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 branch. It is available starting from perl 5.10.
727 Normally capture buffers in a pattern are numbered sequentially,
728 from left to right. Inside this construct that behaviour is
729 overridden so that the capture buffers are shared between all the
730 branches and take their values from the branch that matched.
732 The numbering within each branch will be as normal, and any buffers
733 following this construct will be numbered as though the construct
734 contained only one branch, that being the one with the most capture
737 This construct will be useful when you want to capture one of a
738 number of alternative matches.
740 Consider the following pattern. The numbers underneath show in
741 which buffer the captured content will be stored.
744 # before ---------------branch-reset----------- after
745 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
748 =item Look-Around Assertions
749 X<look-around assertion> X<lookaround assertion> X<look-around> X<lookaround>
751 Look-around assertions are zero width patterns which match a specific
752 pattern without including it in C<$&>. Positive assertions match when
753 their subpattern matches, negative assertions match when their subpattern
754 fails. Look-behind matches text up to the current match position,
755 look-ahead matches text following the current match position.
760 X<(?=)> X<look-ahead, positive> X<lookahead, positive>
762 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
763 matches a word followed by a tab, without including the tab in C<$&>.
766 X<(?!)> X<look-ahead, negative> X<lookahead, negative>
768 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
769 matches any occurrence of "foo" that isn't followed by "bar". Note
770 however that look-ahead and look-behind are NOT the same thing. You cannot
771 use this for look-behind.
773 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
774 will not do what you want. That's because the C<(?!foo)> is just saying that
775 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
776 match. You would have to do something like C</(?!foo)...bar/> for that. We
777 say "like" because there's the case of your "bar" not having three characters
778 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
779 Sometimes it's still easier just to say:
781 if (/bar/ && $` !~ /foo$/)
783 For look-behind see below.
785 =item C<(?<=pattern)> C<\K>
786 X<(?<=)> X<look-behind, positive> X<lookbehind, positive> X<\K>
788 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
789 matches a word that follows a tab, without including the tab in C<$&>.
790 Works only for fixed-width look-behind.
792 There is a special form of this construct, called C<\K>, which causes the
793 regex engine to "keep" everything it had matched prior to the C<\K> and
794 not include it in C<$&>. This effectively provides variable length
795 look-behind. The use of C<\K> inside of another look-around assertion
796 is allowed, but the behaviour is currently not well defined.
798 For various reasons C<\K> may be signifigantly more efficient than the
799 equivalent C<< (?<=...) >> construct, and it is especially useful in
800 situations where you want to efficiently remove something following
801 something else in a string. For instance
805 can be rewritten as the much more efficient
809 =item C<(?<!pattern)>
810 X<(?<!)> X<look-behind, negative> X<lookbehind, negative>
812 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
813 matches any occurrence of "foo" that does not follow "bar". Works
814 only for fixed-width look-behind.
818 =item C<(?'NAME'pattern)>
820 =item C<< (?<NAME>pattern) >>
821 X<< (?<NAME>) >> X<(?'NAME')> X<named capture> X<capture>
823 A named capture buffer. Identical in every respect to normal capturing
824 parentheses C<()> but for the additional fact that C<%+> may be used after
825 a succesful match to refer to a named buffer. See C<perlvar> for more
826 details on the C<%+> hash.
828 If multiple distinct capture buffers have the same name then the
829 $+{NAME} will refer to the leftmost defined buffer in the match.
831 The forms C<(?'NAME'pattern)> and C<< (?<NAME>pattern) >> are equivalent.
833 B<NOTE:> While the notation of this construct is the same as the similar
834 function in .NET regexes, the behavior is not. In Perl the buffers are
835 numbered sequentially regardless of being named or not. Thus in the
840 $+{foo} will be the same as $2, and $3 will contain 'z' instead of
841 the opposite which is what a .NET regex hacker might expect.
843 Currently NAME is restricted to simple identifiers only.
844 In other words, it must match C</^[_A-Za-z][_A-Za-z0-9]*\z/> or
845 its Unicode extension (see L<utf8>),
846 though it isn't extended by the locale (see L<perllocale>).
848 B<NOTE:> In order to make things easier for programmers with experience
849 with the Python or PCRE regex engines, the pattern C<< (?PE<lt>NAMEE<gt>pattern) >>
850 may be used instead of C<< (?<NAME>pattern) >>; however this form does not
851 support the use of single quotes as a delimiter for the name. This is
852 only available in Perl 5.10 or later.
854 =item C<< \k<NAME> >>
856 =item C<< \k'NAME' >>
858 Named backreference. Similar to numeric backreferences, except that
859 the group is designated by name and not number. If multiple groups
860 have the same name then it refers to the leftmost defined group in
863 It is an error to refer to a name not defined by a C<< (?<NAME>) >>
864 earlier in the pattern.
866 Both forms are equivalent.
868 B<NOTE:> In order to make things easier for programmers with experience
869 with the Python or PCRE regex engines, the pattern C<< (?P=NAME) >>
870 may be used instead of C<< \k<NAME> >> in Perl 5.10 or later.
873 X<(?{})> X<regex, code in> X<regexp, code in> X<regular expression, code in>
875 B<WARNING>: This extended regular expression feature is considered
876 experimental, and may be changed without notice. Code executed that
877 has side effects may not perform identically from version to version
878 due to the effect of future optimisations in the regex engine.
880 This zero-width assertion evaluates any embedded Perl code. It
881 always succeeds, and its C<code> is not interpolated. Currently,
882 the rules to determine where the C<code> ends are somewhat convoluted.
884 This feature can be used together with the special variable C<$^N> to
885 capture the results of submatches in variables without having to keep
886 track of the number of nested parentheses. For example:
888 $_ = "The brown fox jumps over the lazy dog";
889 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
890 print "color = $color, animal = $animal\n";
892 Inside the C<(?{...})> block, C<$_> refers to the string the regular
893 expression is matching against. You can also use C<pos()> to know what is
894 the current position of matching within this string.
896 The C<code> is properly scoped in the following sense: If the assertion
897 is backtracked (compare L<"Backtracking">), all changes introduced after
898 C<local>ization are undone, so that
902 (?{ $cnt = 0 }) # Initialize $cnt.
906 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
910 (?{ $res = $cnt }) # On success copy to non-localized
914 will set C<$res = 4>. Note that after the match, C<$cnt> returns to the globally
915 introduced value, because the scopes that restrict C<local> operators
918 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
919 switch. If I<not> used in this way, the result of evaluation of
920 C<code> is put into the special variable C<$^R>. This happens
921 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
922 inside the same regular expression.
924 The assignment to C<$^R> above is properly localized, so the old
925 value of C<$^R> is restored if the assertion is backtracked; compare
928 Due to an unfortunate implementation issue, the Perl code contained in these
929 blocks is treated as a compile time closure that can have seemingly bizarre
930 consequences when used with lexically scoped variables inside of subroutines
931 or loops. There are various workarounds for this, including simply using
932 global variables instead. If you are using this construct and strange results
933 occur then check for the use of lexically scoped variables.
935 For reasons of security, this construct is forbidden if the regular
936 expression involves run-time interpolation of variables, unless the
937 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
938 variables contain results of C<qr//> operator (see
939 L<perlop/"qr/STRING/imosx">).
941 This restriction is due to the wide-spread and remarkably convenient
942 custom of using run-time determined strings as patterns. For example:
948 Before Perl knew how to execute interpolated code within a pattern,
949 this operation was completely safe from a security point of view,
950 although it could raise an exception from an illegal pattern. If
951 you turn on the C<use re 'eval'>, though, it is no longer secure,
952 so you should only do so if you are also using taint checking.
953 Better yet, use the carefully constrained evaluation within a Safe
954 compartment. See L<perlsec> for details about both these mechanisms.
956 Because Perl's regex engine is currently not re-entrant, interpolated
957 code may not invoke the regex engine either directly with C<m//> or C<s///>),
958 or indirectly with functions such as C<split>.
960 =item C<(??{ code })>
962 X<regex, postponed> X<regexp, postponed> X<regular expression, postponed>
964 B<WARNING>: This extended regular expression feature is considered
965 experimental, and may be changed without notice. Code executed that
966 has side effects may not perform identically from version to version
967 due to the effect of future optimisations in the regex engine.
969 This is a "postponed" regular subexpression. The C<code> is evaluated
970 at run time, at the moment this subexpression may match. The result
971 of evaluation is considered as a regular expression and matched as
972 if it were inserted instead of this construct. Note that this means
973 that the contents of capture buffers defined inside an eval'ed pattern
974 are not available outside of the pattern, and vice versa, there is no
975 way for the inner pattern to refer to a capture buffer defined outside.
978 ('a' x 100)=~/(??{'(.)' x 100})/
980 B<will> match, it will B<not> set $1.
982 The C<code> is not interpolated. As before, the rules to determine
983 where the C<code> ends are currently somewhat convoluted.
985 The following pattern matches a parenthesized group:
990 (?> [^()]+ ) # Non-parens without backtracking
992 (??{ $re }) # Group with matching parens
997 See also C<(?PARNO)> for a different, more efficient way to accomplish
1000 Because perl's regex engine is not currently re-entrant, delayed
1001 code may not invoke the regex engine either directly with C<m//> or C<s///>),
1002 or indirectly with functions such as C<split>.
1004 Recursing deeper than 50 times without consuming any input string will
1005 result in a fatal error. The maximum depth is compiled into perl, so
1006 changing it requires a custom build.
1008 =item C<(?PARNO)> C<(?-PARNO)> C<(?+PARNO)> C<(?R)> C<(?0)>
1009 X<(?PARNO)> X<(?1)> X<(?R)> X<(?0)> X<(?-1)> X<(?+1)> X<(?-PARNO)> X<(?+PARNO)>
1010 X<regex, recursive> X<regexp, recursive> X<regular expression, recursive>
1011 X<regex, relative recursion>
1013 Similar to C<(??{ code })> except it does not involve compiling any code,
1014 instead it treats the contents of a capture buffer as an independent
1015 pattern that must match at the current position. Capture buffers
1016 contained by the pattern will have the value as determined by the
1017 outermost recursion.
1019 PARNO is a sequence of digits (not starting with 0) whose value reflects
1020 the paren-number of the capture buffer to recurse to. C<(?R)> recurses to
1021 the beginning of the whole pattern. C<(?0)> is an alternate syntax for
1022 C<(?R)>. If PARNO is preceded by a plus or minus sign then it is assumed
1023 to be relative, with negative numbers indicating preceding capture buffers
1024 and positive ones following. Thus C<(?-1)> refers to the most recently
1025 declared buffer, and C<(?+1)> indicates the next buffer to be declared.
1026 Note that the counting for relative recursion differs from that of
1027 relative backreferences, in that with recursion unclosed buffers B<are>
1030 The following pattern matches a function foo() which may contain
1031 balanced parentheses as the argument.
1033 $re = qr{ ( # paren group 1 (full function)
1035 ( # paren group 2 (parens)
1037 ( # paren group 3 (contents of parens)
1039 (?> [^()]+ ) # Non-parens without backtracking
1041 (?2) # Recurse to start of paren group 2
1049 If the pattern was used as follows
1051 'foo(bar(baz)+baz(bop))'=~/$re/
1052 and print "\$1 = $1\n",
1056 the output produced should be the following:
1058 $1 = foo(bar(baz)+baz(bop))
1059 $2 = (bar(baz)+baz(bop))
1060 $3 = bar(baz)+baz(bop)
1062 If there is no corresponding capture buffer defined, then it is a
1063 fatal error. Recursing deeper than 50 times without consuming any input
1064 string will also result in a fatal error. The maximum depth is compiled
1065 into perl, so changing it requires a custom build.
1067 The following shows how using negative indexing can make it
1068 easier to embed recursive patterns inside of a C<qr//> construct
1071 my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1072 if (/foo $parens \s+ + \s+ bar $parens/x) {
1073 # do something here...
1076 B<Note> that this pattern does not behave the same way as the equivalent
1077 PCRE or Python construct of the same form. In Perl you can backtrack into
1078 a recursed group, in PCRE and Python the recursed into group is treated
1079 as atomic. Also, modifiers are resolved at compile time, so constructs
1080 like (?i:(?1)) or (?:(?i)(?1)) do not affect how the sub-pattern will
1086 Recurse to a named subpattern. Identical to C<(?PARNO)> except that the
1087 parenthesis to recurse to is determined by name. If multiple parentheses have
1088 the same name, then it recurses to the leftmost.
1090 It is an error to refer to a name that is not declared somewhere in the
1093 B<NOTE:> In order to make things easier for programmers with experience
1094 with the Python or PCRE regex engines the pattern C<< (?P>NAME) >>
1095 may be used instead of C<< (?&NAME) >> in Perl 5.10 or later.
1097 =item C<(?(condition)yes-pattern|no-pattern)>
1100 =item C<(?(condition)yes-pattern)>
1102 Conditional expression. C<(condition)> should be either an integer in
1103 parentheses (which is valid if the corresponding pair of parentheses
1104 matched), a look-ahead/look-behind/evaluate zero-width assertion, a
1105 name in angle brackets or single quotes (which is valid if a buffer
1106 with the given name matched), or the special symbol (R) (true when
1107 evaluated inside of recursion or eval). Additionally the R may be
1108 followed by a number, (which will be true when evaluated when recursing
1109 inside of the appropriate group), or by C<&NAME>, in which case it will
1110 be true only when evaluated during recursion in the named group.
1112 Here's a summary of the possible predicates:
1118 Checks if the numbered capturing buffer has matched something.
1120 =item (<NAME>) ('NAME')
1122 Checks if a buffer with the given name has matched something.
1126 Treats the code block as the condition.
1130 Checks if the expression has been evaluated inside of recursion.
1134 Checks if the expression has been evaluated while executing directly
1135 inside of the n-th capture group. This check is the regex equivalent of
1137 if ((caller(0))[3] eq 'subname') { ... }
1139 In other words, it does not check the full recursion stack.
1143 Similar to C<(R1)>, this predicate checks to see if we're executing
1144 directly inside of the leftmost group with a given name (this is the same
1145 logic used by C<(?&NAME)> to disambiguate). It does not check the full
1146 stack, but only the name of the innermost active recursion.
1150 In this case, the yes-pattern is never directly executed, and no
1151 no-pattern is allowed. Similar in spirit to C<(?{0})> but more efficient.
1152 See below for details.
1163 matches a chunk of non-parentheses, possibly included in parentheses
1166 A special form is the C<(DEFINE)> predicate, which never executes directly
1167 its yes-pattern, and does not allow a no-pattern. This allows to define
1168 subpatterns which will be executed only by using the recursion mechanism.
1169 This way, you can define a set of regular expression rules that can be
1170 bundled into any pattern you choose.
1172 It is recommended that for this usage you put the DEFINE block at the
1173 end of the pattern, and that you name any subpatterns defined within it.
1175 Also, it's worth noting that patterns defined this way probably will
1176 not be as efficient, as the optimiser is not very clever about
1179 An example of how this might be used is as follows:
1181 /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
1187 Note that capture buffers matched inside of recursion are not accessible
1188 after the recursion returns, so the extra layer of capturing buffers is
1189 necessary. Thus C<$+{NAME_PAT}> would not be defined even though
1190 C<$+{NAME}> would be.
1192 =item C<< (?>pattern) >>
1193 X<backtrack> X<backtracking> X<atomic> X<possessive>
1195 An "independent" subexpression, one which matches the substring
1196 that a I<standalone> C<pattern> would match if anchored at the given
1197 position, and it matches I<nothing other than this substring>. This
1198 construct is useful for optimizations of what would otherwise be
1199 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
1200 It may also be useful in places where the "grab all you can, and do not
1201 give anything back" semantic is desirable.
1203 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
1204 (anchored at the beginning of string, as above) will match I<all>
1205 characters C<a> at the beginning of string, leaving no C<a> for
1206 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
1207 since the match of the subgroup C<a*> is influenced by the following
1208 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
1209 C<a*ab> will match fewer characters than a standalone C<a*>, since
1210 this makes the tail match.
1212 An effect similar to C<< (?>pattern) >> may be achieved by writing
1213 C<(?=(pattern))\1>. This matches the same substring as a standalone
1214 C<a+>, and the following C<\1> eats the matched string; it therefore
1215 makes a zero-length assertion into an analogue of C<< (?>...) >>.
1216 (The difference between these two constructs is that the second one
1217 uses a capturing group, thus shifting ordinals of backreferences
1218 in the rest of a regular expression.)
1220 Consider this pattern:
1231 That will efficiently match a nonempty group with matching parentheses
1232 two levels deep or less. However, if there is no such group, it
1233 will take virtually forever on a long string. That's because there
1234 are so many different ways to split a long string into several
1235 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
1236 to a subpattern of the above pattern. Consider how the pattern
1237 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
1238 seconds, but that each extra letter doubles this time. This
1239 exponential performance will make it appear that your program has
1240 hung. However, a tiny change to this pattern
1244 (?> [^()]+ ) # change x+ above to (?> x+ )
1251 which uses C<< (?>...) >> matches exactly when the one above does (verifying
1252 this yourself would be a productive exercise), but finishes in a fourth
1253 the time when used on a similar string with 1000000 C<a>s. Be aware,
1254 however, that this pattern currently triggers a warning message under
1255 the C<use warnings> pragma or B<-w> switch saying it
1256 C<"matches null string many times in regex">.
1258 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
1259 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
1260 This was only 4 times slower on a string with 1000000 C<a>s.
1262 The "grab all you can, and do not give anything back" semantic is desirable
1263 in many situations where on the first sight a simple C<()*> looks like
1264 the correct solution. Suppose we parse text with comments being delimited
1265 by C<#> followed by some optional (horizontal) whitespace. Contrary to
1266 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
1267 the comment delimiter, because it may "give up" some whitespace if
1268 the remainder of the pattern can be made to match that way. The correct
1269 answer is either one of these:
1274 For example, to grab non-empty comments into $1, one should use either
1277 / (?> \# [ \t]* ) ( .+ ) /x;
1278 / \# [ \t]* ( [^ \t] .* ) /x;
1280 Which one you pick depends on which of these expressions better reflects
1281 the above specification of comments.
1283 In some literature this construct is called "atomic matching" or
1284 "possessive matching".
1286 Possessive quantifiers are equivalent to putting the item they are applied
1287 to inside of one of these constructs. The following equivalences apply:
1289 Quantifier Form Bracketing Form
1290 --------------- ---------------
1294 PAT{min,max}+ (?>PAT{min,max})
1298 =head2 Special Backtracking Control Verbs
1300 B<WARNING:> These patterns are experimental and subject to change or
1301 removal in a future version of Perl. Their usage in production code should
1302 be noted to avoid problems during upgrades.
1304 These special patterns are generally of the form C<(*VERB:ARG)>. Unless
1305 otherwise stated the ARG argument is optional; in some cases, it is
1308 Any pattern containing a special backtracking verb that allows an argument
1309 has the special behaviour that when executed it sets the current packages'
1310 C<$REGERROR> and C<$REGMARK> variables. When doing so the following
1313 On failure, the C<$REGERROR> variable will be set to the ARG value of the
1314 verb pattern, if the verb was involved in the failure of the match. If the
1315 ARG part of the pattern was omitted, then C<$REGERROR> will be set to the
1316 name of the last C<(*MARK:NAME)> pattern executed, or to TRUE if there was
1317 none. Also, the C<$REGMARK> variable will be set to FALSE.
1319 On a successful match, the C<$REGERROR> variable will be set to FALSE, and
1320 the C<$REGMARK> variable will be set to the name of the last
1321 C<(*MARK:NAME)> pattern executed. See the explanation for the
1322 C<(*MARK:NAME)> verb below for more details.
1324 B<NOTE:> C<$REGERROR> and C<$REGMARK> are not magic variables like C<$1>
1325 and most other regex related variables. They are not local to a scope, nor
1326 readonly, but instead are volatile package variables similar to C<$AUTOLOAD>.
1327 Use C<local> to localize changes to them to a specific scope if necessary.
1329 If a pattern does not contain a special backtracking verb that allows an
1330 argument, then C<$REGERROR> and C<$REGMARK> are not touched at all.
1334 =item Verbs that take an argument
1338 =item C<(*PRUNE)> C<(*PRUNE:NAME)>
1339 X<(*PRUNE)> X<(*PRUNE:NAME)> X<\v>
1341 This zero-width pattern prunes the backtracking tree at the current point
1342 when backtracked into on failure. Consider the pattern C<A (*PRUNE) B>,
1343 where A and B are complex patterns. Until the C<(*PRUNE)> verb is reached,
1344 A may backtrack as necessary to match. Once it is reached, matching
1345 continues in B, which may also backtrack as necessary; however, should B
1346 not match, then no further backtracking will take place, and the pattern
1347 will fail outright at the current starting position.
1349 As a shortcut, C<\v> is exactly equivalent to C<(*PRUNE)>.
1351 The following example counts all the possible matching strings in a
1352 pattern (without actually matching any of them).
1354 'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
1355 print "Count=$count\n";
1370 If we add a C<(*PRUNE)> before the count like the following
1372 'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
1373 print "Count=$count\n";
1375 we prevent backtracking and find the count of the longest matching
1376 at each matching startpoint like so:
1383 Any number of C<(*PRUNE)> assertions may be used in a pattern.
1385 See also C<< (?>pattern) >> and possessive quantifiers for other ways to
1386 control backtracking. In some cases, the use of C<(*PRUNE)> can be
1387 replaced with a C<< (?>pattern) >> with no functional difference; however,
1388 C<(*PRUNE)> can be used to handle cases that cannot be expressed using a
1389 C<< (?>pattern) >> alone.
1392 =item C<(*SKIP)> C<(*SKIP:NAME)>
1395 This zero-width pattern is similar to C<(*PRUNE)>, except that on
1396 failure it also signifies that whatever text that was matched leading up
1397 to the C<(*SKIP)> pattern being executed cannot be part of I<any> match
1398 of this pattern. This effectively means that the regex engine "skips" forward
1399 to this position on failure and tries to match again, (assuming that
1400 there is sufficient room to match).
1402 As a shortcut C<\V> is exactly equivalent to C<(*SKIP)>.
1404 The name of the C<(*SKIP:NAME)> pattern has special significance. If a
1405 C<(*MARK:NAME)> was encountered while matching, then it is that position
1406 which is used as the "skip point". If no C<(*MARK)> of that name was
1407 encountered, then the C<(*SKIP)> operator has no effect. When used
1408 without a name the "skip point" is where the match point was when
1409 executing the (*SKIP) pattern.
1411 Compare the following to the examples in C<(*PRUNE)>, note the string
1414 'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
1415 print "Count=$count\n";
1423 Once the 'aaab' at the start of the string has matched, and the C<(*SKIP)>
1424 executed, the next startpoint will be where the cursor was when the
1425 C<(*SKIP)> was executed.
1427 =item C<(*MARK:NAME)> C<(*:NAME)>
1428 X<(*MARK)> C<(*MARK:NAME)> C<(*:NAME)>
1430 This zero-width pattern can be used to mark the point reached in a string
1431 when a certain part of the pattern has been successfully matched. This
1432 mark may be given a name. A later C<(*SKIP)> pattern will then skip
1433 forward to that point if backtracked into on failure. Any number of
1434 C<(*MARK)> patterns are allowed, and the NAME portion is optional and may
1437 In addition to interacting with the C<(*SKIP)> pattern, C<(*MARK:NAME)>
1438 can be used to "label" a pattern branch, so that after matching, the
1439 program can determine which branches of the pattern were involved in the
1442 When a match is successful, the C<$REGMARK> variable will be set to the
1443 name of the most recently executed C<(*MARK:NAME)> that was involved
1446 This can be used to determine which branch of a pattern was matched
1447 without using a seperate capture buffer for each branch, which in turn
1448 can result in a performance improvement, as perl cannot optimize
1449 C</(?:(x)|(y)|(z))/> as efficiently as something like
1450 C</(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/>.
1452 When a match has failed, and unless another verb has been involved in
1453 failing the match and has provided its own name to use, the C<$REGERROR>
1454 variable will be set to the name of the most recently executed
1457 See C<(*SKIP)> for more details.
1459 As a shortcut C<(*MARK:NAME)> can be written C<(*:NAME)>.
1461 =item C<(*THEN)> C<(*THEN:NAME)>
1463 This is similar to the "cut group" operator C<::> from Perl6. Like
1464 C<(*PRUNE)>, this verb always matches, and when backtracked into on
1465 failure, it causes the regex engine to try the next alternation in the
1466 innermost enclosing group (capturing or otherwise).
1468 Its name comes from the observation that this operation combined with the
1469 alternation operator (C<|>) can be used to create what is essentially a
1470 pattern-based if/then/else block:
1472 ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
1474 Note that if this operator is used and NOT inside of an alternation then
1475 it acts exactly like the C<(*PRUNE)> operator.
1485 / ( A (*THEN) B | C (*THEN) D ) /
1489 / ( A (*PRUNE) B | C (*PRUNE) D ) /
1491 as after matching the A but failing on the B the C<(*THEN)> verb will
1492 backtrack and try C; but the C<(*PRUNE)> verb will simply fail.
1497 This is the Perl6 "commit pattern" C<< <commit> >> or C<:::>. It's a
1498 zero-width pattern similar to C<(*SKIP)>, except that when backtracked
1499 into on failure it causes the match to fail outright. No further attempts
1500 to find a valid match by advancing the start pointer will occur again.
1503 'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
1504 print "Count=$count\n";
1511 In other words, once the C<(*COMMIT)> has been entered, and if the pattern
1512 does not match, the regex engine will not try any further matching on the
1517 =item Verbs without an argument
1521 =item C<(*FAIL)> C<(*F)>
1524 This pattern matches nothing and always fails. It can be used to force the
1525 engine to backtrack. It is equivalent to C<(?!)>, but easier to read. In
1526 fact, C<(?!)> gets optimised into C<(*FAIL)> internally.
1528 It is probably useful only when combined with C<(?{})> or C<(??{})>.
1533 B<WARNING:> This feature is highly experimental. It is not recommended
1534 for production code.
1536 This pattern matches nothing and causes the end of successful matching at
1537 the point at which the C<(*ACCEPT)> pattern was encountered, regardless of
1538 whether there is actually more to match in the string. When inside of a
1539 nested pattern, such as recursion, or in a subpattern dynamically generated
1540 via C<(??{})>, only the innermost pattern is ended immediately.
1542 If the C<(*ACCEPT)> is inside of capturing buffers then the buffers are
1543 marked as ended at the point at which the C<(*ACCEPT)> was encountered.
1546 'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
1548 will match, and C<$1> will be C<AB> and C<$2> will be C<B>, C<$3> will not
1549 be set. If another branch in the inner parentheses were matched, such as in the
1550 string 'ACDE', then the C<D> and C<E> would have to be matched as well.
1557 X<backtrack> X<backtracking>
1559 NOTE: This section presents an abstract approximation of regular
1560 expression behavior. For a more rigorous (and complicated) view of
1561 the rules involved in selecting a match among possible alternatives,
1562 see L<Combining RE Pieces>.
1564 A fundamental feature of regular expression matching involves the
1565 notion called I<backtracking>, which is currently used (when needed)
1566 by all regular non-possessive expression quantifiers, namely C<*>, C<*?>, C<+>,
1567 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
1568 internally, but the general principle outlined here is valid.
1570 For a regular expression to match, the I<entire> regular expression must
1571 match, not just part of it. So if the beginning of a pattern containing a
1572 quantifier succeeds in a way that causes later parts in the pattern to
1573 fail, the matching engine backs up and recalculates the beginning
1574 part--that's why it's called backtracking.
1576 Here is an example of backtracking: Let's say you want to find the
1577 word following "foo" in the string "Food is on the foo table.":
1579 $_ = "Food is on the foo table.";
1580 if ( /\b(foo)\s+(\w+)/i ) {
1581 print "$2 follows $1.\n";
1584 When the match runs, the first part of the regular expression (C<\b(foo)>)
1585 finds a possible match right at the beginning of the string, and loads up
1586 $1 with "Foo". However, as soon as the matching engine sees that there's
1587 no whitespace following the "Foo" that it had saved in $1, it realizes its
1588 mistake and starts over again one character after where it had the
1589 tentative match. This time it goes all the way until the next occurrence
1590 of "foo". The complete regular expression matches this time, and you get
1591 the expected output of "table follows foo."
1593 Sometimes minimal matching can help a lot. Imagine you'd like to match
1594 everything between "foo" and "bar". Initially, you write something
1597 $_ = "The food is under the bar in the barn.";
1598 if ( /foo(.*)bar/ ) {
1602 Which perhaps unexpectedly yields:
1604 got <d is under the bar in the >
1606 That's because C<.*> was greedy, so you get everything between the
1607 I<first> "foo" and the I<last> "bar". Here it's more effective
1608 to use minimal matching to make sure you get the text between a "foo"
1609 and the first "bar" thereafter.
1611 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
1612 got <d is under the >
1614 Here's another example. Let's say you'd like to match a number at the end
1615 of a string, and you also want to keep the preceding part of the match.
1618 $_ = "I have 2 numbers: 53147";
1619 if ( /(.*)(\d*)/ ) { # Wrong!
1620 print "Beginning is <$1>, number is <$2>.\n";
1623 That won't work at all, because C<.*> was greedy and gobbled up the
1624 whole string. As C<\d*> can match on an empty string the complete
1625 regular expression matched successfully.
1627 Beginning is <I have 2 numbers: 53147>, number is <>.
1629 Here are some variants, most of which don't work:
1631 $_ = "I have 2 numbers: 53147";
1644 printf "%-12s ", $pat;
1646 print "<$1> <$2>\n";
1652 That will print out:
1654 (.*)(\d*) <I have 2 numbers: 53147> <>
1655 (.*)(\d+) <I have 2 numbers: 5314> <7>
1657 (.*?)(\d+) <I have > <2>
1658 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
1659 (.*?)(\d+)$ <I have 2 numbers: > <53147>
1660 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
1661 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
1663 As you see, this can be a bit tricky. It's important to realize that a
1664 regular expression is merely a set of assertions that gives a definition
1665 of success. There may be 0, 1, or several different ways that the
1666 definition might succeed against a particular string. And if there are
1667 multiple ways it might succeed, you need to understand backtracking to
1668 know which variety of success you will achieve.
1670 When using look-ahead assertions and negations, this can all get even
1671 trickier. Imagine you'd like to find a sequence of non-digits not
1672 followed by "123". You might try to write that as
1675 if ( /^\D*(?!123)/ ) { # Wrong!
1676 print "Yup, no 123 in $_\n";
1679 But that isn't going to match; at least, not the way you're hoping. It
1680 claims that there is no 123 in the string. Here's a clearer picture of
1681 why that pattern matches, contrary to popular expectations:
1686 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
1687 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
1689 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
1690 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
1698 You might have expected test 3 to fail because it seems to a more
1699 general purpose version of test 1. The important difference between
1700 them is that test 3 contains a quantifier (C<\D*>) and so can use
1701 backtracking, whereas test 1 will not. What's happening is
1702 that you've asked "Is it true that at the start of $x, following 0 or more
1703 non-digits, you have something that's not 123?" If the pattern matcher had
1704 let C<\D*> expand to "ABC", this would have caused the whole pattern to
1707 The search engine will initially match C<\D*> with "ABC". Then it will
1708 try to match C<(?!123> with "123", which fails. But because
1709 a quantifier (C<\D*>) has been used in the regular expression, the
1710 search engine can backtrack and retry the match differently
1711 in the hope of matching the complete regular expression.
1713 The pattern really, I<really> wants to succeed, so it uses the
1714 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
1715 time. Now there's indeed something following "AB" that is not
1716 "123". It's "C123", which suffices.
1718 We can deal with this by using both an assertion and a negation.
1719 We'll say that the first part in $1 must be followed both by a digit
1720 and by something that's not "123". Remember that the look-aheads
1721 are zero-width expressions--they only look, but don't consume any
1722 of the string in their match. So rewriting this way produces what
1723 you'd expect; that is, case 5 will fail, but case 6 succeeds:
1725 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
1726 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
1730 In other words, the two zero-width assertions next to each other work as though
1731 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
1732 matches only if you're at the beginning of the line AND the end of the
1733 line simultaneously. The deeper underlying truth is that juxtaposition in
1734 regular expressions always means AND, except when you write an explicit OR
1735 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
1736 although the attempted matches are made at different positions because "a"
1737 is not a zero-width assertion, but a one-width assertion.
1739 B<WARNING>: Particularly complicated regular expressions can take
1740 exponential time to solve because of the immense number of possible
1741 ways they can use backtracking to try for a match. For example, without
1742 internal optimizations done by the regular expression engine, this will
1743 take a painfully long time to run:
1745 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
1747 And if you used C<*>'s in the internal groups instead of limiting them
1748 to 0 through 5 matches, then it would take forever--or until you ran
1749 out of stack space. Moreover, these internal optimizations are not
1750 always applicable. For example, if you put C<{0,5}> instead of C<*>
1751 on the external group, no current optimization is applicable, and the
1752 match takes a long time to finish.
1754 A powerful tool for optimizing such beasts is what is known as an
1755 "independent group",
1756 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
1757 zero-length look-ahead/look-behind assertions will not backtrack to make
1758 the tail match, since they are in "logical" context: only
1759 whether they match is considered relevant. For an example
1760 where side-effects of look-ahead I<might> have influenced the
1761 following match, see L<C<< (?>pattern) >>>.
1763 =head2 Version 8 Regular Expressions
1764 X<regular expression, version 8> X<regex, version 8> X<regexp, version 8>
1766 In case you're not familiar with the "regular" Version 8 regex
1767 routines, here are the pattern-matching rules not described above.
1769 Any single character matches itself, unless it is a I<metacharacter>
1770 with a special meaning described here or above. You can cause
1771 characters that normally function as metacharacters to be interpreted
1772 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
1773 character; "\\" matches a "\"). This escape mechanism is also required
1774 for the character used as the pattern delimiter.
1776 A series of characters matches that series of characters in the target
1777 string, so the pattern C<blurfl> would match "blurfl" in the target
1780 You can specify a character class, by enclosing a list of characters
1781 in C<[]>, which will match any character from the list. If the
1782 first character after the "[" is "^", the class matches any character not
1783 in the list. Within a list, the "-" character specifies a
1784 range, so that C<a-z> represents all characters between "a" and "z",
1785 inclusive. If you want either "-" or "]" itself to be a member of a
1786 class, put it at the start of the list (possibly after a "^"), or
1787 escape it with a backslash. "-" is also taken literally when it is
1788 at the end of the list, just before the closing "]". (The
1789 following all specify the same class of three characters: C<[-az]>,
1790 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1791 specifies a class containing twenty-six characters, even on EBCDIC-based
1792 character sets.) Also, if you try to use the character
1793 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1794 a range, the "-" is understood literally.
1796 Note also that the whole range idea is rather unportable between
1797 character sets--and even within character sets they may cause results
1798 you probably didn't expect. A sound principle is to use only ranges
1799 that begin from and end at either alphabetics of equal case ([a-e],
1800 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1801 spell out the character sets in full.
1803 Characters may be specified using a metacharacter syntax much like that
1804 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1805 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1806 of octal digits, matches the character whose coded character set value
1807 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1808 matches the character whose numeric value is I<nn>. The expression \cI<x>
1809 matches the character control-I<x>. Finally, the "." metacharacter
1810 matches any character except "\n" (unless you use C</s>).
1812 You can specify a series of alternatives for a pattern using "|" to
1813 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1814 or "foe" in the target string (as would C<f(e|i|o)e>). The
1815 first alternative includes everything from the last pattern delimiter
1816 ("(", "[", or the beginning of the pattern) up to the first "|", and
1817 the last alternative contains everything from the last "|" to the next
1818 pattern delimiter. That's why it's common practice to include
1819 alternatives in parentheses: to minimize confusion about where they
1822 Alternatives are tried from left to right, so the first
1823 alternative found for which the entire expression matches, is the one that
1824 is chosen. This means that alternatives are not necessarily greedy. For
1825 example: when matching C<foo|foot> against "barefoot", only the "foo"
1826 part will match, as that is the first alternative tried, and it successfully
1827 matches the target string. (This might not seem important, but it is
1828 important when you are capturing matched text using parentheses.)
1830 Also remember that "|" is interpreted as a literal within square brackets,
1831 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1833 Within a pattern, you may designate subpatterns for later reference
1834 by enclosing them in parentheses, and you may refer back to the
1835 I<n>th subpattern later in the pattern using the metacharacter
1836 \I<n>. Subpatterns are numbered based on the left to right order
1837 of their opening parenthesis. A backreference matches whatever
1838 actually matched the subpattern in the string being examined, not
1839 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1840 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1841 1 matched "0x", even though the rule C<0|0x> could potentially match
1842 the leading 0 in the second number.
1844 =head2 Warning on \1 Instead of $1
1846 Some people get too used to writing things like:
1848 $pattern =~ s/(\W)/\\\1/g;
1850 This is grandfathered for the RHS of a substitute to avoid shocking the
1851 B<sed> addicts, but it's a dirty habit to get into. That's because in
1852 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1853 the usual double-quoted string means a control-A. The customary Unix
1854 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1855 of doing that, you get yourself into trouble if you then add an C</e>
1858 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1864 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1865 C<${1}000>. The operation of interpolation should not be confused
1866 with the operation of matching a backreference. Certainly they mean two
1867 different things on the I<left> side of the C<s///>.
1869 =head2 Repeated Patterns Matching a Zero-length Substring
1871 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1873 Regular expressions provide a terse and powerful programming language. As
1874 with most other power tools, power comes together with the ability
1877 A common abuse of this power stems from the ability to make infinite
1878 loops using regular expressions, with something as innocuous as:
1880 'foo' =~ m{ ( o? )* }x;
1882 The C<o?> matches at the beginning of C<'foo'>, and since the position
1883 in the string is not moved by the match, C<o?> would match again and again
1884 because of the C<*> modifier. Another common way to create a similar cycle
1885 is with the looping modifier C<//g>:
1887 @matches = ( 'foo' =~ m{ o? }xg );
1891 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1893 or the loop implied by split().
1895 However, long experience has shown that many programming tasks may
1896 be significantly simplified by using repeated subexpressions that
1897 may match zero-length substrings. Here's a simple example being:
1899 @chars = split //, $string; # // is not magic in split
1900 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1902 Thus Perl allows such constructs, by I<forcefully breaking
1903 the infinite loop>. The rules for this are different for lower-level
1904 loops given by the greedy modifiers C<*+{}>, and for higher-level
1905 ones like the C</g> modifier or split() operator.
1907 The lower-level loops are I<interrupted> (that is, the loop is
1908 broken) when Perl detects that a repeated expression matched a
1909 zero-length substring. Thus
1911 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1913 is made equivalent to
1915 m{ (?: NON_ZERO_LENGTH )*
1920 The higher level-loops preserve an additional state between iterations:
1921 whether the last match was zero-length. To break the loop, the following
1922 match after a zero-length match is prohibited to have a length of zero.
1923 This prohibition interacts with backtracking (see L<"Backtracking">),
1924 and so the I<second best> match is chosen if the I<best> match is of
1932 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1933 match given by non-greedy C<??> is the zero-length match, and the I<second
1934 best> match is what is matched by C<\w>. Thus zero-length matches
1935 alternate with one-character-long matches.
1937 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1938 position one notch further in the string.
1940 The additional state of being I<matched with zero-length> is associated with
1941 the matched string, and is reset by each assignment to pos().
1942 Zero-length matches at the end of the previous match are ignored
1945 =head2 Combining RE Pieces
1947 Each of the elementary pieces of regular expressions which were described
1948 before (such as C<ab> or C<\Z>) could match at most one substring
1949 at the given position of the input string. However, in a typical regular
1950 expression these elementary pieces are combined into more complicated
1951 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1952 (in these examples C<S> and C<T> are regular subexpressions).
1954 Such combinations can include alternatives, leading to a problem of choice:
1955 if we match a regular expression C<a|ab> against C<"abc">, will it match
1956 substring C<"a"> or C<"ab">? One way to describe which substring is
1957 actually matched is the concept of backtracking (see L<"Backtracking">).
1958 However, this description is too low-level and makes you think
1959 in terms of a particular implementation.
1961 Another description starts with notions of "better"/"worse". All the
1962 substrings which may be matched by the given regular expression can be
1963 sorted from the "best" match to the "worst" match, and it is the "best"
1964 match which is chosen. This substitutes the question of "what is chosen?"
1965 by the question of "which matches are better, and which are worse?".
1967 Again, for elementary pieces there is no such question, since at most
1968 one match at a given position is possible. This section describes the
1969 notion of better/worse for combining operators. In the description
1970 below C<S> and C<T> are regular subexpressions.
1976 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1977 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1978 which can be matched by C<T>.
1980 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
1983 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
1984 C<B> is better match for C<T> than C<B'>.
1988 When C<S> can match, it is a better match than when only C<T> can match.
1990 Ordering of two matches for C<S> is the same as for C<S>. Similar for
1991 two matches for C<T>.
1993 =item C<S{REPEAT_COUNT}>
1995 Matches as C<SSS...S> (repeated as many times as necessary).
1999 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
2001 =item C<S{min,max}?>
2003 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
2005 =item C<S?>, C<S*>, C<S+>
2007 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
2009 =item C<S??>, C<S*?>, C<S+?>
2011 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
2015 Matches the best match for C<S> and only that.
2017 =item C<(?=S)>, C<(?<=S)>
2019 Only the best match for C<S> is considered. (This is important only if
2020 C<S> has capturing parentheses, and backreferences are used somewhere
2021 else in the whole regular expression.)
2023 =item C<(?!S)>, C<(?<!S)>
2025 For this grouping operator there is no need to describe the ordering, since
2026 only whether or not C<S> can match is important.
2028 =item C<(??{ EXPR })>, C<(?PARNO)>
2030 The ordering is the same as for the regular expression which is
2031 the result of EXPR, or the pattern contained by capture buffer PARNO.
2033 =item C<(?(condition)yes-pattern|no-pattern)>
2035 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
2036 already determined. The ordering of the matches is the same as for the
2037 chosen subexpression.
2041 The above recipes describe the ordering of matches I<at a given position>.
2042 One more rule is needed to understand how a match is determined for the
2043 whole regular expression: a match at an earlier position is always better
2044 than a match at a later position.
2046 =head2 Creating Custom RE Engines
2048 Overloaded constants (see L<overload>) provide a simple way to extend
2049 the functionality of the RE engine.
2051 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
2052 matches at a boundary between whitespace characters and non-whitespace
2053 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
2054 at these positions, so we want to have each C<\Y|> in the place of the
2055 more complicated version. We can create a module C<customre> to do
2063 die "No argument to customre::import allowed" if @_;
2064 overload::constant 'qr' => \&convert;
2067 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2069 # We must also take care of not escaping the legitimate \\Y|
2070 # sequence, hence the presence of '\\' in the conversion rules.
2071 my %rules = ( '\\' => '\\\\',
2072 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2078 { $rules{$1} or invalid($re,$1) }sgex;
2082 Now C<use customre> enables the new escape in constant regular
2083 expressions, i.e., those without any runtime variable interpolations.
2084 As documented in L<overload>, this conversion will work only over
2085 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
2086 part of this regular expression needs to be converted explicitly
2087 (but only if the special meaning of C<\Y|> should be enabled inside $re):
2092 $re = customre::convert $re;
2095 =head1 PCRE/Python Support
2097 As of Perl 5.10 Perl supports several Python/PCRE specific extensions
2098 to the regex syntax. While Perl programmers are encouraged to use the
2099 Perl specific syntax, the following are legal in Perl 5.10:
2103 =item C<< (?PE<lt>NAMEE<gt>pattern) >>
2105 Define a named capture buffer. Equivalent to C<< (?<NAME>pattern) >>.
2107 =item C<< (?P=NAME) >>
2109 Backreference to a named capture buffer. Equivalent to C<< \g{NAME} >>.
2111 =item C<< (?P>NAME) >>
2113 Subroutine call to a named capture buffer. Equivalent to C<< (?&NAME) >>.
2119 This document varies from difficult to understand to completely
2120 and utterly opaque. The wandering prose riddled with jargon is
2121 hard to fathom in several places.
2123 This document needs a rewrite that separates the tutorial content
2124 from the reference content.
2132 L<perlop/"Regexp Quote-Like Operators">.
2134 L<perlop/"Gory details of parsing quoted constructs">.
2144 I<Mastering Regular Expressions> by Jeffrey Friedl, published
2145 by O'Reilly and Associates.