/[pcre]/code/trunk/doc/pcre.3
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 43 - (hide annotations) (download)
Sat Feb 24 21:39:21 2007 UTC (7 years, 5 months ago) by nigel
File size: 78779 byte(s)
Load pcre-3.0 into code/trunk.

1 nigel 41 .TH PCRE 3
2     .SH NAME
3     pcre - Perl-compatible regular expressions.
4     .SH SYNOPSIS
5     .B #include <pcre.h>
6     .PP
7     .SM
8     .br
9     .B pcre *pcre_compile(const char *\fIpattern\fR, int \fIoptions\fR,
10     .ti +5n
11     .B const char **\fIerrptr\fR, int *\fIerroffset\fR,
12     .ti +5n
13     .B const unsigned char *\fItableptr\fR);
14     .PP
15     .br
16     .B pcre_extra *pcre_study(const pcre *\fIcode\fR, int \fIoptions\fR,
17     .ti +5n
18     .B const char **\fIerrptr\fR);
19     .PP
20     .br
21     .B int pcre_exec(const pcre *\fIcode\fR, "const pcre_extra *\fIextra\fR,"
22     .ti +5n
23     .B "const char *\fIsubject\fR," int \fIlength\fR, int \fIstartoffset\fR,
24     .ti +5n
25     .B int \fIoptions\fR, int *\fIovector\fR, int \fIovecsize\fR);
26     .PP
27     .br
28     .B int pcre_copy_substring(const char *\fIsubject\fR, int *\fIovector\fR,
29     .ti +5n
30     .B int \fIstringcount\fR, int \fIstringnumber\fR, char *\fIbuffer\fR,
31     .ti +5n
32     .B int \fIbuffersize\fR);
33     .PP
34     .br
35     .B int pcre_get_substring(const char *\fIsubject\fR, int *\fIovector\fR,
36     .ti +5n
37     .B int \fIstringcount\fR, int \fIstringnumber\fR,
38     .ti +5n
39     .B const char **\fIstringptr\fR);
40     .PP
41     .br
42     .B int pcre_get_substring_list(const char *\fIsubject\fR,
43     .ti +5n
44     .B int *\fIovector\fR, int \fIstringcount\fR, "const char ***\fIlistptr\fR);"
45     .PP
46     .br
47     .B const unsigned char *pcre_maketables(void);
48     .PP
49     .br
50 nigel 43 .B int pcre_fullinfo(const pcre *\fIcode\fR, "const pcre_extra *\fIextra\fR,"
51     .ti +5n
52     .B int \fIwhat\fR, void *\fIwhere\fR);
53     .PP
54     .br
55 nigel 41 .B int pcre_info(const pcre *\fIcode\fR, int *\fIoptptr\fR, int
56     .B *\fIfirstcharptr\fR);
57     .PP
58     .br
59     .B char *pcre_version(void);
60     .PP
61     .br
62     .B void *(*pcre_malloc)(size_t);
63     .PP
64     .br
65     .B void (*pcre_free)(void *);
66    
67    
68    
69     .SH DESCRIPTION
70     The PCRE library is a set of functions that implement regular expression
71     pattern matching using the same syntax and semantics as Perl 5, with just a few
72 nigel 43 differences (see below). The current implementation corresponds to Perl 5.005,
73     with some additional features from the Perl development release.
74 nigel 41
75     PCRE has its own native API, which is described in this document. There is also
76 nigel 43 a set of wrapper functions that correspond to the POSIX regular expression API.
77     These are described in the \fBpcreposix\fR documentation.
78 nigel 41
79     The native API function prototypes are defined in the header file \fBpcre.h\fR,
80     and on Unix systems the library itself is called \fBlibpcre.a\fR, so can be
81     accessed by adding \fB-lpcre\fR to the command for linking an application which
82 nigel 43 calls it. The header file defines the macros PCRE_MAJOR and PCRE_MINOR to
83     contain the major and minor release numbers for the library. Applications can
84     use these to include support for different releases.
85 nigel 41
86     The functions \fBpcre_compile()\fR, \fBpcre_study()\fR, and \fBpcre_exec()\fR
87     are used for compiling and matching regular expressions, while
88     \fBpcre_copy_substring()\fR, \fBpcre_get_substring()\fR, and
89     \fBpcre_get_substring_list()\fR are convenience functions for extracting
90     captured substrings from a matched subject string. The function
91     \fBpcre_maketables()\fR is used (optionally) to build a set of character tables
92     in the current locale for passing to \fBpcre_compile()\fR.
93    
94 nigel 43 The function \fBpcre_fullinfo()\fR is used to find out information about a
95     compiled pattern; \fBpcre_info()\fR is an obsolete version which returns only
96     some of the available information, but is retained for backwards compatibility.
97     The function \fBpcre_version()\fR returns a pointer to a string containing the
98     version of PCRE and its date of release.
99 nigel 41
100     The global variables \fBpcre_malloc\fR and \fBpcre_free\fR initially contain
101     the entry points of the standard \fBmalloc()\fR and \fBfree()\fR functions
102     respectively. PCRE calls the memory management functions via these variables,
103     so a calling program can replace them if it wishes to intercept the calls. This
104     should be done before calling any PCRE functions.
105    
106    
107     .SH MULTI-THREADING
108     The PCRE functions can be used in multi-threading applications, with the
109     proviso that the memory management functions pointed to by \fBpcre_malloc\fR
110     and \fBpcre_free\fR are shared by all threads.
111    
112     The compiled form of a regular expression is not altered during matching, so
113     the same compiled pattern can safely be used by several threads at once.
114    
115    
116     .SH COMPILING A PATTERN
117     The function \fBpcre_compile()\fR is called to compile a pattern into an
118     internal form. The pattern is a C string terminated by a binary zero, and
119     is passed in the argument \fIpattern\fR. A pointer to a single block of memory
120     that is obtained via \fBpcre_malloc\fR is returned. This contains the
121     compiled code and related data. The \fBpcre\fR type is defined for this for
122     convenience, but in fact \fBpcre\fR is just a typedef for \fBvoid\fR, since the
123     contents of the block are not externally defined. It is up to the caller to
124     free the memory when it is no longer required.
125     .PP
126     The size of a compiled pattern is roughly proportional to the length of the
127     pattern string, except that each character class (other than those containing
128     just a single character, negated or not) requires 33 bytes, and repeat
129     quantifiers with a minimum greater than one or a bounded maximum cause the
130     relevant portions of the compiled pattern to be replicated.
131     .PP
132     The \fIoptions\fR argument contains independent bits that affect the
133     compilation. It should be zero if no options are required. Some of the options,
134     in particular, those that are compatible with Perl, can also be set and unset
135     from within the pattern (see the detailed description of regular expressions
136     below). For these options, the contents of the \fIoptions\fR argument specifies
137     their initial settings at the start of compilation and execution. The
138     PCRE_ANCHORED option can be set at the time of matching as well as at compile
139     time.
140     .PP
141     If \fIerrptr\fR is NULL, \fBpcre_compile()\fR returns NULL immediately.
142     Otherwise, if compilation of a pattern fails, \fBpcre_compile()\fR returns
143     NULL, and sets the variable pointed to by \fIerrptr\fR to point to a textual
144     error message. The offset from the start of the pattern to the character where
145     the error was discovered is placed in the variable pointed to by
146     \fIerroffset\fR, which must not be NULL. If it is, an immediate error is given.
147     .PP
148     If the final argument, \fItableptr\fR, is NULL, PCRE uses a default set of
149     character tables which are built when it is compiled, using the default C
150     locale. Otherwise, \fItableptr\fR must be the result of a call to
151     \fBpcre_maketables()\fR. See the section on locale support below.
152     .PP
153     The following option bits are defined in the header file:
154    
155     PCRE_ANCHORED
156    
157     If this bit is set, the pattern is forced to be "anchored", that is, it is
158     constrained to match only at the start of the string which is being searched
159     (the "subject string"). This effect can also be achieved by appropriate
160     constructs in the pattern itself, which is the only way to do it in Perl.
161    
162     PCRE_CASELESS
163    
164     If this bit is set, letters in the pattern match both upper and lower case
165     letters. It is equivalent to Perl's /i option.
166    
167     PCRE_DOLLAR_ENDONLY
168    
169     If this bit is set, a dollar metacharacter in the pattern matches only at the
170     end of the subject string. Without this option, a dollar also matches
171     immediately before the final character if it is a newline (but not before any
172     other newlines). The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is
173     set. There is no equivalent to this option in Perl.
174    
175     PCRE_DOTALL
176    
177     If this bit is set, a dot metacharater in the pattern matches all characters,
178     including newlines. Without it, newlines are excluded. This option is
179     equivalent to Perl's /s option. A negative class such as [^a] always matches a
180     newline character, independent of the setting of this option.
181    
182     PCRE_EXTENDED
183    
184     If this bit is set, whitespace data characters in the pattern are totally
185     ignored except when escaped or inside a character class, and characters between
186     an unescaped # outside a character class and the next newline character,
187     inclusive, are also ignored. This is equivalent to Perl's /x option, and makes
188     it possible to include comments inside complicated patterns. Note, however,
189     that this applies only to data characters. Whitespace characters may never
190     appear within special character sequences in a pattern, for example within the
191     sequence (?( which introduces a conditional subpattern.
192    
193     PCRE_EXTRA
194    
195 nigel 43 This option was invented in order to turn on additional functionality of PCRE
196     that is incompatible with Perl, but it is currently of very little use. When
197     set, any backslash in a pattern that is followed by a letter that has no
198 nigel 41 special meaning causes an error, thus reserving these combinations for future
199     expansion. By default, as in Perl, a backslash followed by a letter with no
200     special meaning is treated as a literal. There are at present no other features
201 nigel 43 controlled by this option. It can also be set by a (?X) option setting within a
202     pattern.
203 nigel 41
204     PCRE_MULTILINE
205    
206     By default, PCRE treats the subject string as consisting of a single "line" of
207     characters (even if it actually contains several newlines). The "start of line"
208     metacharacter (^) matches only at the start of the string, while the "end of
209     line" metacharacter ($) matches only at the end of the string, or before a
210     terminating newline (unless PCRE_DOLLAR_ENDONLY is set). This is the same as
211     Perl.
212    
213     When PCRE_MULTILINE it is set, the "start of line" and "end of line" constructs
214     match immediately following or immediately before any newline in the subject
215     string, respectively, as well as at the very start and end. This is equivalent
216     to Perl's /m option. If there are no "\\n" characters in a subject string, or
217     no occurrences of ^ or $ in a pattern, setting PCRE_MULTILINE has no
218     effect.
219    
220     PCRE_UNGREEDY
221    
222     This option inverts the "greediness" of the quantifiers so that they are not
223     greedy by default, but become greedy if followed by "?". It is not compatible
224     with Perl. It can also be set by a (?U) option setting within the pattern.
225    
226    
227     .SH STUDYING A PATTERN
228     When a pattern is going to be used several times, it is worth spending more
229     time analyzing it in order to speed up the time taken for matching. The
230     function \fBpcre_study()\fR takes a pointer to a compiled pattern as its first
231     argument, and returns a pointer to a \fBpcre_extra\fR block (another \fBvoid\fR
232     typedef) containing additional information about the pattern; this can be
233     passed to \fBpcre_exec()\fR. If no additional information is available, NULL
234     is returned.
235    
236     The second argument contains option bits. At present, no options are defined
237     for \fBpcre_study()\fR, and this argument should always be zero.
238    
239     The third argument for \fBpcre_study()\fR is a pointer to an error message. If
240     studying succeeds (even if no data is returned), the variable it points to is
241     set to NULL. Otherwise it points to a textual error message.
242    
243     At present, studying a pattern is useful only for non-anchored patterns that do
244     not have a single fixed starting character. A bitmap of possible starting
245     characters is created.
246    
247    
248     .SH LOCALE SUPPORT
249     PCRE handles caseless matching, and determines whether characters are letters,
250     digits, or whatever, by reference to a set of tables. The library contains a
251     default set of tables which is created in the default C locale when PCRE is
252     compiled. This is used when the final argument of \fBpcre_compile()\fR is NULL,
253     and is sufficient for many applications.
254    
255     An alternative set of tables can, however, be supplied. Such tables are built
256     by calling the \fBpcre_maketables()\fR function, which has no arguments, in the
257     relevant locale. The result can then be passed to \fBpcre_compile()\fR as often
258     as necessary. For example, to build and use tables that are appropriate for the
259     French locale (where accented characters with codes greater than 128 are
260     treated as letters), the following code could be used:
261    
262     setlocale(LC_CTYPE, "fr");
263     tables = pcre_maketables();
264     re = pcre_compile(..., tables);
265    
266     The tables are built in memory that is obtained via \fBpcre_malloc\fR. The
267     pointer that is passed to \fBpcre_compile\fR is saved with the compiled
268     pattern, and the same tables are used via this pointer by \fBpcre_study()\fR
269     and \fBpcre_exec()\fR. Thus for any single pattern, compilation, studying and
270     matching all happen in the same locale, but different patterns can be compiled
271     in different locales. It is the caller's responsibility to ensure that the
272     memory containing the tables remains available for as long as it is needed.
273    
274    
275     .SH INFORMATION ABOUT A PATTERN
276 nigel 43 The \fBpcre_fullinfo()\fR function returns information about a compiled
277     pattern. It replaces the obsolete \fBpcre_info()\fR function, which is
278     nevertheless retained for backwards compability (and is documented below).
279 nigel 41
280 nigel 43 The first argument for \fBpcre_fullinfo()\fR is a pointer to the compiled
281     pattern. The second argument is the result of \fBpcre_study()\fR, or NULL if
282     the pattern was not studied. The third argument specifies which piece of
283     information is required, while the fourth argument is a pointer to a variable
284     to receive the data. The yield of the function is zero for success, or one of
285     the following negative numbers:
286    
287 nigel 41 PCRE_ERROR_NULL the argument \fIcode\fR was NULL
288 nigel 43 the argument \fIwhere\fR was NULL
289 nigel 41 PCRE_ERROR_BADMAGIC the "magic number" was not found
290 nigel 43 PCRE_ERROR_BADOPTION the value of \fIwhat\fR was invalid
291 nigel 41
292 nigel 43 The possible values for the third argument are defined in \fBpcre.h\fR, and are
293     as follows:
294    
295     PCRE_INFO_OPTIONS
296    
297     Return a copy of the options with which the pattern was compiled. The fourth
298     argument should point to au \fBunsigned long int\fR variable. These option bits
299 nigel 41 are those specified in the call to \fBpcre_compile()\fR, modified by any
300     top-level option settings within the pattern itself, and with the PCRE_ANCHORED
301 nigel 43 bit forcibly set if the form of the pattern implies that it can match only at
302     the start of a subject string.
303 nigel 41
304 nigel 43 PCRE_INFO_SIZE
305 nigel 41
306 nigel 43 Return the size of the compiled pattern, that is, the value that was passed as
307     the argument to \fBpcre_malloc()\fR when PCRE was getting memory in which to
308     place the compiled data. The fourth argument should point to a \fBsize_t\fR
309     variable.
310    
311     PCRE_INFO_CAPTURECOUNT
312    
313     Return the number of capturing subpatterns in the pattern. The fourth argument
314     should point to an \fbint\fR variable.
315    
316     PCRE_INFO_BACKREFMAX
317    
318     Return the number of the highest back reference in the pattern. The fourth
319     argument should point to an \fBint\fR variable. Zero is returned if there are
320     no back references.
321    
322     PCRE_INFO_FIRSTCHAR
323    
324     Return information about the first character of any matched string, for a
325     non-anchored pattern. If there is a fixed first character, e.g. from a pattern
326     such as (cat|cow|coyote), then it is returned in the integer pointed to by
327     \fIwhere\fR. Otherwise, if either
328    
329 nigel 41 (a) the pattern was compiled with the PCRE_MULTILINE option, and every branch
330     starts with "^", or
331    
332     (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not set
333     (if it were set, the pattern would be anchored),
334    
335     then -1 is returned, indicating that the pattern matches only at the
336     start of a subject string or after any "\\n" within the string. Otherwise -2 is
337 nigel 43 returned. For anchored patterns, -2 is returned.
338 nigel 41
339 nigel 43 PCRE_INFO_FIRSTTABLE
340 nigel 41
341 nigel 43 If the pattern was studied, and this resulted in the construction of a 256-bit
342     table indicating a fixed set of characters for the first character in any
343     matching string, a pointer to the table is returned. Otherwise NULL is
344     returned. The fourth argument should point to an \fBunsigned char *\fR
345     variable.
346    
347     PCRE_INFO_LASTLITERAL
348    
349     For a non-anchored pattern, return the value of the rightmost literal character
350     which must exist in any matched string, other than at its start. The fourth
351     argument should point to an \fBint\fR variable. If there is no such character,
352     or if the pattern is anchored, -1 is returned. For example, for the pattern
353     /a\\d+z\\d+/ the returned value is 'z'.
354    
355     The \fBpcre_info()\fR function is now obsolete because its interface is too
356     restrictive to return all the available data about a compiled pattern. New
357     programs should use \fBpcre_fullinfo()\fR instead. The yield of
358     \fBpcre_info()\fR is the number of capturing subpatterns, or one of the
359     following negative numbers:
360    
361     PCRE_ERROR_NULL the argument \fIcode\fR was NULL
362     PCRE_ERROR_BADMAGIC the "magic number" was not found
363    
364     If the \fIoptptr\fR argument is not NULL, a copy of the options with which the
365     pattern was compiled is placed in the integer it points to (see
366     PCRE_INFO_OPTIONS above).
367    
368     If the pattern is not anchored and the \fIfirstcharptr\fR argument is not NULL,
369     it is used to pass back information about the first character of any matched
370     string (see PCRE_INFO_FIRSTCHAR above).
371    
372    
373 nigel 41 .SH MATCHING A PATTERN
374     The function \fBpcre_exec()\fR is called to match a subject string against a
375     pre-compiled pattern, which is passed in the \fIcode\fR argument. If the
376     pattern has been studied, the result of the study should be passed in the
377     \fIextra\fR argument. Otherwise this must be NULL.
378    
379     The PCRE_ANCHORED option can be passed in the \fIoptions\fR argument, whose
380     unused bits must be zero. However, if a pattern was compiled with
381     PCRE_ANCHORED, or turned out to be anchored by virtue of its contents, it
382     cannot be made unachored at matching time.
383    
384     There are also three further options that can be set only at matching time:
385    
386     PCRE_NOTBOL
387    
388     The first character of the string is not the beginning of a line, so the
389     circumflex metacharacter should not match before it. Setting this without
390     PCRE_MULTILINE (at compile time) causes circumflex never to match.
391    
392     PCRE_NOTEOL
393    
394     The end of the string is not the end of a line, so the dollar metacharacter
395     should not match it nor (except in multiline mode) a newline immediately before
396     it. Setting this without PCRE_MULTILINE (at compile time) causes dollar never
397     to match.
398    
399     PCRE_NOTEMPTY
400    
401     An empty string is not considered to be a valid match if this option is set. If
402     there are alternatives in the pattern, they are tried. If all the alternatives
403     match the empty string, the entire match fails. For example, if the pattern
404    
405     a?b?
406    
407     is applied to a string not beginning with "a" or "b", it matches the empty
408     string at the start of the subject. With PCRE_NOTEMPTY set, this match is not
409     valid, so PCRE searches further into the string for occurrences of "a" or "b".
410    
411     Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a special case
412     of a pattern match of the empty string within its \fBsplit()\fR function, and
413     when using the /g modifier. It is possible to emulate Perl's behaviour after
414     matching a null string by first trying the match again at the same offset with
415     PCRE_NOTEMPTY set, and then if that fails by advancing the starting offset (see
416     below) and trying an ordinary match again.
417    
418     The subject string is passed as a pointer in \fIsubject\fR, a length in
419     \fIlength\fR, and a starting offset in \fIstartoffset\fR. Unlike the pattern
420     string, it may contain binary zero characters. When the starting offset is
421     zero, the search for a match starts at the beginning of the subject, and this
422     is by far the most common case.
423    
424     A non-zero starting offset is useful when searching for another match in the
425     same subject by calling \fBpcre_exec()\fR again after a previous success.
426     Setting \fIstartoffset\fR differs from just passing over a shortened string and
427     setting PCRE_NOTBOL in the case of a pattern that begins with any kind of
428     lookbehind. For example, consider the pattern
429    
430     \\Biss\\B
431    
432     which finds occurrences of "iss" in the middle of words. (\\B matches only if
433     the current position in the subject is not a word boundary.) When applied to
434     the string "Mississipi" the first call to \fBpcre_exec()\fR finds the first
435     occurrence. If \fBpcre_exec()\fR is called again with just the remainder of the
436     subject, namely "issipi", it does not match, because \\B is always false at the
437     start of the subject, which is deemed to be a word boundary. However, if
438     \fBpcre_exec()\fR is passed the entire string again, but with \fIstartoffset\fR
439     set to 4, it finds the second occurrence of "iss" because it is able to look
440     behind the starting point to discover that it is preceded by a letter.
441    
442     If a non-zero starting offset is passed when the pattern is anchored, one
443     attempt to match at the given offset is tried. This can only succeed if the
444     pattern does not require the match to be at the start of the subject.
445    
446     In general, a pattern matches a certain portion of the subject, and in
447     addition, further substrings from the subject may be picked out by parts of the
448     pattern. Following the usage in Jeffrey Friedl's book, this is called
449     "capturing" in what follows, and the phrase "capturing subpattern" is used for
450     a fragment of a pattern that picks out a substring. PCRE supports several other
451     kinds of parenthesized subpattern that do not cause substrings to be captured.
452    
453     Captured substrings are returned to the caller via a vector of integer offsets
454     whose address is passed in \fIovector\fR. The number of elements in the vector
455     is passed in \fIovecsize\fR. The first two-thirds of the vector is used to pass
456     back captured substrings, each substring using a pair of integers. The
457     remaining third of the vector is used as workspace by \fBpcre_exec()\fR while
458     matching capturing subpatterns, and is not available for passing back
459     information. The length passed in \fIovecsize\fR should always be a multiple of
460     three. If it is not, it is rounded down.
461    
462     When a match has been successful, information about captured substrings is
463     returned in pairs of integers, starting at the beginning of \fIovector\fR, and
464     continuing up to two-thirds of its length at the most. The first element of a
465     pair is set to the offset of the first character in a substring, and the second
466     is set to the offset of the first character after the end of a substring. The
467     first pair, \fIovector[0]\fR and \fIovector[1]\fR, identify the portion of the
468     subject string matched by the entire pattern. The next pair is used for the
469     first capturing subpattern, and so on. The value returned by \fBpcre_exec()\fR
470     is the number of pairs that have been set. If there are no capturing
471     subpatterns, the return value from a successful match is 1, indicating that
472     just the first pair of offsets has been set.
473    
474     Some convenience functions are provided for extracting the captured substrings
475     as separate strings. These are described in the following section.
476    
477     It is possible for an capturing subpattern number \fIn+1\fR to match some
478     part of the subject when subpattern \fIn\fR has not been used at all. For
479     example, if the string "abc" is matched against the pattern (a|(z))(bc)
480     subpatterns 1 and 3 are matched, but 2 is not. When this happens, both offset
481     values corresponding to the unused subpattern are set to -1.
482    
483     If a capturing subpattern is matched repeatedly, it is the last portion of the
484     string that it matched that gets returned.
485    
486     If the vector is too small to hold all the captured substrings, it is used as
487     far as possible (up to two-thirds of its length), and the function returns a
488     value of zero. In particular, if the substring offsets are not of interest,
489     \fBpcre_exec()\fR may be called with \fIovector\fR passed as NULL and
490     \fIovecsize\fR as zero. However, if the pattern contains back references and
491     the \fIovector\fR isn't big enough to remember the related substrings, PCRE has
492     to get additional memory for use during matching. Thus it is usually advisable
493     to supply an \fIovector\fR.
494    
495     Note that \fBpcre_info()\fR can be used to find out how many capturing
496     subpatterns there are in a compiled pattern. The smallest size for
497     \fIovector\fR that will allow for \fIn\fR captured substrings in addition to
498     the offsets of the substring matched by the whole pattern is (\fIn\fR+1)*3.
499    
500     If \fBpcre_exec()\fR fails, it returns a negative number. The following are
501     defined in the header file:
502    
503     PCRE_ERROR_NOMATCH (-1)
504    
505     The subject string did not match the pattern.
506    
507     PCRE_ERROR_NULL (-2)
508    
509     Either \fIcode\fR or \fIsubject\fR was passed as NULL, or \fIovector\fR was
510     NULL and \fIovecsize\fR was not zero.
511    
512     PCRE_ERROR_BADOPTION (-3)
513    
514     An unrecognized bit was set in the \fIoptions\fR argument.
515    
516     PCRE_ERROR_BADMAGIC (-4)
517    
518     PCRE stores a 4-byte "magic number" at the start of the compiled code, to catch
519     the case when it is passed a junk pointer. This is the error it gives when the
520     magic number isn't present.
521    
522     PCRE_ERROR_UNKNOWN_NODE (-5)
523    
524     While running the pattern match, an unknown item was encountered in the
525     compiled pattern. This error could be caused by a bug in PCRE or by overwriting
526     of the compiled pattern.
527    
528     PCRE_ERROR_NOMEMORY (-6)
529    
530     If a pattern contains back references, but the \fIovector\fR that is passed to
531     \fBpcre_exec()\fR is not big enough to remember the referenced substrings, PCRE
532     gets a block of memory at the start of matching to use for this purpose. If the
533     call via \fBpcre_malloc()\fR fails, this error is given. The memory is freed at
534     the end of matching.
535    
536    
537     .SH EXTRACTING CAPTURED SUBSTRINGS
538     Captured substrings can be accessed directly by using the offsets returned by
539     \fBpcre_exec()\fR in \fIovector\fR. For convenience, the functions
540     \fBpcre_copy_substring()\fR, \fBpcre_get_substring()\fR, and
541     \fBpcre_get_substring_list()\fR are provided for extracting captured substrings
542     as new, separate, zero-terminated strings. A substring that contains a binary
543     zero is correctly extracted and has a further zero added on the end, but the
544     result does not, of course, function as a C string.
545    
546     The first three arguments are the same for all three functions: \fIsubject\fR
547     is the subject string which has just been successfully matched, \fIovector\fR
548     is a pointer to the vector of integer offsets that was passed to
549     \fBpcre_exec()\fR, and \fIstringcount\fR is the number of substrings that
550     were captured by the match, including the substring that matched the entire
551     regular expression. This is the value returned by \fBpcre_exec\fR if it
552     is greater than zero. If \fBpcre_exec()\fR returned zero, indicating that it
553     ran out of space in \fIovector\fR, then the value passed as
554     \fIstringcount\fR should be the size of the vector divided by three.
555    
556     The functions \fBpcre_copy_substring()\fR and \fBpcre_get_substring()\fR
557     extract a single substring, whose number is given as \fIstringnumber\fR. A
558     value of zero extracts the substring that matched the entire pattern, while
559     higher values extract the captured substrings. For \fBpcre_copy_substring()\fR,
560     the string is placed in \fIbuffer\fR, whose length is given by
561     \fIbuffersize\fR, while for \fBpcre_get_substring()\fR a new block of store is
562     obtained via \fBpcre_malloc\fR, and its address is returned via
563     \fIstringptr\fR. The yield of the function is the length of the string, not
564     including the terminating zero, or one of
565    
566     PCRE_ERROR_NOMEMORY (-6)
567    
568     The buffer was too small for \fBpcre_copy_substring()\fR, or the attempt to get
569     memory failed for \fBpcre_get_substring()\fR.
570    
571     PCRE_ERROR_NOSUBSTRING (-7)
572    
573     There is no substring whose number is \fIstringnumber\fR.
574    
575     The \fBpcre_get_substring_list()\fR function extracts all available substrings
576     and builds a list of pointers to them. All this is done in a single block of
577     memory which is obtained via \fBpcre_malloc\fR. The address of the memory block
578     is returned via \fIlistptr\fR, which is also the start of the list of string
579     pointers. The end of the list is marked by a NULL pointer. The yield of the
580     function is zero if all went well, or
581    
582     PCRE_ERROR_NOMEMORY (-6)
583    
584     if the attempt to get the memory block failed.
585    
586     When any of these functions encounter a substring that is unset, which can
587     happen when capturing subpattern number \fIn+1\fR matches some part of the
588     subject, but subpattern \fIn\fR has not been used at all, they return an empty
589     string. This can be distinguished from a genuine zero-length substring by
590     inspecting the appropriate offset in \fIovector\fR, which is negative for unset
591     substrings.
592    
593    
594    
595     .SH LIMITATIONS
596     There are some size limitations in PCRE but it is hoped that they will never in
597     practice be relevant.
598     The maximum length of a compiled pattern is 65539 (sic) bytes.
599     All values in repeating quantifiers must be less than 65536.
600     The maximum number of capturing subpatterns is 99.
601     The maximum number of all parenthesized subpatterns, including capturing
602     subpatterns, assertions, and other types of subpattern, is 200.
603    
604     The maximum length of a subject string is the largest positive number that an
605     integer variable can hold. However, PCRE uses recursion to handle subpatterns
606     and indefinite repetition. This means that the available stack space may limit
607     the size of a subject string that can be processed by certain patterns.
608    
609    
610     .SH DIFFERENCES FROM PERL
611     The differences described here are with respect to Perl 5.005.
612    
613     1. By default, a whitespace character is any character that the C library
614     function \fBisspace()\fR recognizes, though it is possible to compile PCRE with
615     alternative character type tables. Normally \fBisspace()\fR matches space,
616     formfeed, newline, carriage return, horizontal tab, and vertical tab. Perl 5
617     no longer includes vertical tab in its set of whitespace characters. The \\v
618     escape that was in the Perl documentation for a long time was never in fact
619     recognized. However, the character itself was treated as whitespace at least
620     up to 5.002. In 5.004 and 5.005 it does not match \\s.
621    
622     2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl permits
623     them, but they do not mean what you might think. For example, (?!a){3} does
624     not assert that the next three characters are not "a". It just asserts that the
625     next character is not "a" three times.
626    
627     3. Capturing subpatterns that occur inside negative lookahead assertions are
628     counted, but their entries in the offsets vector are never set. Perl sets its
629     numerical variables from any such patterns that are matched before the
630     assertion fails to match something (thereby succeeding), but only if the
631     negative lookahead assertion contains just one branch.
632    
633     4. Though binary zero characters are supported in the subject string, they are
634     not allowed in a pattern string because it is passed as a normal C string,
635     terminated by zero. The escape sequence "\\0" can be used in the pattern to
636     represent a binary zero.
637    
638     5. The following Perl escape sequences are not supported: \\l, \\u, \\L, \\U,
639     \\E, \\Q. In fact these are implemented by Perl's general string-handling and
640     are not part of its pattern matching engine.
641    
642     6. The Perl \\G assertion is not supported as it is not relevant to single
643     pattern matches.
644    
645 nigel 43 7. Fairly obviously, PCRE does not support the (?{code}) and (?p{code})
646     constructions. However, there is some experimental support for recursive
647     patterns using the non-Perl item (?R).
648 nigel 41
649     8. There are at the time of writing some oddities in Perl 5.005_02 concerned
650     with the settings of captured strings when part of a pattern is repeated. For
651     example, matching "aba" against the pattern /^(a(b)?)+$/ sets $2 to the value
652     "b", but matching "aabbaa" against /^(aa(bb)?)+$/ leaves $2 unset. However, if
653     the pattern is changed to /^(aa(b(b))?)+$/ then $2 (and $3) get set.
654    
655     In Perl 5.004 $2 is set in both cases, and that is also true of PCRE. If in the
656     future Perl changes to a consistent state that is different, PCRE may change to
657     follow.
658    
659     9. Another as yet unresolved discrepancy is that in Perl 5.005_02 the pattern
660     /^(a)?(?(1)a|b)+$/ matches the string "a", whereas in PCRE it does not.
661     However, in both Perl and PCRE /^(a)?a/ matched against "a" leaves $1 unset.
662    
663     10. PCRE provides some extensions to the Perl regular expression facilities:
664    
665     (a) Although lookbehind assertions must match fixed length strings, each
666     alternative branch of a lookbehind assertion can match a different length of
667     string. Perl 5.005 requires them all to have the same length.
668    
669     (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $ meta-
670     character matches only at the very end of the string.
671    
672     (c) If PCRE_EXTRA is set, a backslash followed by a letter with no special
673     meaning is faulted.
674    
675     (d) If PCRE_UNGREEDY is set, the greediness of the repetition quantifiers is
676     inverted, that is, by default they are not greedy, but if followed by a
677     question mark they are.
678    
679     (e) PCRE_ANCHORED can be used to force a pattern to be tried only at the start
680     of the subject.
681    
682     (f) The PCRE_NOTBOL, PCRE_NOTEOL, and PCRE_NOTEMPTY options for
683     \fBpcre_exec()\fR have no Perl equivalents.
684    
685 nigel 43 (g) The (?R) construct allows for recursive pattern matching (Perl 5.6 can do
686     this using the (?p{code}) construct, which PCRE cannot of course support.)
687 nigel 41
688 nigel 43
689 nigel 41 .SH REGULAR EXPRESSION DETAILS
690     The syntax and semantics of the regular expressions supported by PCRE are
691     described below. Regular expressions are also described in the Perl
692     documentation and in a number of other books, some of which have copious
693     examples. Jeffrey Friedl's "Mastering Regular Expressions", published by
694 nigel 43 O'Reilly (ISBN 1-56592-257), covers them in great detail. The description
695 nigel 41 here is intended as reference documentation.
696    
697     A regular expression is a pattern that is matched against a subject string from
698     left to right. Most characters stand for themselves in a pattern, and match the
699     corresponding characters in the subject. As a trivial example, the pattern
700    
701     The quick brown fox
702    
703     matches a portion of a subject string that is identical to itself. The power of
704     regular expressions comes from the ability to include alternatives and
705     repetitions in the pattern. These are encoded in the pattern by the use of
706     \fImeta-characters\fR, which do not stand for themselves but instead are
707     interpreted in some special way.
708    
709     There are two different sets of meta-characters: those that are recognized
710     anywhere in the pattern except within square brackets, and those that are
711     recognized in square brackets. Outside square brackets, the meta-characters are
712     as follows:
713    
714     \\ general escape character with several uses
715     ^ assert start of subject (or line, in multiline mode)
716     $ assert end of subject (or line, in multiline mode)
717     . match any character except newline (by default)
718     [ start character class definition
719     | start of alternative branch
720     ( start subpattern
721     ) end subpattern
722     ? extends the meaning of (
723     also 0 or 1 quantifier
724     also quantifier minimizer
725     * 0 or more quantifier
726     + 1 or more quantifier
727     { start min/max quantifier
728    
729     Part of a pattern that is in square brackets is called a "character class". In
730     a character class the only meta-characters are:
731    
732     \\ general escape character
733     ^ negate the class, but only if the first character
734     - indicates character range
735     ] terminates the character class
736    
737     The following sections describe the use of each of the meta-characters.
738    
739    
740     .SH BACKSLASH
741     The backslash character has several uses. Firstly, if it is followed by a
742     non-alphameric character, it takes away any special meaning that character may
743     have. This use of backslash as an escape character applies both inside and
744     outside character classes.
745    
746     For example, if you want to match a "*" character, you write "\\*" in the
747     pattern. This applies whether or not the following character would otherwise be
748     interpreted as a meta-character, so it is always safe to precede a
749     non-alphameric with "\\" to specify that it stands for itself. In particular,
750     if you want to match a backslash, you write "\\\\".
751    
752     If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
753     pattern (other than in a character class) and characters between a "#" outside
754     a character class and the next newline character are ignored. An escaping
755     backslash can be used to include a whitespace or "#" character as part of the
756     pattern.
757    
758     A second use of backslash provides a way of encoding non-printing characters
759     in patterns in a visible manner. There is no restriction on the appearance of
760     non-printing characters, apart from the binary zero that terminates a pattern,
761     but when a pattern is being prepared by text editing, it is usually easier to
762     use one of the following escape sequences than the binary character it
763     represents:
764    
765     \\a alarm, that is, the BEL character (hex 07)
766     \\cx "control-x", where x is any character
767     \\e escape (hex 1B)
768     \\f formfeed (hex 0C)
769     \\n newline (hex 0A)
770     \\r carriage return (hex 0D)
771     \\t tab (hex 09)
772     \\xhh character with hex code hh
773     \\ddd character with octal code ddd, or backreference
774    
775     The precise effect of "\\cx" is as follows: if "x" is a lower case letter, it
776     is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
777     Thus "\\cz" becomes hex 1A, but "\\c{" becomes hex 3B, while "\\c;" becomes hex
778     7B.
779    
780     After "\\x", up to two hexadecimal digits are read (letters can be in upper or
781     lower case).
782    
783     After "\\0" up to two further octal digits are read. In both cases, if there
784     are fewer than two digits, just those that are present are used. Thus the
785     sequence "\\0\\x\\07" specifies two binary zeros followed by a BEL character.
786     Make sure you supply two digits after the initial zero if the character that
787     follows is itself an octal digit.
788    
789     The handling of a backslash followed by a digit other than 0 is complicated.
790     Outside a character class, PCRE reads it and any following digits as a decimal
791     number. If the number is less than 10, or if there have been at least that many
792     previous capturing left parentheses in the expression, the entire sequence is
793     taken as a \fIback reference\fR. A description of how this works is given
794     later, following the discussion of parenthesized subpatterns.
795    
796     Inside a character class, or if the decimal number is greater than 9 and there
797     have not been that many capturing subpatterns, PCRE re-reads up to three octal
798     digits following the backslash, and generates a single byte from the least
799     significant 8 bits of the value. Any subsequent digits stand for themselves.
800     For example:
801    
802     \\040 is another way of writing a space
803     \\40 is the same, provided there are fewer than 40
804     previous capturing subpatterns
805     \\7 is always a back reference
806     \\11 might be a back reference, or another way of
807     writing a tab
808     \\011 is always a tab
809     \\0113 is a tab followed by the character "3"
810     \\113 is the character with octal code 113 (since there
811     can be no more than 99 back references)
812     \\377 is a byte consisting entirely of 1 bits
813     \\81 is either a back reference, or a binary zero
814     followed by the two characters "8" and "1"
815    
816     Note that octal values of 100 or greater must not be introduced by a leading
817     zero, because no more than three octal digits are ever read.
818    
819     All the sequences that define a single byte value can be used both inside and
820     outside character classes. In addition, inside a character class, the sequence
821     "\\b" is interpreted as the backspace character (hex 08). Outside a character
822     class it has a different meaning (see below).
823    
824     The third use of backslash is for specifying generic character types:
825    
826     \\d any decimal digit
827     \\D any character that is not a decimal digit
828     \\s any whitespace character
829     \\S any character that is not a whitespace character
830     \\w any "word" character
831     \\W any "non-word" character
832    
833     Each pair of escape sequences partitions the complete set of characters into
834     two disjoint sets. Any given character matches one, and only one, of each pair.
835    
836     A "word" character is any letter or digit or the underscore character, that is,
837     any character which can be part of a Perl "word". The definition of letters and
838     digits is controlled by PCRE's character tables, and may vary if locale-
839     specific matching is taking place (see "Locale support" above). For example, in
840     the "fr" (French) locale, some character codes greater than 128 are used for
841     accented letters, and these are matched by \\w.
842    
843     These character type sequences can appear both inside and outside character
844     classes. They each match one character of the appropriate type. If the current
845     matching point is at the end of the subject string, all of them fail, since
846     there is no character to match.
847    
848     The fourth use of backslash is for certain simple assertions. An assertion
849     specifies a condition that has to be met at a particular point in a match,
850     without consuming any characters from the subject string. The use of
851     subpatterns for more complicated assertions is described below. The backslashed
852     assertions are
853    
854     \\b word boundary
855     \\B not a word boundary
856     \\A start of subject (independent of multiline mode)
857     \\Z end of subject or newline at end (independent of multiline mode)
858     \\z end of subject (independent of multiline mode)
859    
860     These assertions may not appear in character classes (but note that "\\b" has a
861     different meaning, namely the backspace character, inside a character class).
862    
863     A word boundary is a position in the subject string where the current character
864     and the previous character do not both match \\w or \\W (i.e. one matches
865     \\w and the other matches \\W), or the start or end of the string if the
866     first or last character matches \\w, respectively.
867    
868     The \\A, \\Z, and \\z assertions differ from the traditional circumflex and
869     dollar (described below) in that they only ever match at the very start and end
870     of the subject string, whatever options are set. They are not affected by the
871     PCRE_NOTBOL or PCRE_NOTEOL options. If the \fIstartoffset\fR argument of
872     \fBpcre_exec()\fR is non-zero, \\A can never match. The difference between \\Z
873     and \\z is that \\Z matches before a newline that is the last character of the
874     string as well as at the end of the string, whereas \\z matches only at the
875     end.
876    
877    
878     .SH CIRCUMFLEX AND DOLLAR
879     Outside a character class, in the default matching mode, the circumflex
880     character is an assertion which is true only if the current matching point is
881     at the start of the subject string. If the \fIstartoffset\fR argument of
882     \fBpcre_exec()\fR is non-zero, circumflex can never match. Inside a character
883     class, circumflex has an entirely different meaning (see below).
884    
885     Circumflex need not be the first character of the pattern if a number of
886     alternatives are involved, but it should be the first thing in each alternative
887     in which it appears if the pattern is ever to match that branch. If all
888     possible alternatives start with a circumflex, that is, if the pattern is
889     constrained to match only at the start of the subject, it is said to be an
890     "anchored" pattern. (There are also other constructs that can cause a pattern
891     to be anchored.)
892    
893     A dollar character is an assertion which is true only if the current matching
894     point is at the end of the subject string, or immediately before a newline
895     character that is the last character in the string (by default). Dollar need
896     not be the last character of the pattern if a number of alternatives are
897     involved, but it should be the last item in any branch in which it appears.
898     Dollar has no special meaning in a character class.
899    
900     The meaning of dollar can be changed so that it matches only at the very end of
901     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile or matching
902     time. This does not affect the \\Z assertion.
903    
904     The meanings of the circumflex and dollar characters are changed if the
905     PCRE_MULTILINE option is set. When this is the case, they match immediately
906     after and immediately before an internal "\\n" character, respectively, in
907     addition to matching at the start and end of the subject string. For example,
908     the pattern /^abc$/ matches the subject string "def\\nabc" in multiline mode,
909     but not otherwise. Consequently, patterns that are anchored in single line mode
910     because all branches start with "^" are not anchored in multiline mode, and a
911     match for circumflex is possible when the \fIstartoffset\fR argument of
912     \fBpcre_exec()\fR is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
913     PCRE_MULTILINE is set.
914    
915     Note that the sequences \\A, \\Z, and \\z can be used to match the start and
916     end of the subject in both modes, and if all branches of a pattern start with
917     \\A is it always anchored, whether PCRE_MULTILINE is set or not.
918    
919    
920     .SH FULL STOP (PERIOD, DOT)
921     Outside a character class, a dot in the pattern matches any one character in
922     the subject, including a non-printing character, but not (by default) newline.
923     If the PCRE_DOTALL option is set, then dots match newlines as well. The
924     handling of dot is entirely independent of the handling of circumflex and
925     dollar, the only relationship being that they both involve newline characters.
926     Dot has no special meaning in a character class.
927    
928    
929     .SH SQUARE BRACKETS
930     An opening square bracket introduces a character class, terminated by a closing
931     square bracket. A closing square bracket on its own is not special. If a
932     closing square bracket is required as a member of the class, it should be the
933     first data character in the class (after an initial circumflex, if present) or
934     escaped with a backslash.
935    
936     A character class matches a single character in the subject; the character must
937     be in the set of characters defined by the class, unless the first character in
938     the class is a circumflex, in which case the subject character must not be in
939     the set defined by the class. If a circumflex is actually required as a member
940     of the class, ensure it is not the first character, or escape it with a
941     backslash.
942    
943     For example, the character class [aeiou] matches any lower case vowel, while
944     [^aeiou] matches any character that is not a lower case vowel. Note that a
945     circumflex is just a convenient notation for specifying the characters which
946     are in the class by enumerating those that are not. It is not an assertion: it
947     still consumes a character from the subject string, and fails if the current
948     pointer is at the end of the string.
949    
950     When caseless matching is set, any letters in a class represent both their
951     upper case and lower case versions, so for example, a caseless [aeiou] matches
952     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
953     caseful version would.
954    
955     The newline character is never treated in any special way in character classes,
956     whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class
957     such as [^a] will always match a newline.
958    
959     The minus (hyphen) character can be used to specify a range of characters in a
960     character class. For example, [d-m] matches any letter between d and m,
961     inclusive. If a minus character is required in a class, it must be escaped with
962     a backslash or appear in a position where it cannot be interpreted as
963     indicating a range, typically as the first or last character in the class.
964    
965     It is not possible to have the literal character "]" as the end character of a
966     range. A pattern such as [W-]46] is interpreted as a class of two characters
967     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
968     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
969     the end of range, so [W-\\]46] is interpreted as a single class containing a
970     range followed by two separate characters. The octal or hexadecimal
971     representation of "]" can also be used to end a range.
972    
973     Ranges operate in ASCII collating sequence. They can also be used for
974     characters specified numerically, for example [\\000-\\037]. If a range that
975     includes letters is used when caseless matching is set, it matches the letters
976     in either case. For example, [W-c] is equivalent to [][\\^_`wxyzabc], matched
977     caselessly, and if character tables for the "fr" locale are in use,
978     [\\xc8-\\xcb] matches accented E characters in both cases.
979    
980     The character types \\d, \\D, \\s, \\S, \\w, and \\W may also appear in a
981     character class, and add the characters that they match to the class. For
982     example, [\\dABCDEF] matches any hexadecimal digit. A circumflex can
983     conveniently be used with the upper case character types to specify a more
984     restricted set of characters than the matching lower case type. For example,
985     the class [^\\W_] matches any letter or digit, but not underscore.
986    
987     All non-alphameric characters other than \\, -, ^ (at the start) and the
988     terminating ] are non-special in character classes, but it does no harm if they
989     are escaped.
990    
991    
992 nigel 43 .SH POSIX CHARACTER CLASSES
993     Perl 5.6 (not yet released at the time of writing) is going to support the
994     POSIX notation for character classes, which uses names enclosed by [: and :]
995     within the enclosing square brackets. PCRE supports this notation. For example,
996    
997     [01[:alpha:]%]
998    
999     matches "0", "1", any alphabetic character, or "%". The supported class names
1000     are
1001    
1002     alnum letters and digits
1003     alpha letters
1004     ascii character codes 0 - 127
1005     cntrl control characters
1006     digit decimal digits (same as \\d)
1007     graph printing characters, excluding space
1008     lower lower case letters
1009     print printing characters, including space
1010     punct printing characters, excluding letters and digits
1011     space white space (same as \\s)
1012     upper upper case letters
1013     word "word" characters (same as \\w)
1014     xdigit hexadecimal digits
1015    
1016     The names "ascii" and "word" are Perl extensions. Another Perl extension is
1017     negation, which is indicated by a ^ character after the colon. For example,
1018    
1019     [12[:^digit:]]
1020    
1021     matches "1", "2", or any non-digit. PCRE (and Perl) also recogize the POSIX
1022     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1023     supported, and an error is given if they are encountered.
1024    
1025    
1026 nigel 41 .SH VERTICAL BAR
1027     Vertical bar characters are used to separate alternative patterns. For example,
1028     the pattern
1029    
1030     gilbert|sullivan
1031    
1032     matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1033     and an empty alternative is permitted (matching the empty string).
1034     The matching process tries each alternative in turn, from left to right,
1035     and the first one that succeeds is used. If the alternatives are within a
1036     subpattern (defined below), "succeeds" means matching the rest of the main
1037     pattern as well as the alternative in the subpattern.
1038    
1039    
1040     .SH INTERNAL OPTION SETTING
1041     The settings of PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and PCRE_EXTENDED
1042     can be changed from within the pattern by a sequence of Perl option letters
1043     enclosed between "(?" and ")". The option letters are
1044    
1045     i for PCRE_CASELESS
1046     m for PCRE_MULTILINE
1047     s for PCRE_DOTALL
1048     x for PCRE_EXTENDED
1049    
1050     For example, (?im) sets caseless, multiline matching. It is also possible to
1051     unset these options by preceding the letter with a hyphen, and a combined
1052     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1053     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1054     permitted. If a letter appears both before and after the hyphen, the option is
1055     unset.
1056    
1057     The scope of these option changes depends on where in the pattern the setting
1058     occurs. For settings that are outside any subpattern (defined below), the
1059     effect is the same as if the options were set or unset at the start of
1060     matching. The following patterns all behave in exactly the same way:
1061    
1062     (?i)abc
1063     a(?i)bc
1064     ab(?i)c
1065     abc(?i)
1066    
1067     which in turn is the same as compiling the pattern abc with PCRE_CASELESS set.
1068     In other words, such "top level" settings apply to the whole pattern (unless
1069     there are other changes inside subpatterns). If there is more than one setting
1070     of the same option at top level, the rightmost setting is used.
1071    
1072     If an option change occurs inside a subpattern, the effect is different. This
1073     is a change of behaviour in Perl 5.005. An option change inside a subpattern
1074     affects only that part of the subpattern that follows it, so
1075    
1076     (a(?i)b)c
1077    
1078     matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1079     By this means, options can be made to have different settings in different
1080     parts of the pattern. Any changes made in one alternative do carry on
1081     into subsequent branches within the same subpattern. For example,
1082    
1083     (a(?i)b|c)
1084    
1085     matches "ab", "aB", "c", and "C", even though when matching "C" the first
1086     branch is abandoned before the option setting. This is because the effects of
1087     option settings happen at compile time. There would be some very weird
1088     behaviour otherwise.
1089    
1090     The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the
1091     same way as the Perl-compatible options by using the characters U and X
1092     respectively. The (?X) flag setting is special in that it must always occur
1093     earlier in the pattern than any of the additional features it turns on, even
1094     when it is at top level. It is best put at the start.
1095    
1096    
1097     .SH SUBPATTERNS
1098     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1099     Marking part of a pattern as a subpattern does two things:
1100    
1101     1. It localizes a set of alternatives. For example, the pattern
1102    
1103     cat(aract|erpillar|)
1104    
1105     matches one of the words "cat", "cataract", or "caterpillar". Without the
1106     parentheses, it would match "cataract", "erpillar" or the empty string.
1107    
1108     2. It sets up the subpattern as a capturing subpattern (as defined above).
1109     When the whole pattern matches, that portion of the subject string that matched
1110     the subpattern is passed back to the caller via the \fIovector\fR argument of
1111     \fBpcre_exec()\fR. Opening parentheses are counted from left to right (starting
1112     from 1) to obtain the numbers of the capturing subpatterns.
1113    
1114     For example, if the string "the red king" is matched against the pattern
1115    
1116     the ((red|white) (king|queen))
1117    
1118     the captured substrings are "red king", "red", and "king", and are numbered 1,
1119     2, and 3.
1120    
1121     The fact that plain parentheses fulfil two functions is not always helpful.
1122     There are often times when a grouping subpattern is required without a
1123     capturing requirement. If an opening parenthesis is followed by "?:", the
1124     subpattern does not do any capturing, and is not counted when computing the
1125     number of any subsequent capturing subpatterns. For example, if the string "the
1126     white queen" is matched against the pattern
1127    
1128     the ((?:red|white) (king|queen))
1129    
1130     the captured substrings are "white queen" and "queen", and are numbered 1 and
1131     2. The maximum number of captured substrings is 99, and the maximum number of
1132     all subpatterns, both capturing and non-capturing, is 200.
1133    
1134     As a convenient shorthand, if any option settings are required at the start of
1135     a non-capturing subpattern, the option letters may appear between the "?" and
1136     the ":". Thus the two patterns
1137    
1138     (?i:saturday|sunday)
1139     (?:(?i)saturday|sunday)
1140    
1141     match exactly the same set of strings. Because alternative branches are tried
1142     from left to right, and options are not reset until the end of the subpattern
1143     is reached, an option setting in one branch does affect subsequent branches, so
1144     the above patterns match "SUNDAY" as well as "Saturday".
1145    
1146    
1147     .SH REPETITION
1148     Repetition is specified by quantifiers, which can follow any of the following
1149     items:
1150    
1151     a single character, possibly escaped
1152     the . metacharacter
1153     a character class
1154     a back reference (see next section)
1155     a parenthesized subpattern (unless it is an assertion - see below)
1156    
1157     The general repetition quantifier specifies a minimum and maximum number of
1158     permitted matches, by giving the two numbers in curly brackets (braces),
1159     separated by a comma. The numbers must be less than 65536, and the first must
1160     be less than or equal to the second. For example:
1161    
1162     z{2,4}
1163    
1164     matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1165     character. If the second number is omitted, but the comma is present, there is
1166     no upper limit; if the second number and the comma are both omitted, the
1167     quantifier specifies an exact number of required matches. Thus
1168    
1169     [aeiou]{3,}
1170    
1171     matches at least 3 successive vowels, but may match many more, while
1172    
1173     \\d{8}
1174    
1175     matches exactly 8 digits. An opening curly bracket that appears in a position
1176     where a quantifier is not allowed, or one that does not match the syntax of a
1177     quantifier, is taken as a literal character. For example, {,6} is not a
1178     quantifier, but a literal string of four characters.
1179    
1180     The quantifier {0} is permitted, causing the expression to behave as if the
1181     previous item and the quantifier were not present.
1182    
1183     For convenience (and historical compatibility) the three most common
1184     quantifiers have single-character abbreviations:
1185    
1186     * is equivalent to {0,}
1187     + is equivalent to {1,}
1188     ? is equivalent to {0,1}
1189    
1190     It is possible to construct infinite loops by following a subpattern that can
1191     match no characters with a quantifier that has no upper limit, for example:
1192    
1193     (a?)*
1194    
1195     Earlier versions of Perl and PCRE used to give an error at compile time for
1196     such patterns. However, because there are cases where this can be useful, such
1197     patterns are now accepted, but if any repetition of the subpattern does in fact
1198     match no characters, the loop is forcibly broken.
1199    
1200     By default, the quantifiers are "greedy", that is, they match as much as
1201     possible (up to the maximum number of permitted times), without causing the
1202     rest of the pattern to fail. The classic example of where this gives problems
1203     is in trying to match comments in C programs. These appear between the
1204     sequences /* and */ and within the sequence, individual * and / characters may
1205     appear. An attempt to match C comments by applying the pattern
1206    
1207     /\\*.*\\*/
1208    
1209     to the string
1210    
1211     /* first command */ not comment /* second comment */
1212    
1213     fails, because it matches the entire string due to the greediness of the .*
1214     item.
1215    
1216     However, if a quantifier is followed by a question mark, then it ceases to be
1217     greedy, and instead matches the minimum number of times possible, so the
1218     pattern
1219    
1220     /\\*.*?\\*/
1221    
1222     does the right thing with the C comments. The meaning of the various
1223     quantifiers is not otherwise changed, just the preferred number of matches.
1224     Do not confuse this use of question mark with its use as a quantifier in its
1225     own right. Because it has two uses, it can sometimes appear doubled, as in
1226    
1227     \\d??\\d
1228    
1229     which matches one digit by preference, but can match two if that is the only
1230     way the rest of the pattern matches.
1231    
1232     If the PCRE_UNGREEDY option is set (an option which is not available in Perl)
1233     then the quantifiers are not greedy by default, but individual ones can be made
1234     greedy by following them with a question mark. In other words, it inverts the
1235     default behaviour.
1236    
1237     When a parenthesized subpattern is quantified with a minimum repeat count that
1238     is greater than 1 or with a limited maximum, more store is required for the
1239     compiled pattern, in proportion to the size of the minimum or maximum.
1240    
1241     If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1242     to Perl's /s) is set, thus allowing the . to match newlines, then the pattern
1243     is implicitly anchored, because whatever follows will be tried against every
1244     character position in the subject string, so there is no point in retrying the
1245     overall match at any position after the first. PCRE treats such a pattern as
1246     though it were preceded by \\A. In cases where it is known that the subject
1247     string contains no newlines, it is worth setting PCRE_DOTALL when the pattern
1248     begins with .* in order to obtain this optimization, or alternatively using ^
1249     to indicate anchoring explicitly.
1250    
1251     When a capturing subpattern is repeated, the value captured is the substring
1252     that matched the final iteration. For example, after
1253    
1254     (tweedle[dume]{3}\\s*)+
1255    
1256     has matched "tweedledum tweedledee" the value of the captured substring is
1257     "tweedledee". However, if there are nested capturing subpatterns, the
1258     corresponding captured values may have been set in previous iterations. For
1259     example, after
1260    
1261     /(a|(b))+/
1262    
1263     matches "aba" the value of the second captured substring is "b".
1264    
1265    
1266     .SH BACK REFERENCES
1267     Outside a character class, a backslash followed by a digit greater than 0 (and
1268     possibly further digits) is a back reference to a capturing subpattern earlier
1269     (i.e. to its left) in the pattern, provided there have been that many previous
1270     capturing left parentheses.
1271    
1272     However, if the decimal number following the backslash is less than 10, it is
1273     always taken as a back reference, and causes an error only if there are not
1274     that many capturing left parentheses in the entire pattern. In other words, the
1275     parentheses that are referenced need not be to the left of the reference for
1276     numbers less than 10. See the section entitled "Backslash" above for further
1277     details of the handling of digits following a backslash.
1278    
1279     A back reference matches whatever actually matched the capturing subpattern in
1280     the current subject string, rather than anything matching the subpattern
1281     itself. So the pattern
1282    
1283     (sens|respons)e and \\1ibility
1284    
1285     matches "sense and sensibility" and "response and responsibility", but not
1286     "sense and responsibility". If caseful matching is in force at the time of the
1287     back reference, then the case of letters is relevant. For example,
1288    
1289     ((?i)rah)\\s+\\1
1290    
1291     matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1292     capturing subpattern is matched caselessly.
1293    
1294     There may be more than one back reference to the same subpattern. If a
1295     subpattern has not actually been used in a particular match, then any back
1296     references to it always fail. For example, the pattern
1297    
1298     (a|(bc))\\2
1299    
1300     always fails if it starts to match "a" rather than "bc". Because there may be
1301     up to 99 back references, all digits following the backslash are taken
1302     as part of a potential back reference number. If the pattern continues with a
1303     digit character, then some delimiter must be used to terminate the back
1304     reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1305     Otherwise an empty comment can be used.
1306    
1307     A back reference that occurs inside the parentheses to which it refers fails
1308     when the subpattern is first used, so, for example, (a\\1) never matches.
1309     However, such references can be useful inside repeated subpatterns. For
1310     example, the pattern
1311    
1312     (a|b\\1)+
1313    
1314     matches any number of "a"s and also "aba", "ababaa" etc. At each iteration of
1315     the subpattern, the back reference matches the character string corresponding
1316     to the previous iteration. In order for this to work, the pattern must be such
1317     that the first iteration does not need to match the back reference. This can be
1318     done using alternation, as in the example above, or by a quantifier with a
1319     minimum of zero.
1320    
1321    
1322     .SH ASSERTIONS
1323     An assertion is a test on the characters following or preceding the current
1324     matching point that does not actually consume any characters. The simple
1325     assertions coded as \\b, \\B, \\A, \\Z, \\z, ^ and $ are described above. More
1326     complicated assertions are coded as subpatterns. There are two kinds: those
1327     that look ahead of the current position in the subject string, and those that
1328     look behind it.
1329    
1330     An assertion subpattern is matched in the normal way, except that it does not
1331     cause the current matching position to be changed. Lookahead assertions start
1332     with (?= for positive assertions and (?! for negative assertions. For example,
1333    
1334     \\w+(?=;)
1335    
1336     matches a word followed by a semicolon, but does not include the semicolon in
1337     the match, and
1338    
1339     foo(?!bar)
1340    
1341     matches any occurrence of "foo" that is not followed by "bar". Note that the
1342     apparently similar pattern
1343    
1344     (?!foo)bar
1345    
1346     does not find an occurrence of "bar" that is preceded by something other than
1347     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1348     (?!foo) is always true when the next three characters are "bar". A
1349     lookbehind assertion is needed to achieve this effect.
1350    
1351     Lookbehind assertions start with (?<= for positive assertions and (?<! for
1352     negative assertions. For example,
1353    
1354     (?<!foo)bar
1355    
1356     does find an occurrence of "bar" that is not preceded by "foo". The contents of
1357     a lookbehind assertion are restricted such that all the strings it matches must
1358     have a fixed length. However, if there are several alternatives, they do not
1359     all have to have the same fixed length. Thus
1360    
1361     (?<=bullock|donkey)
1362    
1363     is permitted, but
1364    
1365     (?<!dogs?|cats?)
1366    
1367     causes an error at compile time. Branches that match different length strings
1368     are permitted only at the top level of a lookbehind assertion. This is an
1369     extension compared with Perl 5.005, which requires all branches to match the
1370     same length of string. An assertion such as
1371    
1372     (?<=ab(c|de))
1373    
1374     is not permitted, because its single top-level branch can match two different
1375     lengths, but it is acceptable if rewritten to use two top-level branches:
1376    
1377     (?<=abc|abde)
1378    
1379     The implementation of lookbehind assertions is, for each alternative, to
1380     temporarily move the current position back by the fixed width and then try to
1381     match. If there are insufficient characters before the current position, the
1382     match is deemed to fail. Lookbehinds in conjunction with once-only subpatterns
1383     can be particularly useful for matching at the ends of strings; an example is
1384     given at the end of the section on once-only subpatterns.
1385    
1386     Several assertions (of any sort) may occur in succession. For example,
1387    
1388     (?<=\\d{3})(?<!999)foo
1389    
1390     matches "foo" preceded by three digits that are not "999". Notice that each of
1391     the assertions is applied independently at the same point in the subject
1392     string. First there is a check that the previous three characters are all
1393     digits, then there is a check that the same three characters are not "999".
1394     This pattern does \fInot\fR match "foo" preceded by six characters, the first
1395     of which are digits and the last three of which are not "999". For example, it
1396     doesn't match "123abcfoo". A pattern to do that is
1397    
1398     (?<=\\d{3}...)(?<!999)foo
1399    
1400     This time the first assertion looks at the preceding six characters, checking
1401     that the first three are digits, and then the second assertion checks that the
1402     preceding three characters are not "999".
1403    
1404     Assertions can be nested in any combination. For example,
1405    
1406     (?<=(?<!foo)bar)baz
1407    
1408     matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1409     preceded by "foo", while
1410    
1411     (?<=\\d{3}(?!999)...)foo
1412    
1413     is another pattern which matches "foo" preceded by three digits and any three
1414     characters that are not "999".
1415    
1416     Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1417     because it makes no sense to assert the same thing several times. If any kind
1418     of assertion contains capturing subpatterns within it, these are counted for
1419     the purposes of numbering the capturing subpatterns in the whole pattern.
1420     However, substring capturing is carried out only for positive assertions,
1421     because it does not make sense for negative assertions.
1422    
1423     Assertions count towards the maximum of 200 parenthesized subpatterns.
1424    
1425    
1426     .SH ONCE-ONLY SUBPATTERNS
1427     With both maximizing and minimizing repetition, failure of what follows
1428     normally causes the repeated item to be re-evaluated to see if a different
1429     number of repeats allows the rest of the pattern to match. Sometimes it is
1430     useful to prevent this, either to change the nature of the match, or to cause
1431     it fail earlier than it otherwise might, when the author of the pattern knows
1432     there is no point in carrying on.
1433    
1434     Consider, for example, the pattern \\d+foo when applied to the subject line
1435    
1436     123456bar
1437    
1438     After matching all 6 digits and then failing to match "foo", the normal
1439     action of the matcher is to try again with only 5 digits matching the \\d+
1440     item, and then with 4, and so on, before ultimately failing. Once-only
1441     subpatterns provide the means for specifying that once a portion of the pattern
1442     has matched, it is not to be re-evaluated in this way, so the matcher would
1443     give up immediately on failing to match "foo" the first time. The notation is
1444     another kind of special parenthesis, starting with (?> as in this example:
1445    
1446     (?>\\d+)bar
1447    
1448     This kind of parenthesis "locks up" the part of the pattern it contains once
1449     it has matched, and a failure further into the pattern is prevented from
1450     backtracking into it. Backtracking past it to previous items, however, works as
1451     normal.
1452    
1453     An alternative description is that a subpattern of this type matches the string
1454     of characters that an identical standalone pattern would match, if anchored at
1455     the current point in the subject string.
1456    
1457     Once-only subpatterns are not capturing subpatterns. Simple cases such as the
1458     above example can be thought of as a maximizing repeat that must swallow
1459     everything it can. So, while both \\d+ and \\d+? are prepared to adjust the
1460     number of digits they match in order to make the rest of the pattern match,
1461     (?>\\d+) can only match an entire sequence of digits.
1462    
1463     This construction can of course contain arbitrarily complicated subpatterns,
1464     and it can be nested.
1465    
1466     Once-only subpatterns can be used in conjunction with lookbehind assertions to
1467     specify efficient matching at the end of the subject string. Consider a simple
1468     pattern such as
1469    
1470     abcd$
1471    
1472 nigel 43 when applied to a long string which does not match. Because matching proceeds
1473     from left to right, PCRE will look for each "a" in the subject and then see if
1474     what follows matches the rest of the pattern. If the pattern is specified as
1475 nigel 41
1476     ^.*abcd$
1477    
1478 nigel 43 then the initial .* matches the entire string at first, but when this fails
1479     (because there is no following "a"), it backtracks to match all but the last
1480     character, then all but the last two characters, and so on. Once again the
1481     search for "a" covers the entire string, from right to left, so we are no
1482     better off. However, if the pattern is written as
1483 nigel 41
1484     ^(?>.*)(?<=abcd)
1485    
1486     then there can be no backtracking for the .* item; it can match only the entire
1487     string. The subsequent lookbehind assertion does a single test on the last four
1488     characters. If it fails, the match fails immediately. For long strings, this
1489     approach makes a significant difference to the processing time.
1490    
1491 nigel 43 When a pattern contains an unlimited repeat inside a subpattern that can itself
1492     be repeated an unlimited number of times, the use of a once-only subpattern is
1493     the only way to avoid some failing matches taking a very long time indeed.
1494     The pattern
1495 nigel 41
1496 nigel 43 (\\D+|<\\d+>)*[!?]
1497    
1498     matches an unlimited number of substrings that either consist of non-digits, or
1499     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1500     quickly. However, if it is applied to
1501    
1502     aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1503    
1504     it takes a long time before reporting failure. This is because the string can
1505     be divided between the two repeats in a large number of ways, and all have to
1506     be tried. (The example used [!?] rather than a single character at the end,
1507     because both PCRE and Perl have an optimization that allows for fast failure
1508     when a single character is used. They remember the last single character that
1509     is required for a match, and fail early if it is not present in the string.)
1510     If the pattern is changed to
1511    
1512     ((?>\\D+)|<\\d+>)*[!?]
1513    
1514     sequences of non-digits cannot be broken, and failure happens quickly.
1515    
1516    
1517 nigel 41 .SH CONDITIONAL SUBPATTERNS
1518     It is possible to cause the matching process to obey a subpattern
1519     conditionally or to choose between two alternative subpatterns, depending on
1520     the result of an assertion, or whether a previous capturing subpattern matched
1521     or not. The two possible forms of conditional subpattern are
1522    
1523     (?(condition)yes-pattern)
1524     (?(condition)yes-pattern|no-pattern)
1525    
1526     If the condition is satisfied, the yes-pattern is used; otherwise the
1527     no-pattern (if present) is used. If there are more than two alternatives in the
1528     subpattern, a compile-time error occurs.
1529    
1530     There are two kinds of condition. If the text between the parentheses consists
1531     of a sequence of digits, then the condition is satisfied if the capturing
1532     subpattern of that number has previously matched. Consider the following
1533     pattern, which contains non-significant white space to make it more readable
1534     (assume the PCRE_EXTENDED option) and to divide it into three parts for ease
1535     of discussion:
1536    
1537     ( \\( )? [^()]+ (?(1) \\) )
1538    
1539     The first part matches an optional opening parenthesis, and if that
1540     character is present, sets it as the first captured substring. The second part
1541     matches one or more characters that are not parentheses. The third part is a
1542     conditional subpattern that tests whether the first set of parentheses matched
1543     or not. If they did, that is, if subject started with an opening parenthesis,
1544     the condition is true, and so the yes-pattern is executed and a closing
1545     parenthesis is required. Otherwise, since no-pattern is not present, the
1546     subpattern matches nothing. In other words, this pattern matches a sequence of
1547     non-parentheses, optionally enclosed in parentheses.
1548    
1549     If the condition is not a sequence of digits, it must be an assertion. This may
1550     be a positive or negative lookahead or lookbehind assertion. Consider this
1551     pattern, again containing non-significant white space, and with the two
1552     alternatives on the second line:
1553    
1554     (?(?=[^a-z]*[a-z])
1555     \\d{2}-[a-z]{3}-\\d{2} | \\d{2}-\\d{2}-\\d{2} )
1556    
1557     The condition is a positive lookahead assertion that matches an optional
1558     sequence of non-letters followed by a letter. In other words, it tests for the
1559     presence of at least one letter in the subject. If a letter is found, the
1560     subject is matched against the first alternative; otherwise it is matched
1561     against the second. This pattern matches strings in one of the two forms
1562     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1563    
1564    
1565     .SH COMMENTS
1566     The sequence (?# marks the start of a comment which continues up to the next
1567     closing parenthesis. Nested parentheses are not permitted. The characters
1568     that make up a comment play no part in the pattern matching at all.
1569    
1570     If the PCRE_EXTENDED option is set, an unescaped # character outside a
1571     character class introduces a comment that continues up to the next newline
1572     character in the pattern.
1573    
1574    
1575 nigel 43 .SH RECURSIVE PATTERNS
1576     Consider the problem of matching a string in parentheses, allowing for
1577     unlimited nested parentheses. Without the use of recursion, the best that can
1578     be done is to use a pattern that matches up to some fixed depth of nesting. It
1579     is not possible to handle an arbitrary nesting depth. Perl 5.6 has provided an
1580     experimental facility that allows regular expressions to recurse (amongst other
1581     things). It does this by interpolating Perl code in the expression at run time,
1582     and the code can refer to the expression itself. A Perl pattern to solve the
1583     parentheses problem can be created like this:
1584    
1585     $re = qr{\\( (?: (?>[^()]+) | (?p{$re}) )* \\)}x;
1586    
1587     The (?p{...}) item interpolates Perl code at run time, and in this case refers
1588     recursively to the pattern in which it appears. Obviously, PCRE cannot support
1589     the interpolation of Perl code. Instead, the special item (?R) is provided for
1590     the specific case of recursion. This PCRE pattern solves the parentheses
1591     problem (assume the PCRE_EXTENDED option is set so that white space is
1592     ignored):
1593    
1594     \\( ( (?>[^()]+) | (?R) )* \\)
1595    
1596     First it matches an opening parenthesis. Then it matches any number of
1597     substrings which can either be a sequence of non-parentheses, or a recursive
1598     match of the pattern itself (i.e. a correctly parenthesized substring). Finally
1599     there is a closing parenthesis.
1600    
1601     This particular example pattern contains nested unlimited repeats, and so the
1602     use of a once-only subpattern for matching strings of non-parentheses is
1603     important when applying the pattern to strings that do not match. For example,
1604     when it is applied to
1605    
1606     (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1607    
1608     it yields "no match" quickly. However, if a once-only subpattern is not used,
1609     the match runs for a very long time indeed because there are so many different
1610     ways the + and * repeats can carve up the subject, and all have to be tested
1611     before failure can be reported.
1612    
1613     The values set for any capturing subpatterns are those from the outermost level
1614     of the recursion at which the subpattern value is set. If the pattern above is
1615     matched against
1616    
1617     (ab(cd)ef)
1618    
1619     the value for the capturing parentheses is "ef", which is the last value taken
1620     on at the top level. If additional parentheses are added, giving
1621    
1622     \\( ( ( (?>[^()]+) | (?R) )* ) \\)
1623     ^ ^
1624     ^ ^
1625     then the string they capture is "ab(cd)ef", the contents of the top level
1626     parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1627     has to obtain extra memory to store data during a recursion, which it does by
1628     using \fBpcre_malloc\fR, freeing it via \fBpcre_free\fR afterwards. If no
1629     memory can be obtained, it saves data for the first 15 capturing parentheses
1630     only, as there is no way to give an out-of-memory error from within a
1631     recursion.
1632    
1633    
1634 nigel 41 .SH PERFORMANCE
1635     Certain items that may appear in patterns are more efficient than others. It is
1636     more efficient to use a character class like [aeiou] than a set of alternatives
1637     such as (a|e|i|o|u). In general, the simplest construction that provides the
1638     required behaviour is usually the most efficient. Jeffrey Friedl's book
1639     contains a lot of discussion about optimizing regular expressions for efficient
1640     performance.
1641    
1642     When a pattern begins with .* and the PCRE_DOTALL option is set, the pattern is
1643     implicitly anchored by PCRE, since it can match only at the start of a subject
1644     string. However, if PCRE_DOTALL is not set, PCRE cannot make this optimization,
1645     because the . metacharacter does not then match a newline, and if the subject
1646     string contains newlines, the pattern may match from the character immediately
1647     following one of them instead of from the very start. For example, the pattern
1648    
1649     (.*) second
1650    
1651     matches the subject "first\\nand second" (where \\n stands for a newline
1652     character) with the first captured substring being "and". In order to do this,
1653     PCRE has to retry the match starting after every newline in the subject.
1654    
1655     If you are using such a pattern with subject strings that do not contain
1656     newlines, the best performance is obtained by setting PCRE_DOTALL, or starting
1657     the pattern with ^.* to indicate explicit anchoring. That saves PCRE from
1658     having to scan along the subject looking for a newline to restart at.
1659    
1660     Beware of patterns that contain nested indefinite repeats. These can take a
1661     long time to run when applied to a string that does not match. Consider the
1662     pattern fragment
1663    
1664     (a+)*
1665    
1666     This can match "aaaa" in 33 different ways, and this number increases very
1667     rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4
1668     times, and for each of those cases other than 0, the + repeats can match
1669     different numbers of times.) When the remainder of the pattern is such that the
1670     entire match is going to fail, PCRE has in principle to try every possible
1671     variation, and this can take an extremely long time.
1672    
1673     An optimization catches some of the more simple cases such as
1674    
1675     (a+)*b
1676    
1677     where a literal character follows. Before embarking on the standard matching
1678     procedure, PCRE checks that there is a "b" later in the subject string, and if
1679     there is not, it fails the match immediately. However, when there is no
1680     following literal this optimization cannot be used. You can see the difference
1681     by comparing the behaviour of
1682    
1683     (a+)*\\d
1684    
1685     with the pattern above. The former gives a failure almost instantly when
1686     applied to a whole line of "a" characters, whereas the latter takes an
1687     appreciable time with strings longer than about 20 characters.
1688    
1689     .SH AUTHOR
1690     Philip Hazel <ph10@cam.ac.uk>
1691     .br
1692     University Computing Service,
1693     .br
1694     New Museums Site,
1695     .br
1696     Cambridge CB2 3QG, England.
1697     .br
1698     Phone: +44 1223 334714
1699    
1700 nigel 43 Last updated: 27 January 2000
1701 nigel 41 .br
1702 nigel 43 Copyright (c) 1997-2000 University of Cambridge.

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12