ph10 930 attempt that started at the letter "X" in the subject. Subsequent match
2744     attempts starting at "P" and then with an empty string do not get as far as the
2745     (*MARK) item, but nevertheless do not reset it.
2746     .P
2747     If you are interested in (*MARK) values after failed matches, you should
2748 ph10 975 probably set the PCRE_NO_START_OPTIMIZE option
2749 ph10 930 .\" HTML <a href="#nooptimize">
2750     .\" </a>
2751 ph10 975 (see above)
2752 ph10 930 .\"
2753     to ensure that the match is always attempted.
2754 ph10 510 .
2755     .
2756 ph10 210 .SS "Verbs that act after backtracking"
2757     .rs
2758     .sp
2759 ph10 211 The following verbs do nothing when they are encountered. Matching continues
2760 ph10 510 with what follows, but if there is no subsequent match, causing a backtrack to
2761     the verb, a failure is forced. That is, backtracking cannot pass to the left of
2762     the verb. However, when one of these verbs appears inside an atomic group, its
2763     effect is confined to that group, because once the group has been matched,
2764     there is never any backtracking into it. In this situation, backtracking can
2765 ph10 512 "jump back" to the left of the entire atomic group. (Remember also, as stated
2766 ph10 510 above, that this localization also applies in subroutine calls and assertions.)
2767     .P
2768     These verbs differ in exactly what kind of failure occurs when backtracking
2769     reaches them.
2770 ph10 210 .sp
2771     (*COMMIT)
2772     .sp
2773 ph10 510 This verb, which may not be followed by a name, causes the whole match to fail
2774     outright if the rest of the pattern does not match. Even if the pattern is
2775     unanchored, no further attempts to find a match by advancing the starting point
2776     take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2777     finding a match at the current starting point, or not at all. For example:
2778 ph10 210 .sp
2779     a+(*COMMIT)b
2780     .sp
2781 ph10 211 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2782 ph10 512 dynamic anchor, or "I've started, so I must finish." The name of the most
2783     recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2784 ph10 510 match failure.
2785     .P
2786 ph10 512 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2787     unless PCRE's start-of-match optimizations are turned off, as shown in this
2788 ph10 510 \fBpcretest\fP example:
2789 ph10 210 .sp
2790 ph10 836 re> /(*COMMIT)abc/
2791     data> xyzabc
2792 ph10 510 0: abc
2793     xyzabc\eY
2794     No match
2795 ph10 210 .sp
2796 ph10 512 PCRE knows that any match must start with "a", so the optimization skips along
2797 ph10 510 the subject to "a" before running the first match attempt, which succeeds. When
2798     the optimization is disabled by the \eY escape in the second subject, the match
2799 ph10 512 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2800 ph10 510 starting points.
2801 ph10 210 .sp
2802 ph10 510 (*PRUNE) or (*PRUNE:NAME)
2803     .sp
2804 ph10 512 This verb causes the match to fail at the current starting position in the
2805 ph10 510 subject if the rest of the pattern does not match. If the pattern is
2806     unanchored, the normal "bumpalong" advance to the next starting character then
2807     happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2808     reached, or when matching to the right of (*PRUNE), but if there is no match to
2809     the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2810     (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2811     but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2812 ph10 836 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2813     anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2814 ph10 510 .sp
2815 ph10 210 (*SKIP)
2816     .sp
2817 ph10 510 This verb, when given without a name, is like (*PRUNE), except that if the
2818     pattern is unanchored, the "bumpalong" advance is not to the next character,
2819     but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2820     signifies that whatever text was matched leading up to it cannot be part of a
2821     successful match. Consider:
2822 ph10 210 .sp
2823     a+(*SKIP)b
2824     .sp
2825 ph10 211 If the subject is "aaaac...", after the first match attempt fails (starting at
2826 ph10 210 the first character in the string), the starting point skips on to start the
2827 ph10 211 next attempt at "c". Note that a possessive quantifer does not have the same
2828 ph10 456 effect as this example; although it would suppress backtracking during the
2829 ph10 210 first match attempt, the second attempt would start at the second character
2830     instead of skipping on to "c".
2831     .sp
2832 ph10 510 (*SKIP:NAME)
2833 ph10 211 .sp
2834 ph10 512 When (*SKIP) has an associated name, its behaviour is modified. If the
2835     following pattern fails to match, the previous path through the pattern is
2836     searched for the most recent (*MARK) that has the same name. If one is found,
2837     the "bumpalong" advance is to the subject position that corresponds to that
2838     (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2839 ph10 836 matching name is found, the (*SKIP) is ignored.
2840 ph10 510 .sp
2841     (*THEN) or (*THEN:NAME)
2842     .sp
2843 ph10 716 This verb causes a skip to the next innermost alternative if the rest of the
2844     pattern does not match. That is, it cancels pending backtracking, but only
2845     within the current alternative. Its name comes from the observation that it can
2846     be used for a pattern-based if-then-else block:
2847 ph10 210 .sp
2848     ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2849     .sp
2850 ph10 211 If the COND1 pattern matches, FOO is tried (and possibly further items after
2851 ph10 716 the end of the group if FOO succeeds); on failure, the matcher skips to the
2852 ph10 510 second alternative and tries COND2, without backtracking into COND1. The
2853 ph10 836 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2854     If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2855 ph10 551 .P
2856 ph10 716 Note that a subpattern that does not contain a | character is just a part of
2857     the enclosing alternative; it is not a nested alternation with only one
2858     alternative. The effect of (*THEN) extends beyond such a subpattern to the
2859     enclosing alternative. Consider this pattern, where A, B, etc. are complex
2860     pattern fragments that do not contain any | characters at this level:
2861     .sp
2862     A (B(*THEN)C) | D
2863     .sp
2864 ph10 733 If A and B are matched, but there is a failure in C, matching does not
2865 ph10 716 backtrack into A; instead it moves to the next alternative, that is, D.
2866     However, if the subpattern containing (*THEN) is given an alternative, it
2867     behaves differently:
2868     .sp
2869     A (B(*THEN)C | (*FAIL)) | D
2870     .sp
2871     The effect of (*THEN) is now confined to the inner subpattern. After a failure
2872 ph10 733 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2873     because there are no more alternatives to try. In this case, matching does now
2874 ph10 716 backtrack into A.
2875 ph10 551 .P
2876 ph10 733 Note also that a conditional subpattern is not considered as having two
2877     alternatives, because only one is ever used. In other words, the | character in
2878 ph10 716 a conditional subpattern has a different meaning. Ignoring white space,
2879     consider:
2880 ph10 551 .sp
2881 ph10 716 ^.*? (?(?=a) a | b(*THEN)c )
2882     .sp
2883 ph10 733 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2884     it initially matches zero characters. The condition (?=a) then fails, the
2885 ph10 716 character "b" is matched, but "c" is not. At this point, matching does not
2886     backtrack to .*? as might perhaps be expected from the presence of the |
2887     character. The conditional subpattern is part of the single alternative that
2888 ph10 733 comprises the whole pattern, and so the match fails. (If there was a backtrack
2889 ph10 716 into .*?, allowing it to match "b", the match would succeed.)
2890     .P
2891     The verbs just described provide four different "strengths" of control when
2892     subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2893     next alternative. (*PRUNE) comes next, failing the match at the current
2894     starting position, but allowing an advance to the next character (for an
2895     unanchored pattern). (*SKIP) is similar, except that the advance may be more
2896     than one character. (*COMMIT) is the strongest, causing the entire match to
2897     fail.
2898     .P
2899     If more than one such verb is present in a pattern, the "strongest" one wins.
2900     For example, consider this pattern, where A, B, etc. are complex pattern
2901     fragments:
2902     .sp
2903 ph10 551 (A(*COMMIT)B(*THEN)C|D)
2904     .sp
2905     Once A has matched, PCRE is committed to this match, at the current starting
2906     position. If subsequently B matches, but C does not, the normal (*THEN) action
2907 ph10 716 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2908 ph10 551 overrides.
2909 ph10 210 .
2910 ph10 551 .
2911 nigel 93 .SH "SEE ALSO"
2912     .rs
2913     .sp
2914 ph10 461 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2915 ph10 859 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP.
2916 ph10 99 .
2917     .
2918     .SH AUTHOR
2919     .rs
2920     .sp
2921     .nf
2922     Philip Hazel
2923     University Computing Service
2924     Cambridge CB2 3QH, England.
2925     .fi
2926     .
2927     .
2928     .SH REVISION
2929     .rs
2930     .sp
2931     .nf
2932 ph10 1001 Last updated: 08 August 2012
2933 ph10 859 Copyright (c) 1997-2012 University of Cambridge.
2934 ph10 99 .fi


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

ViewVC Help
Powered by ViewVC 1.1.12