2 X<regular expression> X<regex> X<regexp>
4 perlre - Perl regular expressions
8 This page describes the syntax of regular expressions in Perl.
10 If you haven't used regular expressions before, a quick-start
11 introduction is available in L<perlrequick>, and a longer tutorial
12 introduction is available in L<perlretut>.
14 For reference on how regular expressions are used in matching
15 operations, plus various examples of the same, see discussions of
16 C<m//>, C<s///>, C<qr//> and C<??> in L<perlop/"Regexp Quote-Like
22 Matching operations can have various modifiers. Modifiers
23 that relate to the interpretation of the regular expression inside
24 are listed below. Modifiers that alter the way a regular expression
25 is used by Perl are detailed in L<perlop/"Regexp Quote-Like Operators"> and
26 L<perlop/"Gory details of parsing quoted constructs">.
31 X</i> X<regex, case-insensitive> X<regexp, case-insensitive>
32 X<regular expression, case-insensitive>
34 Do case-insensitive pattern matching.
36 If C<use locale> is in effect, the case map is taken from the current
37 locale. See L<perllocale>.
40 X</m> X<regex, multiline> X<regexp, multiline> X<regular expression, multiline>
42 Treat string as multiple lines. That is, change "^" and "$" from matching
43 the start or end of the string to matching the start or end of any
44 line anywhere within the string.
47 X</s> X<regex, single-line> X<regexp, single-line>
48 X<regular expression, single-line>
50 Treat string as single line. That is, change "." to match any character
51 whatsoever, even a newline, which normally it would not match.
53 Used together, as /ms, they let the "." match any character whatsoever,
54 while still allowing "^" and "$" to match, respectively, just after
55 and just before newlines within the string.
60 Extend your pattern's legibility by permitting whitespace and comments.
64 These are usually written as "the C</x> modifier", even though the delimiter
65 in question might not really be a slash. Any of these
66 modifiers may also be embedded within the regular expression itself using
67 the C<(?...)> construct. See below.
69 The C</x> modifier itself needs a little more explanation. It tells
70 the regular expression parser to ignore whitespace that is neither
71 backslashed nor within a character class. You can use this to break up
72 your regular expression into (slightly) more readable parts. The C<#>
73 character is also treated as a metacharacter introducing a comment,
74 just as in ordinary Perl code. This also means that if you want real
75 whitespace or C<#> characters in the pattern (outside a character
76 class, where they are unaffected by C</x>), then you'll either have to
77 escape them (using backslashes or C<\Q...\E>) or encode them using octal
78 or hex escapes. Taken together, these features go a long way towards
79 making Perl's regular expressions more readable. Note that you have to
80 be careful not to include the pattern delimiter in the comment--perl has
81 no way of knowing you did not intend to close the pattern early. See
82 the C-comment deletion code in L<perlop>. Also note that anything inside
83 a C<\Q...\E> stays unaffected by C</x>.
86 =head2 Regular Expressions
90 The patterns used in Perl pattern matching evolved from the ones supplied in
91 the Version 8 regex routines. (The routines are derived
92 (distantly) from Henry Spencer's freely redistributable reimplementation
93 of the V8 routines.) See L<Version 8 Regular Expressions> for
96 In particular the following metacharacters have their standard I<egrep>-ish
99 X<\> X<^> X<.> X<$> X<|> X<(> X<()> X<[> X<[]>
102 \ Quote the next metacharacter
103 ^ Match the beginning of the line
104 . Match any character (except newline)
105 $ Match the end of the line (or before newline at the end)
110 By default, the "^" character is guaranteed to match only the
111 beginning of the string, the "$" character only the end (or before the
112 newline at the end), and Perl does certain optimizations with the
113 assumption that the string contains only one line. Embedded newlines
114 will not be matched by "^" or "$". You may, however, wish to treat a
115 string as a multi-line buffer, such that the "^" will match after any
116 newline within the string (except if the newline is the last character in
117 the string), and "$" will match before any newline. At the
118 cost of a little more overhead, you can do this by using the /m modifier
119 on the pattern match operator. (Older programs did this by setting C<$*>,
120 but this practice has been removed in perl 5.9.)
123 To simplify multi-line substitutions, the "." character never matches a
124 newline unless you use the C</s> modifier, which in effect tells Perl to pretend
125 the string is a single line--even if it isn't.
130 The following standard quantifiers are recognized:
131 X<metacharacter> X<quantifier> X<*> X<+> X<?> X<{n}> X<{n,}> X<{n,m}>
133 * Match 0 or more times
134 + Match 1 or more times
136 {n} Match exactly n times
137 {n,} Match at least n times
138 {n,m} Match at least n but not more than m times
140 (If a curly bracket occurs in any other context, it is treated
141 as a regular character. In particular, the lower bound
142 is not optional.) The "*" modifier is equivalent to C<{0,}>, the "+"
143 modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
144 to integral values less than a preset limit defined when perl is built.
145 This is usually 32766 on the most common platforms. The actual limit can
146 be seen in the error message generated by code such as this:
148 $_ **= $_ , / {$_} / for 2 .. 42;
150 By default, a quantified subpattern is "greedy", that is, it will match as
151 many times as possible (given a particular starting location) while still
152 allowing the rest of the pattern to match. If you want it to match the
153 minimum number of times possible, follow the quantifier with a "?". Note
154 that the meanings don't change, just the "greediness":
155 X<metacharacter> X<greedy> X<greediness>
156 X<?> X<*?> X<+?> X<??> X<{n}?> X<{n,}?> X<{n,m}?>
158 *? Match 0 or more times, not greedily
159 +? Match 1 or more times, not greedily
160 ?? Match 0 or 1 time, not greedily
161 {n}? Match exactly n times, not greedily
162 {n,}? Match at least n times, not greedily
163 {n,m}? Match at least n but not more than m times, not greedily
165 By default, when a quantified subpattern does not allow the rest of the
166 overall pattern to match, Perl will backtrack. However, this behaviour is
167 sometimes undesirable. Thus Perl provides the "possessive" quantifier form
170 *+ Match 0 or more times and give nothing back
171 ++ Match 1 or more times and give nothing back
172 ?+ Match 0 or 1 time and give nothing back
173 {n}+ Match exactly n times and give nothing back (redundant)
174 {n,}+ Match at least n times and give nothing back
175 {n,m}+ Match at least n but not more than m times and give nothing back
181 will never match, as the C<a++> will gobble up all the C<a>'s in the
182 string and won't leave any for the remaining part of the pattern. This
183 feature can be extremely useful to give perl hints about where it
184 shouldn't backtrack. For instance, the typical "match a double-quoted
185 string" problem can be most efficiently performed when written as:
187 /"(?:[^"\\]++|\\.)*+"/
189 as we know that if the final quote does not match, backtracking will not
190 help. See the independent subexpression C<< (?>...) >> for more details;
191 possessive quantifiers are just syntactic sugar for that construct. For
192 instance the above example could also be written as follows:
194 /"(?>(?:(?>[^"\\]+)|\\.)*)"/
196 =head3 Escape sequences
198 Because patterns are processed as double quoted strings, the following
200 X<\t> X<\n> X<\r> X<\f> X<\e> X<\a> X<\l> X<\u> X<\L> X<\U> X<\E> X<\Q>
201 X<\0> X<\c> X<\N> X<\x>
207 \a alarm (bell) (BEL)
208 \e escape (think troff) (ESC)
209 \033 octal char (example: ESC)
210 \x1B hex char (example: ESC)
211 \x{263a} wide hex char (example: Unicode SMILEY)
212 \cK control char (example: VT)
214 \l lowercase next char (think vi)
215 \u uppercase next char (think vi)
216 \L lowercase till \E (think vi)
217 \U uppercase till \E (think vi)
218 \E end case modification (think vi)
219 \Q quote (disable) pattern metacharacters till \E
221 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
222 and C<\U> is taken from the current locale. See L<perllocale>. For
223 documentation of C<\N{name}>, see L<charnames>.
225 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
226 An unescaped C<$> or C<@> interpolates the corresponding variable,
227 while escaping will cause the literal string C<\$> to be matched.
228 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
230 =head3 Character classes
232 In addition, Perl defines the following:
233 X<\w> X<\W> X<\s> X<\S> X<\d> X<\D> X<\X> X<\p> X<\P> X<\C>
234 X<\g> X<\k> X<\N> X<\K> X<\v> X<\V>
235 X<word> X<whitespace> X<character class> X<backreference>
237 \w Match a "word" character (alphanumeric plus "_")
238 \W Match a non-"word" character
239 \s Match a whitespace character
240 \S Match a non-whitespace character
241 \d Match a digit character
242 \D Match a non-digit character
243 \pP Match P, named property. Use \p{Prop} for longer names.
245 \X Match eXtended Unicode "combining character sequence",
246 equivalent to (?:\PM\pM*)
247 \C Match a single C char (octet) even under Unicode.
248 NOTE: breaks up characters into their UTF-8 bytes,
249 so you may end up with malformed pieces of UTF-8.
250 Unsupported in lookbehind.
251 \1 Backreference to a specific group.
252 '1' may actually be any positive integer.
253 \g1 Backreference to a specific or previous group,
254 \g{-1} number may be negative indicating a previous buffer and may
255 optionally be wrapped in curly brackets for safer parsing.
256 \g{name} Named backreference
257 \k<name> Named backreference
258 \N{name} Named unicode character, or unicode escape
259 \x12 Hexadecimal escape sequence
260 \x{1234} Long hexadecimal escape sequence
261 \K Keep the stuff left of the \K, don't include it in $&
262 \v Shortcut for (*PRUNE)
263 \V Shortcut for (*SKIP)
265 A C<\w> matches a single alphanumeric character (an alphabetic
266 character, or a decimal digit) or C<_>, not a whole word. Use C<\w+>
267 to match a string of Perl-identifier characters (which isn't the same
268 as matching an English word). If C<use locale> is in effect, the list
269 of alphabetic characters generated by C<\w> is taken from the current
270 locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
271 C<\d>, and C<\D> within character classes, but they aren't usable
272 as either end of a range. If any of them precedes or follows a "-",
273 the "-" is understood literally. If Unicode is in effect, C<\s> matches
274 also "\x{85}", "\x{2028}, and "\x{2029}". See L<perlunicode> for more
275 details about C<\pP>, C<\PP>, C<\X> and the possibility of defining
276 your own C<\p> and C<\P> properties, and L<perluniintro> about Unicode
280 The POSIX character class syntax
285 is also available. Note that the C<[> and C<]> brackets are I<literal>;
286 they must always be used within a character class expression.
289 $string =~ /[[:alpha:]]/;
291 # this is not, and will generate a warning:
292 $string =~ /[:alpha:]/;
294 The available classes and their backslash equivalents (if available) are
297 X<alpha> X<alnum> X<ascii> X<blank> X<cntrl> X<digit> X<graph>
298 X<lower> X<print> X<punct> X<space> X<upper> X<word> X<xdigit>
319 A GNU extension equivalent to C<[ \t]>, "all horizontal whitespace".
323 Not exactly equivalent to C<\s> since the C<[[:space:]]> includes
324 also the (very rare) "vertical tabulator", "\cK" or chr(11) in ASCII.
328 A Perl extension, see above.
332 For example use C<[:upper:]> to match all the uppercase characters.
333 Note that the C<[]> are part of the C<[::]> construct, not part of the
334 whole character class. For example:
338 matches zero, one, any alphabetic character, and the percent sign.
340 The following equivalences to Unicode \p{} constructs and equivalent
341 backslash character classes (if available), will hold:
342 X<character class> X<\p> X<\p{}>
344 [[:...:]] \p{...} backslash
362 For example C<[[:lower:]]> and C<\p{IsLower}> are equivalent.
364 If the C<utf8> pragma is not used but the C<locale> pragma is, the
365 classes correlate with the usual isalpha(3) interface (except for
368 The assumedly non-obviously named classes are:
375 Any control character. Usually characters that don't produce output as
376 such but instead control the terminal somehow: for example newline and
377 backspace are control characters. All characters with ord() less than
378 32 are usually classified as control characters (assuming ASCII,
379 the ISO Latin character sets, and Unicode), as is the character with
380 the ord() value of 127 (C<DEL>).
385 Any alphanumeric or punctuation (special) character.
390 Any alphanumeric or punctuation (special) character or the space character.
395 Any punctuation (special) character.
400 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
401 work just fine) it is included for completeness.
405 You can negate the [::] character classes by prefixing the class name
406 with a '^'. This is a Perl extension. For example:
407 X<character class, negation>
409 POSIX traditional Unicode
411 [[:^digit:]] \D \P{IsDigit}
412 [[:^space:]] \S \P{IsSpace}
413 [[:^word:]] \W \P{IsWord}
415 Perl respects the POSIX standard in that POSIX character classes are
416 only supported within a character class. The POSIX character classes
417 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
418 use them will cause an error.
422 Perl defines the following zero-width assertions:
423 X<zero-width assertion> X<assertion> X<regex, zero-width assertion>
424 X<regexp, zero-width assertion>
425 X<regular expression, zero-width assertion>
426 X<\b> X<\B> X<\A> X<\Z> X<\z> X<\G>
428 \b Match a word boundary
429 \B Match except at a word boundary
430 \A Match only at beginning of string
431 \Z Match only at end of string, or before newline at the end
432 \z Match only at end of string
433 \G Match only at pos() (e.g. at the end-of-match position
436 A word boundary (C<\b>) is a spot between two characters
437 that has a C<\w> on one side of it and a C<\W> on the other side
438 of it (in either order), counting the imaginary characters off the
439 beginning and end of the string as matching a C<\W>. (Within
440 character classes C<\b> represents backspace rather than a word
441 boundary, just as it normally does in any double-quoted string.)
442 The C<\A> and C<\Z> are just like "^" and "$", except that they
443 won't match multiple times when the C</m> modifier is used, while
444 "^" and "$" will match at every internal line boundary. To match
445 the actual end of the string and not ignore an optional trailing
447 X<\b> X<\A> X<\Z> X<\z> X</m>
449 The C<\G> assertion can be used to chain global matches (using
450 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
451 It is also useful when writing C<lex>-like scanners, when you have
452 several patterns that you want to match against consequent substrings
453 of your string, see the previous reference. The actual location
454 where C<\G> will match can also be influenced by using C<pos()> as
455 an lvalue: see L<perlfunc/pos>. Note that the rule for zero-length
456 matches is modified somewhat, in that contents to the left of C<\G> is
457 not counted when determining the length of the match. Thus the following
458 will not match forever:
467 It will print 'A' and then terminate, as it considers the match to
468 be zero-width, and thus will not match at the same position twice in a
471 It is worth noting that C<\G> improperly used can result in an infinite
472 loop. Take care when using patterns that include C<\G> in an alternation.
474 =head3 Capture buffers
476 The bracketing construct C<( ... )> creates capture buffers. To refer
477 to the current contents of a buffer later on, within the same pattern,
478 use \1 for the first, \2 for the second, and so on.
479 Outside the match use "$" instead of "\". (The
480 \<digit> notation works in certain circumstances outside
481 the match. See the warning below about \1 vs $1 for details.)
482 Referring back to another part of the match is called a
484 X<regex, capture buffer> X<regexp, capture buffer>
485 X<regular expression, capture buffer> X<backreference>
487 There is no limit to the number of captured substrings that you may
488 use. However Perl also uses \10, \11, etc. as aliases for \010,
489 \011, etc. (Recall that 0 means octal, so \011 is the character at
490 number 9 in your coded character set; which would be the 10th character,
491 a horizontal tab under ASCII.) Perl resolves this
492 ambiguity by interpreting \10 as a backreference only if at least 10
493 left parentheses have opened before it. Likewise \11 is a
494 backreference only if at least 11 left parentheses have opened
495 before it. And so on. \1 through \9 are always interpreted as
498 X<\g{1}> X<\g{-1}> X<\g{name}> X<relative backreference> X<named backreference>
499 In order to provide a safer and easier way to construct patterns using
500 backreferences, Perl 5.10 provides the C<\g{N}> notation. The curly
501 brackets are optional, however omitting them is less safe as the meaning
502 of the pattern can be changed by text (such as digits) following it.
503 When N is a positive integer the C<\g{N}> notation is exactly equivalent
504 to using normal backreferences. When N is a negative integer then it is
505 a relative backreference referring to the previous N'th capturing group.
506 When the bracket form is used and N is not an integer, it is treated as a
507 reference to a named buffer.
509 Thus C<\g{-1}> refers to the last buffer, C<\g{-2}> refers to the
510 buffer before that. For example:
516 \g{-1} # backref to buffer 3
517 \g{-3} # backref to buffer 1
521 and would match the same as C</(Y) ( (X) \3 \1 )/x>.
523 Additionally, as of Perl 5.10 you may use named capture buffers and named
524 backreferences. The notation is C<< (?<name>...) >> to declare and C<< \k<name> >>
525 to reference. You may also use apostrophes instead of angle brackets to delimit the
526 name; and you may use the bracketed C<< \g{name} >> backreference syntax.
527 It's possible to refer to a named capture buffer by absolute and relative number as well.
528 Outside the pattern, a named capture buffer is available via the C<%+> hash.
529 When different buffers within the same pattern have the same name, C<$+{name}>
530 and C<< \k<name> >> refer to the leftmost defined group. (Thus it's possible
531 to do things with named capture buffers that would otherwise require C<(??{})>
533 X<named capture buffer> X<regular expression, named capture buffer>
534 X<%+> X<$+{name}> X<\k{name}>
538 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
540 /(.)\1/ # find first doubled char
541 and print "'$1' is the first doubled character\n";
543 /(?<char>.)\k<char>/ # ... a different way
544 and print "'$+{char}' is the first doubled character\n";
546 /(?'char'.)\1/ # ... mix and match
547 and print "'$1' is the first doubled character\n";
549 if (/Time: (..):(..):(..)/) { # parse out values
555 Several special variables also refer back to portions of the previous
556 match. C<$+> returns whatever the last bracket match matched.
557 C<$&> returns the entire matched string. (At one point C<$0> did
558 also, but now it returns the name of the program.) C<$`> returns
559 everything before the matched string. C<$'> returns everything
560 after the matched string. And C<$^N> contains whatever was matched by
561 the most-recently closed group (submatch). C<$^N> can be used in
562 extended patterns (see below), for example to assign a submatch to a
564 X<$+> X<$^N> X<$&> X<$`> X<$'>
566 The numbered match variables ($1, $2, $3, etc.) and the related punctuation
567 set (C<$+>, C<$&>, C<$`>, C<$'>, and C<$^N>) are all dynamically scoped
568 until the end of the enclosing block or until the next successful
569 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
570 X<$+> X<$^N> X<$&> X<$`> X<$'>
571 X<$1> X<$2> X<$3> X<$4> X<$5> X<$6> X<$7> X<$8> X<$9>
574 B<NOTE>: Failed matches in Perl do not reset the match variables,
575 which makes it easier to write code that tests for a series of more
576 specific cases and remembers the best match.
578 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
579 C<$'> anywhere in the program, it has to provide them for every
580 pattern match. This may substantially slow your program. Perl
581 uses the same mechanism to produce $1, $2, etc, so you also pay a
582 price for each pattern that contains capturing parentheses. (To
583 avoid this cost while retaining the grouping behaviour, use the
584 extended regular expression C<(?: ... )> instead.) But if you never
585 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
586 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
587 if you can, but if you can't (and some algorithms really appreciate
588 them), once you've used them once, use them at will, because you've
589 already paid the price. As of 5.005, C<$&> is not so costly as the
593 As a workaround for this problem, Perl 5.10 introduces C<${^PREMATCH}>,
594 C<${^MATCH}> and C<${^POSTMATCH}>, which are equivalent to C<$`>, C<$&>
595 and C<$'>, B<except> that they are only guaranteed to be defined after a
596 successful match that was executed with the C</k> (keep-copy) modifier.
597 The use of these variables incurs no global performance penalty, unlike
598 their punctuation char equivalents, however at the trade-off that you
599 have to tell perl when you want to use them.
602 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
603 C<\w>, C<\n>. Unlike some other regular expression languages, there
604 are no backslashed symbols that aren't alphanumeric. So anything
605 that looks like \\, \(, \), \<, \>, \{, or \} is always
606 interpreted as a literal character, not a metacharacter. This was
607 once used in a common idiom to disable or quote the special meanings
608 of regular expression metacharacters in a string that you want to
609 use for a pattern. Simply quote all non-"word" characters:
611 $pattern =~ s/(\W)/\\$1/g;
613 (If C<use locale> is set, then this depends on the current locale.)
614 Today it is more common to use the quotemeta() function or the C<\Q>
615 metaquoting escape sequence to disable all metacharacters' special
618 /$unquoted\Q$quoted\E$unquoted/
620 Beware that if you put literal backslashes (those not inside
621 interpolated variables) between C<\Q> and C<\E>, double-quotish
622 backslash interpolation may lead to confusing results. If you
623 I<need> to use literal backslashes within C<\Q...\E>,
624 consult L<perlop/"Gory details of parsing quoted constructs">.
626 =head2 Extended Patterns
628 Perl also defines a consistent extension syntax for features not
629 found in standard tools like B<awk> and B<lex>. The syntax is a
630 pair of parentheses with a question mark as the first thing within
631 the parentheses. The character after the question mark indicates
634 The stability of these extensions varies widely. Some have been
635 part of the core language for many years. Others are experimental
636 and may change without warning or be completely removed. Check
637 the documentation on an individual feature to verify its current
640 A question mark was chosen for this and for the minimal-matching
641 construct because 1) question marks are rare in older regular
642 expressions, and 2) whenever you see one, you should stop and
643 "question" exactly what is going on. That's psychology...
650 A comment. The text is ignored. If the C</x> modifier enables
651 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
652 the comment as soon as it sees a C<)>, so there is no way to put a literal
655 =item C<(?kimsx-imsx)>
658 One or more embedded pattern-match modifiers, to be turned on (or
659 turned off, if preceded by C<->) for the remainder of the pattern or
660 the remainder of the enclosing pattern group (if any). This is
661 particularly useful for dynamic patterns, such as those read in from a
662 configuration file, taken from an argument, or specified in a table
663 somewhere. Consider the case where some patterns want to be case
664 sensitive and some do not: The case insensitive ones merely need to
665 include C<(?i)> at the front of the pattern. For example:
668 if ( /$pattern/i ) { }
672 $pattern = "(?i)foobar";
673 if ( /$pattern/ ) { }
675 These modifiers are restored at the end of the enclosing group. For example,
679 will match C<blah> in any case, some spaces, and an exact (I<including the case>!)
680 repetition of the previous word, assuming the C</x> modifier, and no C</i>
681 modifier outside this group.
683 Note that the C<k> modifier is special in that it can only be enabled,
684 not disabled, and that its presence anywhere in a pattern has a global
685 effect. Thus C<(?-k)> and C<(?-k:...)> are meaningless and will warn
686 when executed under C<use warnings>.
691 =item C<(?imsx-imsx:pattern)>
693 This is for clustering, not capturing; it groups subexpressions like
694 "()", but doesn't make backreferences as "()" does. So
696 @fields = split(/\b(?:a|b|c)\b/)
700 @fields = split(/\b(a|b|c)\b/)
702 but doesn't spit out extra fields. It's also cheaper not to capture
703 characters if you don't need to.
705 Any letters between C<?> and C<:> act as flags modifiers as with
706 C<(?imsx-imsx)>. For example,
708 /(?s-i:more.*than).*million/i
710 is equivalent to the more verbose
712 /(?:(?s-i)more.*than).*million/i
715 X<(?|)> X<Branch reset>
717 This is the "branch reset" pattern, which has the special property
718 that the capture buffers are numbered from the same starting point
721 Normally capture buffers in a pattern are number sequentially, left
722 to right in the pattern. Inside of this construct this behaviour is
723 overriden so that the captures buffers in each branch share the same
724 numbers. The numbering in each branch will be as normal, and any
725 buffers following the use of this pattern will be numbered as though
726 the construct contained only one branch, that being the one with the
727 most capture buffers in it.
729 Consider the following pattern. The numbers underneath are which
730 buffer number the captured content will be stored in.
733 # before ---------------branch-reset----------- after
734 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
737 =item Look-Around Assertions
738 X<look-around assertion> X<lookaround assertion> X<look-around> X<lookaround>
740 Look-around assertions are zero width patterns which match a specific
741 pattern without including it in C<$&>. Positive assertions match when
742 their subpattern matches, negative assertions match when their subpattern
743 fails. Look-behind matches text up to the current match position,
744 look-ahead matches text following the current match position.
749 X<(?=)> X<look-ahead, positive> X<lookahead, positive>
751 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
752 matches a word followed by a tab, without including the tab in C<$&>.
755 X<(?!)> X<look-ahead, negative> X<lookahead, negative>
757 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
758 matches any occurrence of "foo" that isn't followed by "bar". Note
759 however that look-ahead and look-behind are NOT the same thing. You cannot
760 use this for look-behind.
762 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
763 will not do what you want. That's because the C<(?!foo)> is just saying that
764 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
765 match. You would have to do something like C</(?!foo)...bar/> for that. We
766 say "like" because there's the case of your "bar" not having three characters
767 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
768 Sometimes it's still easier just to say:
770 if (/bar/ && $` !~ /foo$/)
772 For look-behind see below.
774 =item C<(?<=pattern)> C<\K>
775 X<(?<=)> X<look-behind, positive> X<lookbehind, positive> X<\K>
777 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
778 matches a word that follows a tab, without including the tab in C<$&>.
779 Works only for fixed-width look-behind.
781 There is a special form of this construct, called C<\K>, which causes the
782 regex engine to "keep" everything it had matched prior to the C<\K> and
783 not include it in C<$&>. This effectively provides variable length
784 look-behind. The use of C<\K> inside of another look-around assertion
785 is allowed, but the behaviour is currently not well defined.
787 For various reasons C<\K> may be signifigantly more efficient than the
788 equivalent C<< (?<=...) >> construct, and it is especially useful in
789 situations where you want to efficiently remove something following
790 something else in a string. For instance
794 can be rewritten as the much more efficient
798 =item C<(?<!pattern)>
799 X<(?<!)> X<look-behind, negative> X<lookbehind, negative>
801 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
802 matches any occurrence of "foo" that does not follow "bar". Works
803 only for fixed-width look-behind.
807 =item C<(?'NAME'pattern)>
809 =item C<< (?<NAME>pattern) >>
810 X<< (?<NAME>) >> X<(?'NAME')> X<named capture> X<capture>
812 A named capture buffer. Identical in every respect to normal capturing
813 parentheses C<()> but for the additional fact that C<%+> may be used after
814 a succesful match to refer to a named buffer. See C<perlvar> for more
815 details on the C<%+> hash.
817 If multiple distinct capture buffers have the same name then the
818 $+{NAME} will refer to the leftmost defined buffer in the match.
820 The forms C<(?'NAME'pattern)> and C<< (?<NAME>pattern) >> are equivalent.
822 B<NOTE:> While the notation of this construct is the same as the similar
823 function in .NET regexes, the behavior is not. In Perl the buffers are
824 numbered sequentially regardless of being named or not. Thus in the
829 $+{foo} will be the same as $2, and $3 will contain 'z' instead of
830 the opposite which is what a .NET regex hacker might expect.
832 Currently NAME is restricted to simple identifiers only.
833 In other words, it must match C</^[_A-Za-z][_A-Za-z0-9]*\z/> or
834 its Unicode extension (see L<utf8>),
835 though it isn't extended by the locale (see L<perllocale>).
837 B<NOTE:> In order to make things easier for programmers with experience
838 with the Python or PCRE regex engines, the pattern C<< (?PE<lt>NAMEE<gt>pattern) >>
839 may be used instead of C<< (?<NAME>pattern) >>; however this form does not
840 support the use of single quotes as a delimiter for the name. This is
841 only available in Perl 5.10 or later.
843 =item C<< \k<NAME> >>
845 =item C<< \k'NAME' >>
847 Named backreference. Similar to numeric backreferences, except that
848 the group is designated by name and not number. If multiple groups
849 have the same name then it refers to the leftmost defined group in
852 It is an error to refer to a name not defined by a C<< (?<NAME>) >>
853 earlier in the pattern.
855 Both forms are equivalent.
857 B<NOTE:> In order to make things easier for programmers with experience
858 with the Python or PCRE regex engines, the pattern C<< (?P=NAME) >>
859 may be used instead of C<< \k<NAME> >> in Perl 5.10 or later.
862 X<(?{})> X<regex, code in> X<regexp, code in> X<regular expression, code in>
864 B<WARNING>: This extended regular expression feature is considered
865 experimental, and may be changed without notice. Code executed that
866 has side effects may not perform identically from version to version
867 due to the effect of future optimisations in the regex engine.
869 This zero-width assertion evaluates any embedded Perl code. It
870 always succeeds, and its C<code> is not interpolated. Currently,
871 the rules to determine where the C<code> ends are somewhat convoluted.
873 This feature can be used together with the special variable C<$^N> to
874 capture the results of submatches in variables without having to keep
875 track of the number of nested parentheses. For example:
877 $_ = "The brown fox jumps over the lazy dog";
878 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
879 print "color = $color, animal = $animal\n";
881 Inside the C<(?{...})> block, C<$_> refers to the string the regular
882 expression is matching against. You can also use C<pos()> to know what is
883 the current position of matching within this string.
885 The C<code> is properly scoped in the following sense: If the assertion
886 is backtracked (compare L<"Backtracking">), all changes introduced after
887 C<local>ization are undone, so that
891 (?{ $cnt = 0 }) # Initialize $cnt.
895 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
899 (?{ $res = $cnt }) # On success copy to non-localized
903 will set C<$res = 4>. Note that after the match, C<$cnt> returns to the globally
904 introduced value, because the scopes that restrict C<local> operators
907 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
908 switch. If I<not> used in this way, the result of evaluation of
909 C<code> is put into the special variable C<$^R>. This happens
910 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
911 inside the same regular expression.
913 The assignment to C<$^R> above is properly localized, so the old
914 value of C<$^R> is restored if the assertion is backtracked; compare
917 Due to an unfortunate implementation issue, the Perl code contained in these
918 blocks is treated as a compile time closure that can have seemingly bizarre
919 consequences when used with lexically scoped variables inside of subroutines
920 or loops. There are various workarounds for this, including simply using
921 global variables instead. If you are using this construct and strange results
922 occur then check for the use of lexically scoped variables.
924 For reasons of security, this construct is forbidden if the regular
925 expression involves run-time interpolation of variables, unless the
926 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
927 variables contain results of C<qr//> operator (see
928 L<perlop/"qr/STRING/imosx">).
930 This restriction is due to the wide-spread and remarkably convenient
931 custom of using run-time determined strings as patterns. For example:
937 Before Perl knew how to execute interpolated code within a pattern,
938 this operation was completely safe from a security point of view,
939 although it could raise an exception from an illegal pattern. If
940 you turn on the C<use re 'eval'>, though, it is no longer secure,
941 so you should only do so if you are also using taint checking.
942 Better yet, use the carefully constrained evaluation within a Safe
943 compartment. See L<perlsec> for details about both these mechanisms.
945 Because Perl's regex engine is currently not re-entrant, interpolated
946 code may not invoke the regex engine either directly with C<m//> or C<s///>),
947 or indirectly with functions such as C<split>.
949 =item C<(??{ code })>
951 X<regex, postponed> X<regexp, postponed> X<regular expression, postponed>
953 B<WARNING>: This extended regular expression feature is considered
954 experimental, and may be changed without notice. Code executed that
955 has side effects may not perform identically from version to version
956 due to the effect of future optimisations in the regex engine.
958 This is a "postponed" regular subexpression. The C<code> is evaluated
959 at run time, at the moment this subexpression may match. The result
960 of evaluation is considered as a regular expression and matched as
961 if it were inserted instead of this construct. Note that this means
962 that the contents of capture buffers defined inside an eval'ed pattern
963 are not available outside of the pattern, and vice versa, there is no
964 way for the inner pattern to refer to a capture buffer defined outside.
967 ('a' x 100)=~/(??{'(.)' x 100})/
969 B<will> match, it will B<not> set $1.
971 The C<code> is not interpolated. As before, the rules to determine
972 where the C<code> ends are currently somewhat convoluted.
974 The following pattern matches a parenthesized group:
979 (?> [^()]+ ) # Non-parens without backtracking
981 (??{ $re }) # Group with matching parens
986 See also C<(?PARNO)> for a different, more efficient way to accomplish
989 Because perl's regex engine is not currently re-entrant, delayed
990 code may not invoke the regex engine either directly with C<m//> or C<s///>),
991 or indirectly with functions such as C<split>.
993 Recursing deeper than 50 times without consuming any input string will
994 result in a fatal error. The maximum depth is compiled into perl, so
995 changing it requires a custom build.
997 =item C<(?PARNO)> C<(?-PARNO)> C<(?+PARNO)> C<(?R)> C<(?0)>
998 X<(?PARNO)> X<(?1)> X<(?R)> X<(?0)> X<(?-1)> X<(?+1)> X<(?-PARNO)> X<(?+PARNO)>
999 X<regex, recursive> X<regexp, recursive> X<regular expression, recursive>
1000 X<regex, relative recursion>
1002 Similar to C<(??{ code })> except it does not involve compiling any code,
1003 instead it treats the contents of a capture buffer as an independent
1004 pattern that must match at the current position. Capture buffers
1005 contained by the pattern will have the value as determined by the
1006 outermost recursion.
1008 PARNO is a sequence of digits (not starting with 0) whose value reflects
1009 the paren-number of the capture buffer to recurse to. C<(?R)> recurses to
1010 the beginning of the whole pattern. C<(?0)> is an alternate syntax for
1011 C<(?R)>. If PARNO is preceded by a plus or minus sign then it is assumed
1012 to be relative, with negative numbers indicating preceding capture buffers
1013 and positive ones following. Thus C<(?-1)> refers to the most recently
1014 declared buffer, and C<(?+1)> indicates the next buffer to be declared.
1015 Note that the counting for relative recursion differs from that of
1016 relative backreferences, in that with recursion unclosed buffers B<are>
1019 The following pattern matches a function foo() which may contain
1020 balanced parentheses as the argument.
1022 $re = qr{ ( # paren group 1 (full function)
1024 ( # paren group 2 (parens)
1026 ( # paren group 3 (contents of parens)
1028 (?> [^()]+ ) # Non-parens without backtracking
1030 (?2) # Recurse to start of paren group 2
1038 If the pattern was used as follows
1040 'foo(bar(baz)+baz(bop))'=~/$re/
1041 and print "\$1 = $1\n",
1045 the output produced should be the following:
1047 $1 = foo(bar(baz)+baz(bop))
1048 $2 = (bar(baz)+baz(bop))
1049 $3 = bar(baz)+baz(bop)
1051 If there is no corresponding capture buffer defined, then it is a
1052 fatal error. Recursing deeper than 50 times without consuming any input
1053 string will also result in a fatal error. The maximum depth is compiled
1054 into perl, so changing it requires a custom build.
1056 The following shows how using negative indexing can make it
1057 easier to embed recursive patterns inside of a C<qr//> construct
1060 my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1061 if (/foo $parens \s+ + \s+ bar $parens/x) {
1062 # do something here...
1065 B<Note> that this pattern does not behave the same way as the equivalent
1066 PCRE or Python construct of the same form. In Perl you can backtrack into
1067 a recursed group, in PCRE and Python the recursed into group is treated
1068 as atomic. Also, modifiers are resolved at compile time, so constructs
1069 like (?i:(?1)) or (?:(?i)(?1)) do not affect how the sub-pattern will
1075 Recurse to a named subpattern. Identical to C<(?PARNO)> except that the
1076 parenthesis to recurse to is determined by name. If multiple parentheses have
1077 the same name, then it recurses to the leftmost.
1079 It is an error to refer to a name that is not declared somewhere in the
1082 B<NOTE:> In order to make things easier for programmers with experience
1083 with the Python or PCRE regex engines the pattern C<< (?P>NAME) >>
1084 may be used instead of C<< (?&NAME) >> in Perl 5.10 or later.
1086 =item C<(?(condition)yes-pattern|no-pattern)>
1089 =item C<(?(condition)yes-pattern)>
1091 Conditional expression. C<(condition)> should be either an integer in
1092 parentheses (which is valid if the corresponding pair of parentheses
1093 matched), a look-ahead/look-behind/evaluate zero-width assertion, a
1094 name in angle brackets or single quotes (which is valid if a buffer
1095 with the given name matched), or the special symbol (R) (true when
1096 evaluated inside of recursion or eval). Additionally the R may be
1097 followed by a number, (which will be true when evaluated when recursing
1098 inside of the appropriate group), or by C<&NAME>, in which case it will
1099 be true only when evaluated during recursion in the named group.
1101 Here's a summary of the possible predicates:
1107 Checks if the numbered capturing buffer has matched something.
1109 =item (<NAME>) ('NAME')
1111 Checks if a buffer with the given name has matched something.
1115 Treats the code block as the condition.
1119 Checks if the expression has been evaluated inside of recursion.
1123 Checks if the expression has been evaluated while executing directly
1124 inside of the n-th capture group. This check is the regex equivalent of
1126 if ((caller(0))[3] eq 'subname') { ... }
1128 In other words, it does not check the full recursion stack.
1132 Similar to C<(R1)>, this predicate checks to see if we're executing
1133 directly inside of the leftmost group with a given name (this is the same
1134 logic used by C<(?&NAME)> to disambiguate). It does not check the full
1135 stack, but only the name of the innermost active recursion.
1139 In this case, the yes-pattern is never directly executed, and no
1140 no-pattern is allowed. Similar in spirit to C<(?{0})> but more efficient.
1141 See below for details.
1152 matches a chunk of non-parentheses, possibly included in parentheses
1155 A special form is the C<(DEFINE)> predicate, which never executes directly
1156 its yes-pattern, and does not allow a no-pattern. This allows to define
1157 subpatterns which will be executed only by using the recursion mechanism.
1158 This way, you can define a set of regular expression rules that can be
1159 bundled into any pattern you choose.
1161 It is recommended that for this usage you put the DEFINE block at the
1162 end of the pattern, and that you name any subpatterns defined within it.
1164 Also, it's worth noting that patterns defined this way probably will
1165 not be as efficient, as the optimiser is not very clever about
1168 An example of how this might be used is as follows:
1170 /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
1176 Note that capture buffers matched inside of recursion are not accessible
1177 after the recursion returns, so the extra layer of capturing buffers is
1178 necessary. Thus C<$+{NAME_PAT}> would not be defined even though
1179 C<$+{NAME}> would be.
1181 =item C<< (?>pattern) >>
1182 X<backtrack> X<backtracking> X<atomic> X<possessive>
1184 An "independent" subexpression, one which matches the substring
1185 that a I<standalone> C<pattern> would match if anchored at the given
1186 position, and it matches I<nothing other than this substring>. This
1187 construct is useful for optimizations of what would otherwise be
1188 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
1189 It may also be useful in places where the "grab all you can, and do not
1190 give anything back" semantic is desirable.
1192 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
1193 (anchored at the beginning of string, as above) will match I<all>
1194 characters C<a> at the beginning of string, leaving no C<a> for
1195 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
1196 since the match of the subgroup C<a*> is influenced by the following
1197 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
1198 C<a*ab> will match fewer characters than a standalone C<a*>, since
1199 this makes the tail match.
1201 An effect similar to C<< (?>pattern) >> may be achieved by writing
1202 C<(?=(pattern))\1>. This matches the same substring as a standalone
1203 C<a+>, and the following C<\1> eats the matched string; it therefore
1204 makes a zero-length assertion into an analogue of C<< (?>...) >>.
1205 (The difference between these two constructs is that the second one
1206 uses a capturing group, thus shifting ordinals of backreferences
1207 in the rest of a regular expression.)
1209 Consider this pattern:
1220 That will efficiently match a nonempty group with matching parentheses
1221 two levels deep or less. However, if there is no such group, it
1222 will take virtually forever on a long string. That's because there
1223 are so many different ways to split a long string into several
1224 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
1225 to a subpattern of the above pattern. Consider how the pattern
1226 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
1227 seconds, but that each extra letter doubles this time. This
1228 exponential performance will make it appear that your program has
1229 hung. However, a tiny change to this pattern
1233 (?> [^()]+ ) # change x+ above to (?> x+ )
1240 which uses C<< (?>...) >> matches exactly when the one above does (verifying
1241 this yourself would be a productive exercise), but finishes in a fourth
1242 the time when used on a similar string with 1000000 C<a>s. Be aware,
1243 however, that this pattern currently triggers a warning message under
1244 the C<use warnings> pragma or B<-w> switch saying it
1245 C<"matches null string many times in regex">.
1247 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
1248 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
1249 This was only 4 times slower on a string with 1000000 C<a>s.
1251 The "grab all you can, and do not give anything back" semantic is desirable
1252 in many situations where on the first sight a simple C<()*> looks like
1253 the correct solution. Suppose we parse text with comments being delimited
1254 by C<#> followed by some optional (horizontal) whitespace. Contrary to
1255 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
1256 the comment delimiter, because it may "give up" some whitespace if
1257 the remainder of the pattern can be made to match that way. The correct
1258 answer is either one of these:
1263 For example, to grab non-empty comments into $1, one should use either
1266 / (?> \# [ \t]* ) ( .+ ) /x;
1267 / \# [ \t]* ( [^ \t] .* ) /x;
1269 Which one you pick depends on which of these expressions better reflects
1270 the above specification of comments.
1272 In some literature this construct is called "atomic matching" or
1273 "possessive matching".
1275 Possessive quantifiers are equivalent to putting the item they are applied
1276 to inside of one of these constructs. The following equivalences apply:
1278 Quantifier Form Bracketing Form
1279 --------------- ---------------
1283 PAT{min,max}+ (?>PAT{min,max})
1287 =head2 Special Backtracking Control Verbs
1289 B<WARNING:> These patterns are experimental and subject to change or
1290 removal in a future version of Perl. Their usage in production code should
1291 be noted to avoid problems during upgrades.
1293 These special patterns are generally of the form C<(*VERB:ARG)>. Unless
1294 otherwise stated the ARG argument is optional; in some cases, it is
1297 Any pattern containing a special backtracking verb that allows an argument
1298 has the special behaviour that when executed it sets the current packages'
1299 C<$REGERROR> and C<$REGMARK> variables. When doing so the following
1302 On failure, the C<$REGERROR> variable will be set to the ARG value of the
1303 verb pattern, if the verb was involved in the failure of the match. If the
1304 ARG part of the pattern was omitted, then C<$REGERROR> will be set to the
1305 name of the last C<(*MARK:NAME)> pattern executed, or to TRUE if there was
1306 none. Also, the C<$REGMARK> variable will be set to FALSE.
1308 On a successful match, the C<$REGERROR> variable will be set to FALSE, and
1309 the C<$REGMARK> variable will be set to the name of the last
1310 C<(*MARK:NAME)> pattern executed. See the explanation for the
1311 C<(*MARK:NAME)> verb below for more details.
1313 B<NOTE:> C<$REGERROR> and C<$REGMARK> are not magic variables like C<$1>
1314 and most other regex related variables. They are not local to a scope, nor
1315 readonly, but instead are volatile package variables similar to C<$AUTOLOAD>.
1316 Use C<local> to localize changes to them to a specific scope if necessary.
1318 If a pattern does not contain a special backtracking verb that allows an
1319 argument, then C<$REGERROR> and C<$REGMARK> are not touched at all.
1323 =item Verbs that take an argument
1327 =item C<(*PRUNE)> C<(*PRUNE:NAME)>
1328 X<(*PRUNE)> X<(*PRUNE:NAME)> X<\v>
1330 This zero-width pattern prunes the backtracking tree at the current point
1331 when backtracked into on failure. Consider the pattern C<A (*PRUNE) B>,
1332 where A and B are complex patterns. Until the C<(*PRUNE)> verb is reached,
1333 A may backtrack as necessary to match. Once it is reached, matching
1334 continues in B, which may also backtrack as necessary; however, should B
1335 not match, then no further backtracking will take place, and the pattern
1336 will fail outright at the current starting position.
1338 As a shortcut, C<\v> is exactly equivalent to C<(*PRUNE)>.
1340 The following example counts all the possible matching strings in a
1341 pattern (without actually matching any of them).
1343 'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
1344 print "Count=$count\n";
1359 If we add a C<(*PRUNE)> before the count like the following
1361 'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
1362 print "Count=$count\n";
1364 we prevent backtracking and find the count of the longest matching
1365 at each matching startpoint like so:
1372 Any number of C<(*PRUNE)> assertions may be used in a pattern.
1374 See also C<< (?>pattern) >> and possessive quantifiers for other ways to
1375 control backtracking. In some cases, the use of C<(*PRUNE)> can be
1376 replaced with a C<< (?>pattern) >> with no functional difference; however,
1377 C<(*PRUNE)> can be used to handle cases that cannot be expressed using a
1378 C<< (?>pattern) >> alone.
1381 =item C<(*SKIP)> C<(*SKIP:NAME)>
1384 This zero-width pattern is similar to C<(*PRUNE)>, except that on
1385 failure it also signifies that whatever text that was matched leading up
1386 to the C<(*SKIP)> pattern being executed cannot be part of I<any> match
1387 of this pattern. This effectively means that the regex engine "skips" forward
1388 to this position on failure and tries to match again, (assuming that
1389 there is sufficient room to match).
1391 As a shortcut C<\V> is exactly equivalent to C<(*SKIP)>.
1393 The name of the C<(*SKIP:NAME)> pattern has special significance. If a
1394 C<(*MARK:NAME)> was encountered while matching, then it is that position
1395 which is used as the "skip point". If no C<(*MARK)> of that name was
1396 encountered, then the C<(*SKIP)> operator has no effect. When used
1397 without a name the "skip point" is where the match point was when
1398 executing the (*SKIP) pattern.
1400 Compare the following to the examples in C<(*PRUNE)>, note the string
1403 'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
1404 print "Count=$count\n";
1412 Once the 'aaab' at the start of the string has matched, and the C<(*SKIP)>
1413 executed, the next startpoint will be where the cursor was when the
1414 C<(*SKIP)> was executed.
1416 =item C<(*MARK:NAME)> C<(*:NAME)>
1417 X<(*MARK)> C<(*MARK:NAME)> C<(*:NAME)>
1419 This zero-width pattern can be used to mark the point reached in a string
1420 when a certain part of the pattern has been successfully matched. This
1421 mark may be given a name. A later C<(*SKIP)> pattern will then skip
1422 forward to that point if backtracked into on failure. Any number of
1423 C<(*MARK)> patterns are allowed, and the NAME portion is optional and may
1426 In addition to interacting with the C<(*SKIP)> pattern, C<(*MARK:NAME)>
1427 can be used to "label" a pattern branch, so that after matching, the
1428 program can determine which branches of the pattern were involved in the
1431 When a match is successful, the C<$REGMARK> variable will be set to the
1432 name of the most recently executed C<(*MARK:NAME)> that was involved
1435 This can be used to determine which branch of a pattern was matched
1436 without using a seperate capture buffer for each branch, which in turn
1437 can result in a performance improvement, as perl cannot optimize
1438 C</(?:(x)|(y)|(z))/> as efficiently as something like
1439 C</(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/>.
1441 When a match has failed, and unless another verb has been involved in
1442 failing the match and has provided its own name to use, the C<$REGERROR>
1443 variable will be set to the name of the most recently executed
1446 See C<(*SKIP)> for more details.
1448 As a shortcut C<(*MARK:NAME)> can be written C<(*:NAME)>.
1450 =item C<(*THEN)> C<(*THEN:NAME)>
1452 This is similar to the "cut group" operator C<::> from Perl6. Like
1453 C<(*PRUNE)>, this verb always matches, and when backtracked into on
1454 failure, it causes the regex engine to try the next alternation in the
1455 innermost enclosing group (capturing or otherwise).
1457 Its name comes from the observation that this operation combined with the
1458 alternation operator (C<|>) can be used to create what is essentially a
1459 pattern-based if/then/else block:
1461 ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
1463 Note that if this operator is used and NOT inside of an alternation then
1464 it acts exactly like the C<(*PRUNE)> operator.
1474 / ( A (*THEN) B | C (*THEN) D ) /
1478 / ( A (*PRUNE) B | C (*PRUNE) D ) /
1480 as after matching the A but failing on the B the C<(*THEN)> verb will
1481 backtrack and try C; but the C<(*PRUNE)> verb will simply fail.
1486 This is the Perl6 "commit pattern" C<< <commit> >> or C<:::>. It's a
1487 zero-width pattern similar to C<(*SKIP)>, except that when backtracked
1488 into on failure it causes the match to fail outright. No further attempts
1489 to find a valid match by advancing the start pointer will occur again.
1492 'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
1493 print "Count=$count\n";
1500 In other words, once the C<(*COMMIT)> has been entered, and if the pattern
1501 does not match, the regex engine will not try any further matching on the
1506 =item Verbs without an argument
1510 =item C<(*FAIL)> C<(*F)>
1513 This pattern matches nothing and always fails. It can be used to force the
1514 engine to backtrack. It is equivalent to C<(?!)>, but easier to read. In
1515 fact, C<(?!)> gets optimised into C<(*FAIL)> internally.
1517 It is probably useful only when combined with C<(?{})> or C<(??{})>.
1522 B<WARNING:> This feature is highly experimental. It is not recommended
1523 for production code.
1525 This pattern matches nothing and causes the end of successful matching at
1526 the point at which the C<(*ACCEPT)> pattern was encountered, regardless of
1527 whether there is actually more to match in the string. When inside of a
1528 nested pattern, such as recursion, or in a subpattern dynamically generated
1529 via C<(??{})>, only the innermost pattern is ended immediately.
1531 If the C<(*ACCEPT)> is inside of capturing buffers then the buffers are
1532 marked as ended at the point at which the C<(*ACCEPT)> was encountered.
1535 'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
1537 will match, and C<$1> will be C<AB> and C<$2> will be C<B>, C<$3> will not
1538 be set. If another branch in the inner parentheses were matched, such as in the
1539 string 'ACDE', then the C<D> and C<E> would have to be matched as well.
1546 X<backtrack> X<backtracking>
1548 NOTE: This section presents an abstract approximation of regular
1549 expression behavior. For a more rigorous (and complicated) view of
1550 the rules involved in selecting a match among possible alternatives,
1551 see L<Combining RE Pieces>.
1553 A fundamental feature of regular expression matching involves the
1554 notion called I<backtracking>, which is currently used (when needed)
1555 by all regular non-possessive expression quantifiers, namely C<*>, C<*?>, C<+>,
1556 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
1557 internally, but the general principle outlined here is valid.
1559 For a regular expression to match, the I<entire> regular expression must
1560 match, not just part of it. So if the beginning of a pattern containing a
1561 quantifier succeeds in a way that causes later parts in the pattern to
1562 fail, the matching engine backs up and recalculates the beginning
1563 part--that's why it's called backtracking.
1565 Here is an example of backtracking: Let's say you want to find the
1566 word following "foo" in the string "Food is on the foo table.":
1568 $_ = "Food is on the foo table.";
1569 if ( /\b(foo)\s+(\w+)/i ) {
1570 print "$2 follows $1.\n";
1573 When the match runs, the first part of the regular expression (C<\b(foo)>)
1574 finds a possible match right at the beginning of the string, and loads up
1575 $1 with "Foo". However, as soon as the matching engine sees that there's
1576 no whitespace following the "Foo" that it had saved in $1, it realizes its
1577 mistake and starts over again one character after where it had the
1578 tentative match. This time it goes all the way until the next occurrence
1579 of "foo". The complete regular expression matches this time, and you get
1580 the expected output of "table follows foo."
1582 Sometimes minimal matching can help a lot. Imagine you'd like to match
1583 everything between "foo" and "bar". Initially, you write something
1586 $_ = "The food is under the bar in the barn.";
1587 if ( /foo(.*)bar/ ) {
1591 Which perhaps unexpectedly yields:
1593 got <d is under the bar in the >
1595 That's because C<.*> was greedy, so you get everything between the
1596 I<first> "foo" and the I<last> "bar". Here it's more effective
1597 to use minimal matching to make sure you get the text between a "foo"
1598 and the first "bar" thereafter.
1600 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
1601 got <d is under the >
1603 Here's another example. Let's say you'd like to match a number at the end
1604 of a string, and you also want to keep the preceding part of the match.
1607 $_ = "I have 2 numbers: 53147";
1608 if ( /(.*)(\d*)/ ) { # Wrong!
1609 print "Beginning is <$1>, number is <$2>.\n";
1612 That won't work at all, because C<.*> was greedy and gobbled up the
1613 whole string. As C<\d*> can match on an empty string the complete
1614 regular expression matched successfully.
1616 Beginning is <I have 2 numbers: 53147>, number is <>.
1618 Here are some variants, most of which don't work:
1620 $_ = "I have 2 numbers: 53147";
1633 printf "%-12s ", $pat;
1635 print "<$1> <$2>\n";
1641 That will print out:
1643 (.*)(\d*) <I have 2 numbers: 53147> <>
1644 (.*)(\d+) <I have 2 numbers: 5314> <7>
1646 (.*?)(\d+) <I have > <2>
1647 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
1648 (.*?)(\d+)$ <I have 2 numbers: > <53147>
1649 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
1650 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
1652 As you see, this can be a bit tricky. It's important to realize that a
1653 regular expression is merely a set of assertions that gives a definition
1654 of success. There may be 0, 1, or several different ways that the
1655 definition might succeed against a particular string. And if there are
1656 multiple ways it might succeed, you need to understand backtracking to
1657 know which variety of success you will achieve.
1659 When using look-ahead assertions and negations, this can all get even
1660 trickier. Imagine you'd like to find a sequence of non-digits not
1661 followed by "123". You might try to write that as
1664 if ( /^\D*(?!123)/ ) { # Wrong!
1665 print "Yup, no 123 in $_\n";
1668 But that isn't going to match; at least, not the way you're hoping. It
1669 claims that there is no 123 in the string. Here's a clearer picture of
1670 why that pattern matches, contrary to popular expectations:
1675 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
1676 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
1678 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
1679 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
1687 You might have expected test 3 to fail because it seems to a more
1688 general purpose version of test 1. The important difference between
1689 them is that test 3 contains a quantifier (C<\D*>) and so can use
1690 backtracking, whereas test 1 will not. What's happening is
1691 that you've asked "Is it true that at the start of $x, following 0 or more
1692 non-digits, you have something that's not 123?" If the pattern matcher had
1693 let C<\D*> expand to "ABC", this would have caused the whole pattern to
1696 The search engine will initially match C<\D*> with "ABC". Then it will
1697 try to match C<(?!123> with "123", which fails. But because
1698 a quantifier (C<\D*>) has been used in the regular expression, the
1699 search engine can backtrack and retry the match differently
1700 in the hope of matching the complete regular expression.
1702 The pattern really, I<really> wants to succeed, so it uses the
1703 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
1704 time. Now there's indeed something following "AB" that is not
1705 "123". It's "C123", which suffices.
1707 We can deal with this by using both an assertion and a negation.
1708 We'll say that the first part in $1 must be followed both by a digit
1709 and by something that's not "123". Remember that the look-aheads
1710 are zero-width expressions--they only look, but don't consume any
1711 of the string in their match. So rewriting this way produces what
1712 you'd expect; that is, case 5 will fail, but case 6 succeeds:
1714 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
1715 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
1719 In other words, the two zero-width assertions next to each other work as though
1720 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
1721 matches only if you're at the beginning of the line AND the end of the
1722 line simultaneously. The deeper underlying truth is that juxtaposition in
1723 regular expressions always means AND, except when you write an explicit OR
1724 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
1725 although the attempted matches are made at different positions because "a"
1726 is not a zero-width assertion, but a one-width assertion.
1728 B<WARNING>: Particularly complicated regular expressions can take
1729 exponential time to solve because of the immense number of possible
1730 ways they can use backtracking to try for a match. For example, without
1731 internal optimizations done by the regular expression engine, this will
1732 take a painfully long time to run:
1734 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
1736 And if you used C<*>'s in the internal groups instead of limiting them
1737 to 0 through 5 matches, then it would take forever--or until you ran
1738 out of stack space. Moreover, these internal optimizations are not
1739 always applicable. For example, if you put C<{0,5}> instead of C<*>
1740 on the external group, no current optimization is applicable, and the
1741 match takes a long time to finish.
1743 A powerful tool for optimizing such beasts is what is known as an
1744 "independent group",
1745 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
1746 zero-length look-ahead/look-behind assertions will not backtrack to make
1747 the tail match, since they are in "logical" context: only
1748 whether they match is considered relevant. For an example
1749 where side-effects of look-ahead I<might> have influenced the
1750 following match, see L<C<< (?>pattern) >>>.
1752 =head2 Version 8 Regular Expressions
1753 X<regular expression, version 8> X<regex, version 8> X<regexp, version 8>
1755 In case you're not familiar with the "regular" Version 8 regex
1756 routines, here are the pattern-matching rules not described above.
1758 Any single character matches itself, unless it is a I<metacharacter>
1759 with a special meaning described here or above. You can cause
1760 characters that normally function as metacharacters to be interpreted
1761 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
1762 character; "\\" matches a "\"). This escape mechanism is also required
1763 for the character used as the pattern delimiter.
1765 A series of characters matches that series of characters in the target
1766 string, so the pattern C<blurfl> would match "blurfl" in the target
1769 You can specify a character class, by enclosing a list of characters
1770 in C<[]>, which will match any character from the list. If the
1771 first character after the "[" is "^", the class matches any character not
1772 in the list. Within a list, the "-" character specifies a
1773 range, so that C<a-z> represents all characters between "a" and "z",
1774 inclusive. If you want either "-" or "]" itself to be a member of a
1775 class, put it at the start of the list (possibly after a "^"), or
1776 escape it with a backslash. "-" is also taken literally when it is
1777 at the end of the list, just before the closing "]". (The
1778 following all specify the same class of three characters: C<[-az]>,
1779 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1780 specifies a class containing twenty-six characters, even on EBCDIC-based
1781 character sets.) Also, if you try to use the character
1782 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1783 a range, the "-" is understood literally.
1785 Note also that the whole range idea is rather unportable between
1786 character sets--and even within character sets they may cause results
1787 you probably didn't expect. A sound principle is to use only ranges
1788 that begin from and end at either alphabetics of equal case ([a-e],
1789 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1790 spell out the character sets in full.
1792 Characters may be specified using a metacharacter syntax much like that
1793 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1794 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1795 of octal digits, matches the character whose coded character set value
1796 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1797 matches the character whose numeric value is I<nn>. The expression \cI<x>
1798 matches the character control-I<x>. Finally, the "." metacharacter
1799 matches any character except "\n" (unless you use C</s>).
1801 You can specify a series of alternatives for a pattern using "|" to
1802 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1803 or "foe" in the target string (as would C<f(e|i|o)e>). The
1804 first alternative includes everything from the last pattern delimiter
1805 ("(", "[", or the beginning of the pattern) up to the first "|", and
1806 the last alternative contains everything from the last "|" to the next
1807 pattern delimiter. That's why it's common practice to include
1808 alternatives in parentheses: to minimize confusion about where they
1811 Alternatives are tried from left to right, so the first
1812 alternative found for which the entire expression matches, is the one that
1813 is chosen. This means that alternatives are not necessarily greedy. For
1814 example: when matching C<foo|foot> against "barefoot", only the "foo"
1815 part will match, as that is the first alternative tried, and it successfully
1816 matches the target string. (This might not seem important, but it is
1817 important when you are capturing matched text using parentheses.)
1819 Also remember that "|" is interpreted as a literal within square brackets,
1820 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1822 Within a pattern, you may designate subpatterns for later reference
1823 by enclosing them in parentheses, and you may refer back to the
1824 I<n>th subpattern later in the pattern using the metacharacter
1825 \I<n>. Subpatterns are numbered based on the left to right order
1826 of their opening parenthesis. A backreference matches whatever
1827 actually matched the subpattern in the string being examined, not
1828 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1829 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1830 1 matched "0x", even though the rule C<0|0x> could potentially match
1831 the leading 0 in the second number.
1833 =head2 Warning on \1 Instead of $1
1835 Some people get too used to writing things like:
1837 $pattern =~ s/(\W)/\\\1/g;
1839 This is grandfathered for the RHS of a substitute to avoid shocking the
1840 B<sed> addicts, but it's a dirty habit to get into. That's because in
1841 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1842 the usual double-quoted string means a control-A. The customary Unix
1843 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1844 of doing that, you get yourself into trouble if you then add an C</e>
1847 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1853 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1854 C<${1}000>. The operation of interpolation should not be confused
1855 with the operation of matching a backreference. Certainly they mean two
1856 different things on the I<left> side of the C<s///>.
1858 =head2 Repeated Patterns Matching a Zero-length Substring
1860 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1862 Regular expressions provide a terse and powerful programming language. As
1863 with most other power tools, power comes together with the ability
1866 A common abuse of this power stems from the ability to make infinite
1867 loops using regular expressions, with something as innocuous as:
1869 'foo' =~ m{ ( o? )* }x;
1871 The C<o?> matches at the beginning of C<'foo'>, and since the position
1872 in the string is not moved by the match, C<o?> would match again and again
1873 because of the C<*> modifier. Another common way to create a similar cycle
1874 is with the looping modifier C<//g>:
1876 @matches = ( 'foo' =~ m{ o? }xg );
1880 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1882 or the loop implied by split().
1884 However, long experience has shown that many programming tasks may
1885 be significantly simplified by using repeated subexpressions that
1886 may match zero-length substrings. Here's a simple example being:
1888 @chars = split //, $string; # // is not magic in split
1889 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1891 Thus Perl allows such constructs, by I<forcefully breaking
1892 the infinite loop>. The rules for this are different for lower-level
1893 loops given by the greedy modifiers C<*+{}>, and for higher-level
1894 ones like the C</g> modifier or split() operator.
1896 The lower-level loops are I<interrupted> (that is, the loop is
1897 broken) when Perl detects that a repeated expression matched a
1898 zero-length substring. Thus
1900 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1902 is made equivalent to
1904 m{ (?: NON_ZERO_LENGTH )*
1909 The higher level-loops preserve an additional state between iterations:
1910 whether the last match was zero-length. To break the loop, the following
1911 match after a zero-length match is prohibited to have a length of zero.
1912 This prohibition interacts with backtracking (see L<"Backtracking">),
1913 and so the I<second best> match is chosen if the I<best> match is of
1921 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1922 match given by non-greedy C<??> is the zero-length match, and the I<second
1923 best> match is what is matched by C<\w>. Thus zero-length matches
1924 alternate with one-character-long matches.
1926 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1927 position one notch further in the string.
1929 The additional state of being I<matched with zero-length> is associated with
1930 the matched string, and is reset by each assignment to pos().
1931 Zero-length matches at the end of the previous match are ignored
1934 =head2 Combining RE Pieces
1936 Each of the elementary pieces of regular expressions which were described
1937 before (such as C<ab> or C<\Z>) could match at most one substring
1938 at the given position of the input string. However, in a typical regular
1939 expression these elementary pieces are combined into more complicated
1940 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1941 (in these examples C<S> and C<T> are regular subexpressions).
1943 Such combinations can include alternatives, leading to a problem of choice:
1944 if we match a regular expression C<a|ab> against C<"abc">, will it match
1945 substring C<"a"> or C<"ab">? One way to describe which substring is
1946 actually matched is the concept of backtracking (see L<"Backtracking">).
1947 However, this description is too low-level and makes you think
1948 in terms of a particular implementation.
1950 Another description starts with notions of "better"/"worse". All the
1951 substrings which may be matched by the given regular expression can be
1952 sorted from the "best" match to the "worst" match, and it is the "best"
1953 match which is chosen. This substitutes the question of "what is chosen?"
1954 by the question of "which matches are better, and which are worse?".
1956 Again, for elementary pieces there is no such question, since at most
1957 one match at a given position is possible. This section describes the
1958 notion of better/worse for combining operators. In the description
1959 below C<S> and C<T> are regular subexpressions.
1965 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1966 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1967 which can be matched by C<T>.
1969 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
1972 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
1973 C<B> is better match for C<T> than C<B'>.
1977 When C<S> can match, it is a better match than when only C<T> can match.
1979 Ordering of two matches for C<S> is the same as for C<S>. Similar for
1980 two matches for C<T>.
1982 =item C<S{REPEAT_COUNT}>
1984 Matches as C<SSS...S> (repeated as many times as necessary).
1988 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
1990 =item C<S{min,max}?>
1992 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
1994 =item C<S?>, C<S*>, C<S+>
1996 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
1998 =item C<S??>, C<S*?>, C<S+?>
2000 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
2004 Matches the best match for C<S> and only that.
2006 =item C<(?=S)>, C<(?<=S)>
2008 Only the best match for C<S> is considered. (This is important only if
2009 C<S> has capturing parentheses, and backreferences are used somewhere
2010 else in the whole regular expression.)
2012 =item C<(?!S)>, C<(?<!S)>
2014 For this grouping operator there is no need to describe the ordering, since
2015 only whether or not C<S> can match is important.
2017 =item C<(??{ EXPR })>, C<(?PARNO)>
2019 The ordering is the same as for the regular expression which is
2020 the result of EXPR, or the pattern contained by capture buffer PARNO.
2022 =item C<(?(condition)yes-pattern|no-pattern)>
2024 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
2025 already determined. The ordering of the matches is the same as for the
2026 chosen subexpression.
2030 The above recipes describe the ordering of matches I<at a given position>.
2031 One more rule is needed to understand how a match is determined for the
2032 whole regular expression: a match at an earlier position is always better
2033 than a match at a later position.
2035 =head2 Creating Custom RE Engines
2037 Overloaded constants (see L<overload>) provide a simple way to extend
2038 the functionality of the RE engine.
2040 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
2041 matches at a boundary between whitespace characters and non-whitespace
2042 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
2043 at these positions, so we want to have each C<\Y|> in the place of the
2044 more complicated version. We can create a module C<customre> to do
2052 die "No argument to customre::import allowed" if @_;
2053 overload::constant 'qr' => \&convert;
2056 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2058 # We must also take care of not escaping the legitimate \\Y|
2059 # sequence, hence the presence of '\\' in the conversion rules.
2060 my %rules = ( '\\' => '\\\\',
2061 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2067 { $rules{$1} or invalid($re,$1) }sgex;
2071 Now C<use customre> enables the new escape in constant regular
2072 expressions, i.e., those without any runtime variable interpolations.
2073 As documented in L<overload>, this conversion will work only over
2074 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
2075 part of this regular expression needs to be converted explicitly
2076 (but only if the special meaning of C<\Y|> should be enabled inside $re):
2081 $re = customre::convert $re;
2084 =head1 PCRE/Python Support
2086 As of Perl 5.10 Perl supports several Python/PCRE specific extensions
2087 to the regex syntax. While Perl programmers are encouraged to use the
2088 Perl specific syntax, the following are legal in Perl 5.10:
2092 =item C<< (?PE<lt>NAMEE<gt>pattern) >>
2094 Define a named capture buffer. Equivalent to C<< (?<NAME>pattern) >>.
2096 =item C<< (?P=NAME) >>
2098 Backreference to a named capture buffer. Equivalent to C<< \g{NAME} >>.
2100 =item C<< (?P>NAME) >>
2102 Subroutine call to a named capture buffer. Equivalent to C<< (?&NAME) >>.
2108 This document varies from difficult to understand to completely
2109 and utterly opaque. The wandering prose riddled with jargon is
2110 hard to fathom in several places.
2112 This document needs a rewrite that separates the tutorial content
2113 from the reference content.
2121 L<perlop/"Regexp Quote-Like Operators">.
2123 L<perlop/"Gory details of parsing quoted constructs">.
2133 I<Mastering Regular Expressions> by Jeffrey Friedl, published
2134 by O'Reilly and Associates.