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

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

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 41 by nigel, Sat Feb 24 21:39:17 2007 UTC revision 63 by nigel, Sat Feb 24 21:40:03 2007 UTC
# Line 1  Line 1 
1  .TH PCRE 3  .TH PCRE 3
2  .SH NAME  .SH NAME
3  pcre - Perl-compatible regular expressions.  PCRE - Perl-compatible regular expressions
 .SH SYNOPSIS  
 .B #include <pcre.h>  
 .PP  
 .SM  
 .br  
 .B pcre *pcre_compile(const char *\fIpattern\fR, int \fIoptions\fR,  
 .ti +5n  
 .B const char **\fIerrptr\fR, int *\fIerroffset\fR,  
 .ti +5n  
 .B const unsigned char *\fItableptr\fR);  
 .PP  
 .br  
 .B pcre_extra *pcre_study(const pcre *\fIcode\fR, int \fIoptions\fR,  
 .ti +5n  
 .B const char **\fIerrptr\fR);  
 .PP  
 .br  
 .B int pcre_exec(const pcre *\fIcode\fR, "const pcre_extra *\fIextra\fR,"  
 .ti +5n  
 .B "const char *\fIsubject\fR," int \fIlength\fR, int \fIstartoffset\fR,  
 .ti +5n  
 .B int \fIoptions\fR, int *\fIovector\fR, int \fIovecsize\fR);  
 .PP  
 .br  
 .B int pcre_copy_substring(const char *\fIsubject\fR, int *\fIovector\fR,  
 .ti +5n  
 .B int \fIstringcount\fR, int \fIstringnumber\fR, char *\fIbuffer\fR,  
 .ti +5n  
 .B int \fIbuffersize\fR);  
 .PP  
 .br  
 .B int pcre_get_substring(const char *\fIsubject\fR, int *\fIovector\fR,  
 .ti +5n  
 .B int \fIstringcount\fR, int \fIstringnumber\fR,  
 .ti +5n  
 .B const char **\fIstringptr\fR);  
 .PP  
 .br  
 .B int pcre_get_substring_list(const char *\fIsubject\fR,  
 .ti +5n  
 .B int *\fIovector\fR, int \fIstringcount\fR, "const char ***\fIlistptr\fR);"  
 .PP  
 .br  
 .B const unsigned char *pcre_maketables(void);  
 .PP  
 .br  
 .B int pcre_info(const pcre *\fIcode\fR, int *\fIoptptr\fR, int  
 .B *\fIfirstcharptr\fR);  
 .PP  
 .br  
 .B char *pcre_version(void);  
 .PP  
 .br  
 .B void *(*pcre_malloc)(size_t);  
 .PP  
 .br  
 .B void (*pcre_free)(void *);  
   
   
   
4  .SH DESCRIPTION  .SH DESCRIPTION
5    .rs
6    .sp
7  The PCRE library is a set of functions that implement regular expression  The PCRE library is a set of functions that implement regular expression
8  pattern matching using the same syntax and semantics as Perl 5, with just a few  pattern matching using the same syntax and semantics as Perl, with just a few
9  differences (see below). The current implementation corresponds to Perl 5.005.  differences. The current implementation of PCRE (release 4.x) corresponds
10    approximately with Perl 5.8, including support for UTF-8 encoded strings.
11  PCRE has its own native API, which is described in this document. There is also  However, this support has to be explicitly enabled; it is not the default.
12  a set of wrapper functions that correspond to the POSIX API. These are  
13  described in the \fBpcreposix\fR documentation.  PCRE is written in C and released as a C library. However, a number of people
14    have written wrappers and interfaces of various kinds. A C++ class is included
15  The native API function prototypes are defined in the header file \fBpcre.h\fR,  in these contributions, which can be found in the \fIContrib\fR directory at
16  and on Unix systems the library itself is called \fBlibpcre.a\fR, so can be  the primary FTP site, which is:
17  accessed by adding \fB-lpcre\fR to the command for linking an application which  
18  calls it.  .\" HTML <a href="ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre">
19    .\" </a>
20  The functions \fBpcre_compile()\fR, \fBpcre_study()\fR, and \fBpcre_exec()\fR  ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
21  are used for compiling and matching regular expressions, while  
22  \fBpcre_copy_substring()\fR, \fBpcre_get_substring()\fR, and  Details of exactly which Perl regular expression features are and are not
23  \fBpcre_get_substring_list()\fR are convenience functions for extracting  supported by PCRE are given in separate documents. See the
24  captured substrings from a matched subject string. The function  .\" HREF
25  \fBpcre_maketables()\fR is used (optionally) to build a set of character tables  \fBpcrepattern\fR
26  in the current locale for passing to \fBpcre_compile()\fR.  .\"
27    and
28  The function \fBpcre_info()\fR is used to find out information about a compiled  .\" HREF
29  pattern, while the function \fBpcre_version()\fR returns a pointer to a string  \fBpcrecompat\fR
30  containing the version of PCRE and its date of release.  .\"
31    pages.
32  The global variables \fBpcre_malloc\fR and \fBpcre_free\fR initially contain  
33  the entry points of the standard \fBmalloc()\fR and \fBfree()\fR functions  Some features of PCRE can be included, excluded, or changed when the library is
34  respectively. PCRE calls the memory management functions via these variables,  built. The
35  so a calling program can replace them if it wishes to intercept the calls. This  .\" HREF
36  should be done before calling any PCRE functions.  \fBpcre_config()\fR
37    .\"
38    function makes it possible for a client to discover which features are
39  .SH MULTI-THREADING  available. Documentation about building PCRE for various operating systems can
40  The PCRE functions can be used in multi-threading applications, with the  be found in the \fBREADME\fR file in the source distribution.
41  proviso that the memory management functions pointed to by \fBpcre_malloc\fR  
42  and \fBpcre_free\fR are shared by all threads.  .SH USER DOCUMENTATION
43    .rs
44  The compiled form of a regular expression is not altered during matching, so  .sp
45  the same compiled pattern can safely be used by several threads at once.  The user documentation for PCRE has been split up into a number of different
46    sections. In the "man" format, each of these is a separate "man page". In the
47    HTML format, each is a separate page, linked from the index page. In the plain
48  .SH COMPILING A PATTERN  text format, all the sections are concatenated, for ease of searching. The
49  The function \fBpcre_compile()\fR is called to compile a pattern into an  sections are as follows:
50  internal form. The pattern is a C string terminated by a binary zero, and  
51  is passed in the argument \fIpattern\fR. A pointer to a single block of memory    pcre              this document
52  that is obtained via \fBpcre_malloc\fR is returned. This contains the    pcreapi           details of PCRE's native API
53  compiled code and related data. The \fBpcre\fR type is defined for this for    pcrebuild         options for building PCRE
54  convenience, but in fact \fBpcre\fR is just a typedef for \fBvoid\fR, since the    pcrecallout       details of the callout feature
55  contents of the block are not externally defined. It is up to the caller to    pcrecompat        discussion of Perl compatibility
56  free the memory when it is no longer required.    pcregrep          description of the \fBpcregrep\fR command
57  .PP    pcrepattern       syntax and semantics of supported
58  The size of a compiled pattern is roughly proportional to the length of the                        regular expressions
59  pattern string, except that each character class (other than those containing    pcreperform       discussion of performance issues
60  just a single character, negated or not) requires 33 bytes, and repeat    pcreposix         the POSIX-compatible API
61  quantifiers with a minimum greater than one or a bounded maximum cause the    pcresample        discussion of the sample program
62  relevant portions of the compiled pattern to be replicated.    pcretest          the \fBpcretest\fR testing command
 .PP  
 The \fIoptions\fR argument contains independent bits that affect the  
 compilation. It should be zero if no options are required. Some of the options,  
 in particular, those that are compatible with Perl, can also be set and unset  
 from within the pattern (see the detailed description of regular expressions  
 below). For these options, the contents of the \fIoptions\fR argument specifies  
 their initial settings at the start of compilation and execution. The  
 PCRE_ANCHORED option can be set at the time of matching as well as at compile  
 time.  
 .PP  
 If \fIerrptr\fR is NULL, \fBpcre_compile()\fR returns NULL immediately.  
 Otherwise, if compilation of a pattern fails, \fBpcre_compile()\fR returns  
 NULL, and sets the variable pointed to by \fIerrptr\fR to point to a textual  
 error message. The offset from the start of the pattern to the character where  
 the error was discovered is placed in the variable pointed to by  
 \fIerroffset\fR, which must not be NULL. If it is, an immediate error is given.  
 .PP  
 If the final argument, \fItableptr\fR, is NULL, PCRE uses a default set of  
 character tables which are built when it is compiled, using the default C  
 locale. Otherwise, \fItableptr\fR must be the result of a call to  
 \fBpcre_maketables()\fR. See the section on locale support below.  
 .PP  
 The following option bits are defined in the header file:  
   
   PCRE_ANCHORED  
   
 If this bit is set, the pattern is forced to be "anchored", that is, it is  
 constrained to match only at the start of the string which is being searched  
 (the "subject string"). This effect can also be achieved by appropriate  
 constructs in the pattern itself, which is the only way to do it in Perl.  
   
   PCRE_CASELESS  
   
 If this bit is set, letters in the pattern match both upper and lower case  
 letters. It is equivalent to Perl's /i option.  
   
   PCRE_DOLLAR_ENDONLY  
   
 If this bit is set, a dollar metacharacter in the pattern matches only at the  
 end of the subject string. Without this option, a dollar also matches  
 immediately before the final character if it is a newline (but not before any  
 other newlines). The PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is  
 set. There is no equivalent to this option in Perl.  
   
   PCRE_DOTALL  
   
 If this bit is set, a dot metacharater in the pattern matches all characters,  
 including newlines. Without it, newlines are excluded. This option is  
 equivalent to Perl's /s option. A negative class such as [^a] always matches a  
 newline character, independent of the setting of this option.  
   
   PCRE_EXTENDED  
   
 If this bit is set, whitespace data characters in the pattern are totally  
 ignored except when escaped or inside a character class, and characters between  
 an unescaped # outside a character class and the next newline character,  
 inclusive, are also ignored. This is equivalent to Perl's /x option, and makes  
 it possible to include comments inside complicated patterns. Note, however,  
 that this applies only to data characters. Whitespace characters may never  
 appear within special character sequences in a pattern, for example within the  
 sequence (?( which introduces a conditional subpattern.  
   
   PCRE_EXTRA  
   
 This option turns on additional functionality of PCRE that is incompatible with  
 Perl. Any backslash in a pattern that is followed by a letter that has no  
 special meaning causes an error, thus reserving these combinations for future  
 expansion. By default, as in Perl, a backslash followed by a letter with no  
 special meaning is treated as a literal. There are at present no other features  
 controlled by this option.  
   
   PCRE_MULTILINE  
   
 By default, PCRE treats the subject string as consisting of a single "line" of  
 characters (even if it actually contains several newlines). The "start of line"  
 metacharacter (^) matches only at the start of the string, while the "end of  
 line" metacharacter ($) matches only at the end of the string, or before a  
 terminating newline (unless PCRE_DOLLAR_ENDONLY is set). This is the same as  
 Perl.  
   
 When PCRE_MULTILINE it is set, the "start of line" and "end of line" constructs  
 match immediately following or immediately before any newline in the subject  
 string, respectively, as well as at the very start and end. This is equivalent  
 to Perl's /m option. If there are no "\\n" characters in a subject string, or  
 no occurrences of ^ or $ in a pattern, setting PCRE_MULTILINE has no  
 effect.  
   
   PCRE_UNGREEDY  
   
 This option inverts the "greediness" of the quantifiers so that they are not  
 greedy by default, but become greedy if followed by "?". It is not compatible  
 with Perl. It can also be set by a (?U) option setting within the pattern.  
   
   
 .SH STUDYING A PATTERN  
 When a pattern is going to be used several times, it is worth spending more  
 time analyzing it in order to speed up the time taken for matching. The  
 function \fBpcre_study()\fR takes a pointer to a compiled pattern as its first  
 argument, and returns a pointer to a \fBpcre_extra\fR block (another \fBvoid\fR  
 typedef) containing additional information about the pattern; this can be  
 passed to \fBpcre_exec()\fR. If no additional information is available, NULL  
 is returned.  
   
 The second argument contains option bits. At present, no options are defined  
 for \fBpcre_study()\fR, and this argument should always be zero.  
   
 The third argument for \fBpcre_study()\fR is a pointer to an error message. If  
 studying succeeds (even if no data is returned), the variable it points to is  
 set to NULL. Otherwise it points to a textual error message.  
   
 At present, studying a pattern is useful only for non-anchored patterns that do  
 not have a single fixed starting character. A bitmap of possible starting  
 characters is created.  
   
   
 .SH LOCALE SUPPORT  
 PCRE handles caseless matching, and determines whether characters are letters,  
 digits, or whatever, by reference to a set of tables. The library contains a  
 default set of tables which is created in the default C locale when PCRE is  
 compiled. This is used when the final argument of \fBpcre_compile()\fR is NULL,  
 and is sufficient for many applications.  
   
 An alternative set of tables can, however, be supplied. Such tables are built  
 by calling the \fBpcre_maketables()\fR function, which has no arguments, in the  
 relevant locale. The result can then be passed to \fBpcre_compile()\fR as often  
 as necessary. For example, to build and use tables that are appropriate for the  
 French locale (where accented characters with codes greater than 128 are  
 treated as letters), the following code could be used:  
   
   setlocale(LC_CTYPE, "fr");  
   tables = pcre_maketables();  
   re = pcre_compile(..., tables);  
   
 The tables are built in memory that is obtained via \fBpcre_malloc\fR. The  
 pointer that is passed to \fBpcre_compile\fR is saved with the compiled  
 pattern, and the same tables are used via this pointer by \fBpcre_study()\fR  
 and \fBpcre_exec()\fR. Thus for any single pattern, compilation, studying and  
 matching all happen in the same locale, but different patterns can be compiled  
 in different locales. It is the caller's responsibility to ensure that the  
 memory containing the tables remains available for as long as it is needed.  
   
   
 .SH INFORMATION ABOUT A PATTERN  
 The \fBpcre_info()\fR function returns information about a compiled pattern.  
 Its yield is the number of capturing subpatterns, or one of the following  
 negative numbers:  
   
   PCRE_ERROR_NULL       the argument \fIcode\fR was NULL  
   PCRE_ERROR_BADMAGIC   the "magic number" was not found  
   
 If the \fIoptptr\fR argument is not NULL, a copy of the options with which the  
 pattern was compiled is placed in the integer it points to. These option bits  
 are those specified in the call to \fBpcre_compile()\fR, modified by any  
 top-level option settings within the pattern itself, and with the PCRE_ANCHORED  
 bit set if the form of the pattern implies that it can match only at the start  
 of a subject string.  
   
 If the pattern is not anchored and the \fIfirstcharptr\fR argument is not NULL,  
 it is used to pass back information about the first character of any matched  
 string. If there is a fixed first character, e.g. from a pattern such as  
 (cat|cow|coyote), then it is returned in the integer pointed to by  
 \fIfirstcharptr\fR. Otherwise, if either  
   
 (a) the pattern was compiled with the PCRE_MULTILINE option, and every branch  
 starts with "^", or  
   
 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not set  
 (if it were set, the pattern would be anchored),  
   
 then -1 is returned, indicating that the pattern matches only at the  
 start of a subject string or after any "\\n" within the string. Otherwise -2 is  
 returned.  
   
   
 .SH MATCHING A PATTERN  
 The function \fBpcre_exec()\fR is called to match a subject string against a  
 pre-compiled pattern, which is passed in the \fIcode\fR argument. If the  
 pattern has been studied, the result of the study should be passed in the  
 \fIextra\fR argument. Otherwise this must be NULL.  
   
 The PCRE_ANCHORED option can be passed in the \fIoptions\fR argument, whose  
 unused bits must be zero. However, if a pattern was compiled with  
 PCRE_ANCHORED, or turned out to be anchored by virtue of its contents, it  
 cannot be made unachored at matching time.  
   
 There are also three further options that can be set only at matching time:  
   
   PCRE_NOTBOL  
   
 The first character of the string is not the beginning of a line, so the  
 circumflex metacharacter should not match before it. Setting this without  
 PCRE_MULTILINE (at compile time) causes circumflex never to match.  
   
   PCRE_NOTEOL  
   
 The end of the string is not the end of a line, so the dollar metacharacter  
 should not match it nor (except in multiline mode) a newline immediately before  
 it. Setting this without PCRE_MULTILINE (at compile time) causes dollar never  
 to match.  
   
   PCRE_NOTEMPTY  
   
 An empty string is not considered to be a valid match if this option is set. If  
 there are alternatives in the pattern, they are tried. If all the alternatives  
 match the empty string, the entire match fails. For example, if the pattern  
   
   a?b?  
   
 is applied to a string not beginning with "a" or "b", it matches the empty  
 string at the start of the subject. With PCRE_NOTEMPTY set, this match is not  
 valid, so PCRE searches further into the string for occurrences of "a" or "b".  
   
 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a special case  
 of a pattern match of the empty string within its \fBsplit()\fR function, and  
 when using the /g modifier. It is possible to emulate Perl's behaviour after  
 matching a null string by first trying the match again at the same offset with  
 PCRE_NOTEMPTY set, and then if that fails by advancing the starting offset (see  
 below) and trying an ordinary match again.  
   
 The subject string is passed as a pointer in \fIsubject\fR, a length in  
 \fIlength\fR, and a starting offset in \fIstartoffset\fR. Unlike the pattern  
 string, it may contain binary zero characters. When the starting offset is  
 zero, the search for a match starts at the beginning of the subject, and this  
 is by far the most common case.  
   
 A non-zero starting offset is useful when searching for another match in the  
 same subject by calling \fBpcre_exec()\fR again after a previous success.  
 Setting \fIstartoffset\fR differs from just passing over a shortened string and  
 setting PCRE_NOTBOL in the case of a pattern that begins with any kind of  
 lookbehind. For example, consider the pattern  
   
   \\Biss\\B  
   
 which finds occurrences of "iss" in the middle of words. (\\B matches only if  
 the current position in the subject is not a word boundary.) When applied to  
 the string "Mississipi" the first call to \fBpcre_exec()\fR finds the first  
 occurrence. If \fBpcre_exec()\fR is called again with just the remainder of the  
 subject, namely "issipi", it does not match, because \\B is always false at the  
 start of the subject, which is deemed to be a word boundary. However, if  
 \fBpcre_exec()\fR is passed the entire string again, but with \fIstartoffset\fR  
 set to 4, it finds the second occurrence of "iss" because it is able to look  
 behind the starting point to discover that it is preceded by a letter.  
   
 If a non-zero starting offset is passed when the pattern is anchored, one  
 attempt to match at the given offset is tried. This can only succeed if the  
 pattern does not require the match to be at the start of the subject.  
   
 In general, a pattern matches a certain portion of the subject, and in  
 addition, further substrings from the subject may be picked out by parts of the  
 pattern. Following the usage in Jeffrey Friedl's book, this is called  
 "capturing" in what follows, and the phrase "capturing subpattern" is used for  
 a fragment of a pattern that picks out a substring. PCRE supports several other  
 kinds of parenthesized subpattern that do not cause substrings to be captured.  
   
 Captured substrings are returned to the caller via a vector of integer offsets  
 whose address is passed in \fIovector\fR. The number of elements in the vector  
 is passed in \fIovecsize\fR. The first two-thirds of the vector is used to pass  
 back captured substrings, each substring using a pair of integers. The  
 remaining third of the vector is used as workspace by \fBpcre_exec()\fR while  
 matching capturing subpatterns, and is not available for passing back  
 information. The length passed in \fIovecsize\fR should always be a multiple of  
 three. If it is not, it is rounded down.  
   
 When a match has been successful, information about captured substrings is  
 returned in pairs of integers, starting at the beginning of \fIovector\fR, and  
 continuing up to two-thirds of its length at the most. The first element of a  
 pair is set to the offset of the first character in a substring, and the second  
 is set to the offset of the first character after the end of a substring. The  
 first pair, \fIovector[0]\fR and \fIovector[1]\fR, identify the portion of the  
 subject string matched by the entire pattern. The next pair is used for the  
 first capturing subpattern, and so on. The value returned by \fBpcre_exec()\fR  
 is the number of pairs that have been set. If there are no capturing  
 subpatterns, the return value from a successful match is 1, indicating that  
 just the first pair of offsets has been set.  
   
 Some convenience functions are provided for extracting the captured substrings  
 as separate strings. These are described in the following section.  
   
 It is possible for an capturing subpattern number \fIn+1\fR to match some  
 part of the subject when subpattern \fIn\fR has not been used at all. For  
 example, if the string "abc" is matched against the pattern (a|(z))(bc)  
 subpatterns 1 and 3 are matched, but 2 is not. When this happens, both offset  
 values corresponding to the unused subpattern are set to -1.  
   
 If a capturing subpattern is matched repeatedly, it is the last portion of the  
 string that it matched that gets returned.  
   
 If the vector is too small to hold all the captured substrings, it is used as  
 far as possible (up to two-thirds of its length), and the function returns a  
 value of zero. In particular, if the substring offsets are not of interest,  
 \fBpcre_exec()\fR may be called with \fIovector\fR passed as NULL and  
 \fIovecsize\fR as zero. However, if the pattern contains back references and  
 the \fIovector\fR isn't big enough to remember the related substrings, PCRE has  
 to get additional memory for use during matching. Thus it is usually advisable  
 to supply an \fIovector\fR.  
   
 Note that \fBpcre_info()\fR can be used to find out how many capturing  
 subpatterns there are in a compiled pattern. The smallest size for  
 \fIovector\fR that will allow for \fIn\fR captured substrings in addition to  
 the offsets of the substring matched by the whole pattern is (\fIn\fR+1)*3.  
   
 If \fBpcre_exec()\fR fails, it returns a negative number. The following are  
 defined in the header file:  
   
   PCRE_ERROR_NOMATCH        (-1)  
   
 The subject string did not match the pattern.  
   
   PCRE_ERROR_NULL           (-2)  
   
 Either \fIcode\fR or \fIsubject\fR was passed as NULL, or \fIovector\fR was  
 NULL and \fIovecsize\fR was not zero.  
   
   PCRE_ERROR_BADOPTION      (-3)  
   
 An unrecognized bit was set in the \fIoptions\fR argument.  
   
   PCRE_ERROR_BADMAGIC       (-4)  
   
 PCRE stores a 4-byte "magic number" at the start of the compiled code, to catch  
 the case when it is passed a junk pointer. This is the error it gives when the  
 magic number isn't present.  
   
   PCRE_ERROR_UNKNOWN_NODE   (-5)  
   
 While running the pattern match, an unknown item was encountered in the  
 compiled pattern. This error could be caused by a bug in PCRE or by overwriting  
 of the compiled pattern.  
   
   PCRE_ERROR_NOMEMORY       (-6)  
   
 If a pattern contains back references, but the \fIovector\fR that is passed to  
 \fBpcre_exec()\fR is not big enough to remember the referenced substrings, PCRE  
 gets a block of memory at the start of matching to use for this purpose. If the  
 call via \fBpcre_malloc()\fR fails, this error is given. The memory is freed at  
 the end of matching.  
   
   
 .SH EXTRACTING CAPTURED SUBSTRINGS  
 Captured substrings can be accessed directly by using the offsets returned by  
 \fBpcre_exec()\fR in \fIovector\fR. For convenience, the functions  
 \fBpcre_copy_substring()\fR, \fBpcre_get_substring()\fR, and  
 \fBpcre_get_substring_list()\fR are provided for extracting captured substrings  
 as new, separate, zero-terminated strings. A substring that contains a binary  
 zero is correctly extracted and has a further zero added on the end, but the  
 result does not, of course, function as a C string.  
   
 The first three arguments are the same for all three functions: \fIsubject\fR  
 is the subject string which has just been successfully matched, \fIovector\fR  
 is a pointer to the vector of integer offsets that was passed to  
 \fBpcre_exec()\fR, and \fIstringcount\fR is the number of substrings that  
 were captured by the match, including the substring that matched the entire  
 regular expression. This is the value returned by \fBpcre_exec\fR if it  
 is greater than zero. If \fBpcre_exec()\fR returned zero, indicating that it  
 ran out of space in \fIovector\fR, then the value passed as  
 \fIstringcount\fR should be the size of the vector divided by three.  
   
 The functions \fBpcre_copy_substring()\fR and \fBpcre_get_substring()\fR  
 extract a single substring, whose number is given as \fIstringnumber\fR. A  
 value of zero extracts the substring that matched the entire pattern, while  
 higher values extract the captured substrings. For \fBpcre_copy_substring()\fR,  
 the string is placed in \fIbuffer\fR, whose length is given by  
 \fIbuffersize\fR, while for \fBpcre_get_substring()\fR a new block of store is  
 obtained via \fBpcre_malloc\fR, and its address is returned via  
 \fIstringptr\fR. The yield of the function is the length of the string, not  
 including the terminating zero, or one of  
   
   PCRE_ERROR_NOMEMORY       (-6)  
   
 The buffer was too small for \fBpcre_copy_substring()\fR, or the attempt to get  
 memory failed for \fBpcre_get_substring()\fR.  
   
   PCRE_ERROR_NOSUBSTRING    (-7)  
   
 There is no substring whose number is \fIstringnumber\fR.  
   
 The \fBpcre_get_substring_list()\fR function extracts all available substrings  
 and builds a list of pointers to them. All this is done in a single block of  
 memory which is obtained via \fBpcre_malloc\fR. The address of the memory block  
 is returned via \fIlistptr\fR, which is also the start of the list of string  
 pointers. The end of the list is marked by a NULL pointer. The yield of the  
 function is zero if all went well, or  
   
   PCRE_ERROR_NOMEMORY       (-6)  
   
 if the attempt to get the memory block failed.  
   
 When any of these functions encounter a substring that is unset, which can  
 happen when capturing subpattern number \fIn+1\fR matches some part of the  
 subject, but subpattern \fIn\fR has not been used at all, they return an empty  
 string. This can be distinguished from a genuine zero-length substring by  
 inspecting the appropriate offset in \fIovector\fR, which is negative for unset  
 substrings.  
   
63    
64    In addition, in the "man" and HTML formats, there is a short page for each
65    library function, listing its arguments and results.
66    
67  .SH LIMITATIONS  .SH LIMITATIONS
68    .rs
69    .sp
70  There are some size limitations in PCRE but it is hoped that they will never in  There are some size limitations in PCRE but it is hoped that they will never in
71  practice be relevant.  practice be relevant.
72  The maximum length of a compiled pattern is 65539 (sic) bytes.  
73    The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE is
74    compiled with the default internal linkage size of 2. If you want to process
75    regular expressions that are truly enormous, you can compile PCRE with an
76    internal linkage size of 3 or 4 (see the \fBREADME\fR file in the source
77    distribution and the
78    .\" HREF
79    \fBpcrebuild\fR
80    .\"
81    documentation for details). If these cases the limit is substantially larger.
82    However, the speed of execution will be slower.
83    
84  All values in repeating quantifiers must be less than 65536.  All values in repeating quantifiers must be less than 65536.
85  The maximum number of capturing subpatterns is 99.  The maximum number of capturing subpatterns is 65535.
86  The maximum number of all parenthesized subpatterns, including capturing  
87    There is no limit to the number of non-capturing subpatterns, but the maximum
88    depth of nesting of all kinds of parenthesized subpattern, including capturing
89  subpatterns, assertions, and other types of subpattern, is 200.  subpatterns, assertions, and other types of subpattern, is 200.
90    
91  The maximum length of a subject string is the largest positive number that an  The maximum length of a subject string is the largest positive number that an
# Line 528  integer variable can hold. However, PCRE Line 93  integer variable can hold. However, PCRE
93  and indefinite repetition. This means that the available stack space may limit  and indefinite repetition. This means that the available stack space may limit
94  the size of a subject string that can be processed by certain patterns.  the size of a subject string that can be processed by certain patterns.
95    
96    .\" HTML <a name="utf8support"></a>
97    .SH UTF-8 SUPPORT
98    .rs
99    .sp
100    Starting at release 3.3, PCRE has had some support for character strings
101    encoded in the UTF-8 format. For release 4.0 this has been greatly extended to
102    cover most common requirements.
103    
104    In order process UTF-8 strings, you must build PCRE to include UTF-8 support in
105    the code, and, in addition, you must call
106    .\" HREF
107    \fBpcre_compile()\fR
108    .\"
109    with the PCRE_UTF8 option flag. When you do this, both the pattern and any
110    subject strings that are matched against it are treated as UTF-8 strings
111    instead of just strings of bytes.
112    
113    If you compile PCRE with UTF-8 support, but do not use it at run time, the
114    library will be a bit bigger, but the additional run time overhead is limited
115    to testing the PCRE_UTF8 flag in several places, so should not be very large.
116    
117    The following comments apply when PCRE is running in UTF-8 mode:
118    
119    1. PCRE assumes that the strings it is given contain valid UTF-8 codes. It does
120    not diagnose invalid UTF-8 strings. If you pass invalid UTF-8 strings to PCRE,
121    the results are undefined.
122    
123    2. In a pattern, the escape sequence \\x{...}, where the contents of the braces
124    is a string of hexadecimal digits, is interpreted as a UTF-8 character whose
125    code number is the given hexadecimal number, for example: \\x{1234}. If a
126    non-hexadecimal digit appears between the braces, the item is not recognized.
127    This escape sequence can be used either as a literal, or within a character
128    class.
129    
130    3. The original hexadecimal escape sequence, \\xhh, matches a two-byte UTF-8
131    character if the value is greater than 127.
132    
133    4. Repeat quantifiers apply to complete UTF-8 characters, not to individual
134    bytes, for example: \\x{100}{3}.
135    
136    5. The dot metacharacter matches one UTF-8 character instead of a single byte.
137    
138    6. The escape sequence \\C can be used to match a single byte in UTF-8 mode,
139    but its use can lead to some strange effects.
140    
141    7. The character escapes \\b, \\B, \\d, \\D, \\s, \\S, \\w, and \\W correctly
142    test characters of any code value, but the characters that PCRE recognizes as
143    digits, spaces, or word characters remain the same set as before, all with
144    values less than 256.
145    
146    8. Case-insensitive matching applies only to characters whose values are less
147    than 256. PCRE does not support the notion of "case" for higher-valued
148    characters.
149    
150  .SH DIFFERENCES FROM PERL  9. PCRE does not support the use of Unicode tables and properties or the Perl
151  The differences described here are with respect to Perl 5.005.  escapes \\p, \\P, and \\X.
   
 1. By default, a whitespace character is any character that the C library  
 function \fBisspace()\fR recognizes, though it is possible to compile PCRE with  
 alternative character type tables. Normally \fBisspace()\fR matches space,  
 formfeed, newline, carriage return, horizontal tab, and vertical tab. Perl 5  
 no longer includes vertical tab in its set of whitespace characters. The \\v  
 escape that was in the Perl documentation for a long time was never in fact  
 recognized. However, the character itself was treated as whitespace at least  
 up to 5.002. In 5.004 and 5.005 it does not match \\s.  
   
 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl permits  
 them, but they do not mean what you might think. For example, (?!a){3} does  
 not assert that the next three characters are not "a". It just asserts that the  
 next character is not "a" three times.  
   
 3. Capturing subpatterns that occur inside negative lookahead assertions are  
 counted, but their entries in the offsets vector are never set. Perl sets its  
 numerical variables from any such patterns that are matched before the  
 assertion fails to match something (thereby succeeding), but only if the  
 negative lookahead assertion contains just one branch.  
   
 4. Though binary zero characters are supported in the subject string, they are  
 not allowed in a pattern string because it is passed as a normal C string,  
 terminated by zero. The escape sequence "\\0" can be used in the pattern to  
 represent a binary zero.  
   
 5. The following Perl escape sequences are not supported: \\l, \\u, \\L, \\U,  
 \\E, \\Q. In fact these are implemented by Perl's general string-handling and  
 are not part of its pattern matching engine.  
   
 6. The Perl \\G assertion is not supported as it is not relevant to single  
 pattern matches.  
   
 7. Fairly obviously, PCRE does not support the (?{code}) construction.  
   
 8. There are at the time of writing some oddities in Perl 5.005_02 concerned  
 with the settings of captured strings when part of a pattern is repeated. For  
 example, matching "aba" against the pattern /^(a(b)?)+$/ sets $2 to the value  
 "b", but matching "aabbaa" against /^(aa(bb)?)+$/ leaves $2 unset. However, if  
 the pattern is changed to /^(aa(b(b))?)+$/ then $2 (and $3) get set.  
   
 In Perl 5.004 $2 is set in both cases, and that is also true of PCRE. If in the  
 future Perl changes to a consistent state that is different, PCRE may change to  
 follow.  
   
 9. Another as yet unresolved discrepancy is that in Perl 5.005_02 the pattern  
 /^(a)?(?(1)a|b)+$/ matches the string "a", whereas in PCRE it does not.  
 However, in both Perl and PCRE /^(a)?a/ matched against "a" leaves $1 unset.  
   
 10. PCRE provides some extensions to the Perl regular expression facilities:  
   
 (a) Although lookbehind assertions must match fixed length strings, each  
 alternative branch of a lookbehind assertion can match a different length of  
 string. Perl 5.005 requires them all to have the same length.  
   
 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $ meta-  
 character matches only at the very end of the string.  
   
 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no special  
 meaning is faulted.  
   
 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quantifiers is  
 inverted, that is, by default they are not greedy, but if followed by a  
 question mark they are.  
   
 (e) PCRE_ANCHORED can be used to force a pattern to be tried only at the start  
 of the subject.  
   
 (f) The PCRE_NOTBOL, PCRE_NOTEOL, and PCRE_NOTEMPTY options for  
 \fBpcre_exec()\fR have no Perl equivalents.  
   
   
 .SH REGULAR EXPRESSION DETAILS  
 The syntax and semantics of the regular expressions supported by PCRE are  
 described below. Regular expressions are also described in the Perl  
 documentation and in a number of other books, some of which have copious  
 examples. Jeffrey Friedl's "Mastering Regular Expressions", published by  
 O'Reilly (ISBN 1-56592-257-3), covers them in great detail. The description  
 here is intended as reference documentation.  
   
 A regular expression is a pattern that is matched against a subject string from  
 left to right. Most characters stand for themselves in a pattern, and match the  
 corresponding characters in the subject. As a trivial example, the pattern  
   
   The quick brown fox  
   
 matches a portion of a subject string that is identical to itself. The power of  
 regular expressions comes from the ability to include alternatives and  
 repetitions in the pattern. These are encoded in the pattern by the use of  
 \fImeta-characters\fR, which do not stand for themselves but instead are  
 interpreted in some special way.  
   
 There are two different sets of meta-characters: those that are recognized  
 anywhere in the pattern except within square brackets, and those that are  
 recognized in square brackets. Outside square brackets, the meta-characters are  
 as follows:  
   
   \\      general escape character with several uses  
   ^      assert start of subject (or line, in multiline mode)  
   $      assert end of subject (or line, in multiline mode)  
   .      match any character except newline (by default)  
   [      start character class definition  
   |      start of alternative branch  
   (      start subpattern  
   )      end subpattern  
   ?      extends the meaning of (  
          also 0 or 1 quantifier  
          also quantifier minimizer  
   *      0 or more quantifier  
   +      1 or more quantifier  
   {      start min/max quantifier  
   
 Part of a pattern that is in square brackets is called a "character class". In  
 a character class the only meta-characters are:  
   
   \\      general escape character  
   ^      negate the class, but only if the first character  
   -      indicates character range  
   ]      terminates the character class  
   
 The following sections describe the use of each of the meta-characters.  
   
   
 .SH BACKSLASH  
 The backslash character has several uses. Firstly, if it is followed by a  
 non-alphameric character, it takes away any special meaning that character may  
 have. This use of backslash as an escape character applies both inside and  
 outside character classes.  
   
 For example, if you want to match a "*" character, you write "\\*" in the  
 pattern. This applies whether or not the following character would otherwise be  
 interpreted as a meta-character, so it is always safe to precede a  
 non-alphameric with "\\" to specify that it stands for itself. In particular,  
 if you want to match a backslash, you write "\\\\".  
   
 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the  
 pattern (other than in a character class) and characters between a "#" outside  
 a character class and the next newline character are ignored. An escaping  
 backslash can be used to include a whitespace or "#" character as part of the  
 pattern.  
   
 A second use of backslash provides a way of encoding non-printing characters  
 in patterns in a visible manner. There is no restriction on the appearance of  
 non-printing characters, apart from the binary zero that terminates a pattern,  
 but when a pattern is being prepared by text editing, it is usually easier to  
 use one of the following escape sequences than the binary character it  
 represents:  
   
   \\a     alarm, that is, the BEL character (hex 07)  
   \\cx    "control-x", where x is any character  
   \\e     escape (hex 1B)  
   \\f     formfeed (hex 0C)  
   \\n     newline (hex 0A)  
   \\r     carriage return (hex 0D)  
   \\t     tab (hex 09)  
   \\xhh   character with hex code hh  
   \\ddd   character with octal code ddd, or backreference  
   
 The precise effect of "\\cx" is as follows: if "x" is a lower case letter, it  
 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.  
 Thus "\\cz" becomes hex 1A, but "\\c{" becomes hex 3B, while "\\c;" becomes hex  
 7B.  
   
 After "\\x", up to two hexadecimal digits are read (letters can be in upper or  
 lower case).  
   
 After "\\0" up to two further octal digits are read. In both cases, if there  
 are fewer than two digits, just those that are present are used. Thus the  
 sequence "\\0\\x\\07" specifies two binary zeros followed by a BEL character.  
 Make sure you supply two digits after the initial zero if the character that  
 follows is itself an octal digit.  
   
 The handling of a backslash followed by a digit other than 0 is complicated.  
 Outside a character class, PCRE reads it and any following digits as a decimal  
 number. If the number is less than 10, or if there have been at least that many  
 previous capturing left parentheses in the expression, the entire sequence is  
 taken as a \fIback reference\fR. A description of how this works is given  
 later, following the discussion of parenthesized subpatterns.  
   
 Inside a character class, or if the decimal number is greater than 9 and there  
 have not been that many capturing subpatterns, PCRE re-reads up to three octal  
 digits following the backslash, and generates a single byte from the least  
 significant 8 bits of the value. Any subsequent digits stand for themselves.  
 For example:  
   
   \\040   is another way of writing a space  
   \\40    is the same, provided there are fewer than 40  
             previous capturing subpatterns  
   \\7     is always a back reference  
   \\11    might be a back reference, or another way of  
             writing a tab  
   \\011   is always a tab  
   \\0113  is a tab followed by the character "3"  
   \\113   is the character with octal code 113 (since there  
             can be no more than 99 back references)  
   \\377   is a byte consisting entirely of 1 bits  
   \\81    is either a back reference, or a binary zero  
             followed by the two characters "8" and "1"  
   
 Note that octal values of 100 or greater must not be introduced by a leading  
 zero, because no more than three octal digits are ever read.  
   
 All the sequences that define a single byte value can be used both inside and  
 outside character classes. In addition, inside a character class, the sequence  
 "\\b" is interpreted as the backspace character (hex 08). Outside a character  
 class it has a different meaning (see below).  
   
 The third use of backslash is for specifying generic character types:  
   
   \\d     any decimal digit  
   \\D     any character that is not a decimal digit  
   \\s     any whitespace character  
   \\S     any character that is not a whitespace character  
   \\w     any "word" character  
   \\W     any "non-word" character  
   
 Each pair of escape sequences partitions the complete set of characters into  
 two disjoint sets. Any given character matches one, and only one, of each pair.  
   
 A "word" character is any letter or digit or the underscore character, that is,  
 any character which can be part of a Perl "word". The definition of letters and  
 digits is controlled by PCRE's character tables, and may vary if locale-  
 specific matching is taking place (see "Locale support" above). For example, in  
 the "fr" (French) locale, some character codes greater than 128 are used for  
 accented letters, and these are matched by \\w.  
   
 These character type sequences can appear both inside and outside character  
 classes. They each match one character of the appropriate type. If the current  
 matching point is at the end of the subject string, all of them fail, since  
 there is no character to match.  
   
 The fourth use of backslash is for certain simple assertions. An assertion  
 specifies a condition that has to be met at a particular point in a match,  
 without consuming any characters from the subject string. The use of  
 subpatterns for more complicated assertions is described below. The backslashed  
 assertions are  
   
   \\b     word boundary  
   \\B     not a word boundary  
   \\A     start of subject (independent of multiline mode)  
   \\Z     end of subject or newline at end (independent of multiline mode)  
   \\z     end of subject (independent of multiline mode)  
   
 These assertions may not appear in character classes (but note that "\\b" has a  
 different meaning, namely the backspace character, inside a character class).  
   
 A word boundary is a position in the subject string where the current character  
 and the previous character do not both match \\w or \\W (i.e. one matches  
 \\w and the other matches \\W), or the start or end of the string if the  
 first or last character matches \\w, respectively.  
   
 The \\A, \\Z, and \\z assertions differ from the traditional circumflex and  
 dollar (described below) in that they only ever match at the very start and end  
 of the subject string, whatever options are set. They are not affected by the  
 PCRE_NOTBOL or PCRE_NOTEOL options. If the \fIstartoffset\fR argument of  
 \fBpcre_exec()\fR is non-zero, \\A can never match. The difference between \\Z  
 and \\z is that \\Z matches before a newline that is the last character of the  
 string as well as at the end of the string, whereas \\z matches only at the  
 end.  
   
   
 .SH CIRCUMFLEX AND DOLLAR  
 Outside a character class, in the default matching mode, the circumflex  
 character is an assertion which is true only if the current matching point is  
 at the start of the subject string. If the \fIstartoffset\fR argument of  
 \fBpcre_exec()\fR is non-zero, circumflex can never match. Inside a character  
 class, circumflex has an entirely different meaning (see below).  
   
 Circumflex need not be the first character of the pattern if a number of  
 alternatives are involved, but it should be the first thing in each alternative  
 in which it appears if the pattern is ever to match that branch. If all  
 possible alternatives start with a circumflex, that is, if the pattern is  
 constrained to match only at the start of the subject, it is said to be an  
 "anchored" pattern. (There are also other constructs that can cause a pattern  
 to be anchored.)  
   
 A dollar character is an assertion which is true only if the current matching  
 point is at the end of the subject string, or immediately before a newline  
 character that is the last character in the string (by default). Dollar need  
 not be the last character of the pattern if a number of alternatives are  
 involved, but it should be the last item in any branch in which it appears.  
 Dollar has no special meaning in a character class.  
   
 The meaning of dollar can be changed so that it matches only at the very end of  
 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile or matching  
 time. This does not affect the \\Z assertion.  
   
 The meanings of the circumflex and dollar characters are changed if the  
 PCRE_MULTILINE option is set. When this is the case, they match immediately  
 after and immediately before an internal "\\n" character, respectively, in  
 addition to matching at the start and end of the subject string. For example,  
 the pattern /^abc$/ matches the subject string "def\\nabc" in multiline mode,  
 but not otherwise. Consequently, patterns that are anchored in single line mode  
 because all branches start with "^" are not anchored in multiline mode, and a  
 match for circumflex is possible when the \fIstartoffset\fR argument of  
 \fBpcre_exec()\fR is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if  
 PCRE_MULTILINE is set.  
   
 Note that the sequences \\A, \\Z, and \\z can be used to match the start and  
 end of the subject in both modes, and if all branches of a pattern start with  
 \\A is it always anchored, whether PCRE_MULTILINE is set or not.  
   
   
 .SH FULL STOP (PERIOD, DOT)  
 Outside a character class, a dot in the pattern matches any one character in  
 the subject, including a non-printing character, but not (by default) newline.  
 If the PCRE_DOTALL option is set, then dots match newlines as well. The  
 handling of dot is entirely independent of the handling of circumflex and  
 dollar, the only relationship being that they both involve newline characters.  
 Dot has no special meaning in a character class.  
   
   
 .SH SQUARE BRACKETS  
 An opening square bracket introduces a character class, terminated by a closing  
 square bracket. A closing square bracket on its own is not special. If a  
 closing square bracket is required as a member of the class, it should be the  
 first data character in the class (after an initial circumflex, if present) or  
 escaped with a backslash.  
   
 A character class matches a single character in the subject; the character must  
 be in the set of characters defined by the class, unless the first character in  
 the class is a circumflex, in which case the subject character must not be in  
 the set defined by the class. If a circumflex is actually required as a member  
 of the class, ensure it is not the first character, or escape it with a  
 backslash.  
   
 For example, the character class [aeiou] matches any lower case vowel, while  
 [^aeiou] matches any character that is not a lower case vowel. Note that a  
 circumflex is just a convenient notation for specifying the characters which  
 are in the class by enumerating those that are not. It is not an assertion: it  
 still consumes a character from the subject string, and fails if the current  
 pointer is at the end of the string.  
   
 When caseless matching is set, any letters in a class represent both their  
 upper case and lower case versions, so for example, a caseless [aeiou] matches  
 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a  
 caseful version would.  
   
 The newline character is never treated in any special way in character classes,  
 whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE options is. A class  
 such as [^a] will always match a newline.  
   
 The minus (hyphen) character can be used to specify a range of characters in a  
 character class. For example, [d-m] matches any letter between d and m,  
 inclusive. If a minus character is required in a class, it must be escaped with  
 a backslash or appear in a position where it cannot be interpreted as  
 indicating a range, typically as the first or last character in the class.  
   
 It is not possible to have the literal character "]" as the end character of a  
 range. A pattern such as [W-]46] is interpreted as a class of two characters  
 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or  
 "-46]". However, if the "]" is escaped with a backslash it is interpreted as  
 the end of range, so [W-\\]46] is interpreted as a single class containing a  
 range followed by two separate characters. The octal or hexadecimal  
 representation of "]" can also be used to end a range.  
   
 Ranges operate in ASCII collating sequence. They can also be used for  
 characters specified numerically, for example [\\000-\\037]. If a range that  
 includes letters is used when caseless matching is set, it matches the letters  
 in either case. For example, [W-c] is equivalent to [][\\^_`wxyzabc], matched  
 caselessly, and if character tables for the "fr" locale are in use,  
 [\\xc8-\\xcb] matches accented E characters in both cases.  
   
 The character types \\d, \\D, \\s, \\S, \\w, and \\W may also appear in a  
 character class, and add the characters that they match to the class. For  
 example, [\\dABCDEF] matches any hexadecimal digit. A circumflex can  
 conveniently be used with the upper case character types to specify a more  
 restricted set of characters than the matching lower case type. For example,  
 the class [^\\W_] matches any letter or digit, but not underscore.  
   
 All non-alphameric characters other than \\, -, ^ (at the start) and the  
 terminating ] are non-special in character classes, but it does no harm if they  
 are escaped.  
   
   
 .SH VERTICAL BAR  
 Vertical bar characters are used to separate alternative patterns. For example,  
 the pattern  
   
   gilbert|sullivan  
   
 matches either "gilbert" or "sullivan". Any number of alternatives may appear,  
 and an empty alternative is permitted (matching the empty string).  
 The matching process tries each alternative in turn, from left to right,  
 and the first one that succeeds is used. If the alternatives are within a  
 subpattern (defined below), "succeeds" means matching the rest of the main  
 pattern as well as the alternative in the subpattern.  
   
   
 .SH INTERNAL OPTION SETTING  
 The settings of PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and PCRE_EXTENDED  
 can be changed from within the pattern by a sequence of Perl option letters  
 enclosed between "(?" and ")". The option letters are  
   
   i  for PCRE_CASELESS  
   m  for PCRE_MULTILINE  
   s  for PCRE_DOTALL  
   x  for PCRE_EXTENDED  
   
 For example, (?im) sets caseless, multiline matching. It is also possible to  
 unset these options by preceding the letter with a hyphen, and a combined  
 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and  
 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also  
 permitted. If a letter appears both before and after the hyphen, the option is  
 unset.  
   
 The scope of these option changes depends on where in the pattern the setting  
 occurs. For settings that are outside any subpattern (defined below), the  
 effect is the same as if the options were set or unset at the start of  
 matching. The following patterns all behave in exactly the same way:  
   
   (?i)abc  
   a(?i)bc  
   ab(?i)c  
   abc(?i)  
   
 which in turn is the same as compiling the pattern abc with PCRE_CASELESS set.  
 In other words, such "top level" settings apply to the whole pattern (unless  
 there are other changes inside subpatterns). If there is more than one setting  
 of the same option at top level, the rightmost setting is used.  
   
 If an option change occurs inside a subpattern, the effect is different. This  
 is a change of behaviour in Perl 5.005. An option change inside a subpattern  
 affects only that part of the subpattern that follows it, so  
   
   (a(?i)b)c  
   
 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).  
 By this means, options can be made to have different settings in different  
 parts of the pattern. Any changes made in one alternative do carry on  
 into subsequent branches within the same subpattern. For example,  
   
   (a(?i)b|c)  
   
 matches "ab", "aB", "c", and "C", even though when matching "C" the first  
 branch is abandoned before the option setting. This is because the effects of  
 option settings happen at compile time. There would be some very weird  
 behaviour otherwise.  
   
 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed in the  
 same way as the Perl-compatible options by using the characters U and X  
 respectively. The (?X) flag setting is special in that it must always occur  
 earlier in the pattern than any of the additional features it turns on, even  
 when it is at top level. It is best put at the start.  
   
   
 .SH SUBPATTERNS  
 Subpatterns are delimited by parentheses (round brackets), which can be nested.  
 Marking part of a pattern as a subpattern does two things:  
   
 1. It localizes a set of alternatives. For example, the pattern  
   
   cat(aract|erpillar|)  
   
 matches one of the words "cat", "cataract", or "caterpillar". Without the  
 parentheses, it would match "cataract", "erpillar" or the empty string.  
   
 2. It sets up the subpattern as a capturing subpattern (as defined above).  
 When the whole pattern matches, that portion of the subject string that matched  
 the subpattern is passed back to the caller via the \fIovector\fR argument of  
 \fBpcre_exec()\fR. Opening parentheses are counted from left to right (starting  
 from 1) to obtain the numbers of the capturing subpatterns.  
   
 For example, if the string "the red king" is matched against the pattern  
   
   the ((red|white) (king|queen))  
   
 the captured substrings are "red king", "red", and "king", and are numbered 1,  
 2, and 3.  
   
 The fact that plain parentheses fulfil two functions is not always helpful.  
 There are often times when a grouping subpattern is required without a  
 capturing requirement. If an opening parenthesis is followed by "?:", the  
 subpattern does not do any capturing, and is not counted when computing the  
 number of any subsequent capturing subpatterns. For example, if the string "the  
 white queen" is matched against the pattern  
   
   the ((?:red|white) (king|queen))  
   
 the captured substrings are "white queen" and "queen", and are numbered 1 and  
 2. The maximum number of captured substrings is 99, and the maximum number of  
 all subpatterns, both capturing and non-capturing, is 200.  
   
 As a convenient shorthand, if any option settings are required at the start of  
 a non-capturing subpattern, the option letters may appear between the "?" and  
 the ":". Thus the two patterns  
   
   (?i:saturday|sunday)  
   (?:(?i)saturday|sunday)  
   
 match exactly the same set of strings. Because alternative branches are tried  
 from left to right, and options are not reset until the end of the subpattern  
 is reached, an option setting in one branch does affect subsequent branches, so  
 the above patterns match "SUNDAY" as well as "Saturday".  
   
   
 .SH REPETITION  
 Repetition is specified by quantifiers, which can follow any of the following  
 items:  
   
   a single character, possibly escaped  
   the . metacharacter  
   a character class  
   a back reference (see next section)  
   a parenthesized subpattern (unless it is an assertion - see below)  
   
 The general repetition quantifier specifies a minimum and maximum number of  
 permitted matches, by giving the two numbers in curly brackets (braces),  
 separated by a comma. The numbers must be less than 65536, and the first must  
 be less than or equal to the second. For example:  
   
   z{2,4}  
   
 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special  
 character. If the second number is omitted, but the comma is present, there is  
 no upper limit; if the second number and the comma are both omitted, the  
 quantifier specifies an exact number of required matches. Thus  
   
   [aeiou]{3,}  
   
 matches at least 3 successive vowels, but may match many more, while  
   
   \\d{8}  
   
 matches exactly 8 digits. An opening curly bracket that appears in a position  
 where a quantifier is not allowed, or one that does not match the syntax of a  
 quantifier, is taken as a literal character. For example, {,6} is not a  
 quantifier, but a literal string of four characters.  
   
 The quantifier {0} is permitted, causing the expression to behave as if the  
 previous item and the quantifier were not present.  
   
 For convenience (and historical compatibility) the three most common  
 quantifiers have single-character abbreviations:  
   
   *    is equivalent to {0,}  
   +    is equivalent to {1,}  
   ?    is equivalent to {0,1}  
   
 It is possible to construct infinite loops by following a subpattern that can  
 match no characters with a quantifier that has no upper limit, for example:  
   
   (a?)*  
   
 Earlier versions of Perl and PCRE used to give an error at compile time for  
 such patterns. However, because there are cases where this can be useful, such  
 patterns are now accepted, but if any repetition of the subpattern does in fact  
 match no characters, the loop is forcibly broken.  
   
 By default, the quantifiers are "greedy", that is, they match as much as  
 possible (up to the maximum number of permitted times), without causing the  
 rest of the pattern to fail. The classic example of where this gives problems  
 is in trying to match comments in C programs. These appear between the  
 sequences /* and */ and within the sequence, individual * and / characters may  
 appear. An attempt to match C comments by applying the pattern  
   
   /\\*.*\\*/  
   
 to the string  
   
   /* first command */  not comment  /* second comment */  
   
 fails, because it matches the entire string due to the greediness of the .*  
 item.  
   
 However, if a quantifier is followed by a question mark, then it ceases to be  
 greedy, and instead matches the minimum number of times possible, so the  
 pattern  
   
   /\\*.*?\\*/  
   
 does the right thing with the C comments. The meaning of the various  
 quantifiers is not otherwise changed, just the preferred number of matches.  
 Do not confuse this use of question mark with its use as a quantifier in its  
 own right. Because it has two uses, it can sometimes appear doubled, as in  
   
   \\d??\\d  
   
 which matches one digit by preference, but can match two if that is the only  
 way the rest of the pattern matches.  
   
 If the PCRE_UNGREEDY option is set (an option which is not available in Perl)  
 then the quantifiers are not greedy by default, but individual ones can be made  
 greedy by following them with a question mark. In other words, it inverts the  
 default behaviour.  
   
 When a parenthesized subpattern is quantified with a minimum repeat count that  
 is greater than 1 or with a limited maximum, more store is required for the  
 compiled pattern, in proportion to the size of the minimum or maximum.  
   
 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent  
 to Perl's /s) is set, thus allowing the . to match newlines, then the pattern  
 is implicitly anchored, because whatever follows will be tried against every  
 character position in the subject string, so there is no point in retrying the  
 overall match at any position after the first. PCRE treats such a pattern as  
 though it were preceded by \\A. In cases where it is known that the subject  
 string contains no newlines, it is worth setting PCRE_DOTALL when the pattern  
 begins with .* in order to obtain this optimization, or alternatively using ^  
 to indicate anchoring explicitly.  
   
 When a capturing subpattern is repeated, the value captured is the substring  
 that matched the final iteration. For example, after  
   
   (tweedle[dume]{3}\\s*)+  
   
 has matched "tweedledum tweedledee" the value of the captured substring is  
 "tweedledee". However, if there are nested capturing subpatterns, the  
 corresponding captured values may have been set in previous iterations. For  
 example, after  
   
   /(a|(b))+/  
   
 matches "aba" the value of the second captured substring is "b".  
   
   
 .SH BACK REFERENCES  
 Outside a character class, a backslash followed by a digit greater than 0 (and  
 possibly further digits) is a back reference to a capturing subpattern earlier  
 (i.e. to its left) in the pattern, provided there have been that many previous  
 capturing left parentheses.  
   
 However, if the decimal number following the backslash is less than 10, it is  
 always taken as a back reference, and causes an error only if there are not  
 that many capturing left parentheses in the entire pattern. In other words, the  
 parentheses that are referenced need not be to the left of the reference for  
 numbers less than 10. See the section entitled "Backslash" above for further  
 details of the handling of digits following a backslash.  
   
 A back reference matches whatever actually matched the capturing subpattern in  
 the current subject string, rather than anything matching the subpattern  
 itself. So the pattern  
   
   (sens|respons)e and \\1ibility  
   
 matches "sense and sensibility" and "response and responsibility", but not  
 "sense and responsibility". If caseful matching is in force at the time of the  
 back reference, then the case of letters is relevant. For example,  
   
   ((?i)rah)\\s+\\1  
   
 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original  
 capturing subpattern is matched caselessly.  
   
 There may be more than one back reference to the same subpattern. If a  
 subpattern has not actually been used in a particular match, then any back  
 references to it always fail. For example, the pattern  
   
   (a|(bc))\\2  
   
 always fails if it starts to match "a" rather than "bc". Because there may be  
 up to 99 back references, all digits following the backslash are taken  
 as part of a potential back reference number. If the pattern continues with a  
 digit character, then some delimiter must be used to terminate the back  
 reference. If the PCRE_EXTENDED option is set, this can be whitespace.  
 Otherwise an empty comment can be used.  
   
 A back reference that occurs inside the parentheses to which it refers fails  
 when the subpattern is first used, so, for example, (a\\1) never matches.  
 However, such references can be useful inside repeated subpatterns. For  
 example, the pattern  
   
   (a|b\\1)+  
   
 matches any number of "a"s and also "aba", "ababaa" etc. At each iteration of  
 the subpattern, the back reference matches the character string corresponding  
 to the previous iteration. In order for this to work, the pattern must be such  
 that the first iteration does not need to match the back reference. This can be  
 done using alternation, as in the example above, or by a quantifier with a  
 minimum of zero.  
   
   
 .SH ASSERTIONS  
 An assertion is a test on the characters following or preceding the current  
 matching point that does not actually consume any characters. The simple  
 assertions coded as \\b, \\B, \\A, \\Z, \\z, ^ and $ are described above. More  
 complicated assertions are coded as subpatterns. There are two kinds: those  
 that look ahead of the current position in the subject string, and those that  
 look behind it.  
   
 An assertion subpattern is matched in the normal way, except that it does not  
 cause the current matching position to be changed. Lookahead assertions start  
 with (?= for positive assertions and (?! for negative assertions. For example,  
   
   \\w+(?=;)  
   
 matches a word followed by a semicolon, but does not include the semicolon in  
 the match, and  
   
   foo(?!bar)  
   
 matches any occurrence of "foo" that is not followed by "bar". Note that the  
 apparently similar pattern  
   
   (?!foo)bar  
   
 does not find an occurrence of "bar" that is preceded by something other than  
 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion  
 (?!foo) is always true when the next three characters are "bar". A  
 lookbehind assertion is needed to achieve this effect.  
   
 Lookbehind assertions start with (?<= for positive assertions and (?<! for  
 negative assertions. For example,  
   
   (?<!foo)bar  
   
 does find an occurrence of "bar" that is not preceded by "foo". The contents of  
 a lookbehind assertion are restricted such that all the strings it matches must  
 have a fixed length. However, if there are several alternatives, they do not  
 all have to have the same fixed length. Thus  
   
   (?<=bullock|donkey)  
   
 is permitted, but  
   
   (?<!dogs?|cats?)  
   
 causes an error at compile time. Branches that match different length strings  
 are permitted only at the top level of a lookbehind assertion. This is an  
 extension compared with Perl 5.005, which requires all branches to match the  
 same length of string. An assertion such as  
   
   (?<=ab(c|de))  
   
 is not permitted, because its single top-level branch can match two different  
 lengths, but it is acceptable if rewritten to use two top-level branches:  
   
   (?<=abc|abde)  
   
 The implementation of lookbehind assertions is, for each alternative, to  
 temporarily move the current position back by the fixed width and then try to  
 match. If there are insufficient characters before the current position, the  
 match is deemed to fail. Lookbehinds in conjunction with once-only subpatterns  
 can be particularly useful for matching at the ends of strings; an example is  
 given at the end of the section on once-only subpatterns.  
   
 Several assertions (of any sort) may occur in succession. For example,  
   
   (?<=\\d{3})(?<!999)foo  
   
 matches "foo" preceded by three digits that are not "999". Notice that each of  
 the assertions is applied independently at the same point in the subject  
 string. First there is a check that the previous three characters are all  
 digits, then there is a check that the same three characters are not "999".  
 This pattern does \fInot\fR match "foo" preceded by six characters, the first  
 of which are digits and the last three of which are not "999". For example, it  
 doesn't match "123abcfoo". A pattern to do that is  
   
   (?<=\\d{3}...)(?<!999)foo  
   
 This time the first assertion looks at the preceding six characters, checking  
 that the first three are digits, and then the second assertion checks that the  
 preceding three characters are not "999".  
   
 Assertions can be nested in any combination. For example,  
   
   (?<=(?<!foo)bar)baz  
   
 matches an occurrence of "baz" that is preceded by "bar" which in turn is not  
 preceded by "foo", while  
   
   (?<=\\d{3}(?!999)...)foo  
   
 is another pattern which matches "foo" preceded by three digits and any three  
 characters that are not "999".  
   
 Assertion subpatterns are not capturing subpatterns, and may not be repeated,  
 because it makes no sense to assert the same thing several times. If any kind  
 of assertion contains capturing subpatterns within it, these are counted for  
 the purposes of numbering the capturing subpatterns in the whole pattern.  
 However, substring capturing is carried out only for positive assertions,  
 because it does not make sense for negative assertions.  
   
 Assertions count towards the maximum of 200 parenthesized subpatterns.  
   
   
 .SH ONCE-ONLY SUBPATTERNS  
 With both maximizing and minimizing repetition, failure of what follows  
 normally causes the repeated item to be re-evaluated to see if a different  
 number of repeats allows the rest of the pattern to match. Sometimes it is  
 useful to prevent this, either to change the nature of the match, or to cause  
 it fail earlier than it otherwise might, when the author of the pattern knows  
 there is no point in carrying on.  
   
 Consider, for example, the pattern \\d+foo when applied to the subject line  
   
   123456bar  
   
 After matching all 6 digits and then failing to match "foo", the normal  
 action of the matcher is to try again with only 5 digits matching the \\d+  
 item, and then with 4, and so on, before ultimately failing. Once-only  
 subpatterns provide the means for specifying that once a portion of the pattern  
 has matched, it is not to be re-evaluated in this way, so the matcher would  
 give up immediately on failing to match "foo" the first time. The notation is  
 another kind of special parenthesis, starting with (?> as in this example:  
   
   (?>\\d+)bar  
   
 This kind of parenthesis "locks up" the  part of the pattern it contains once  
 it has matched, and a failure further into the pattern is prevented from  
 backtracking into it. Backtracking past it to previous items, however, works as  
 normal.  
   
 An alternative description is that a subpattern of this type matches the string  
 of characters that an identical standalone pattern would match, if anchored at  
 the current point in the subject string.  
   
 Once-only subpatterns are not capturing subpatterns. Simple cases such as the  
 above example can be thought of as a maximizing repeat that must swallow  
 everything it can. So, while both \\d+ and \\d+? are prepared to adjust the  
 number of digits they match in order to make the rest of the pattern match,  
 (?>\\d+) can only match an entire sequence of digits.  
   
 This construction can of course contain arbitrarily complicated subpatterns,  
 and it can be nested.  
   
 Once-only subpatterns can be used in conjunction with lookbehind assertions to  
 specify efficient matching at the end of the subject string. Consider a simple  
 pattern such as  
   
   abcd$  
   
 when applied to a long string which does not match it. Because matching  
 proceeds from left to right, PCRE will look for each "a" in the subject and  
 then see if what follows matches the rest of the pattern. If the pattern is  
 specified as  
   
   ^.*abcd$  
   
 then the initial .* matches the entire string at first, but when this fails, it  
 backtracks to match all but the last character, then all but the last two  
 characters, and so on. Once again the search for "a" covers the entire string,  
 from right to left, so we are no better off. However, if the pattern is written  
 as  
   
   ^(?>.*)(?<=abcd)  
   
 then there can be no backtracking for the .* item; it can match only the entire  
 string. The subsequent lookbehind assertion does a single test on the last four  
 characters. If it fails, the match fails immediately. For long strings, this  
 approach makes a significant difference to the processing time.  
   
   
 .SH CONDITIONAL SUBPATTERNS  
 It is possible to cause the matching process to obey a subpattern  
 conditionally or to choose between two alternative subpatterns, depending on  
 the result of an assertion, or whether a previous capturing subpattern matched  
 or not. The two possible forms of conditional subpattern are  
   
   (?(condition)yes-pattern)  
   (?(condition)yes-pattern|no-pattern)  
   
 If the condition is satisfied, the yes-pattern is used; otherwise the  
 no-pattern (if present) is used. If there are more than two alternatives in the  
 subpattern, a compile-time error occurs.  
   
 There are two kinds of condition. If the text between the parentheses consists  
 of a sequence of digits, then the condition is satisfied if the capturing  
 subpattern of that number has previously matched. Consider the following  
 pattern, which contains non-significant white space to make it more readable  
 (assume the PCRE_EXTENDED option) and to divide it into three parts for ease  
 of discussion:  
   
   ( \\( )?    [^()]+    (?(1) \\) )  
   
 The first part matches an optional opening parenthesis, and if that  
 character is present, sets it as the first captured substring. The second part  
 matches one or more characters that are not parentheses. The third part is a  
 conditional subpattern that tests whether the first set of parentheses matched  
 or not. If they did, that is, if subject started with an opening parenthesis,  
 the condition is true, and so the yes-pattern is executed and a closing  
 parenthesis is required. Otherwise, since no-pattern is not present, the  
 subpattern matches nothing. In other words, this pattern matches a sequence of  
 non-parentheses, optionally enclosed in parentheses.  
   
 If the condition is not a sequence of digits, it must be an assertion. This may  
 be a positive or negative lookahead or lookbehind assertion. Consider this  
 pattern, again containing non-significant white space, and with the two  
 alternatives on the second line:  
   
   (?(?=[^a-z]*[a-z])  
   \\d{2}-[a-z]{3}-\\d{2}  |  \\d{2}-\\d{2}-\\d{2} )  
   
 The condition is a positive lookahead assertion that matches an optional  
 sequence of non-letters followed by a letter. In other words, it tests for the  
 presence of at least one letter in the subject. If a letter is found, the  
 subject is matched against the first alternative; otherwise it is matched  
 against the second. This pattern matches strings in one of the two forms  
 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.  
   
   
 .SH COMMENTS  
 The sequence (?# marks the start of a comment which continues up to the next  
 closing parenthesis. Nested parentheses are not permitted. The characters  
 that make up a comment play no part in the pattern matching at all.  
   
 If the PCRE_EXTENDED option is set, an unescaped # character outside a  
 character class introduces a comment that continues up to the next newline  
 character in the pattern.  
   
   
 .SH PERFORMANCE  
 Certain items that may appear in patterns are more efficient than others. It is  
 more efficient to use a character class like [aeiou] than a set of alternatives  
 such as (a|e|i|o|u). In general, the simplest construction that provides the  
 required behaviour is usually the most efficient. Jeffrey Friedl's book  
 contains a lot of discussion about optimizing regular expressions for efficient  
 performance.  
   
 When a pattern begins with .* and the PCRE_DOTALL option is set, the pattern is  
 implicitly anchored by PCRE, since it can match only at the start of a subject  
 string. However, if PCRE_DOTALL is not set, PCRE cannot make this optimization,  
 because the . metacharacter does not then match a newline, and if the subject  
 string contains newlines, the pattern may match from the character immediately  
 following one of them instead of from the very start. For example, the pattern  
   
   (.*) second  
   
 matches the subject "first\\nand second" (where \\n stands for a newline  
 character) with the first captured substring being "and". In order to do this,  
 PCRE has to retry the match starting after every newline in the subject.  
   
 If you are using such a pattern with subject strings that do not contain  
 newlines, the best performance is obtained by setting PCRE_DOTALL, or starting  
 the pattern with ^.* to indicate explicit anchoring. That saves PCRE from  
 having to scan along the subject looking for a newline to restart at.  
   
 Beware of patterns that contain nested indefinite repeats. These can take a  
 long time to run when applied to a string that does not match. Consider the  
 pattern fragment  
   
   (a+)*  
   
 This can match "aaaa" in 33 different ways, and this number increases very  
 rapidly as the string gets longer. (The * repeat can match 0, 1, 2, 3, or 4  
 times, and for each of those cases other than 0, the + repeats can match  
 different numbers of times.) When the remainder of the pattern is such that the  
 entire match is going to fail, PCRE has in principle to try every possible  
 variation, and this can take an extremely long time.  
   
 An optimization catches some of the more simple cases such as  
   
   (a+)*b  
   
 where a literal character follows. Before embarking on the standard matching  
 procedure, PCRE checks that there is a "b" later in the subject string, and if  
 there is not, it fails the match immediately. However, when there is no  
 following literal this optimization cannot be used. You can see the difference  
 by comparing the behaviour of  
   
   (a+)*\\d  
   
 with the pattern above. The former gives a failure almost instantly when  
 applied to a whole line of "a" characters, whereas the latter takes an  
 appreciable time with strings longer than about 20 characters.  
152    
153  .SH AUTHOR  .SH AUTHOR
154    .rs
155    .sp
156  Philip Hazel <ph10@cam.ac.uk>  Philip Hazel <ph10@cam.ac.uk>
157  .br  .br
158  University Computing Service,  University Computing Service,
159  .br  .br
 New Museums Site,  
 .br  
160  Cambridge CB2 3QG, England.  Cambridge CB2 3QG, England.
161  .br  .br
162  Phone: +44 1223 334714  Phone: +44 1223 334714
163    
164  Last updated: 29 July 1999  .in 0
165    Last updated: 04 February 2003
166  .br  .br
167  Copyright (c) 1997-1999 University of Cambridge.  Copyright (c) 1997-2003 University of Cambridge.
Legend:
Removed from v.41  
changed lines
  Added in v.63
  ViewVC Help
Powered by ViewVC 1.1.5