Seth Woolley's Man Viewer

pcre(3) - PCRE, pcre - Perl-compatible regular expressions - man 3 pcre

([section] manual, -k keyword, -K [section] search, -f whatis)
man plain no title

PCRE(3)                                                                PCRE(3)



NAME
       pcre - Perl-compatible regular expressions.

SYNOPSIS
       #include <pcre.h>

       pcre *pcre_compile(const char *pattern, int options,
            const char **errptr, int *erroffset,
            const unsigned char *tableptr);

       pcre_extra *pcre_study(const pcre *code, int options,
            const char **errptr);

       int pcre_exec(const pcre *code, const pcre_extra *extra,
            const char *subject, int length, int startoffset,
            int options, int *ovector, int ovecsize);

       int pcre_copy_substring(const char *subject, int *ovector,
            int stringcount, int stringnumber, char *buffer,
            int buffersize);

       int pcre_get_substring(const char *subject, int *ovector,
            int stringcount, int stringnumber,
            const char **stringptr);

       int pcre_get_substring_list(const char *subject,
            int *ovector, int stringcount, const char ***listptr);

       void pcre_free_substring(const char *stringptr);

       void pcre_free_substring_list(const char **stringptr);

       const unsigned char *pcre_maketables(void);

       int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
            int what, void *where);

       int pcre_info(const pcre *code, int *optptr, int *firstcharptr);

       char *pcre_version(void);

       void *(*pcre_malloc)(size_t);

       void (*pcre_free)(void *);




DESCRIPTION
       The  PCRE  library is a set(7,n,1 builtins) of functions that implement regular expres-
       sion pattern matching using the same syntax and semantics  as  Perl  5,
       with  just  a  few  differences (see below). The current implementation
       corresponds to Perl 5.005, with some  additional  features  from  later
       versions. This includes some experimental, incomplete support for UTF-8
       encoded strings. Details of exactly what is and what is  not  supported
       are given below.

       PCRE has its own native API, which is described in(1,8) this document. There
       is also a set(7,n,1 builtins) of wrapper functions that correspond to the POSIX regular
       expression API.  These are described in(1,8) the pcreposix documentation.

       The  native  API  function  prototypes  are  defined in(1,8) the header file(1,n)
       pcre.h, and on Unix systems the library itself is called libpcre.a,  so
       can be accessed by adding -lpcre to the command for linking an applica-
       tion which calls it. The header file(1,n) defines the macros PCRE_MAJOR  and
       PCRE_MINOR  to  contain  the  major  and  minor release numbers for the
       library. Applications can use these to include  support  for  different
       releases.

       The  functions  pcre_compile(),  pcre_study(), and pcre_exec() are used
       for compiling and matching regular expressions.

       The   functions   pcre_copy_substring(),   pcre_get_substring(),    and
       pcre_get_substring_list() are convenience functions for extracting cap-
       tured substrings from a matched subject  string(3,n);  pcre_free_substring()
       and  pcre_free_substring_list()  are  also provided, to free the memory
       used for extracted strings.

       The function pcre_maketables() is used (optionally) to build a  set(7,n,1 builtins)  of
       character tables in(1,8) the current locale(3,5,7) for passing to pcre_compile().

       The  function  pcre_fullinfo()  is used to find out information about a
       compiled pattern; pcre_info() is an obsolete version(1,3,5) which returns only
       some  of  the available information, but is retained for backwards com-
       patibility.  The function pcre_version() returns a pointer to a  string(3,n)
       containing the version(1,3,5) of PCRE and its date of release.

       The  global  variables  pcre_malloc and pcre_free initially contain the
       entry points of the standard  malloc()  and  free()  functions  respec-
       tively. PCRE calls the memory management functions via these variables,
       so a calling program can replace them if(3,n) it  wishes  to  intercept  the
       calls. This should be done before calling any PCRE functions.



MULTI-THREADING
       The  PCRE  functions  can be used in(1,8) multi-threading applications, with
       the  proviso  that  the  memory  management  functions  pointed  to  by
       pcre_malloc and pcre_free are shared by all threads.

       The  compiled form of a regular expression is not altered during match-
       ing, so the same compiled pattern can safely be used by several threads
       at once.



COMPILING A PATTERN
       The  function  pcre_compile()  is  called  to compile a pattern into an
       internal form. The pattern is a C string(3,n) terminated by a  binary  zero,
       and  is  passed in(1,8) the argument pattern. A pointer to a single block of
       memory that is obtained via pcre_malloc is returned. This contains  the
       compiled  code  and related data. The pcre type is defined for this for
       convenience, but in(1,8) fact pcre is just a typedef  for  void,  since  the
       contents  of  the  block  are  not  externally defined. It is up to the
       caller to free the memory when it is no longer required.

       The size of a compiled pattern is roughly proportional to the length of
       the  pattern string(3,n), except that each character class (other than those
       containing just a single character, negated or not) requires 33  bytes,
       and  repeat  quantifiers  with  a minimum greater than one or a bounded
       maximum cause the relevant portions  of  the  compiled  pattern  to  be
       replicated.

       The options argument contains independent bits that affect the compila-
       tion. It should be zero  if(3,n)  no  options  are  required.  Some  of  the
       options,  in(1,8)  particular, those that are compatible with Perl, can also
       be set(7,n,1 builtins) and unset from within the pattern (see the detailed  description
       of  regular  expressions below). For these options, the contents of the
       options argument specifies their initial settings at the start of  com-
       pilation and execution. The PCRE_ANCHORED option can be set(7,n,1 builtins) at the time(1,2,n)
       of matching as well as at compile time.

       If errptr is NULL, pcre_compile() returns NULL immediately.  Otherwise,
       if(3,n)  compilation  of  a  pattern fails, pcre_compile() returns NULL, and
       sets the variable pointed to by errptr to point to a textual error(8,n) mes-
       sage.  The  offset from the start of the pattern to the character where
       the error(8,n) was discovered is  placed  in(1,8)  the  variable  pointed  to  by
       erroffset,  which  must  not  be  NULL. If it is, an immediate error(8,n) is
       given.

       If the final argument, tableptr, is NULL, PCRE uses a  default  set(7,n,1 builtins)  of
       character tables which are built when it is compiled, using the default
       C locale.  Otherwise,  tableptr  must  be  the  result  of  a  call  to
       pcre_maketables(). See the section on locale(3,5,7) support below.

       The following option bits are defined in(1,8) the header file:

         PCRE_ANCHORED

       If this bit is set(7,n,1 builtins), the pattern is forced to be "anchored", that is, it
       is constrained to match only at the start of the string(3,n) which is  being
       searched  (the  "subject  string(3,n)"). This effect can also be achieved by
       appropriate constructs in(1,8) the pattern itself, which is the only way  to
       do it in(1,8) Perl.

         PCRE_CASELESS

       If  this  bit is set(7,n,1 builtins), letters in(1,8) 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(7,n,1 builtins), a dollar metacharacter in(1,8) the pattern matches  only
       at  the  end  of the subject string. Without this option, a dollar also
       matches immediately before the final character if(3,n) it is a newline  (but
       not  before  any  other  newlines).  The  PCRE_DOLLAR_ENDONLY option is
       ignored if(3,n) PCRE_MULTILINE is set. There is no equivalent to this option
       in(1,8) Perl.

         PCRE_DOTALL

       If this bit is set(7,n,1 builtins), a dot metacharater in(1,8) the pattern matches all char-
       acters, 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(7,n,1 builtins),  whitespace  data characters in(1,8) 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 char-
       acter sequences in(1,8) a pattern, for example within the sequence (?( which
       introduces a conditional subpattern.

         PCRE_EXTRA

       This  option  was invented in(1,8) order to turn on additional functionality
       of PCRE that is incompatible with Perl, but it  is  currently  of  very
       little  use. When set(7,n,1 builtins), any backslash in(1,8) a pattern that is followed by a
       letter that has no special meaning  causes  an  error(8,n),  thus  reserving
       these  combinations  for  future  expansion.  By default, as in(1,8) 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. It can also be set(7,n,1 builtins) by a (?X) option setting within a pattern.

         PCRE_MULTILINE

       By default, PCRE treats the subject string(3,n) as consisting  of  a  single
       "line"  of  characters (even if(3,n) it actually contains several newlines).
       The "start of line" metacharacter (^) matches only at the start of  the
       string(3,n),  while  the "end of line" metacharacter ($) matches only at the
       end of the string(3,n), or before a terminating  newline  (unless  PCRE_DOL-
       LAR_ENDONLY is set(7,n,1 builtins)). This is the same as Perl.

       When  PCRE_MULTILINE  it  is set(7,n,1 builtins), the "start of line" and "end of line"
       constructs match immediately following or immediately before  any  new-
       line  in(1,8) the subject string(3,n), 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(1,8)  a subject string(3,n), or no occurrences of ^ or $ in(1,8) a pat-
       tern, 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(3,n) followed by "?". It is
       not compatible with Perl. It can also be set(7,n,1 builtins) by a (?U)  option  setting
       within the pattern.

         PCRE_UTF8

       This  option  causes PCRE to regard both the pattern and the subject as
       strings of UTF-8 characters instead of just byte strings.  However,  it
       is  available  only if(3,n) PCRE has been built to include UTF-8 support. If
       not, the use of this option provokes an error.  Support  for  UTF-8  is
       new,  experimental, and incomplete.  Details of exactly what it entails
       are given below.



STUDYING A PATTERN
       When a pattern is going to be used several times, it is worth  spending
       more  time(1,2,n)  analyzing it in(1,8) order to speed up the time(1,2,n) taken for match-
       ing. The function pcre_study() takes a pointer to a compiled pattern as
       its  first  argument,  and  returns  a  pointer  to  a pcre_extra block
       (another void typedef) containing additional information about the pat-
       tern;  this  can be passed to pcre_exec(). If no additional information
       is available, NULL is returned.

       The second argument contains option bits. At present,  no  options  are
       defined for pcre_study(), and this argument should always be zero.

       The  third  argument for pcre_study() is a pointer to an error(8,n) message.
       If studying succeeds (even if(3,n) no data is  returned),  the  variable  it
       points  to  is set(7,n,1 builtins) to NULL. Otherwise it points to a textual error(8,n) mes-
       sage.

       At present, studying a pattern is useful only for non-anchored patterns
       that  do not have a single fixed starting character. A bitmap of possi-
       ble starting characters is created.



LOCALE SUPPORT
       PCRE handles caseless matching, and determines whether  characters  are
       letters,  digits,  or  whatever,  by  reference to a set(7,n,1 builtins) of tables. The
       library contains a default set(7,n,1 builtins)  of  tables  which  is  created  in(1,8)  the
       default  C  locale(3,5,7)  when  PCRE is compiled. This is used when the final
       argument of pcre_compile() is NULL, and is sufficient for many applica-
       tions.

       An alternative set(7,n,1 builtins) of tables can, however, be supplied. Such tables are
       built by calling the pcre_maketables() function,  which  has  no  argu-
       ments,  in(1,8)  the  relevant  locale.  The  result  can  then be passed to
       pcre_compile() as often as necessary. For example,  to  build  and  use
       tables that are appropriate for the French locale(3,5,7) (where accented char-
       acters with codes greater than 128 are treated as letters), the follow-
       ing code could be used:

         setlocale(LC_CTYPE, "fr");
         tables = pcre_maketables();
         re = pcre_compile(..., tables);

       The  tables  are  built in(1,8) memory that is obtained via pcre_malloc. The
       pointer that is passed to pcre_compile is saved with the compiled  pat-
       tern, and the same tables are used via this pointer by pcre_study() and
       pcre_exec(). Thus for any single  pattern,  compilation,  studying  and
       matching  all  happen in(1,8) the same locale(3,5,7), but different patterns can be
       compiled in(1,8) 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.



INFORMATION ABOUT A PATTERN
       The pcre_fullinfo() function returns information about a compiled  pat-
       tern. It replaces the obsolete pcre_info() function, which is neverthe-
       less(1,3) retained for backwards compability (and is documented below).

       The first argument for pcre_fullinfo() is a  pointer  to  the  compiled
       pattern.  The second argument is the result of pcre_study(), or NULL if(3,n)
       the pattern was not studied. The third argument specifies  which  piece
       of information is required, while the fourth argument is a pointer to a
       variable to receive the data. The yield of the  function  is  zero  for
       success, or one of the following negative numbers:

         PCRE_ERROR_NULL       the argument code was NULL
                               the argument where was NULL
         PCRE_ERROR_BADMAGIC   the "magic(4,5) number" was not found
         PCRE_ERROR_BADOPTION  the value of what was invalid

       The  possible  values for the third argument are defined in(1,8) pcre.h, and
       are as follows:

         PCRE_INFO_OPTIONS

       Return a copy of the options with which the pattern was  compiled.  The
       fourth  argument  should  point to au unsigned long int variable. These
       option bits are those specified in(1,8) the call to pcre_compile(), modified
       by  any  top-level  option settings within the pattern itself, and with
       the PCRE_ANCHORED bit forcibly set(7,n,1 builtins) if(3,n) the form of the  pattern  implies
       that it can match only at the start of a subject string.

         PCRE_INFO_SIZE

       Return  the  size  of the compiled pattern, that is, the value that was
       passed as the argument to pcre_malloc() when PCRE was getting memory in(1,8)
       which to place the compiled data. The fourth argument should point to a
       size_t variable.

         PCRE_INFO_CAPTURECOUNT

       Return the number of capturing subpatterns in(1,8) the pattern.  The  fourth
       argument should point to an int variable.

         PCRE_INFO_BACKREFMAX

       Return  the  number  of  the highest back reference in(1,8) the pattern. The
       fourth argument should point to an int variable. Zero  is  returned  if(3,n)
       there are no back references.

         PCRE_INFO_FIRSTCHAR

       Return information about the first character of any matched string(3,n), for
       a non-anchored pattern. If there is a fixed first character, e.g.  from
       a  pattern  such  as  (cat|cow|coyote),  it  is returned in(1,8) the integer
       pointed to by where. Otherwise, if(3,n) 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(7,n,1 builtins) (if(3,n) it were set(7,n,1 builtins), the pattern would be anchored),

       -1 is returned, indicating that the pattern matches only at  the  start
       of  a  subject string(3,n) or after any "\n" within the string. Otherwise -2
       is returned.  For anchored patterns, -2 is returned.

         PCRE_INFO_FIRSTTABLE

       If the pattern was studied, and this resulted in(1,8) the construction of  a
       256-bit  table indicating a fixed set(7,n,1 builtins) of characters for the first char-
       acter in(1,8) any matching string(3,n), a pointer to the table is returned.  Oth-
       erwise  NULL  is  returned.  The  fourth  argument  should  point to an
       unsigned char * variable.

         PCRE_INFO_LASTLITERAL

       For a non-anchored pattern, return the value of the  rightmost  literal
       character  which  must  exist  in(1,8) any matched string(3,n), other than at its
       start. The fourth argument should point to an int variable. If there is
       no  such  character, or if(3,n) the pattern is anchored, -1 is returned. For
       example, for the pattern /a\d+z\d+/ the returned value is 'z'.

       The pcre_info() function is now obsolete because its interface  is  too
       restrictive  to return all the available data about a compiled pattern.
       New  programs  should  use  pcre_fullinfo()  instead.  The   yield   of
       pcre_info()  is the number of capturing subpatterns, or one of the fol-
       lowing negative numbers:

         PCRE_ERROR_NULL       the argument code was NULL
         PCRE_ERROR_BADMAGIC   the "magic(4,5) number" was not found

       If the optptr argument is not NULL, a copy of the  options  with  which
       the  pattern  was  compiled  is placed in(1,8) the integer it points to (see
       PCRE_INFO_OPTIONS above).

       If the pattern is not anchored and the  firstcharptr  argument  is  not
       NULL,  it is used to pass back information about the first character of
       any matched string(3,n) (see PCRE_INFO_FIRSTCHAR above).



MATCHING A PATTERN
       The function pcre_exec() is called to match a subject string(3,n) against  a
       pre-compiled pattern, which is passed in(1,8) the code argument. If the pat-
       tern has been studied, the result of the study should be passed in(1,8)  the
       extra argument. Otherwise this must be NULL.

       The  PCRE_ANCHORED  option can be passed in(1,8) the options argument, whose
       unused bits must be zero. However,  if(3,n)  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(7,n,1 builtins) only  at  matching
       time:

         PCRE_NOTBOL

       The  first  character  of the string(3,n) is not the beginning of a line, so
       the circumflex metacharacter should not match before it.  Setting  this
       without  PCRE_MULTILINE  (at  compile  time(1,2,n)) causes circumflex never to
       match.

         PCRE_NOTEOL

       The end of the string(3,n) is not the end of a line, so the dollar metachar-
       acter  should  not  match  it  nor (except in(1,8) multiline mode) a newline
       immediately before it. Setting this without PCRE_MULTILINE (at  compile
       time(1,2,n)) causes dollar never to match.

         PCRE_NOTEMPTY

       An empty string(3,n) is not considered to be a valid match if(3,n) this option is
       set. If there are alternatives in(1,8) the pattern, they are tried.  If  all
       the  alternatives  match  the empty string(3,n), the entire match fails. For
       example, if(3,n) the pattern

         a?b?

       is applied to a string(3,n) not beginning with "a" or "b",  it  matches  the
       empty  string(3,n) at the start of the subject. With PCRE_NOTEMPTY set(7,n,1 builtins), this
       match is not valid, so PCRE searches further into the string(3,n) for occur-
       rences of "a" or "b".

       Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
       cial case of a pattern match of the empty  string(3,n)  within  its  split(1,n)()
       function,  and  when  using  the /g modifier. It is possible to emulate
       Perl's behaviour after matching a null string(3,n) by first trying the match
       again at the same offset with PCRE_NOTEMPTY set(7,n,1 builtins), and then if(3,n) that fails
       by advancing the starting offset (see below)  and  trying  an  ordinary
       match again.

       The  subject  string(3,n)  is  passed  as  a pointer in(1,8) subject, a length in(1,8)
       length, and a  starting  offset  in(1,8)  startoffset.  Unlike  the  pattern
       string(3,n), 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(1,8) the same subject by calling pcre_exec() again after a previous  suc-
       cess.   Setting  startoffset differs from just passing over a shortened
       string(3,n) and setting PCRE_NOTBOL in(1,8) 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(1,8) the middle of words. (\B matches
       only if(3,n) the current position in(1,8) the subject is not  a  word  boundary.)
       When  applied  to the string(3,n) "Mississipi" the first call to pcre_exec()
       finds the first occurrence. If pcre_exec() 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(3,n) pcre_exec() is passed the entire
       string(3,n) again, but with startoffset set(7,n,1 builtins) to 4, it finds the second occur-
       rence  of "iss" because it is able to look(1,8,3 Search::Dict) behind the starting point to
       discover(1,3,5) 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 suc-
       ceed if(3,n) 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(1,8)
       addition, further substrings from the subject  may  be  picked  out  by
       parts  of  the  pattern.  Following the usage in(1,8) Jeffrey Friedl's book,
       this is called "capturing" in(1,8) what follows, and the  phrase  "capturing
       subpattern"  is  used for a fragment of a pattern that picks out a sub-
       string. 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(1,8) ovector. The number of  elements  in(1,8)
       the vector is passed in(1,8) ovecsize. 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
       pcre_exec() while matching capturing subpatterns, and is not  available
       for  passing  back  information.  The  length passed in(1,8) ovecsize 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(1,8) pairs of integers, starting at the beginning of ovector,
       and continuing up to two-thirds of its length at the  most.  The  first
       element of a pair is set(7,n,1 builtins) to the offset of the first character in(1,8) a sub-
       string(3,n), and the second is set(7,n,1 builtins) to the  offset  of  the  first  character
       after  the  end  of  a  substring. The first pair, ovector[0] and ovec-
       tor[1], identify the portion of  the  subject  string(3,n)  matched  by  the
       entire  pattern.  The next pair is used for the first capturing subpat-
       tern, and so on. The value returned by pcre_exec()  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(1,8)  the  following
       section.

       It  is  possible  for  an capturing subpattern number n+1 to match some
       part of the subject when subpattern n has not been  used  at  all.  For
       example, if(3,n) the string(3,n) "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(7,n,1 builtins) to -1.

       If a capturing subpattern is matched repeatedly, it is the last portion
       of the string(3,n) that it matched that gets(3,n) 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 func-
       tion returns a value of zero. In particular, if(3,n) the  substring  offsets
       are  not  of interest, pcre_exec() may be called with ovector passed as
       NULL and ovecsize as zero. However, if(3,n) the pattern contains back refer-
       ences  and  the  ovector  isn't big enough to remember the related sub-
       strings, PCRE has to get additional memory  for  use  during  matching.
       Thus it is usually advisable to supply an ovector.

       Note  that  pcre_info() can be used to find out how many capturing sub-
       patterns there are in(1,8) a compiled pattern. The smallest size for ovector
       that will allow for n captured substrings in(1,8) addition to the offsets of
       the substring matched by the whole pattern is (n+1)*3.

       If pcre_exec() fails, it returns a negative number. The  following  are
       defined in(1,8) the header file:

         PCRE_ERROR_NOMATCH        (-1)

       The subject string(3,n) did not match the pattern.

         PCRE_ERROR_NULL           (-2)

       Either  code  or  subject  was  passed as NULL, or ovector was NULL and
       ovecsize was not zero.

         PCRE_ERROR_BADOPTION      (-3)

       An unrecognized bit was set(7,n,1 builtins) in(1,8) the options argument.

         PCRE_ERROR_BADMAGIC       (-4)

       PCRE stores a 4-byte "magic(4,5) number" at the start of the compiled  code,
       to  catch  the case when it is passed a junk pointer. This is the error(8,n)
       it gives when the magic(4,5) number isn't present.

         PCRE_ERROR_UNKNOWN_NODE   (-5)

       While running the pattern match, an unknown item was encountered in(1,8) the
       compiled  pattern.  This  error(8,n)  could be caused by a bug in(1,8) PCRE or by
       overwriting of the compiled pattern.

         PCRE_ERROR_NOMEMORY       (-6)

       If a pattern contains back references, but the ovector that  is  passed
       to pcre_exec() is not big enough to remember the referenced substrings,
       PCRE gets(3,n) a block of memory at the start of matching to  use  for  this
       purpose.  If the call via pcre_malloc() fails, this error(8,n) is given. The
       memory is freed at the end of matching.



EXTRACTING CAPTURED SUBSTRINGS
       Captured substrings can be  accessed  directly  by  using  the  offsets
       returned  by  pcre_exec()  in(1,8)  ovector.  For convenience, the functions
       pcre_copy_substring(),    pcre_get_substring(),    and    pcre_get_sub-
       string_list()  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: subject
       is the subject string(3,n) which has just been successfully matched, ovector
       is a pointer to the vector  of  integer  offsets  that  was  passed  to
       pcre_exec(), and stringcount is the number of substrings that were cap-
       tured by the match, including the substring  that  matched  the  entire
       regular  expression.  This  is the value returned by pcre_exec if(3,n) it is
       greater than zero. If pcre_exec() returned zero, indicating that it ran
       out  of space in(1,8) ovector, the value passed as stringcount should be the
       size of the vector divided by three.

       The functions pcre_copy_substring() and pcre_get_substring() extract  a
       single  substring,  whose  number  is given as stringnumber. A value of
       zero extracts the substring that  matched  the  entire  pattern,  while
       higher  values  extract  the  captured  substrings.  For pcre_copy_sub-
       string(3,n)(), the string(3,n) is placed in(1,8) buffer,  whose  length  is  given  by
       buffersize,  while  for  pcre_get_substring()  a new block of memory is
       obtained via pcre_malloc, and its address is  returned  via  stringptr.
       The  yield  of  the function is the length of the string(3,n), not including
       the terminating zero, or one of

         PCRE_ERROR_NOMEMORY       (-6)

       The buffer was too small for pcre_copy_substring(), or the  attempt  to
       get memory failed for pcre_get_substring().

         PCRE_ERROR_NOSUBSTRING    (-7)

       There is no substring whose number is stringnumber.

       The  pcre_get_substring_list()  function  extracts  all  available sub-
       strings and builds a list of pointers to them. All this is  done  in(1,8)  a
       single  block  of memory which is obtained via pcre_malloc. The address
       of the memory block is returned via listptr, which is also the start of
       the  list  of  string(3,n) pointers. The end of the list is marked by a NULL
       pointer. The yield of the function is zero if(3,n) all went well, or

         PCRE_ERROR_NOMEMORY       (-6)

       if(3,n) 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 n+1 matches some part of
       the subject, but subpattern n has not been used at all, they return  an
       empty string. This can be distinguished from a genuine zero-length sub-
       string(3,n) by inspecting the appropriate offset in(1,8) ovector, which is  nega-
       tive for unset substrings.

       The  two convenience functions pcre_free_substring() and pcre_free_sub-
       string_list() can be used to free the memory  returned  by  a  previous
       call  of  pcre_get_substring()  or  pcre_get_substring_list(),  respec-
       tively. They do nothing more than  call  the  function  pointed  to  by
       pcre_free,  which  of course could be called directly from a C program.
       However, PCRE is used in(1,8) some situations where it is linked via a  spe-
       cial  interface  to  another  programming  language  which  cannot  use
       pcre_free directly; it is for these cases that the functions  are  pro-
       vided.



LIMITATIONS
       There  are some size limitations in(1,8) PCRE but it is hoped that they will
       never in(1,8) practice be relevant.  The maximum length of a  compiled  pat-
       tern is 65539 (sic) bytes.  All values in(1,8) repeating quantifiers must be
       less(1,3) than 65536.  The maximum number of capturing  subpatterns  is  99.
       The  maximum number of all parenthesized subpatterns, including captur-
       ing subpatterns, assertions, and other types of subpattern, is 200.

       The maximum length of a subject string(3,n) is the largest  positive  number
       that an integer variable can hold. However, PCRE uses recursion to han-
       dle subpatterns and indefinite repetition. This means that  the  avail-
       able  stack  space  may  limit the size of a subject string(3,n) that can be
       processed by certain patterns.



DIFFERENCES FROM PERL
       The differences described here are with respect to Perl 5.005.

       1. By default, a whitespace character  is  any  character  that  the  C
       library function isspace() recognizes, though it is possible to compile
       PCRE with alternative character type tables. Normally isspace() matches
       space, formfeed, newline, carriage return, horizontal tab, and vertical
       tab. Perl 5 no longer includes vertical tab in(1,8) its  set(7,n,1 builtins)  of  whitespace
       characters. The \v escape that was in(1,8) the Perl documentation for a long
       time(1,2,n) was never in(1,8) 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 asser-
       tions are counted, but their entries in(1,8) 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 succeed-
       ing),  but  only  if(3,n) the negative lookahead assertion contains just one
       branch.

       4. Though binary zero characters are supported in(1,8) the  subject  string(3,n),
       they are not allowed in(1,8) a pattern string(3,n) because it is passed as a nor-
       mal C string(3,n), terminated by zero. The escape sequence "\0" can be  used
       in(1,8) 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-han-
       dling and are not part of its pattern matching engine.

       6. The Perl \G assertion is not supported as it is not relevant to sin-
       gle pattern matches.

       7. Fairly obviously, PCRE does not support the (?{code}) and (?p{code})
       constructions.  However,  there is some experimental support for recur-
       sive patterns using the non-Perl item (?R).

       8. There are at the time(1,2,n) of writing some oddities in(1,8) Perl 5.005_02 con-
       cerned  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 $2 unset. However, if(3,n) the pattern is changed to /^(aa(b(b))?)+$/
       then $2 (and $3) are set.

       In  Perl  5.004 $2 is set(7,n,1 builtins) in(1,8) both cases, and that is also true of PCRE.
       If in(1,8) 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(1,8) Perl 5.005_02 the
       pattern /^(a)?(?(1)a|b)+$/ matches the string(3,n) "a", whereas in(1,8)  PCRE  it
       does  not.  However, in(1,8) both Perl and PCRE /^(a)?a/ matched against "a"
       leaves $1 unset.

       10. PCRE provides some extensions to the Perl regular expression facil-
       ities:

       (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(7,n,1 builtins) and PCRE_MULTILINE is not set(7,n,1 builtins), the  $
       meta- character matches only at the very end of the string.

       (c) If PCRE_EXTRA is set(7,n,1 builtins), a backslash followed by a letter with no spe-
       cial meaning is faulted.

       (d) If PCRE_UNGREEDY is set(7,n,1 builtins), the greediness of the  repetition  quanti-
       fiers is inverted, that is, by default they are not greedy, but if(3,n) fol-
       lowed 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
       pcre_exec() have no Perl equivalents.

       (g) The (?R) construct allows for recursive pattern matching (Perl  5.6
       can do this using the (?p{code}) construct, which PCRE cannot of course
       support.)



REGULAR EXPRESSION DETAILS
       The syntax and semantics of the regular expressions supported  by  PCRE
       are described below. Regular expressions are also described in(1,8) the Perl
       documentation and in(1,8) a number of other books, some of which have  copi-
       ous  examples.  Jeffrey  Friedl's "Mastering Regular Expressions", pub-
       lished by O'Reilly (ISBN 1-56592-257), covers them in(1,8) great detail.

       The description here is intended as reference documentation. The  basic
       operation  of PCRE is on strings of bytes. However, there is the begin-
       nings of some support for UTF-8 character strings. To use this  support
       you  must  configure  PCRE  to include it, and then call pcre_compile()
       with the PCRE_UTF8 option. How this affects  the  pattern  matching  is
       described in(1,8) the final section of this document.

       A  regular  expression  is  a pattern that is matched against a subject
       string(3,n) from left to right. Most characters stand for  themselves  in(1,8)  a
       pattern,  and  match  the corresponding characters in(1,8) the subject. As a
       trivial example, the pattern

         The quick brown fox

       matches a portion of a subject string(3,n) that is identical to itself.  The
       power of regular expressions comes from the ability to include alterna-
       tives and repetitions in(1,8) the pattern. These are encoded in(1,8) the  pattern
       by  the  use  of meta-characters, which do not stand for themselves but
       instead are interpreted in(1,8) some special way.

       There are two different sets of meta-characters: those that are  recog-
       nized  anywhere in(1,8) the pattern except within square brackets, and those
       that are recognized in(1,8) square brackets. Outside  square  brackets,  the
       meta-characters are as follows:

         \      general escape character with several uses
         ^      assert start of subject (or line, in(1,8) multiline mode)
         $      assert end of subject (or line, in(1,8) 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(1,8) 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(3,n) the first character
         -      indicates character range
         ]      terminates the character class

       The following sections describe the use of each of the meta-characters.



BACKSLASH
       The backslash character has several uses. Firstly, if(3,n) 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(3,n) you want to match a "*" character, you  write(1,2)  "\*"  in(1,8)
       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(3,n) you want to match a backslash, you write(1,2) "\\".

       If a pattern is compiled with the PCRE_EXTENDED option,  whitespace  in(1,8)
       the  pattern (other than in(1,8) 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(3,n) non-printing char-
       acters  in(1,8) patterns in(1,8) 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(1,2), 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(3,n) "x" is a lower case let-
       ter, 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(2,n,1 builtins) (letters  can  be  in(1,8)
       upper or lower case).

       After  "\0"  up to two further octal digits are read. In both cases, if(3,n)
       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(3,n) the character that follows is itself an octal digit.

       The handling of a backslash followed by a digit other than 0 is compli-
       cated.  Outside a character class, PCRE reads it and any following dig-
       its  as  a  decimal  number. If the number is less(1,3) than 10, or if(3,n) there
       have been at least that many previous capturing left parentheses in(1,8) the
       expression,  the  entire  sequence  is  taken  as  a  back reference. A
       description of how this works is given later, following the  discussion
       of parenthesized subpatterns.

       Inside  a  character  class, or if(3,n) 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 sin-
       gle 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(7,n,1 builtins) 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 defini-
       tion of letters and digits is controlled by  PCRE's  character  tables,
       and  may vary if(3,n) locale- specific matching is taking place (see "Locale
       support" above). For example, in(1,8) the "fr" (French) locale(3,5,7), some charac-
       ter 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 char-
       acter  classes.  They each match one character of the appropriate type.
       If the current matching point is at the end of the subject string(3,n),  all
       of them fail, since there is no character to match.

       The fourth use of backslash is for certain simple assertions. An asser-
       tion specifies a condition that has to be met at a particular point  in(1,8)
       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(1,8) 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(1,8) the subject string(3,n) 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(3,n) if(3,n) the first or last character matches \w, respectively.

       The  \A,  \Z,  and \z assertions differ from the traditional circumflex
       and dollar (described below) in(1,8) that they only ever match at  the  very
       start and end of the subject string(3,n), whatever options are set. They are
       not  affected  by  the  PCRE_NOTBOL  or  PCRE_NOTEOL  options.  If  the
       startoffset  argument  of  pcre_exec() 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(3,n) as well as at the end of the
       string(3,n), whereas \z matches only at the end.



CIRCUMFLEX AND DOLLAR
       Outside a character class, in(1,8) the default matching mode, the circumflex
       character  is  an  assertion which is true only if(3,n) the current matching
       point is at the start of the subject string. If the  startoffset  argu-
       ment  of  pcre_exec() 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(3,n) a number
       of alternatives are involved, but it should be the first thing in(1,8)  each
       alternative  in(1,8)  which  it appears if(3,n) the pattern is ever to match that
       branch. If all possible alternatives start with a circumflex, that  is,
       if(3,n)  the  pattern  is constrained to match only at the start of the sub-
       ject, 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(3,n) the current
       matching point is at the end of  the  subject  string(3,n),  or  immediately
       before a newline character that is the last character in(1,8) the string(3,n) (by
       default). Dollar need not be the last character of  the  pattern  if(3,n)  a
       number  of alternatives are involved, but it should be the last item in(1,8)
       any branch in(1,8) which it appears.  Dollar has no  special  meaning  in(1,8)  a
       character class.

       The  meaning  of  dollar  can be changed so that it matches only at the
       very end of the string(3,n), 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(3,n) the
       PCRE_MULTILINE option is set. When this is the case, they match immedi-
       ately  after and immediately before an internal "\n" character, respec-
       tively, in(1,8) addition to matching at the start and  end  of  the  subject
       string.  For  example,  the  pattern /^abc$/ matches the subject string(3,n)
       "def\nabc" in(1,8) multiline mode, but not otherwise. Consequently, patterns
       that  are  anchored in(1,8) single line mode because all branches start with
       "^" are not anchored in(1,8) multiline mode, and a match for  circumflex  is
       possible  when the startoffset argument of pcre_exec() is non-zero. The
       PCRE_DOLLAR_ENDONLY option is ignored if(3,n) 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(1,8) both modes, and if(3,n) all branches of a pattern
       start with \A is it always anchored, whether PCRE_MULTILINE is  set(7,n,1 builtins)  or
       not.



FULL STOP (PERIOD, DOT)
       Outside a character class, a dot in(1,8) the pattern matches any one charac-
       ter in(1,8) the subject, including a non-printing  character,  but  not  (by
       default)  newline.   If  the PCRE_DOTALL option is set(7,n,1 builtins), dots match new-
       lines as well. The handling of dot is entirely independent of the  han-
       dling  of  circumflex and dollar, the only relationship being that they
       both involve newline characters. Dot has no special meaning in(1,8) a  char-
       acter class.



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 spe-
       cial. If a closing square bracket is required as a member of the class,
       it should be the first data character in(1,8) the class  (after  an  initial
       circumflex, if(3,n) present) or escaped with a backslash.

       A  character class matches a single character in(1,8) the subject; the char-
       acter must be in(1,8) the set(7,n,1 builtins) of characters defined by the class, unless the
       first character in(1,8) the class is a circumflex, in(1,8) which case the subject
       character must not be in(1,8) the set(7,n,1 builtins) 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(1,8) the class by enumerating those that are not. It
       is not an assertion: it still consumes a  character  from  the  subject
       string(3,n), and fails if(3,n) the current pointer is at the end of the string.

       When  caseless  matching  is set(7,n,1 builtins), any letters in(1,8) 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(1,3,5) would.

       The newline character is never treated in(1,8) any special way in(1,8)  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  charac-
       ters  in(1,8)  a  character  class.  For  example,  [d-m] matches any letter
       between d and m, inclusive. If a  minus  character  is  required  in(1,8)  a
       class,  it  must  be  escaped  with a backslash or appear in(1,8) a position
       where it cannot be interpreted as indicating a range, typically as  the
       first or last character in(1,8) the class.

       It is not possible to have the literal character "]" as the end charac-
       ter of a range. A pattern such as [W-]46] is interpreted as a class  of
       two  characters ("W" and "-") followed by a literal string(3,n) "46]", so it
       would match "W46]" or "-46]". However, if(3,n) the "]"  is  escaped  with  a
       backslash  it is interpreted as the end of range, so [W-\]46] is inter-
       preted 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(1,8) 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(7,n,1 builtins), it matches
       the  letters  in(1,8)  either  case.  For  example,  [W-c]  is equivalent to
       [][\^_`wxyzabc], matched caselessly, and if(3,n) character  tables  for  the
       "fr"  locale(3,5,7)  are  in(1,8) use, [\xc8-\xcb] matches accented E characters in(1,8)
       both cases.

       The character types \d, \D, \s, \S, \w, and \W may  also  appear  in(1,8)  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(7,n,1 builtins) 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(1,8) character classes, but it does no harm
       if(3,n) they are escaped.



POSIX CHARACTER CLASSES
       Perl 5.6 (not yet released at the time(1,2,n) of writing) is going to  support
       the  POSIX notation for character classes, which uses names enclosed by
       [: and :] within the enclosing  square  brackets.  PCRE  supports  this
       notation. For example,

         [01[:alpha:]%]

       matches "0", "1", any alphabetic character, or "%". The supported class
       names are

         alnum    letters and digits
         alpha    letters
         ascii(1,7)    character codes 0 - 127
         cntrl    control characters
         digit    decimal digits (same as \d)
         graph    printing characters, excluding space
         lower    lower case letters
         print    printing characters, including space
         punct    printing characters, excluding letters and digits
         space    white space (same as \s)
         upper    upper case letters
         word     "word" characters (same as \w)
         xdigit   hexadecimal digits

       The names "ascii(1,7)" and "word" are Perl extensions. Another  Perl  exten-
       sion  is negation, which is indicated by a ^ character after the colon.
       For example,

         [12[:^digit:]]

       matches "1", "2", or any non-digit. PCRE (and Perl) also  recogize  the
       POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
       these are not supported, and an error(8,n) is given if(3,n) they are encountered.



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(3,n)).  The matching process tries each  alternative  in(1,8)  turn,  from
       left to right, and the first one that succeeds is used. If the alterna-
       tives are within a subpattern (defined below), "succeeds" means  match-
       ing the rest of the main pattern as well as the alternative in(1,8) the sub-
       pattern.



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 possi-
       ble to unset these options by preceding the letter with a hyphen, and a
       combined  setting and unsetting such as (?im-sx), which sets PCRE_CASE-
       LESS 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(1,8) the  pattern  the
       setting  occurs.  For settings that are outside any subpattern (defined
       below), the effect is the same as if(3,n) the options were set(7,n,1 builtins) or  unset  at
       the start of matching. The following patterns all behave in(1,8) exactly the
       same way:

         (?i)abc
         a(?i)bc
         ab(?i)c
         abc(?i)

       which in(1,8) turn is the same as compiling the pattern abc with  PCRE_CASE-
       LESS 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  differ-
       ent.  This  is  a  change  of behaviour in(1,8) Perl 5.005. An option change
       inside a subpattern affects only that part of the subpattern that  fol-
       lows 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(1,8)  different parts of the pattern. Any changes made in(1,8) 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(1,8)  the same way as the Perl-compatible options by using the characters
       U and X respectively. The (?X) flag setting is special in(1,8) that it  must
       always occur earlier in(1,8) 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.



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(7,n,1 builtins) 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(3,n) that matched the subpattern is passed back to the caller via the
       ovector  argument  of pcre_exec(). Opening parentheses are counted from
       left to right (starting from 1) to obtain the numbers of the  capturing
       subpatterns.

       For  example,  if(3,n) the string(3,n) "the red king" is matched against the pat-
       tern

         the ((red|white) (king|queen))

       the captured substrings are "red king", "red", and "king", and are num-
       bered 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(3,n) the string(3,n) "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 maxi-
       mum  number  of  all  subpatterns, both capturing and non-capturing, is
       200.

       As a convenient shorthand, if(3,n) 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(7,n,1 builtins) of strings. Because alternative branches are
       tried  from  left  to right, and options are not reset(1,7,1 tput) until the end of
       the subpattern is reached, an option setting in(1,8) one branch does  affect
       subsequent  branches,  so  the above patterns match "SUNDAY" as well as
       "Saturday".



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 num-
       ber of permitted matches, by giving the two numbers in(1,8)  curly  brackets
       (braces),  separated  by  a comma. The numbers must be less(1,3) than 65536,
       and the first must be less(1,3) 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(3,n) 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(1,8)  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  exam-
       ple, {,6} is not a quantifier, but a literal string(3,n) of four characters.

       The quantifier {0} is permitted, causing the expression to behave as if(3,n)
       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(8,n) at compile time(1,2,n)
       for  such  patterns. However, because there are cases where this can be
       useful, such patterns are now accepted, but if(3,n) any  repetition  of  the
       subpattern  does in(1,8) fact match no characters, the loop is forcibly bro-
       ken.

       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(1,8) trying to match comments in(1,8) C programs. These
       appear between the sequences /* and */ and within the  sequence,  indi-
       vidual * and / characters may appear. An attempt to match C comments by
       applying the pattern

         /\*.*\*/

       to the string(3,n)

         /* first command */  not comment  /* second comment */

       fails, because it matches the entire string(3,n) owing to the greediness  of
       the .*  item.

       However,  if(3,n)  a quantifier is followed by a question mark, 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(1,8) its own right. Because it has two uses, it can  sometimes
       appear doubled, as in(1,8)

         \d??\d

       which matches one digit by preference, but can match two if(3,n) that is the
       only way the rest of the pattern matches.

       If the PCRE_UNGREEDY option is set(7,n,1 builtins) (an option which is not available in(1,8)
       Perl),  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(1,8) proportion to the size of the
       minimum or maximum.

       If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
       alent  to Perl's /s) is set(7,n,1 builtins), thus allowing the . to match newlines, the
       pattern is implicitly anchored, because whatever follows will be  tried
       against  every character position in(1,8) the subject string(3,n), so there is no
       point in(1,8) 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(3,n) contains no newlines,  it  is
       worth  setting  PCRE_DOTALL when the pattern begins with .* in(1,8) order to
       obtain this optimization, or alternatively using ^ to indicate  anchor-
       ing explicitly.

       When a capturing subpattern is repeated, the value captured is the sub-
       string(3,n) 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(3,n) there are nested capturing subpatterns,
       the corresponding captured values may have been set(7,n,1 builtins) in(1,8) previous  itera-
       tions. For example, after

         /(a|(b))+/

       matches "aba" the value of the second captured substring is "b".



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 sub-
       pattern  earlier (i.e. to its left) in(1,8) the pattern, provided there have
       been that many previous capturing left parentheses.

       However, if(3,n) the decimal number following the backslash is less(1,3) than 10,
       it  is  always  taken  as a back reference, and causes an error(8,n) only if(3,n)
       there are not that many capturing left parentheses in(1,8) the  entire  pat-
       tern.  In  other words, the parentheses that are referenced need not be
       to the left of the reference for numbers less(1,3) than 10. See the  section
       entitled  "Backslash" above for further details of the handling of dig-
       its following a backslash.

       A back reference matches whatever actually matched the  capturing  sub-
       pattern  in(1,8)  the  current subject string(3,n), 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(1,8) force at the
       time(1,2,n) of the back reference, the case of letters is relevant. For  exam-
       ple,

         ((?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(1,8) a particular match, any back
       references to it always fail. For example, the pattern

         (a|(bc))\2

       always fails if(3,n) 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 con-
       tinues with a digit character, some delimiter must be used to terminate
       the back reference.  If the PCRE_EXTENDED option is set(7,n,1 builtins),  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 sub-
       patterns. For example, the pattern

         (a|b\1)+

       matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
       ation  of  the  subpattern,  the  back  reference matches the character
       string(3,n) 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(1,8)
       the example above, or by a quantifier with a minimum of zero.



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(1,8,3 Search::Dict) ahead of the current position in(1,8)
       the subject string(3,n), and those that look(1,8,3 Search::Dict) behind it.

       An assertion subpattern is matched in(1,8) 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 semi-
       colon in(1,8) 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(3,n) there are sev-
       eral alternatives, they do not all have to have the same fixed  length.
       Thus

         (?<=bullock|donkey)

       is permitted, but

         (?<!dogs?|cats?)

       causes  an  error(8,n) 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(3,n) rewritten to use two top-
       level branches:

         (?<=abc|abde)

       The implementation of lookbehind assertions is, for  each  alternative,
       to  temporarily  move(3x,7,3x curs_move)  the current position back by the fixed width and
       then try to match. If there are insufficient characters before the cur-
       rent  position, the match is deemed to fail. Lookbehinds in(1,8) 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(1,3)) may occur in(1,8) 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(1,8)
       the subject string. First there is a  check  that  the  previous  three
       characters  are  all  digits,  and  then there is a check that the same
       three characters are not "999".  This pattern does not match "foo" pre-
       ceded  by  six  characters,  the first of which are digits and the last
       three of which are not "999". For example, it  doesn't  match  "123abc-
       foo". A pattern to do that is

         (?<=\d{3}...)(?<!999)foo

       This  time(1,2,n)  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(1,8) any combination. For example,

         (?<=(?<!foo)bar)baz

       matches  an occurrence of "baz" that is preceded by "bar" which in(1,8) 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  sub-
       patterns in(1,8) 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.



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(3,n) a dif-
       ferent number of repeats allows the rest of the pattern to match. Some-
       times  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(1,8) 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(1,8)  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, start-
       ing with (?> as in(1,8) this example:

         (?>\d+)bar

       This  kind  of  parenthesis "locks up" the  part of the pattern it con-
       tains 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(3,n)  of  characters  that an identical standalone pattern would
       match, if(3,n) anchored at the current point in(1,8) 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(1,8) 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 subpat-
       terns, and it can be nested.

       Once-only subpatterns can be used in(1,8) conjunction with lookbehind asser-
       tions to specify efficient matching at the end of the  subject  string.
       Consider a simple pattern such as

         abcd$

       when  applied  to  a long string(3,n) which does not match. Because matching
       proceeds from left to right, PCRE will look(1,8,3 Search::Dict) for each "a" in(1,8) the subject
       and  then  see  if(3,n) what follows matches the rest of the pattern. If the
       pattern is specified as

         ^.*abcd$

       the initial .* matches the entire string(3,n) at first, but when this  fails
       (because there is no following "a"), 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(3,n), from right to left,
       so we are no better off. However, if(3,n) the pattern is written as

         ^(?>.*)(?<=abcd)

       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.

       When a pattern contains an unlimited repeat inside  a  subpattern  that
       can itself be repeated an unlimited number of times, the use of a once-
       only subpattern is the only way to avoid some failing matches taking  a
       very long time(1,2,n) indeed.  The pattern

         (\D+|<\d+>)*[!?]

       matches  an  unlimited number of substrings that either consist of non-
       digits, or digits enclosed in(1,8) <>, followed by either ! or  ?.  When  it
       matches, it runs quickly. However, if(3,n) it is applied to

         aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

       it  takes  a  long  time(1,2,n)  before reporting failure. This is because the
       string(3,n) can be divided between the two repeats  in(1,8)  a  large  number  of
       ways,  and  all  have to be tried. (The example used [!?] rather than a
       single character at the end, because both PCRE and Perl have  an  opti-
       mization  that allows for fast failure when a single character is used.
       They remember the last single character that is required for  a  match,
       and  fail early if(3,n) it is not present in(1,8) the string.)  If the pattern is
       changed to

         ((?>\D+)|<\d+>)*[!?]

       sequences of non-digits cannot be broken, and failure happens  quickly.



CONDITIONAL SUBPATTERNS
       It  is possible to cause the matching process to obey a subpattern con-
       ditionally or to choose between two alternative subpatterns,  depending
       on  the result of an assertion, or whether a previous capturing subpat-
       tern 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(3,n) present) is used. If there are more  than  two  alterna-
       tives in(1,8) the subpattern, a compile-time error(8,n) occurs.

       There  are  two kinds of condition. If the text between the parentheses
       consists of a sequence of digits, the condition  is  satisfied  if(3,n)  the
       capturing  subpattern of that number has previously matched. The number
       must be greater than zero.  Consider the following pattern, which  con-
       tains  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(3,n) that
       character is present, sets it as the first captured substring. The sec-
       ond  part  matches one or more characters that are not parentheses. The
       third part is a conditional subpattern that tests whether the first set(7,n,1 builtins)
       of parentheses matched or not. If they did, that is, if(3,n) subject started
       with an opening parenthesis, the condition is true, and so the yes-pat-
       tern  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(1,8) 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(1,8) 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(1,8) one of the two forms dd-aaa-dd or dd-dd-dd,  where  aaa  are
       letters and dd are digits.



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(1,8) the pattern matching
       at all.

       If the PCRE_EXTENDED option is set(7,n,1 builtins), an unescaped # character outside  a
       character class introduces a comment that continues up to the next new-
       line character in(1,8) the pattern.



RECURSIVE PATTERNS
       Consider the problem of matching a string(3,n) in(1,8) parentheses, allowing  for
       unlimited  nested  parentheses.  Without the use of recursion, the best
       that can be done is to use a pattern that  matches  up  to  some  fixed
       depth  of  nesting.  It  is not possible to handle an arbitrary nesting
       depth. Perl 5.6 has provided an experimental facility that allows regu-
       lar  expressions  to  recurse  (amongst  other things). It does this by
       interpolating Perl code in(1,8) the expression at run time(1,2,n), and the code can
       refer to the expression itself. A Perl pattern to solve the parentheses
       problem can be created like this:

         $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;

       The (?p{...}) item interpolates Perl code at run time(1,2,n), and in(1,8) this case
       refers  recursively to the pattern in(1,8) which it appears. Obviously, PCRE
       cannot support the interpolation of Perl  code.  Instead,  the  special
       item  (?R)  is  provided  for the specific case of recursion. This PCRE
       pattern solves the parentheses problem (assume the PCRE_EXTENDED option
       is set(7,n,1 builtins) so that white space is ignored):

         \( ( (?>[^()]+) | (?R) )* \)

       First  it matches an opening parenthesis. Then it matches any number of
       substrings which can either be a  sequence  of  non-parentheses,  or  a
       recursive  match  of the pattern itself (i.e. a correctly parenthesized
       substring). Finally there is a closing parenthesis.

       This particular example pattern contains nested unlimited repeats,  and
       so the use of a once-only subpattern for matching strings of non-paren-
       theses is important when applying the pattern to strings  that  do  not
       match. For example, when it is applied to

         (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()

       it yields "no match" quickly. However, if(3,n) a once-only subpattern is not
       used, the match runs for a very long time(1,2,n) indeed because there  are  so
       many  different  ways the + and * repeats can carve up the subject, and
       all have to be tested before failure can be reported.

       The values set(7,n,1 builtins) for any capturing subpatterns are those from the  outer-
       most  level  of  the recursion at which the subpattern value is set. If
       the pattern above is matched against

         (ab(cd)ef)

       the value for the capturing parentheses is  "ef",  which  is  the  last
       value  taken  on at the top level. If additional parentheses are added,
       giving

         \( ( ( (?>[^()]+) | (?R) )* ) \)
            ^                        ^
            ^                        ^ the string(3,n) they capture is  "ab(cd)ef",
       the  contents  of  the top level parentheses. If there are more than 15
       capturing parentheses in(1,8) a pattern, PCRE has to obtain extra memory  to
       store  data  during  a  recursion,  which it does by using pcre_malloc,
       freeing it via pcre_free afterwards. If no memory can be  obtained,  it
       saves  data for the first 15 capturing parentheses only, as there is no
       way to give an out-of-memory error(8,n) from within a recursion.



PERFORMANCE
       Certain items that may appear in(1,8) patterns are more efficient than  oth-
       ers.  It is more efficient to use a character class like [aeiou] than a
       set(7,n,1 builtins) of alternatives such as (a|e|i|o|u). In general, the simplest  con-
       struction  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(7,n,1 builtins), the
       pattern is implicitly anchored by PCRE, since it can match only at  the
       start  of  a  subject  string. However, if(3,n) PCRE_DOTALL is not set(7,n,1 builtins), PCRE
       cannot make this optimization, because the  .  metacharacter  does  not
       then  match a newline, and if(3,n) the subject string(3,n) 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(1,8)
       the subject.

       If you are using such a pattern with subject strings that do  not  con-
       tain 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(1,2,n) to run when applied to a string(3,n) that does not match.
       Consider the pattern fragment

         (a+)*

       This can match "aaaa" in(1,8) 33 different ways, and this  number  increases
       very  rapidly  as the string(3,n) gets(3,n) 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(1,8)
       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(1,8) the sub-
       ject string(3,n), and if(3,n) there is not, it fails the match immediately.  How-
       ever,  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(1,2,n) with strings longer than about 20 characters.



UTF-8 SUPPORT
       Starting  at  release  3.3, PCRE has some support for character strings
       encoded in(1,8) the UTF-8 format. This is incomplete,  and  is  regarded  as
       experimental.  In  order  to use it, you must configure PCRE to include
       UTF-8 support in(1,8) the code, and, in(1,8) addition, you  must  call  pcre_com-
       pile()  with the PCRE_UTF8 option flag. When you do this, both the pat-
       tern and any subject strings that are matched against it are treated as
       UTF-8  strings  instead of just strings of bytes, but only in(1,8) the cases
       that are mentioned below.

       If you compile PCRE with UTF-8 support, but do not use it at run  time(1,2,n),
       the  library will be a bit bigger, but the additional run time(1,2,n) overhead
       is limited to testing the PCRE_UTF8 flag in(1,8) several places,  so  should
       not be very large.

       PCRE assumes that the strings it is given contain valid UTF-8 codes. It
       does not diagnose invalid UTF-8 strings.  If  you  pass  invalid  UTF-8
       strings to PCRE, the results are undefined.

       Running with PCRE_UTF8 set(7,n,1 builtins) causes these changes in(1,8) the way PCRE works:

       1. In a pattern, the escape sequence \x{...}, where the contents of the
       braces is a string(3,n) of hexadecimal digits, is  interpreted  as  a  UTF-8
       character  whose code number is the given hexadecimal number, for exam-
       ple: \x{1234}. This inserts from one to six literal bytes into the pat-
       tern,  using  the  UTF-8  encoding.  If a non-hexadecimal digit appears
       between the braces, the item is not recognized.

       2. The original hexadecimal escape sequence, \xhh, generates a two-byte
       UTF-8 character if(3,n) its value is greater than 127.

       3. Repeat quantifiers are NOT correctly handled if(3,n) they follow a multi-
       byte character. For example, \x{100}* and \xc3+ do  not  work.  If  you
       want  to repeat such characters, you must enclose them in(1,8) non-capturing
       parentheses, for example (?:\x{100}), at present.

       4. The dot metacharacter matches one UTF-8 character instead of a  sin-
       gle byte.

       5. Unlike literal UTF-8 characters, the dot metacharacter followed by a
       repeat quantifier does operate correctly on UTF-8 characters instead of
       single bytes.

       4. Although the \x{...} escape is permitted in(1,8) a character class, char-
       acters whose values are greater than 255 cannot be included in(1,8) a class.

       5.  A class is matched against a UTF-8 character instead of just a sin-
       gle byte, but it can match only characters whose values are  less(1,3)  than
       256. Characters with greater values always fail to match a class.

       6. Repeated classes work correctly on multiple characters.

       7.  Classes  containing  just a single character whose value is greater
       than 127 (but less(1,3) than 256), for example, [\x80] or [^\x{93}], do  not
       work because these are optimized into single byte matches. In the first
       case, of course, the class brackets are just redundant.

       8. Lookbehind assertions move(3x,7,3x curs_move) backwards in(1,8) the subject by a fixed  num-
       ber of characters instead of a fixed number of bytes. Simple cases have
       been tested to work correctly, but there may be hidden gotchas  herein.

       9.  The  character  types  such as \d and \w do not work correctly with
       UTF-8 characters. They continue to test a single byte.

       10. Anything not explicitly mentioned here continues to work  in(1,8)  bytes
       rather than in(1,8) characters.

       The following UTF-8 features of Perl 5.6 are not implemented:

       1. The escape sequence \C to match a single byte.

       2. The use of Unicode tables and properties and escapes \p, \P, and \X.


AUTHOR
       Philip Hazel <ph10@cam.ac.uk>
       University Computing Service,
       New Museums Site,
       Cambridge CB2 3QG, England.
       Phone: +44 1223 334714

       Last updated: 28 August 2000,
         the 250th anniversary of the death of J.S. Bach.
       Copyright (c) 1997-2000 University of Cambridge.



                                                                       PCRE(3)

References for this manual (incoming links)