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</m> X<regex, multiline> X<regexp, multiline> X<regular expression, multiline>
33 Treat string as multiple lines. That is, change "^" and "$" from matching
34 the start or end of the string to matching the start or end of any
35 line anywhere within the string.
38 X</s> X<regex, single-line> X<regexp, single-line>
39 X<regular expression, single-line>
41 Treat string as single line. That is, change "." to match any character
42 whatsoever, even a newline, which normally it would not match.
44 Used together, as /ms, they let the "." match any character whatsoever,
45 while still allowing "^" and "$" to match, respectively, just after
46 and just before newlines within the string.
49 X</i> X<regex, case-insensitive> X<regexp, case-insensitive>
50 X<regular expression, case-insensitive>
52 Do case-insensitive pattern matching.
54 If C<use locale> is in effect, the case map is taken from the current
55 locale. See L<perllocale>.
60 Extend your pattern's legibility by permitting whitespace and comments.
63 X</p> X<regex, preserve> X<regexp, preserve>
65 Preserve the string matched such that ${^PREMATCH}, {$^MATCH}, and
66 ${^POSTMATCH} are available for use after matching.
70 These are usually written as "the C</x> modifier", even though the delimiter
71 in question might not really be a slash. Any of these
72 modifiers may also be embedded within the regular expression itself using
73 the C<(?...)> construct. See below.
75 The C</x> modifier itself needs a little more explanation. It tells
76 the regular expression parser to ignore whitespace that is neither
77 backslashed nor within a character class. You can use this to break up
78 your regular expression into (slightly) more readable parts. The C<#>
79 character is also treated as a metacharacter introducing a comment,
80 just as in ordinary Perl code. This also means that if you want real
81 whitespace or C<#> characters in the pattern (outside a character
82 class, where they are unaffected by C</x>), then you'll either have to
83 escape them (using backslashes or C<\Q...\E>) or encode them using octal
84 or hex escapes. Taken together, these features go a long way towards
85 making Perl's regular expressions more readable. Note that you have to
86 be careful not to include the pattern delimiter in the comment--perl has
87 no way of knowing you did not intend to close the pattern early. See
88 the C-comment deletion code in L<perlop>. Also note that anything inside
89 a C<\Q...\E> stays unaffected by C</x>.
92 =head2 Regular Expressions
96 The patterns used in Perl pattern matching evolved from the ones supplied in
97 the Version 8 regex routines. (The routines are derived
98 (distantly) from Henry Spencer's freely redistributable reimplementation
99 of the V8 routines.) See L<Version 8 Regular Expressions> for
102 In particular the following metacharacters have their standard I<egrep>-ish
105 X<\> X<^> X<.> X<$> X<|> X<(> X<()> X<[> X<[]>
108 \ Quote the next metacharacter
109 ^ Match the beginning of the line
110 . Match any character (except newline)
111 $ Match the end of the line (or before newline at the end)
116 By default, the "^" character is guaranteed to match only the
117 beginning of the string, the "$" character only the end (or before the
118 newline at the end), and Perl does certain optimizations with the
119 assumption that the string contains only one line. Embedded newlines
120 will not be matched by "^" or "$". You may, however, wish to treat a
121 string as a multi-line buffer, such that the "^" will match after any
122 newline within the string (except if the newline is the last character in
123 the string), and "$" will match before any newline. At the
124 cost of a little more overhead, you can do this by using the /m modifier
125 on the pattern match operator. (Older programs did this by setting C<$*>,
126 but this practice has been removed in perl 5.9.)
129 To simplify multi-line substitutions, the "." character never matches a
130 newline unless you use the C</s> modifier, which in effect tells Perl to pretend
131 the string is a single line--even if it isn't.
136 The following standard quantifiers are recognized:
137 X<metacharacter> X<quantifier> X<*> X<+> X<?> X<{n}> X<{n,}> X<{n,m}>
139 * Match 0 or more times
140 + Match 1 or more times
142 {n} Match exactly n times
143 {n,} Match at least n times
144 {n,m} Match at least n but not more than m times
146 (If a curly bracket occurs in any other context, it is treated
147 as a regular character. In particular, the lower bound
148 is not optional.) The "*" modifier is equivalent to C<{0,}>, the "+"
149 modifier to C<{1,}>, and the "?" modifier to C<{0,1}>. n and m are limited
150 to integral values less than a preset limit defined when perl is built.
151 This is usually 32766 on the most common platforms. The actual limit can
152 be seen in the error message generated by code such as this:
154 $_ **= $_ , / {$_} / for 2 .. 42;
156 By default, a quantified subpattern is "greedy", that is, it will match as
157 many times as possible (given a particular starting location) while still
158 allowing the rest of the pattern to match. If you want it to match the
159 minimum number of times possible, follow the quantifier with a "?". Note
160 that the meanings don't change, just the "greediness":
161 X<metacharacter> X<greedy> X<greediness>
162 X<?> X<*?> X<+?> X<??> X<{n}?> X<{n,}?> X<{n,m}?>
164 *? Match 0 or more times, not greedily
165 +? Match 1 or more times, not greedily
166 ?? Match 0 or 1 time, not greedily
167 {n}? Match exactly n times, not greedily
168 {n,}? Match at least n times, not greedily
169 {n,m}? Match at least n but not more than m times, not greedily
171 By default, when a quantified subpattern does not allow the rest of the
172 overall pattern to match, Perl will backtrack. However, this behaviour is
173 sometimes undesirable. Thus Perl provides the "possessive" quantifier form
176 *+ Match 0 or more times and give nothing back
177 ++ Match 1 or more times and give nothing back
178 ?+ Match 0 or 1 time and give nothing back
179 {n}+ Match exactly n times and give nothing back (redundant)
180 {n,}+ Match at least n times and give nothing back
181 {n,m}+ Match at least n but not more than m times and give nothing back
187 will never match, as the C<a++> will gobble up all the C<a>'s in the
188 string and won't leave any for the remaining part of the pattern. This
189 feature can be extremely useful to give perl hints about where it
190 shouldn't backtrack. For instance, the typical "match a double-quoted
191 string" problem can be most efficiently performed when written as:
193 /"(?:[^"\\]++|\\.)*+"/
195 as we know that if the final quote does not match, backtracking will not
196 help. See the independent subexpression C<< (?>...) >> for more details;
197 possessive quantifiers are just syntactic sugar for that construct. For
198 instance the above example could also be written as follows:
200 /"(?>(?:(?>[^"\\]+)|\\.)*)"/
202 =head3 Escape sequences
204 Because patterns are processed as double quoted strings, the following
206 X<\t> X<\n> X<\r> X<\f> X<\e> X<\a> X<\l> X<\u> X<\L> X<\U> X<\E> X<\Q>
207 X<\0> X<\c> X<\N> X<\x>
213 \a alarm (bell) (BEL)
214 \e escape (think troff) (ESC)
215 \033 octal char (example: ESC)
216 \x1B hex char (example: ESC)
217 \x{263a} wide hex char (example: Unicode SMILEY)
218 \cK control char (example: VT)
220 \l lowercase next char (think vi)
221 \u uppercase next char (think vi)
222 \L lowercase till \E (think vi)
223 \U uppercase till \E (think vi)
224 \E end case modification (think vi)
225 \Q quote (disable) pattern metacharacters till \E
227 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
228 and C<\U> is taken from the current locale. See L<perllocale>. For
229 documentation of C<\N{name}>, see L<charnames>.
231 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
232 An unescaped C<$> or C<@> interpolates the corresponding variable,
233 while escaping will cause the literal string C<\$> to be matched.
234 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
236 =head3 Character classes
238 In addition, Perl defines the following:
239 X<\w> X<\W> X<\s> X<\S> X<\d> X<\D> X<\X> X<\p> X<\P> X<\C>
240 X<\g> X<\k> X<\N> X<\K> X<\v> X<\V>
241 X<word> X<whitespace> X<character class> X<backreference>
243 \w Match a "word" character (alphanumeric plus "_")
244 \W Match a non-"word" character
245 \s Match a whitespace character
246 \S Match a non-whitespace character
247 \d Match a digit character
248 \D Match a non-digit character
249 \pP Match P, named property. Use \p{Prop} for longer names.
251 \X Match eXtended Unicode "combining character sequence",
252 equivalent to (?:\PM\pM*)
253 \C Match a single C char (octet) even under Unicode.
254 NOTE: breaks up characters into their UTF-8 bytes,
255 so you may end up with malformed pieces of UTF-8.
256 Unsupported in lookbehind.
257 \1 Backreference to a specific group.
258 '1' may actually be any positive integer.
259 \g1 Backreference to a specific or previous group,
260 \g{-1} number may be negative indicating a previous buffer and may
261 optionally be wrapped in curly brackets for safer parsing.
262 \g{name} Named backreference
263 \k<name> Named backreference
264 \N{name} Named unicode character, or unicode escape
265 \x12 Hexadecimal escape sequence
266 \x{1234} Long hexadecimal escape sequence
267 \K Keep the stuff left of the \K, don't include it in $&
268 \v Shortcut for (*PRUNE)
269 \V Shortcut for (*SKIP)
271 A C<\w> matches a single alphanumeric character (an alphabetic
272 character, or a decimal digit) or C<_>, not a whole word. Use C<\w+>
273 to match a string of Perl-identifier characters (which isn't the same
274 as matching an English word). If C<use locale> is in effect, the list
275 of alphabetic characters generated by C<\w> is taken from the current
276 locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
277 C<\d>, and C<\D> within character classes, but they aren't usable
278 as either end of a range. If any of them precedes or follows a "-",
279 the "-" is understood literally. If Unicode is in effect, C<\s> matches
280 also "\x{85}", "\x{2028}, and "\x{2029}". See L<perlunicode> for more
281 details about C<\pP>, C<\PP>, C<\X> and the possibility of defining
282 your own C<\p> and C<\P> properties, and L<perluniintro> about Unicode
286 The POSIX character class syntax
291 is also available. Note that the C<[> and C<]> brackets are I<literal>;
292 they must always be used within a character class expression.
295 $string =~ /[[:alpha:]]/;
297 # this is not, and will generate a warning:
298 $string =~ /[:alpha:]/;
300 The available classes and their backslash equivalents (if available) are
303 X<alpha> X<alnum> X<ascii> X<blank> X<cntrl> X<digit> X<graph>
304 X<lower> X<print> X<punct> X<space> X<upper> X<word> X<xdigit>
325 A GNU extension equivalent to C<[ \t]>, "all horizontal whitespace".
329 Not exactly equivalent to C<\s> since the C<[[:space:]]> includes
330 also the (very rare) "vertical tabulator", "\cK" or chr(11) in ASCII.
334 A Perl extension, see above.
338 For example use C<[:upper:]> to match all the uppercase characters.
339 Note that the C<[]> are part of the C<[::]> construct, not part of the
340 whole character class. For example:
344 matches zero, one, any alphabetic character, and the percent sign.
346 The following equivalences to Unicode \p{} constructs and equivalent
347 backslash character classes (if available), will hold:
348 X<character class> X<\p> X<\p{}>
350 [[:...:]] \p{...} backslash
368 For example C<[[:lower:]]> and C<\p{IsLower}> are equivalent.
370 If the C<utf8> pragma is not used but the C<locale> pragma is, the
371 classes correlate with the usual isalpha(3) interface (except for
374 The assumedly non-obviously named classes are:
381 Any control character. Usually characters that don't produce output as
382 such but instead control the terminal somehow: for example newline and
383 backspace are control characters. All characters with ord() less than
384 32 are usually classified as control characters (assuming ASCII,
385 the ISO Latin character sets, and Unicode), as is the character with
386 the ord() value of 127 (C<DEL>).
391 Any alphanumeric or punctuation (special) character.
396 Any alphanumeric or punctuation (special) character or the space character.
401 Any punctuation (special) character.
406 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
407 work just fine) it is included for completeness.
411 You can negate the [::] character classes by prefixing the class name
412 with a '^'. This is a Perl extension. For example:
413 X<character class, negation>
415 POSIX traditional Unicode
417 [[:^digit:]] \D \P{IsDigit}
418 [[:^space:]] \S \P{IsSpace}
419 [[:^word:]] \W \P{IsWord}
421 Perl respects the POSIX standard in that POSIX character classes are
422 only supported within a character class. The POSIX character classes
423 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
424 use them will cause an error.
428 Perl defines the following zero-width assertions:
429 X<zero-width assertion> X<assertion> X<regex, zero-width assertion>
430 X<regexp, zero-width assertion>
431 X<regular expression, zero-width assertion>
432 X<\b> X<\B> X<\A> X<\Z> X<\z> X<\G>
434 \b Match a word boundary
435 \B Match except at a word boundary
436 \A Match only at beginning of string
437 \Z Match only at end of string, or before newline at the end
438 \z Match only at end of string
439 \G Match only at pos() (e.g. at the end-of-match position
442 A word boundary (C<\b>) is a spot between two characters
443 that has a C<\w> on one side of it and a C<\W> on the other side
444 of it (in either order), counting the imaginary characters off the
445 beginning and end of the string as matching a C<\W>. (Within
446 character classes C<\b> represents backspace rather than a word
447 boundary, just as it normally does in any double-quoted string.)
448 The C<\A> and C<\Z> are just like "^" and "$", except that they
449 won't match multiple times when the C</m> modifier is used, while
450 "^" and "$" will match at every internal line boundary. To match
451 the actual end of the string and not ignore an optional trailing
453 X<\b> X<\A> X<\Z> X<\z> X</m>
455 The C<\G> assertion can be used to chain global matches (using
456 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
457 It is also useful when writing C<lex>-like scanners, when you have
458 several patterns that you want to match against consequent substrings
459 of your string, see the previous reference. The actual location
460 where C<\G> will match can also be influenced by using C<pos()> as
461 an lvalue: see L<perlfunc/pos>. Note that the rule for zero-length
462 matches is modified somewhat, in that contents to the left of C<\G> is
463 not counted when determining the length of the match. Thus the following
464 will not match forever:
473 It will print 'A' and then terminate, as it considers the match to
474 be zero-width, and thus will not match at the same position twice in a
477 It is worth noting that C<\G> improperly used can result in an infinite
478 loop. Take care when using patterns that include C<\G> in an alternation.
480 =head3 Capture buffers
482 The bracketing construct C<( ... )> creates capture buffers. To refer
483 to the current contents of a buffer later on, within the same pattern,
484 use \1 for the first, \2 for the second, and so on.
485 Outside the match use "$" instead of "\". (The
486 \<digit> notation works in certain circumstances outside
487 the match. See the warning below about \1 vs $1 for details.)
488 Referring back to another part of the match is called a
490 X<regex, capture buffer> X<regexp, capture buffer>
491 X<regular expression, capture buffer> X<backreference>
493 There is no limit to the number of captured substrings that you may
494 use. However Perl also uses \10, \11, etc. as aliases for \010,
495 \011, etc. (Recall that 0 means octal, so \011 is the character at
496 number 9 in your coded character set; which would be the 10th character,
497 a horizontal tab under ASCII.) Perl resolves this
498 ambiguity by interpreting \10 as a backreference only if at least 10
499 left parentheses have opened before it. Likewise \11 is a
500 backreference only if at least 11 left parentheses have opened
501 before it. And so on. \1 through \9 are always interpreted as
504 X<\g{1}> X<\g{-1}> X<\g{name}> X<relative backreference> X<named backreference>
505 In order to provide a safer and easier way to construct patterns using
506 backreferences, Perl 5.10 provides the C<\g{N}> notation. The curly
507 brackets are optional, however omitting them is less safe as the meaning
508 of the pattern can be changed by text (such as digits) following it.
509 When N is a positive integer the C<\g{N}> notation is exactly equivalent
510 to using normal backreferences. When N is a negative integer then it is
511 a relative backreference referring to the previous N'th capturing group.
512 When the bracket form is used and N is not an integer, it is treated as a
513 reference to a named buffer.
515 Thus C<\g{-1}> refers to the last buffer, C<\g{-2}> refers to the
516 buffer before that. For example:
522 \g{-1} # backref to buffer 3
523 \g{-3} # backref to buffer 1
527 and would match the same as C</(Y) ( (X) \3 \1 )/x>.
529 Additionally, as of Perl 5.10 you may use named capture buffers and named
530 backreferences. The notation is C<< (?<name>...) >> to declare and C<< \k<name> >>
531 to reference. You may also use apostrophes instead of angle brackets to delimit the
532 name; and you may use the bracketed C<< \g{name} >> backreference syntax.
533 It's possible to refer to a named capture buffer by absolute and relative number as well.
534 Outside the pattern, a named capture buffer is available via the C<%+> hash.
535 When different buffers within the same pattern have the same name, C<$+{name}>
536 and C<< \k<name> >> refer to the leftmost defined group. (Thus it's possible
537 to do things with named capture buffers that would otherwise require C<(??{})>
539 X<named capture buffer> X<regular expression, named capture buffer>
540 X<%+> X<$+{name}> X<\k{name}>
544 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
546 /(.)\1/ # find first doubled char
547 and print "'$1' is the first doubled character\n";
549 /(?<char>.)\k<char>/ # ... a different way
550 and print "'$+{char}' is the first doubled character\n";
552 /(?'char'.)\1/ # ... mix and match
553 and print "'$1' is the first doubled character\n";
555 if (/Time: (..):(..):(..)/) { # parse out values
561 Several special variables also refer back to portions of the previous
562 match. C<$+> returns whatever the last bracket match matched.
563 C<$&> returns the entire matched string. (At one point C<$0> did
564 also, but now it returns the name of the program.) C<$`> returns
565 everything before the matched string. C<$'> returns everything
566 after the matched string. And C<$^N> contains whatever was matched by
567 the most-recently closed group (submatch). C<$^N> can be used in
568 extended patterns (see below), for example to assign a submatch to a
570 X<$+> X<$^N> X<$&> X<$`> X<$'>
572 The numbered match variables ($1, $2, $3, etc.) and the related punctuation
573 set (C<$+>, C<$&>, C<$`>, C<$'>, and C<$^N>) are all dynamically scoped
574 until the end of the enclosing block or until the next successful
575 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
576 X<$+> X<$^N> X<$&> X<$`> X<$'>
577 X<$1> X<$2> X<$3> X<$4> X<$5> X<$6> X<$7> X<$8> X<$9>
580 B<NOTE>: Failed matches in Perl do not reset the match variables,
581 which makes it easier to write code that tests for a series of more
582 specific cases and remembers the best match.
584 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
585 C<$'> anywhere in the program, it has to provide them for every
586 pattern match. This may substantially slow your program. Perl
587 uses the same mechanism to produce $1, $2, etc, so you also pay a
588 price for each pattern that contains capturing parentheses. (To
589 avoid this cost while retaining the grouping behaviour, use the
590 extended regular expression C<(?: ... )> instead.) But if you never
591 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
592 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
593 if you can, but if you can't (and some algorithms really appreciate
594 them), once you've used them once, use them at will, because you've
595 already paid the price. As of 5.005, C<$&> is not so costly as the
599 As a workaround for this problem, Perl 5.10 introduces C<${^PREMATCH}>,
600 C<${^MATCH}> and C<${^POSTMATCH}>, which are equivalent to C<$`>, C<$&>
601 and C<$'>, B<except> that they are only guaranteed to be defined after a
602 successful match that was executed with the C</p> (preserve) modifier.
603 The use of these variables incurs no global performance penalty, unlike
604 their punctuation char equivalents, however at the trade-off that you
605 have to tell perl when you want to use them.
608 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
609 C<\w>, C<\n>. Unlike some other regular expression languages, there
610 are no backslashed symbols that aren't alphanumeric. So anything
611 that looks like \\, \(, \), \<, \>, \{, or \} is always
612 interpreted as a literal character, not a metacharacter. This was
613 once used in a common idiom to disable or quote the special meanings
614 of regular expression metacharacters in a string that you want to
615 use for a pattern. Simply quote all non-"word" characters:
617 $pattern =~ s/(\W)/\\$1/g;
619 (If C<use locale> is set, then this depends on the current locale.)
620 Today it is more common to use the quotemeta() function or the C<\Q>
621 metaquoting escape sequence to disable all metacharacters' special
624 /$unquoted\Q$quoted\E$unquoted/
626 Beware that if you put literal backslashes (those not inside
627 interpolated variables) between C<\Q> and C<\E>, double-quotish
628 backslash interpolation may lead to confusing results. If you
629 I<need> to use literal backslashes within C<\Q...\E>,
630 consult L<perlop/"Gory details of parsing quoted constructs">.
632 =head2 Extended Patterns
634 Perl also defines a consistent extension syntax for features not
635 found in standard tools like B<awk> and B<lex>. The syntax is a
636 pair of parentheses with a question mark as the first thing within
637 the parentheses. The character after the question mark indicates
640 The stability of these extensions varies widely. Some have been
641 part of the core language for many years. Others are experimental
642 and may change without warning or be completely removed. Check
643 the documentation on an individual feature to verify its current
646 A question mark was chosen for this and for the minimal-matching
647 construct because 1) question marks are rare in older regular
648 expressions, and 2) whenever you see one, you should stop and
649 "question" exactly what is going on. That's psychology...
656 A comment. The text is ignored. If the C</x> modifier enables
657 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
658 the comment as soon as it sees a C<)>, so there is no way to put a literal
661 =item C<(?kimsx-imsx)>
664 One or more embedded pattern-match modifiers, to be turned on (or
665 turned off, if preceded by C<->) for the remainder of the pattern or
666 the remainder of the enclosing pattern group (if any). This is
667 particularly useful for dynamic patterns, such as those read in from a
668 configuration file, taken from an argument, or specified in a table
669 somewhere. Consider the case where some patterns want to be case
670 sensitive and some do not: The case insensitive ones merely need to
671 include C<(?i)> at the front of the pattern. For example:
674 if ( /$pattern/i ) { }
678 $pattern = "(?i)foobar";
679 if ( /$pattern/ ) { }
681 These modifiers are restored at the end of the enclosing group. For example,
685 will match C<blah> in any case, some spaces, and an exact (I<including the case>!)
686 repetition of the previous word, assuming the C</x> modifier, and no C</i>
687 modifier outside this group.
689 Note that the C<k> modifier is special in that it can only be enabled,
690 not disabled, and that its presence anywhere in a pattern has a global
691 effect. Thus C<(?-k)> and C<(?-k:...)> are meaningless and will warn
692 when executed under C<use warnings>.
697 =item C<(?imsx-imsx:pattern)>
699 This is for clustering, not capturing; it groups subexpressions like
700 "()", but doesn't make backreferences as "()" does. So
702 @fields = split(/\b(?:a|b|c)\b/)
706 @fields = split(/\b(a|b|c)\b/)
708 but doesn't spit out extra fields. It's also cheaper not to capture
709 characters if you don't need to.
711 Any letters between C<?> and C<:> act as flags modifiers as with
712 C<(?imsx-imsx)>. For example,
714 /(?s-i:more.*than).*million/i
716 is equivalent to the more verbose
718 /(?:(?s-i)more.*than).*million/i
721 X<(?|)> X<Branch reset>
723 This is the "branch reset" pattern, which has the special property
724 that the capture buffers are numbered from the same starting point
725 in each alternation branch. It is available starting from perl 5.10.
727 Capture buffers are numbered from left to right, but inside this
728 construct the numbering is restarted for each branch.
730 The numbering within each branch will be as normal, and any buffers
731 following this construct will be numbered as though the construct
732 contained only one branch, that being the one with the most capture
735 This construct will be useful when you want to capture one of a
736 number of alternative matches.
738 Consider the following pattern. The numbers underneath show in
739 which buffer the captured content will be stored.
742 # before ---------------branch-reset----------- after
743 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
746 =item Look-Around Assertions
747 X<look-around assertion> X<lookaround assertion> X<look-around> X<lookaround>
749 Look-around assertions are zero width patterns which match a specific
750 pattern without including it in C<$&>. Positive assertions match when
751 their subpattern matches, negative assertions match when their subpattern
752 fails. Look-behind matches text up to the current match position,
753 look-ahead matches text following the current match position.
758 X<(?=)> X<look-ahead, positive> X<lookahead, positive>
760 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
761 matches a word followed by a tab, without including the tab in C<$&>.
764 X<(?!)> X<look-ahead, negative> X<lookahead, negative>
766 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
767 matches any occurrence of "foo" that isn't followed by "bar". Note
768 however that look-ahead and look-behind are NOT the same thing. You cannot
769 use this for look-behind.
771 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
772 will not do what you want. That's because the C<(?!foo)> is just saying that
773 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
774 match. You would have to do something like C</(?!foo)...bar/> for that. We
775 say "like" because there's the case of your "bar" not having three characters
776 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
777 Sometimes it's still easier just to say:
779 if (/bar/ && $` !~ /foo$/)
781 For look-behind see below.
783 =item C<(?<=pattern)> C<\K>
784 X<(?<=)> X<look-behind, positive> X<lookbehind, positive> X<\K>
786 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
787 matches a word that follows a tab, without including the tab in C<$&>.
788 Works only for fixed-width look-behind.
790 There is a special form of this construct, called C<\K>, which causes the
791 regex engine to "keep" everything it had matched prior to the C<\K> and
792 not include it in C<$&>. This effectively provides variable length
793 look-behind. The use of C<\K> inside of another look-around assertion
794 is allowed, but the behaviour is currently not well defined.
796 For various reasons C<\K> may be signifigantly more efficient than the
797 equivalent C<< (?<=...) >> construct, and it is especially useful in
798 situations where you want to efficiently remove something following
799 something else in a string. For instance
803 can be rewritten as the much more efficient
807 =item C<(?<!pattern)>
808 X<(?<!)> X<look-behind, negative> X<lookbehind, negative>
810 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
811 matches any occurrence of "foo" that does not follow "bar". Works
812 only for fixed-width look-behind.
816 =item C<(?'NAME'pattern)>
818 =item C<< (?<NAME>pattern) >>
819 X<< (?<NAME>) >> X<(?'NAME')> X<named capture> X<capture>
821 A named capture buffer. Identical in every respect to normal capturing
822 parentheses C<()> but for the additional fact that C<%+> may be used after
823 a succesful match to refer to a named buffer. See C<perlvar> for more
824 details on the C<%+> hash.
826 If multiple distinct capture buffers have the same name then the
827 $+{NAME} will refer to the leftmost defined buffer in the match.
829 The forms C<(?'NAME'pattern)> and C<< (?<NAME>pattern) >> are equivalent.
831 B<NOTE:> While the notation of this construct is the same as the similar
832 function in .NET regexes, the behavior is not. In Perl the buffers are
833 numbered sequentially regardless of being named or not. Thus in the
838 $+{foo} will be the same as $2, and $3 will contain 'z' instead of
839 the opposite which is what a .NET regex hacker might expect.
841 Currently NAME is restricted to simple identifiers only.
842 In other words, it must match C</^[_A-Za-z][_A-Za-z0-9]*\z/> or
843 its Unicode extension (see L<utf8>),
844 though it isn't extended by the locale (see L<perllocale>).
846 B<NOTE:> In order to make things easier for programmers with experience
847 with the Python or PCRE regex engines, the pattern C<< (?PE<lt>NAMEE<gt>pattern) >>
848 may be used instead of C<< (?<NAME>pattern) >>; however this form does not
849 support the use of single quotes as a delimiter for the name. This is
850 only available in Perl 5.10 or later.
852 =item C<< \k<NAME> >>
854 =item C<< \k'NAME' >>
856 Named backreference. Similar to numeric backreferences, except that
857 the group is designated by name and not number. If multiple groups
858 have the same name then it refers to the leftmost defined group in
861 It is an error to refer to a name not defined by a C<< (?<NAME>) >>
862 earlier in the pattern.
864 Both forms are equivalent.
866 B<NOTE:> In order to make things easier for programmers with experience
867 with the Python or PCRE regex engines, the pattern C<< (?P=NAME) >>
868 may be used instead of C<< \k<NAME> >> in Perl 5.10 or later.
871 X<(?{})> X<regex, code in> X<regexp, code in> X<regular expression, code in>
873 B<WARNING>: This extended regular expression feature is considered
874 experimental, and may be changed without notice. Code executed that
875 has side effects may not perform identically from version to version
876 due to the effect of future optimisations in the regex engine.
878 This zero-width assertion evaluates any embedded Perl code. It
879 always succeeds, and its C<code> is not interpolated. Currently,
880 the rules to determine where the C<code> ends are somewhat convoluted.
882 This feature can be used together with the special variable C<$^N> to
883 capture the results of submatches in variables without having to keep
884 track of the number of nested parentheses. For example:
886 $_ = "The brown fox jumps over the lazy dog";
887 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
888 print "color = $color, animal = $animal\n";
890 Inside the C<(?{...})> block, C<$_> refers to the string the regular
891 expression is matching against. You can also use C<pos()> to know what is
892 the current position of matching within this string.
894 The C<code> is properly scoped in the following sense: If the assertion
895 is backtracked (compare L<"Backtracking">), all changes introduced after
896 C<local>ization are undone, so that
900 (?{ $cnt = 0 }) # Initialize $cnt.
904 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
908 (?{ $res = $cnt }) # On success copy to non-localized
912 will set C<$res = 4>. Note that after the match, C<$cnt> returns to the globally
913 introduced value, because the scopes that restrict C<local> operators
916 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
917 switch. If I<not> used in this way, the result of evaluation of
918 C<code> is put into the special variable C<$^R>. This happens
919 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
920 inside the same regular expression.
922 The assignment to C<$^R> above is properly localized, so the old
923 value of C<$^R> is restored if the assertion is backtracked; compare
926 Due to an unfortunate implementation issue, the Perl code contained in these
927 blocks is treated as a compile time closure that can have seemingly bizarre
928 consequences when used with lexically scoped variables inside of subroutines
929 or loops. There are various workarounds for this, including simply using
930 global variables instead. If you are using this construct and strange results
931 occur then check for the use of lexically scoped variables.
933 For reasons of security, this construct is forbidden if the regular
934 expression involves run-time interpolation of variables, unless the
935 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
936 variables contain results of C<qr//> operator (see
937 L<perlop/"qr/STRING/imosx">).
939 This restriction is due to the wide-spread and remarkably convenient
940 custom of using run-time determined strings as patterns. For example:
946 Before Perl knew how to execute interpolated code within a pattern,
947 this operation was completely safe from a security point of view,
948 although it could raise an exception from an illegal pattern. If
949 you turn on the C<use re 'eval'>, though, it is no longer secure,
950 so you should only do so if you are also using taint checking.
951 Better yet, use the carefully constrained evaluation within a Safe
952 compartment. See L<perlsec> for details about both these mechanisms.
954 Because Perl's regex engine is currently not re-entrant, interpolated
955 code may not invoke the regex engine either directly with C<m//> or C<s///>),
956 or indirectly with functions such as C<split>.
958 =item C<(??{ code })>
960 X<regex, postponed> X<regexp, postponed> X<regular expression, postponed>
962 B<WARNING>: This extended regular expression feature is considered
963 experimental, and may be changed without notice. Code executed that
964 has side effects may not perform identically from version to version
965 due to the effect of future optimisations in the regex engine.
967 This is a "postponed" regular subexpression. The C<code> is evaluated
968 at run time, at the moment this subexpression may match. The result
969 of evaluation is considered as a regular expression and matched as
970 if it were inserted instead of this construct. Note that this means
971 that the contents of capture buffers defined inside an eval'ed pattern
972 are not available outside of the pattern, and vice versa, there is no
973 way for the inner pattern to refer to a capture buffer defined outside.
976 ('a' x 100)=~/(??{'(.)' x 100})/
978 B<will> match, it will B<not> set $1.
980 The C<code> is not interpolated. As before, the rules to determine
981 where the C<code> ends are currently somewhat convoluted.
983 The following pattern matches a parenthesized group:
988 (?> [^()]+ ) # Non-parens without backtracking
990 (??{ $re }) # Group with matching parens
995 See also C<(?PARNO)> for a different, more efficient way to accomplish
998 Because perl's regex engine is not currently re-entrant, delayed
999 code may not invoke the regex engine either directly with C<m//> or C<s///>),
1000 or indirectly with functions such as C<split>.
1002 Recursing deeper than 50 times without consuming any input string will
1003 result in a fatal error. The maximum depth is compiled into perl, so
1004 changing it requires a custom build.
1006 =item C<(?PARNO)> C<(?-PARNO)> C<(?+PARNO)> C<(?R)> C<(?0)>
1007 X<(?PARNO)> X<(?1)> X<(?R)> X<(?0)> X<(?-1)> X<(?+1)> X<(?-PARNO)> X<(?+PARNO)>
1008 X<regex, recursive> X<regexp, recursive> X<regular expression, recursive>
1009 X<regex, relative recursion>
1011 Similar to C<(??{ code })> except it does not involve compiling any code,
1012 instead it treats the contents of a capture buffer as an independent
1013 pattern that must match at the current position. Capture buffers
1014 contained by the pattern will have the value as determined by the
1015 outermost recursion.
1017 PARNO is a sequence of digits (not starting with 0) whose value reflects
1018 the paren-number of the capture buffer to recurse to. C<(?R)> recurses to
1019 the beginning of the whole pattern. C<(?0)> is an alternate syntax for
1020 C<(?R)>. If PARNO is preceded by a plus or minus sign then it is assumed
1021 to be relative, with negative numbers indicating preceding capture buffers
1022 and positive ones following. Thus C<(?-1)> refers to the most recently
1023 declared buffer, and C<(?+1)> indicates the next buffer to be declared.
1024 Note that the counting for relative recursion differs from that of
1025 relative backreferences, in that with recursion unclosed buffers B<are>
1028 The following pattern matches a function foo() which may contain
1029 balanced parentheses as the argument.
1031 $re = qr{ ( # paren group 1 (full function)
1033 ( # paren group 2 (parens)
1035 ( # paren group 3 (contents of parens)
1037 (?> [^()]+ ) # Non-parens without backtracking
1039 (?2) # Recurse to start of paren group 2
1047 If the pattern was used as follows
1049 'foo(bar(baz)+baz(bop))'=~/$re/
1050 and print "\$1 = $1\n",
1054 the output produced should be the following:
1056 $1 = foo(bar(baz)+baz(bop))
1057 $2 = (bar(baz)+baz(bop))
1058 $3 = bar(baz)+baz(bop)
1060 If there is no corresponding capture buffer defined, then it is a
1061 fatal error. Recursing deeper than 50 times without consuming any input
1062 string will also result in a fatal error. The maximum depth is compiled
1063 into perl, so changing it requires a custom build.
1065 The following shows how using negative indexing can make it
1066 easier to embed recursive patterns inside of a C<qr//> construct
1069 my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1070 if (/foo $parens \s+ + \s+ bar $parens/x) {
1071 # do something here...
1074 B<Note> that this pattern does not behave the same way as the equivalent
1075 PCRE or Python construct of the same form. In Perl you can backtrack into
1076 a recursed group, in PCRE and Python the recursed into group is treated
1077 as atomic. Also, modifiers are resolved at compile time, so constructs
1078 like (?i:(?1)) or (?:(?i)(?1)) do not affect how the sub-pattern will
1084 Recurse to a named subpattern. Identical to C<(?PARNO)> except that the
1085 parenthesis to recurse to is determined by name. If multiple parentheses have
1086 the same name, then it recurses to the leftmost.
1088 It is an error to refer to a name that is not declared somewhere in the
1091 B<NOTE:> In order to make things easier for programmers with experience
1092 with the Python or PCRE regex engines the pattern C<< (?P>NAME) >>
1093 may be used instead of C<< (?&NAME) >> in Perl 5.10 or later.
1095 =item C<(?(condition)yes-pattern|no-pattern)>
1098 =item C<(?(condition)yes-pattern)>
1100 Conditional expression. C<(condition)> should be either an integer in
1101 parentheses (which is valid if the corresponding pair of parentheses
1102 matched), a look-ahead/look-behind/evaluate zero-width assertion, a
1103 name in angle brackets or single quotes (which is valid if a buffer
1104 with the given name matched), or the special symbol (R) (true when
1105 evaluated inside of recursion or eval). Additionally the R may be
1106 followed by a number, (which will be true when evaluated when recursing
1107 inside of the appropriate group), or by C<&NAME>, in which case it will
1108 be true only when evaluated during recursion in the named group.
1110 Here's a summary of the possible predicates:
1116 Checks if the numbered capturing buffer has matched something.
1118 =item (<NAME>) ('NAME')
1120 Checks if a buffer with the given name has matched something.
1124 Treats the code block as the condition.
1128 Checks if the expression has been evaluated inside of recursion.
1132 Checks if the expression has been evaluated while executing directly
1133 inside of the n-th capture group. This check is the regex equivalent of
1135 if ((caller(0))[3] eq 'subname') { ... }
1137 In other words, it does not check the full recursion stack.
1141 Similar to C<(R1)>, this predicate checks to see if we're executing
1142 directly inside of the leftmost group with a given name (this is the same
1143 logic used by C<(?&NAME)> to disambiguate). It does not check the full
1144 stack, but only the name of the innermost active recursion.
1148 In this case, the yes-pattern is never directly executed, and no
1149 no-pattern is allowed. Similar in spirit to C<(?{0})> but more efficient.
1150 See below for details.
1161 matches a chunk of non-parentheses, possibly included in parentheses
1164 A special form is the C<(DEFINE)> predicate, which never executes directly
1165 its yes-pattern, and does not allow a no-pattern. This allows to define
1166 subpatterns which will be executed only by using the recursion mechanism.
1167 This way, you can define a set of regular expression rules that can be
1168 bundled into any pattern you choose.
1170 It is recommended that for this usage you put the DEFINE block at the
1171 end of the pattern, and that you name any subpatterns defined within it.
1173 Also, it's worth noting that patterns defined this way probably will
1174 not be as efficient, as the optimiser is not very clever about
1177 An example of how this might be used is as follows:
1179 /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
1185 Note that capture buffers matched inside of recursion are not accessible
1186 after the recursion returns, so the extra layer of capturing buffers is
1187 necessary. Thus C<$+{NAME_PAT}> would not be defined even though
1188 C<$+{NAME}> would be.
1190 =item C<< (?>pattern) >>
1191 X<backtrack> X<backtracking> X<atomic> X<possessive>
1193 An "independent" subexpression, one which matches the substring
1194 that a I<standalone> C<pattern> would match if anchored at the given
1195 position, and it matches I<nothing other than this substring>. This
1196 construct is useful for optimizations of what would otherwise be
1197 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
1198 It may also be useful in places where the "grab all you can, and do not
1199 give anything back" semantic is desirable.
1201 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
1202 (anchored at the beginning of string, as above) will match I<all>
1203 characters C<a> at the beginning of string, leaving no C<a> for
1204 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
1205 since the match of the subgroup C<a*> is influenced by the following
1206 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
1207 C<a*ab> will match fewer characters than a standalone C<a*>, since
1208 this makes the tail match.
1210 An effect similar to C<< (?>pattern) >> may be achieved by writing
1211 C<(?=(pattern))\1>. This matches the same substring as a standalone
1212 C<a+>, and the following C<\1> eats the matched string; it therefore
1213 makes a zero-length assertion into an analogue of C<< (?>...) >>.
1214 (The difference between these two constructs is that the second one
1215 uses a capturing group, thus shifting ordinals of backreferences
1216 in the rest of a regular expression.)
1218 Consider this pattern:
1229 That will efficiently match a nonempty group with matching parentheses
1230 two levels deep or less. However, if there is no such group, it
1231 will take virtually forever on a long string. That's because there
1232 are so many different ways to split a long string into several
1233 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
1234 to a subpattern of the above pattern. Consider how the pattern
1235 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
1236 seconds, but that each extra letter doubles this time. This
1237 exponential performance will make it appear that your program has
1238 hung. However, a tiny change to this pattern
1242 (?> [^()]+ ) # change x+ above to (?> x+ )
1249 which uses C<< (?>...) >> matches exactly when the one above does (verifying
1250 this yourself would be a productive exercise), but finishes in a fourth
1251 the time when used on a similar string with 1000000 C<a>s. Be aware,
1252 however, that this pattern currently triggers a warning message under
1253 the C<use warnings> pragma or B<-w> switch saying it
1254 C<"matches null string many times in regex">.
1256 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
1257 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
1258 This was only 4 times slower on a string with 1000000 C<a>s.
1260 The "grab all you can, and do not give anything back" semantic is desirable
1261 in many situations where on the first sight a simple C<()*> looks like
1262 the correct solution. Suppose we parse text with comments being delimited
1263 by C<#> followed by some optional (horizontal) whitespace. Contrary to
1264 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
1265 the comment delimiter, because it may "give up" some whitespace if
1266 the remainder of the pattern can be made to match that way. The correct
1267 answer is either one of these:
1272 For example, to grab non-empty comments into $1, one should use either
1275 / (?> \# [ \t]* ) ( .+ ) /x;
1276 / \# [ \t]* ( [^ \t] .* ) /x;
1278 Which one you pick depends on which of these expressions better reflects
1279 the above specification of comments.
1281 In some literature this construct is called "atomic matching" or
1282 "possessive matching".
1284 Possessive quantifiers are equivalent to putting the item they are applied
1285 to inside of one of these constructs. The following equivalences apply:
1287 Quantifier Form Bracketing Form
1288 --------------- ---------------
1292 PAT{min,max}+ (?>PAT{min,max})
1296 =head2 Special Backtracking Control Verbs
1298 B<WARNING:> These patterns are experimental and subject to change or
1299 removal in a future version of Perl. Their usage in production code should
1300 be noted to avoid problems during upgrades.
1302 These special patterns are generally of the form C<(*VERB:ARG)>. Unless
1303 otherwise stated the ARG argument is optional; in some cases, it is
1306 Any pattern containing a special backtracking verb that allows an argument
1307 has the special behaviour that when executed it sets the current packages'
1308 C<$REGERROR> and C<$REGMARK> variables. When doing so the following
1311 On failure, the C<$REGERROR> variable will be set to the ARG value of the
1312 verb pattern, if the verb was involved in the failure of the match. If the
1313 ARG part of the pattern was omitted, then C<$REGERROR> will be set to the
1314 name of the last C<(*MARK:NAME)> pattern executed, or to TRUE if there was
1315 none. Also, the C<$REGMARK> variable will be set to FALSE.
1317 On a successful match, the C<$REGERROR> variable will be set to FALSE, and
1318 the C<$REGMARK> variable will be set to the name of the last
1319 C<(*MARK:NAME)> pattern executed. See the explanation for the
1320 C<(*MARK:NAME)> verb below for more details.
1322 B<NOTE:> C<$REGERROR> and C<$REGMARK> are not magic variables like C<$1>
1323 and most other regex related variables. They are not local to a scope, nor
1324 readonly, but instead are volatile package variables similar to C<$AUTOLOAD>.
1325 Use C<local> to localize changes to them to a specific scope if necessary.
1327 If a pattern does not contain a special backtracking verb that allows an
1328 argument, then C<$REGERROR> and C<$REGMARK> are not touched at all.
1332 =item Verbs that take an argument
1336 =item C<(*PRUNE)> C<(*PRUNE:NAME)>
1337 X<(*PRUNE)> X<(*PRUNE:NAME)> X<\v>
1339 This zero-width pattern prunes the backtracking tree at the current point
1340 when backtracked into on failure. Consider the pattern C<A (*PRUNE) B>,
1341 where A and B are complex patterns. Until the C<(*PRUNE)> verb is reached,
1342 A may backtrack as necessary to match. Once it is reached, matching
1343 continues in B, which may also backtrack as necessary; however, should B
1344 not match, then no further backtracking will take place, and the pattern
1345 will fail outright at the current starting position.
1347 As a shortcut, C<\v> is exactly equivalent to C<(*PRUNE)>.
1349 The following example counts all the possible matching strings in a
1350 pattern (without actually matching any of them).
1352 'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
1353 print "Count=$count\n";
1368 If we add a C<(*PRUNE)> before the count like the following
1370 'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
1371 print "Count=$count\n";
1373 we prevent backtracking and find the count of the longest matching
1374 at each matching startpoint like so:
1381 Any number of C<(*PRUNE)> assertions may be used in a pattern.
1383 See also C<< (?>pattern) >> and possessive quantifiers for other ways to
1384 control backtracking. In some cases, the use of C<(*PRUNE)> can be
1385 replaced with a C<< (?>pattern) >> with no functional difference; however,
1386 C<(*PRUNE)> can be used to handle cases that cannot be expressed using a
1387 C<< (?>pattern) >> alone.
1390 =item C<(*SKIP)> C<(*SKIP:NAME)>
1393 This zero-width pattern is similar to C<(*PRUNE)>, except that on
1394 failure it also signifies that whatever text that was matched leading up
1395 to the C<(*SKIP)> pattern being executed cannot be part of I<any> match
1396 of this pattern. This effectively means that the regex engine "skips" forward
1397 to this position on failure and tries to match again, (assuming that
1398 there is sufficient room to match).
1400 As a shortcut C<\V> is exactly equivalent to C<(*SKIP)>.
1402 The name of the C<(*SKIP:NAME)> pattern has special significance. If a
1403 C<(*MARK:NAME)> was encountered while matching, then it is that position
1404 which is used as the "skip point". If no C<(*MARK)> of that name was
1405 encountered, then the C<(*SKIP)> operator has no effect. When used
1406 without a name the "skip point" is where the match point was when
1407 executing the (*SKIP) pattern.
1409 Compare the following to the examples in C<(*PRUNE)>, note the string
1412 'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
1413 print "Count=$count\n";
1421 Once the 'aaab' at the start of the string has matched, and the C<(*SKIP)>
1422 executed, the next startpoint will be where the cursor was when the
1423 C<(*SKIP)> was executed.
1425 =item C<(*MARK:NAME)> C<(*:NAME)>
1426 X<(*MARK)> C<(*MARK:NAME)> C<(*:NAME)>
1428 This zero-width pattern can be used to mark the point reached in a string
1429 when a certain part of the pattern has been successfully matched. This
1430 mark may be given a name. A later C<(*SKIP)> pattern will then skip
1431 forward to that point if backtracked into on failure. Any number of
1432 C<(*MARK)> patterns are allowed, and the NAME portion is optional and may
1435 In addition to interacting with the C<(*SKIP)> pattern, C<(*MARK:NAME)>
1436 can be used to "label" a pattern branch, so that after matching, the
1437 program can determine which branches of the pattern were involved in the
1440 When a match is successful, the C<$REGMARK> variable will be set to the
1441 name of the most recently executed C<(*MARK:NAME)> that was involved
1444 This can be used to determine which branch of a pattern was matched
1445 without using a seperate capture buffer for each branch, which in turn
1446 can result in a performance improvement, as perl cannot optimize
1447 C</(?:(x)|(y)|(z))/> as efficiently as something like
1448 C</(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/>.
1450 When a match has failed, and unless another verb has been involved in
1451 failing the match and has provided its own name to use, the C<$REGERROR>
1452 variable will be set to the name of the most recently executed
1455 See C<(*SKIP)> for more details.
1457 As a shortcut C<(*MARK:NAME)> can be written C<(*:NAME)>.
1459 =item C<(*THEN)> C<(*THEN:NAME)>
1461 This is similar to the "cut group" operator C<::> from Perl6. Like
1462 C<(*PRUNE)>, this verb always matches, and when backtracked into on
1463 failure, it causes the regex engine to try the next alternation in the
1464 innermost enclosing group (capturing or otherwise).
1466 Its name comes from the observation that this operation combined with the
1467 alternation operator (C<|>) can be used to create what is essentially a
1468 pattern-based if/then/else block:
1470 ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
1472 Note that if this operator is used and NOT inside of an alternation then
1473 it acts exactly like the C<(*PRUNE)> operator.
1483 / ( A (*THEN) B | C (*THEN) D ) /
1487 / ( A (*PRUNE) B | C (*PRUNE) D ) /
1489 as after matching the A but failing on the B the C<(*THEN)> verb will
1490 backtrack and try C; but the C<(*PRUNE)> verb will simply fail.
1495 This is the Perl6 "commit pattern" C<< <commit> >> or C<:::>. It's a
1496 zero-width pattern similar to C<(*SKIP)>, except that when backtracked
1497 into on failure it causes the match to fail outright. No further attempts
1498 to find a valid match by advancing the start pointer will occur again.
1501 'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
1502 print "Count=$count\n";
1509 In other words, once the C<(*COMMIT)> has been entered, and if the pattern
1510 does not match, the regex engine will not try any further matching on the
1515 =item Verbs without an argument
1519 =item C<(*FAIL)> C<(*F)>
1522 This pattern matches nothing and always fails. It can be used to force the
1523 engine to backtrack. It is equivalent to C<(?!)>, but easier to read. In
1524 fact, C<(?!)> gets optimised into C<(*FAIL)> internally.
1526 It is probably useful only when combined with C<(?{})> or C<(??{})>.
1531 B<WARNING:> This feature is highly experimental. It is not recommended
1532 for production code.
1534 This pattern matches nothing and causes the end of successful matching at
1535 the point at which the C<(*ACCEPT)> pattern was encountered, regardless of
1536 whether there is actually more to match in the string. When inside of a
1537 nested pattern, such as recursion, or in a subpattern dynamically generated
1538 via C<(??{})>, only the innermost pattern is ended immediately.
1540 If the C<(*ACCEPT)> is inside of capturing buffers then the buffers are
1541 marked as ended at the point at which the C<(*ACCEPT)> was encountered.
1544 'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
1546 will match, and C<$1> will be C<AB> and C<$2> will be C<B>, C<$3> will not
1547 be set. If another branch in the inner parentheses were matched, such as in the
1548 string 'ACDE', then the C<D> and C<E> would have to be matched as well.
1555 X<backtrack> X<backtracking>
1557 NOTE: This section presents an abstract approximation of regular
1558 expression behavior. For a more rigorous (and complicated) view of
1559 the rules involved in selecting a match among possible alternatives,
1560 see L<Combining RE Pieces>.
1562 A fundamental feature of regular expression matching involves the
1563 notion called I<backtracking>, which is currently used (when needed)
1564 by all regular non-possessive expression quantifiers, namely C<*>, C<*?>, C<+>,
1565 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
1566 internally, but the general principle outlined here is valid.
1568 For a regular expression to match, the I<entire> regular expression must
1569 match, not just part of it. So if the beginning of a pattern containing a
1570 quantifier succeeds in a way that causes later parts in the pattern to
1571 fail, the matching engine backs up and recalculates the beginning
1572 part--that's why it's called backtracking.
1574 Here is an example of backtracking: Let's say you want to find the
1575 word following "foo" in the string "Food is on the foo table.":
1577 $_ = "Food is on the foo table.";
1578 if ( /\b(foo)\s+(\w+)/i ) {
1579 print "$2 follows $1.\n";
1582 When the match runs, the first part of the regular expression (C<\b(foo)>)
1583 finds a possible match right at the beginning of the string, and loads up
1584 $1 with "Foo". However, as soon as the matching engine sees that there's
1585 no whitespace following the "Foo" that it had saved in $1, it realizes its
1586 mistake and starts over again one character after where it had the
1587 tentative match. This time it goes all the way until the next occurrence
1588 of "foo". The complete regular expression matches this time, and you get
1589 the expected output of "table follows foo."
1591 Sometimes minimal matching can help a lot. Imagine you'd like to match
1592 everything between "foo" and "bar". Initially, you write something
1595 $_ = "The food is under the bar in the barn.";
1596 if ( /foo(.*)bar/ ) {
1600 Which perhaps unexpectedly yields:
1602 got <d is under the bar in the >
1604 That's because C<.*> was greedy, so you get everything between the
1605 I<first> "foo" and the I<last> "bar". Here it's more effective
1606 to use minimal matching to make sure you get the text between a "foo"
1607 and the first "bar" thereafter.
1609 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
1610 got <d is under the >
1612 Here's another example. Let's say you'd like to match a number at the end
1613 of a string, and you also want to keep the preceding part of the match.
1616 $_ = "I have 2 numbers: 53147";
1617 if ( /(.*)(\d*)/ ) { # Wrong!
1618 print "Beginning is <$1>, number is <$2>.\n";
1621 That won't work at all, because C<.*> was greedy and gobbled up the
1622 whole string. As C<\d*> can match on an empty string the complete
1623 regular expression matched successfully.
1625 Beginning is <I have 2 numbers: 53147>, number is <>.
1627 Here are some variants, most of which don't work:
1629 $_ = "I have 2 numbers: 53147";
1642 printf "%-12s ", $pat;
1644 print "<$1> <$2>\n";
1650 That will print out:
1652 (.*)(\d*) <I have 2 numbers: 53147> <>
1653 (.*)(\d+) <I have 2 numbers: 5314> <7>
1655 (.*?)(\d+) <I have > <2>
1656 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
1657 (.*?)(\d+)$ <I have 2 numbers: > <53147>
1658 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
1659 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
1661 As you see, this can be a bit tricky. It's important to realize that a
1662 regular expression is merely a set of assertions that gives a definition
1663 of success. There may be 0, 1, or several different ways that the
1664 definition might succeed against a particular string. And if there are
1665 multiple ways it might succeed, you need to understand backtracking to
1666 know which variety of success you will achieve.
1668 When using look-ahead assertions and negations, this can all get even
1669 trickier. Imagine you'd like to find a sequence of non-digits not
1670 followed by "123". You might try to write that as
1673 if ( /^\D*(?!123)/ ) { # Wrong!
1674 print "Yup, no 123 in $_\n";
1677 But that isn't going to match; at least, not the way you're hoping. It
1678 claims that there is no 123 in the string. Here's a clearer picture of
1679 why that pattern matches, contrary to popular expectations:
1684 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
1685 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
1687 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
1688 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
1696 You might have expected test 3 to fail because it seems to a more
1697 general purpose version of test 1. The important difference between
1698 them is that test 3 contains a quantifier (C<\D*>) and so can use
1699 backtracking, whereas test 1 will not. What's happening is
1700 that you've asked "Is it true that at the start of $x, following 0 or more
1701 non-digits, you have something that's not 123?" If the pattern matcher had
1702 let C<\D*> expand to "ABC", this would have caused the whole pattern to
1705 The search engine will initially match C<\D*> with "ABC". Then it will
1706 try to match C<(?!123> with "123", which fails. But because
1707 a quantifier (C<\D*>) has been used in the regular expression, the
1708 search engine can backtrack and retry the match differently
1709 in the hope of matching the complete regular expression.
1711 The pattern really, I<really> wants to succeed, so it uses the
1712 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
1713 time. Now there's indeed something following "AB" that is not
1714 "123". It's "C123", which suffices.
1716 We can deal with this by using both an assertion and a negation.
1717 We'll say that the first part in $1 must be followed both by a digit
1718 and by something that's not "123". Remember that the look-aheads
1719 are zero-width expressions--they only look, but don't consume any
1720 of the string in their match. So rewriting this way produces what
1721 you'd expect; that is, case 5 will fail, but case 6 succeeds:
1723 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
1724 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
1728 In other words, the two zero-width assertions next to each other work as though
1729 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
1730 matches only if you're at the beginning of the line AND the end of the
1731 line simultaneously. The deeper underlying truth is that juxtaposition in
1732 regular expressions always means AND, except when you write an explicit OR
1733 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
1734 although the attempted matches are made at different positions because "a"
1735 is not a zero-width assertion, but a one-width assertion.
1737 B<WARNING>: Particularly complicated regular expressions can take
1738 exponential time to solve because of the immense number of possible
1739 ways they can use backtracking to try for a match. For example, without
1740 internal optimizations done by the regular expression engine, this will
1741 take a painfully long time to run:
1743 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
1745 And if you used C<*>'s in the internal groups instead of limiting them
1746 to 0 through 5 matches, then it would take forever--or until you ran
1747 out of stack space. Moreover, these internal optimizations are not
1748 always applicable. For example, if you put C<{0,5}> instead of C<*>
1749 on the external group, no current optimization is applicable, and the
1750 match takes a long time to finish.
1752 A powerful tool for optimizing such beasts is what is known as an
1753 "independent group",
1754 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
1755 zero-length look-ahead/look-behind assertions will not backtrack to make
1756 the tail match, since they are in "logical" context: only
1757 whether they match is considered relevant. For an example
1758 where side-effects of look-ahead I<might> have influenced the
1759 following match, see L<C<< (?>pattern) >>>.
1761 =head2 Version 8 Regular Expressions
1762 X<regular expression, version 8> X<regex, version 8> X<regexp, version 8>
1764 In case you're not familiar with the "regular" Version 8 regex
1765 routines, here are the pattern-matching rules not described above.
1767 Any single character matches itself, unless it is a I<metacharacter>
1768 with a special meaning described here or above. You can cause
1769 characters that normally function as metacharacters to be interpreted
1770 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
1771 character; "\\" matches a "\"). This escape mechanism is also required
1772 for the character used as the pattern delimiter.
1774 A series of characters matches that series of characters in the target
1775 string, so the pattern C<blurfl> would match "blurfl" in the target
1778 You can specify a character class, by enclosing a list of characters
1779 in C<[]>, which will match any character from the list. If the
1780 first character after the "[" is "^", the class matches any character not
1781 in the list. Within a list, the "-" character specifies a
1782 range, so that C<a-z> represents all characters between "a" and "z",
1783 inclusive. If you want either "-" or "]" itself to be a member of a
1784 class, put it at the start of the list (possibly after a "^"), or
1785 escape it with a backslash. "-" is also taken literally when it is
1786 at the end of the list, just before the closing "]". (The
1787 following all specify the same class of three characters: C<[-az]>,
1788 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1789 specifies a class containing twenty-six characters, even on EBCDIC-based
1790 character sets.) Also, if you try to use the character
1791 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1792 a range, the "-" is understood literally.
1794 Note also that the whole range idea is rather unportable between
1795 character sets--and even within character sets they may cause results
1796 you probably didn't expect. A sound principle is to use only ranges
1797 that begin from and end at either alphabetics of equal case ([a-e],
1798 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1799 spell out the character sets in full.
1801 Characters may be specified using a metacharacter syntax much like that
1802 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1803 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1804 of octal digits, matches the character whose coded character set value
1805 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1806 matches the character whose numeric value is I<nn>. The expression \cI<x>
1807 matches the character control-I<x>. Finally, the "." metacharacter
1808 matches any character except "\n" (unless you use C</s>).
1810 You can specify a series of alternatives for a pattern using "|" to
1811 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1812 or "foe" in the target string (as would C<f(e|i|o)e>). The
1813 first alternative includes everything from the last pattern delimiter
1814 ("(", "[", or the beginning of the pattern) up to the first "|", and
1815 the last alternative contains everything from the last "|" to the next
1816 pattern delimiter. That's why it's common practice to include
1817 alternatives in parentheses: to minimize confusion about where they
1820 Alternatives are tried from left to right, so the first
1821 alternative found for which the entire expression matches, is the one that
1822 is chosen. This means that alternatives are not necessarily greedy. For
1823 example: when matching C<foo|foot> against "barefoot", only the "foo"
1824 part will match, as that is the first alternative tried, and it successfully
1825 matches the target string. (This might not seem important, but it is
1826 important when you are capturing matched text using parentheses.)
1828 Also remember that "|" is interpreted as a literal within square brackets,
1829 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1831 Within a pattern, you may designate subpatterns for later reference
1832 by enclosing them in parentheses, and you may refer back to the
1833 I<n>th subpattern later in the pattern using the metacharacter
1834 \I<n>. Subpatterns are numbered based on the left to right order
1835 of their opening parenthesis. A backreference matches whatever
1836 actually matched the subpattern in the string being examined, not
1837 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1838 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1839 1 matched "0x", even though the rule C<0|0x> could potentially match
1840 the leading 0 in the second number.
1842 =head2 Warning on \1 Instead of $1
1844 Some people get too used to writing things like:
1846 $pattern =~ s/(\W)/\\\1/g;
1848 This is grandfathered for the RHS of a substitute to avoid shocking the
1849 B<sed> addicts, but it's a dirty habit to get into. That's because in
1850 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1851 the usual double-quoted string means a control-A. The customary Unix
1852 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1853 of doing that, you get yourself into trouble if you then add an C</e>
1856 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1862 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1863 C<${1}000>. The operation of interpolation should not be confused
1864 with the operation of matching a backreference. Certainly they mean two
1865 different things on the I<left> side of the C<s///>.
1867 =head2 Repeated Patterns Matching a Zero-length Substring
1869 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1871 Regular expressions provide a terse and powerful programming language. As
1872 with most other power tools, power comes together with the ability
1875 A common abuse of this power stems from the ability to make infinite
1876 loops using regular expressions, with something as innocuous as:
1878 'foo' =~ m{ ( o? )* }x;
1880 The C<o?> matches at the beginning of C<'foo'>, and since the position
1881 in the string is not moved by the match, C<o?> would match again and again
1882 because of the C<*> modifier. Another common way to create a similar cycle
1883 is with the looping modifier C<//g>:
1885 @matches = ( 'foo' =~ m{ o? }xg );
1889 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1891 or the loop implied by split().
1893 However, long experience has shown that many programming tasks may
1894 be significantly simplified by using repeated subexpressions that
1895 may match zero-length substrings. Here's a simple example being:
1897 @chars = split //, $string; # // is not magic in split
1898 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1900 Thus Perl allows such constructs, by I<forcefully breaking
1901 the infinite loop>. The rules for this are different for lower-level
1902 loops given by the greedy modifiers C<*+{}>, and for higher-level
1903 ones like the C</g> modifier or split() operator.
1905 The lower-level loops are I<interrupted> (that is, the loop is
1906 broken) when Perl detects that a repeated expression matched a
1907 zero-length substring. Thus
1909 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1911 is made equivalent to
1913 m{ (?: NON_ZERO_LENGTH )*
1918 The higher level-loops preserve an additional state between iterations:
1919 whether the last match was zero-length. To break the loop, the following
1920 match after a zero-length match is prohibited to have a length of zero.
1921 This prohibition interacts with backtracking (see L<"Backtracking">),
1922 and so the I<second best> match is chosen if the I<best> match is of
1930 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1931 match given by non-greedy C<??> is the zero-length match, and the I<second
1932 best> match is what is matched by C<\w>. Thus zero-length matches
1933 alternate with one-character-long matches.
1935 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1936 position one notch further in the string.
1938 The additional state of being I<matched with zero-length> is associated with
1939 the matched string, and is reset by each assignment to pos().
1940 Zero-length matches at the end of the previous match are ignored
1943 =head2 Combining RE Pieces
1945 Each of the elementary pieces of regular expressions which were described
1946 before (such as C<ab> or C<\Z>) could match at most one substring
1947 at the given position of the input string. However, in a typical regular
1948 expression these elementary pieces are combined into more complicated
1949 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1950 (in these examples C<S> and C<T> are regular subexpressions).
1952 Such combinations can include alternatives, leading to a problem of choice:
1953 if we match a regular expression C<a|ab> against C<"abc">, will it match
1954 substring C<"a"> or C<"ab">? One way to describe which substring is
1955 actually matched is the concept of backtracking (see L<"Backtracking">).
1956 However, this description is too low-level and makes you think
1957 in terms of a particular implementation.
1959 Another description starts with notions of "better"/"worse". All the
1960 substrings which may be matched by the given regular expression can be
1961 sorted from the "best" match to the "worst" match, and it is the "best"
1962 match which is chosen. This substitutes the question of "what is chosen?"
1963 by the question of "which matches are better, and which are worse?".
1965 Again, for elementary pieces there is no such question, since at most
1966 one match at a given position is possible. This section describes the
1967 notion of better/worse for combining operators. In the description
1968 below C<S> and C<T> are regular subexpressions.
1974 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1975 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1976 which can be matched by C<T>.
1978 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
1981 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
1982 C<B> is better match for C<T> than C<B'>.
1986 When C<S> can match, it is a better match than when only C<T> can match.
1988 Ordering of two matches for C<S> is the same as for C<S>. Similar for
1989 two matches for C<T>.
1991 =item C<S{REPEAT_COUNT}>
1993 Matches as C<SSS...S> (repeated as many times as necessary).
1997 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
1999 =item C<S{min,max}?>
2001 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
2003 =item C<S?>, C<S*>, C<S+>
2005 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
2007 =item C<S??>, C<S*?>, C<S+?>
2009 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
2013 Matches the best match for C<S> and only that.
2015 =item C<(?=S)>, C<(?<=S)>
2017 Only the best match for C<S> is considered. (This is important only if
2018 C<S> has capturing parentheses, and backreferences are used somewhere
2019 else in the whole regular expression.)
2021 =item C<(?!S)>, C<(?<!S)>
2023 For this grouping operator there is no need to describe the ordering, since
2024 only whether or not C<S> can match is important.
2026 =item C<(??{ EXPR })>, C<(?PARNO)>
2028 The ordering is the same as for the regular expression which is
2029 the result of EXPR, or the pattern contained by capture buffer PARNO.
2031 =item C<(?(condition)yes-pattern|no-pattern)>
2033 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
2034 already determined. The ordering of the matches is the same as for the
2035 chosen subexpression.
2039 The above recipes describe the ordering of matches I<at a given position>.
2040 One more rule is needed to understand how a match is determined for the
2041 whole regular expression: a match at an earlier position is always better
2042 than a match at a later position.
2044 =head2 Creating Custom RE Engines
2046 Overloaded constants (see L<overload>) provide a simple way to extend
2047 the functionality of the RE engine.
2049 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
2050 matches at a boundary between whitespace characters and non-whitespace
2051 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
2052 at these positions, so we want to have each C<\Y|> in the place of the
2053 more complicated version. We can create a module C<customre> to do
2061 die "No argument to customre::import allowed" if @_;
2062 overload::constant 'qr' => \&convert;
2065 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2067 # We must also take care of not escaping the legitimate \\Y|
2068 # sequence, hence the presence of '\\' in the conversion rules.
2069 my %rules = ( '\\' => '\\\\',
2070 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2076 { $rules{$1} or invalid($re,$1) }sgex;
2080 Now C<use customre> enables the new escape in constant regular
2081 expressions, i.e., those without any runtime variable interpolations.
2082 As documented in L<overload>, this conversion will work only over
2083 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
2084 part of this regular expression needs to be converted explicitly
2085 (but only if the special meaning of C<\Y|> should be enabled inside $re):
2090 $re = customre::convert $re;
2093 =head1 PCRE/Python Support
2095 As of Perl 5.10 Perl supports several Python/PCRE specific extensions
2096 to the regex syntax. While Perl programmers are encouraged to use the
2097 Perl specific syntax, the following are legal in Perl 5.10:
2101 =item C<< (?PE<lt>NAMEE<gt>pattern) >>
2103 Define a named capture buffer. Equivalent to C<< (?<NAME>pattern) >>.
2105 =item C<< (?P=NAME) >>
2107 Backreference to a named capture buffer. Equivalent to C<< \g{NAME} >>.
2109 =item C<< (?P>NAME) >>
2111 Subroutine call to a named capture buffer. Equivalent to C<< (?&NAME) >>.
2117 This document varies from difficult to understand to completely
2118 and utterly opaque. The wandering prose riddled with jargon is
2119 hard to fathom in several places.
2121 This document needs a rewrite that separates the tutorial content
2122 from the reference content.
2130 L<perlop/"Regexp Quote-Like Operators">.
2132 L<perlop/"Gory details of parsing quoted constructs">.
2142 I<Mastering Regular Expressions> by Jeffrey Friedl, published
2143 by O'Reilly and Associates.