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.
71 Global matching, and keep the Current position after failed matching.
72 Unlike i, m, s and x, these two flags affect the way the regex is used
73 rather than the regex itself. See
74 L<perlretut/"Using regular expressions in Perl"> for further explanation
75 of the g and c modifiers.
79 These are usually written as "the C</x> modifier", even though the delimiter
80 in question might not really be a slash. Any of these
81 modifiers may also be embedded within the regular expression itself using
82 the C<(?...)> construct. See below.
84 The C</x> modifier itself needs a little more explanation. It tells
85 the regular expression parser to ignore whitespace that is neither
86 backslashed nor within a character class. You can use this to break up
87 your regular expression into (slightly) more readable parts. The C<#>
88 character is also treated as a metacharacter introducing a comment,
89 just as in ordinary Perl code. This also means that if you want real
90 whitespace or C<#> characters in the pattern (outside a character
91 class, where they are unaffected by C</x>), then you'll either have to
92 escape them (using backslashes or C<\Q...\E>) or encode them using octal
93 or hex escapes. Taken together, these features go a long way towards
94 making Perl's regular expressions more readable. Note that you have to
95 be careful not to include the pattern delimiter in the comment--perl has
96 no way of knowing you did not intend to close the pattern early. See
97 the C-comment deletion code in L<perlop>. Also note that anything inside
98 a C<\Q...\E> stays unaffected by C</x>.
101 =head2 Regular Expressions
103 =head3 Metacharacters
105 The patterns used in Perl pattern matching evolved from those supplied in
106 the Version 8 regex routines. (The routines are derived
107 (distantly) from Henry Spencer's freely redistributable reimplementation
108 of the V8 routines.) See L<Version 8 Regular Expressions> for
111 In particular the following metacharacters have their standard I<egrep>-ish
114 X<\> X<^> X<.> X<$> X<|> X<(> X<()> X<[> X<[]>
117 \ Quote the next metacharacter
118 ^ Match the beginning of the line
119 . Match any character (except newline)
120 $ Match the end of the line (or before newline at the end)
125 By default, the "^" character is guaranteed to match only the
126 beginning of the string, the "$" character only the end (or before the
127 newline at the end), and Perl does certain optimizations with the
128 assumption that the string contains only one line. Embedded newlines
129 will not be matched by "^" or "$". You may, however, wish to treat a
130 string as a multi-line buffer, such that the "^" will match after any
131 newline within the string (except if the newline is the last character in
132 the string), and "$" will match before any newline. At the
133 cost of a little more overhead, you can do this by using the /m modifier
134 on the pattern match operator. (Older programs did this by setting C<$*>,
135 but this practice has been removed in perl 5.9.)
138 To simplify multi-line substitutions, the "." character never matches a
139 newline unless you use the C</s> modifier, which in effect tells Perl to pretend
140 the string is a single line--even if it isn't.
145 The following standard quantifiers are recognized:
146 X<metacharacter> X<quantifier> X<*> X<+> X<?> X<{n}> X<{n,}> X<{n,m}>
148 * Match 0 or more times
149 + Match 1 or more times
151 {n} Match exactly n times
152 {n,} Match at least n times
153 {n,m} Match at least n but not more than m times
155 (If a curly bracket occurs in any other context, it is treated
156 as a regular character. In particular, the lower bound
157 is not optional.) The "*" quantifier is equivalent to C<{0,}>, the "+"
158 quantifier to C<{1,}>, and the "?" quantifier to C<{0,1}>. n and m are limited
159 to integral values less than a preset limit defined when perl is built.
160 This is usually 32766 on the most common platforms. The actual limit can
161 be seen in the error message generated by code such as this:
163 $_ **= $_ , / {$_} / for 2 .. 42;
165 By default, a quantified subpattern is "greedy", that is, it will match as
166 many times as possible (given a particular starting location) while still
167 allowing the rest of the pattern to match. If you want it to match the
168 minimum number of times possible, follow the quantifier with a "?". Note
169 that the meanings don't change, just the "greediness":
170 X<metacharacter> X<greedy> X<greediness>
171 X<?> X<*?> X<+?> X<??> X<{n}?> X<{n,}?> X<{n,m}?>
173 *? Match 0 or more times, not greedily
174 +? Match 1 or more times, not greedily
175 ?? Match 0 or 1 time, not greedily
176 {n}? Match exactly n times, not greedily
177 {n,}? Match at least n times, not greedily
178 {n,m}? Match at least n but not more than m times, not greedily
180 By default, when a quantified subpattern does not allow the rest of the
181 overall pattern to match, Perl will backtrack. However, this behaviour is
182 sometimes undesirable. Thus Perl provides the "possessive" quantifier form
185 *+ Match 0 or more times and give nothing back
186 ++ Match 1 or more times and give nothing back
187 ?+ Match 0 or 1 time and give nothing back
188 {n}+ Match exactly n times and give nothing back (redundant)
189 {n,}+ Match at least n times and give nothing back
190 {n,m}+ Match at least n but not more than m times and give nothing back
196 will never match, as the C<a++> will gobble up all the C<a>'s in the
197 string and won't leave any for the remaining part of the pattern. This
198 feature can be extremely useful to give perl hints about where it
199 shouldn't backtrack. For instance, the typical "match a double-quoted
200 string" problem can be most efficiently performed when written as:
202 /"(?:[^"\\]++|\\.)*+"/
204 as we know that if the final quote does not match, backtracking will not
205 help. See the independent subexpression C<< (?>...) >> for more details;
206 possessive quantifiers are just syntactic sugar for that construct. For
207 instance the above example could also be written as follows:
209 /"(?>(?:(?>[^"\\]+)|\\.)*)"/
211 =head3 Escape sequences
213 Because patterns are processed as double quoted strings, the following
215 X<\t> X<\n> X<\r> X<\f> X<\e> X<\a> X<\l> X<\u> X<\L> X<\U> X<\E> X<\Q>
216 X<\0> X<\c> X<\N> X<\x>
222 \a alarm (bell) (BEL)
223 \e escape (think troff) (ESC)
224 \033 octal char (example: ESC)
225 \x1B hex char (example: ESC)
226 \x{263a} long hex char (example: Unicode SMILEY)
227 \cK control char (example: VT)
228 \N{name} named Unicode character
229 \l lowercase next char (think vi)
230 \u uppercase next char (think vi)
231 \L lowercase till \E (think vi)
232 \U uppercase till \E (think vi)
233 \E end case modification (think vi)
234 \Q quote (disable) pattern metacharacters till \E
236 If C<use locale> is in effect, the case map used by C<\l>, C<\L>, C<\u>
237 and C<\U> is taken from the current locale. See L<perllocale>. For
238 documentation of C<\N{name}>, see L<charnames>.
240 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
241 An unescaped C<$> or C<@> interpolates the corresponding variable,
242 while escaping will cause the literal string C<\$> to be matched.
243 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
245 =head3 Character Classes and other Special Escapes
247 In addition, Perl defines the following:
248 X<\w> X<\W> X<\s> X<\S> X<\d> X<\D> X<\X> X<\p> X<\P> X<\C>
249 X<\g> X<\k> X<\N> X<\K> X<\v> X<\V> X<\h> X<\H>
250 X<word> X<whitespace> X<character class> X<backreference>
252 \w Match a "word" character (alphanumeric plus "_")
253 \W Match a non-"word" character
254 \s Match a whitespace character
255 \S Match a non-whitespace character
256 \d Match a digit character
257 \D Match a non-digit character
258 \pP Match P, named property. Use \p{Prop} for longer names.
260 \X Match eXtended Unicode "combining character sequence",
261 equivalent to (?>\PM\pM*)
262 \C Match a single C char (octet) even under Unicode.
263 NOTE: breaks up characters into their UTF-8 bytes,
264 so you may end up with malformed pieces of UTF-8.
265 Unsupported in lookbehind.
266 \1 Backreference to a specific group.
267 '1' may actually be any positive integer.
268 \g1 Backreference to a specific or previous group,
269 \g{-1} number may be negative indicating a previous buffer and may
270 optionally be wrapped in curly brackets for safer parsing.
271 \g{name} Named backreference
272 \k<name> Named backreference
273 \K Keep the stuff left of the \K, don't include it in $&
274 \N Any character but \n
275 \v Vertical whitespace
276 \V Not vertical whitespace
277 \h Horizontal whitespace
278 \H Not horizontal whitespace
281 A C<\w> matches a single alphanumeric character (an alphabetic
282 character, or a decimal digit) or C<_>, not a whole word. Use C<\w+>
283 to match a string of Perl-identifier characters (which isn't the same
284 as matching an English word). If C<use locale> is in effect, the list
285 of alphabetic characters generated by C<\w> is taken from the current
286 locale. See L<perllocale>. You may use C<\w>, C<\W>, C<\s>, C<\S>,
287 C<\d>, and C<\D> within character classes, but they aren't usable
288 as either end of a range. If any of them precedes or follows a "-",
289 the "-" is understood literally. If Unicode is in effect, C<\s> matches
290 also "\x{85}", "\x{2028}", and "\x{2029}". See L<perlunicode> for more
291 details about C<\pP>, C<\PP>, C<\X> and the possibility of defining
292 your own C<\p> and C<\P> properties, and L<perluniintro> about Unicode
296 C<\R> will atomically match a linebreak, including the network line-ending
297 "\x0D\x0A". Specifically, X<\R> is exactly equivalent to
299 (?>\x0D\x0A?|[\x0A-\x0C\x85\x{2028}\x{2029}])
301 B<Note:> C<\R> has no special meaning inside of a character class;
302 use C<\v> instead (vertical whitespace).
305 The POSIX character class syntax
310 is also available. Note that the C<[> and C<]> brackets are I<literal>;
311 they must always be used within a character class expression.
314 $string =~ /[[:alpha:]]/;
316 # this is not, and will generate a warning:
317 $string =~ /[:alpha:]/;
319 The available classes and their backslash equivalents (if available) are
322 X<alpha> X<alnum> X<ascii> X<blank> X<cntrl> X<digit> X<graph>
323 X<lower> X<print> X<punct> X<space> X<upper> X<word> X<xdigit>
344 A GNU extension equivalent to C<[ \t]>, "all horizontal whitespace".
348 Not exactly equivalent to C<\s> since the C<[[:space:]]> includes
349 also the (very rare) "vertical tabulator", "\cK" or chr(11) in ASCII.
353 A Perl extension, see above.
357 For example use C<[:upper:]> to match all the uppercase characters.
358 Note that the C<[]> are part of the C<[::]> construct, not part of the
359 whole character class. For example:
363 matches zero, one, any alphabetic character, and the percent sign.
365 The following equivalences to Unicode \p{} constructs and equivalent
366 backslash character classes (if available), will hold:
367 X<character class> X<\p> X<\p{}>
369 [[:...:]] \p{...} backslash
379 print IsPrint (but see [2] below)
380 punct IsPunct (but see [3] below)
387 For example C<[[:lower:]]> and C<\p{IsLower}> are equivalent.
389 However, the equivalence between C<[[:xxxxx:]]> and C<\p{IsXxxxx}>
396 If the C<utf8> pragma is not used but the C<locale> pragma is, the
397 classes correlate with the usual isalpha(3) interface (except for
400 But if the C<locale> or C<encoding> pragmas are not used and
401 the string is not C<utf8>, then C<[[:xxxxx:]]> (and C<\w>, etc.)
402 will not match characters 0x80-0xff; whereas C<\p{IsXxxxx}> will
403 force the string to C<utf8> and can match these characters
408 C<\p{IsPrint}> matches characters 0x09-0x0d but C<[[:print:]]> does not.
412 C<[[:punct::]]> matches the following but C<\p{IsPunct}> does not,
413 because they are classed as symbols (not punctuation) in Unicode.
421 =item C<+> C<< < >> C<=> C<< > >> C<|> C<~>
427 Modifier symbols (accents)
433 The other named classes are:
440 Any control character. Usually characters that don't produce output as
441 such but instead control the terminal somehow: for example newline and
442 backspace are control characters. All characters with ord() less than
443 32 are usually classified as control characters (assuming ASCII,
444 the ISO Latin character sets, and Unicode), as is the character with
445 the ord() value of 127 (C<DEL>).
450 Any alphanumeric or punctuation (special) character.
455 Any alphanumeric or punctuation (special) character or the space character.
460 Any punctuation (special) character.
465 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
466 work just fine) it is included for completeness.
470 You can negate the [::] character classes by prefixing the class name
471 with a '^'. This is a Perl extension. For example:
472 X<character class, negation>
474 POSIX traditional Unicode
476 [[:^digit:]] \D \P{IsDigit}
477 [[:^space:]] \S \P{IsSpace}
478 [[:^word:]] \W \P{IsWord}
480 Perl respects the POSIX standard in that POSIX character classes are
481 only supported within a character class. The POSIX character classes
482 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
483 use them will cause an error.
487 Perl defines the following zero-width assertions:
488 X<zero-width assertion> X<assertion> X<regex, zero-width assertion>
489 X<regexp, zero-width assertion>
490 X<regular expression, zero-width assertion>
491 X<\b> X<\B> X<\A> X<\Z> X<\z> X<\G>
493 \b Match a word boundary
494 \B Match except at a word boundary
495 \A Match only at beginning of string
496 \Z Match only at end of string, or before newline at the end
497 \z Match only at end of string
498 \G Match only at pos() (e.g. at the end-of-match position
501 A word boundary (C<\b>) is a spot between two characters
502 that has a C<\w> on one side of it and a C<\W> on the other side
503 of it (in either order), counting the imaginary characters off the
504 beginning and end of the string as matching a C<\W>. (Within
505 character classes C<\b> represents backspace rather than a word
506 boundary, just as it normally does in any double-quoted string.)
507 The C<\A> and C<\Z> are just like "^" and "$", except that they
508 won't match multiple times when the C</m> modifier is used, while
509 "^" and "$" will match at every internal line boundary. To match
510 the actual end of the string and not ignore an optional trailing
512 X<\b> X<\A> X<\Z> X<\z> X</m>
514 The C<\G> assertion can be used to chain global matches (using
515 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
516 It is also useful when writing C<lex>-like scanners, when you have
517 several patterns that you want to match against consequent substrings
518 of your string, see the previous reference. The actual location
519 where C<\G> will match can also be influenced by using C<pos()> as
520 an lvalue: see L<perlfunc/pos>. Note that the rule for zero-length
521 matches is modified somewhat, in that contents to the left of C<\G> is
522 not counted when determining the length of the match. Thus the following
523 will not match forever:
532 It will print 'A' and then terminate, as it considers the match to
533 be zero-width, and thus will not match at the same position twice in a
536 It is worth noting that C<\G> improperly used can result in an infinite
537 loop. Take care when using patterns that include C<\G> in an alternation.
539 =head3 Capture buffers
541 The bracketing construct C<( ... )> creates capture buffers. To refer
542 to the current contents of a buffer later on, within the same pattern,
543 use \1 for the first, \2 for the second, and so on.
544 Outside the match use "$" instead of "\". (The
545 \<digit> notation works in certain circumstances outside
546 the match. See the warning below about \1 vs $1 for details.)
547 Referring back to another part of the match is called a
549 X<regex, capture buffer> X<regexp, capture buffer>
550 X<regular expression, capture buffer> X<backreference>
552 There is no limit to the number of captured substrings that you may
553 use. However Perl also uses \10, \11, etc. as aliases for \010,
554 \011, etc. (Recall that 0 means octal, so \011 is the character at
555 number 9 in your coded character set; which would be the 10th character,
556 a horizontal tab under ASCII.) Perl resolves this
557 ambiguity by interpreting \10 as a backreference only if at least 10
558 left parentheses have opened before it. Likewise \11 is a
559 backreference only if at least 11 left parentheses have opened
560 before it. And so on. \1 through \9 are always interpreted as
563 X<\g{1}> X<\g{-1}> X<\g{name}> X<relative backreference> X<named backreference>
564 In order to provide a safer and easier way to construct patterns using
565 backreferences, Perl provides the C<\g{N}> notation (starting with perl
566 5.10.0). The curly brackets are optional, however omitting them is less
567 safe as the meaning of the pattern can be changed by text (such as digits)
568 following it. When N is a positive integer the C<\g{N}> notation is
569 exactly equivalent to using normal backreferences. When N is a negative
570 integer then it is a relative backreference referring to the previous N'th
571 capturing group. When the bracket form is used and N is not an integer, it
572 is treated as a reference to a named buffer.
574 Thus C<\g{-1}> refers to the last buffer, C<\g{-2}> refers to the
575 buffer before that. For example:
581 \g{-1} # backref to buffer 3
582 \g{-3} # backref to buffer 1
586 and would match the same as C</(Y) ( (X) \3 \1 )/x>.
588 Additionally, as of Perl 5.10.0 you may use named capture buffers and named
589 backreferences. The notation is C<< (?<name>...) >> to declare and C<< \k<name> >>
590 to reference. You may also use apostrophes instead of angle brackets to delimit the
591 name; and you may use the bracketed C<< \g{name} >> backreference syntax.
592 It's possible to refer to a named capture buffer by absolute and relative number as well.
593 Outside the pattern, a named capture buffer is available via the C<%+> hash.
594 When different buffers within the same pattern have the same name, C<$+{name}>
595 and C<< \k<name> >> refer to the leftmost defined group. (Thus it's possible
596 to do things with named capture buffers that would otherwise require C<(??{})>
598 X<named capture buffer> X<regular expression, named capture buffer>
599 X<%+> X<$+{name}> X<< \k<name> >>
603 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
605 /(.)\1/ # find first doubled char
606 and print "'$1' is the first doubled character\n";
608 /(?<char>.)\k<char>/ # ... a different way
609 and print "'$+{char}' is the first doubled character\n";
611 /(?'char'.)\1/ # ... mix and match
612 and print "'$1' is the first doubled character\n";
614 if (/Time: (..):(..):(..)/) { # parse out values
620 Several special variables also refer back to portions of the previous
621 match. C<$+> returns whatever the last bracket match matched.
622 C<$&> returns the entire matched string. (At one point C<$0> did
623 also, but now it returns the name of the program.) C<$`> returns
624 everything before the matched string. C<$'> returns everything
625 after the matched string. And C<$^N> contains whatever was matched by
626 the most-recently closed group (submatch). C<$^N> can be used in
627 extended patterns (see below), for example to assign a submatch to a
629 X<$+> X<$^N> X<$&> X<$`> X<$'>
631 The numbered match variables ($1, $2, $3, etc.) and the related punctuation
632 set (C<$+>, C<$&>, C<$`>, C<$'>, and C<$^N>) are all dynamically scoped
633 until the end of the enclosing block or until the next successful
634 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
635 X<$+> X<$^N> X<$&> X<$`> X<$'>
636 X<$1> X<$2> X<$3> X<$4> X<$5> X<$6> X<$7> X<$8> X<$9>
639 B<NOTE>: Failed matches in Perl do not reset the match variables,
640 which makes it easier to write code that tests for a series of more
641 specific cases and remembers the best match.
643 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
644 C<$'> anywhere in the program, it has to provide them for every
645 pattern match. This may substantially slow your program. Perl
646 uses the same mechanism to produce $1, $2, etc, so you also pay a
647 price for each pattern that contains capturing parentheses. (To
648 avoid this cost while retaining the grouping behaviour, use the
649 extended regular expression C<(?: ... )> instead.) But if you never
650 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
651 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
652 if you can, but if you can't (and some algorithms really appreciate
653 them), once you've used them once, use them at will, because you've
654 already paid the price. As of 5.005, C<$&> is not so costly as the
658 As a workaround for this problem, Perl 5.10.0 introduces C<${^PREMATCH}>,
659 C<${^MATCH}> and C<${^POSTMATCH}>, which are equivalent to C<$`>, C<$&>
660 and C<$'>, B<except> that they are only guaranteed to be defined after a
661 successful match that was executed with the C</p> (preserve) modifier.
662 The use of these variables incurs no global performance penalty, unlike
663 their punctuation char equivalents, however at the trade-off that you
664 have to tell perl when you want to use them.
667 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
668 C<\w>, C<\n>. Unlike some other regular expression languages, there
669 are no backslashed symbols that aren't alphanumeric. So anything
670 that looks like \\, \(, \), \<, \>, \{, or \} is always
671 interpreted as a literal character, not a metacharacter. This was
672 once used in a common idiom to disable or quote the special meanings
673 of regular expression metacharacters in a string that you want to
674 use for a pattern. Simply quote all non-"word" characters:
676 $pattern =~ s/(\W)/\\$1/g;
678 (If C<use locale> is set, then this depends on the current locale.)
679 Today it is more common to use the quotemeta() function or the C<\Q>
680 metaquoting escape sequence to disable all metacharacters' special
683 /$unquoted\Q$quoted\E$unquoted/
685 Beware that if you put literal backslashes (those not inside
686 interpolated variables) between C<\Q> and C<\E>, double-quotish
687 backslash interpolation may lead to confusing results. If you
688 I<need> to use literal backslashes within C<\Q...\E>,
689 consult L<perlop/"Gory details of parsing quoted constructs">.
691 =head2 Extended Patterns
693 Perl also defines a consistent extension syntax for features not
694 found in standard tools like B<awk> and B<lex>. The syntax is a
695 pair of parentheses with a question mark as the first thing within
696 the parentheses. The character after the question mark indicates
699 The stability of these extensions varies widely. Some have been
700 part of the core language for many years. Others are experimental
701 and may change without warning or be completely removed. Check
702 the documentation on an individual feature to verify its current
705 A question mark was chosen for this and for the minimal-matching
706 construct because 1) question marks are rare in older regular
707 expressions, and 2) whenever you see one, you should stop and
708 "question" exactly what is going on. That's psychology...
715 A comment. The text is ignored. If the C</x> modifier enables
716 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
717 the comment as soon as it sees a C<)>, so there is no way to put a literal
720 =item C<(?pimsx-imsx)>
723 One or more embedded pattern-match modifiers, to be turned on (or
724 turned off, if preceded by C<->) for the remainder of the pattern or
725 the remainder of the enclosing pattern group (if any). This is
726 particularly useful for dynamic patterns, such as those read in from a
727 configuration file, taken from an argument, or specified in a table
728 somewhere. Consider the case where some patterns want to be case
729 sensitive and some do not: The case insensitive ones merely need to
730 include C<(?i)> at the front of the pattern. For example:
733 if ( /$pattern/i ) { }
737 $pattern = "(?i)foobar";
738 if ( /$pattern/ ) { }
740 These modifiers are restored at the end of the enclosing group. For example,
744 will match C<blah> in any case, some spaces, and an exact (I<including the case>!)
745 repetition of the previous word, assuming the C</x> modifier, and no C</i>
746 modifier outside this group.
748 Note that the C<p> modifier is special in that it can only be enabled,
749 not disabled, and that its presence anywhere in a pattern has a global
750 effect. Thus C<(?-p)> and C<(?-p:...)> are meaningless and will warn
751 when executed under C<use warnings>.
756 =item C<(?imsx-imsx:pattern)>
758 This is for clustering, not capturing; it groups subexpressions like
759 "()", but doesn't make backreferences as "()" does. So
761 @fields = split(/\b(?:a|b|c)\b/)
765 @fields = split(/\b(a|b|c)\b/)
767 but doesn't spit out extra fields. It's also cheaper not to capture
768 characters if you don't need to.
770 Any letters between C<?> and C<:> act as flags modifiers as with
771 C<(?imsx-imsx)>. For example,
773 /(?s-i:more.*than).*million/i
775 is equivalent to the more verbose
777 /(?:(?s-i)more.*than).*million/i
780 X<(?|)> X<Branch reset>
782 This is the "branch reset" pattern, which has the special property
783 that the capture buffers are numbered from the same starting point
784 in each alternation branch. It is available starting from perl 5.10.0.
786 Capture buffers are numbered from left to right, but inside this
787 construct the numbering is restarted for each branch.
789 The numbering within each branch will be as normal, and any buffers
790 following this construct will be numbered as though the construct
791 contained only one branch, that being the one with the most capture
794 This construct will be useful when you want to capture one of a
795 number of alternative matches.
797 Consider the following pattern. The numbers underneath show in
798 which buffer the captured content will be stored.
801 # before ---------------branch-reset----------- after
802 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
805 Note: as of Perl 5.10.0, branch resets interfere with the contents of
806 the C<%+> hash, that holds named captures. Consider using C<%-> instead.
808 =item Look-Around Assertions
809 X<look-around assertion> X<lookaround assertion> X<look-around> X<lookaround>
811 Look-around assertions are zero width patterns which match a specific
812 pattern without including it in C<$&>. Positive assertions match when
813 their subpattern matches, negative assertions match when their subpattern
814 fails. Look-behind matches text up to the current match position,
815 look-ahead matches text following the current match position.
820 X<(?=)> X<look-ahead, positive> X<lookahead, positive>
822 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
823 matches a word followed by a tab, without including the tab in C<$&>.
826 X<(?!)> X<look-ahead, negative> X<lookahead, negative>
828 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
829 matches any occurrence of "foo" that isn't followed by "bar". Note
830 however that look-ahead and look-behind are NOT the same thing. You cannot
831 use this for look-behind.
833 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
834 will not do what you want. That's because the C<(?!foo)> is just saying that
835 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
836 match. You would have to do something like C</(?!foo)...bar/> for that. We
837 say "like" because there's the case of your "bar" not having three characters
838 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
839 Sometimes it's still easier just to say:
841 if (/bar/ && $` !~ /foo$/)
843 For look-behind see below.
845 =item C<(?<=pattern)> C<\K>
846 X<(?<=)> X<look-behind, positive> X<lookbehind, positive> X<\K>
848 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
849 matches a word that follows a tab, without including the tab in C<$&>.
850 Works only for fixed-width look-behind.
852 There is a special form of this construct, called C<\K>, which causes the
853 regex engine to "keep" everything it had matched prior to the C<\K> and
854 not include it in C<$&>. This effectively provides variable length
855 look-behind. The use of C<\K> inside of another look-around assertion
856 is allowed, but the behaviour is currently not well defined.
858 For various reasons C<\K> may be significantly more efficient than the
859 equivalent C<< (?<=...) >> construct, and it is especially useful in
860 situations where you want to efficiently remove something following
861 something else in a string. For instance
865 can be rewritten as the much more efficient
869 =item C<(?<!pattern)>
870 X<(?<!)> X<look-behind, negative> X<lookbehind, negative>
872 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
873 matches any occurrence of "foo" that does not follow "bar". Works
874 only for fixed-width look-behind.
878 =item C<(?'NAME'pattern)>
880 =item C<< (?<NAME>pattern) >>
881 X<< (?<NAME>) >> X<(?'NAME')> X<named capture> X<capture>
883 A named capture buffer. Identical in every respect to normal capturing
884 parentheses C<()> but for the additional fact that C<%+> or C<%-> may be
885 used after a successful match to refer to a named buffer. See C<perlvar>
886 for more details on the C<%+> and C<%-> hashes.
888 If multiple distinct capture buffers have the same name then the
889 $+{NAME} will refer to the leftmost defined buffer in the match.
891 The forms C<(?'NAME'pattern)> and C<< (?<NAME>pattern) >> are equivalent.
893 B<NOTE:> While the notation of this construct is the same as the similar
894 function in .NET regexes, the behavior is not. In Perl the buffers are
895 numbered sequentially regardless of being named or not. Thus in the
900 $+{foo} will be the same as $2, and $3 will contain 'z' instead of
901 the opposite which is what a .NET regex hacker might expect.
903 Currently NAME is restricted to simple identifiers only.
904 In other words, it must match C</^[_A-Za-z][_A-Za-z0-9]*\z/> or
905 its Unicode extension (see L<utf8>),
906 though it isn't extended by the locale (see L<perllocale>).
908 B<NOTE:> In order to make things easier for programmers with experience
909 with the Python or PCRE regex engines, the pattern C<< (?PE<lt>NAMEE<gt>pattern) >>
910 may be used instead of C<< (?<NAME>pattern) >>; however this form does not
911 support the use of single quotes as a delimiter for the name.
913 =item C<< \k<NAME> >>
915 =item C<< \k'NAME' >>
917 Named backreference. Similar to numeric backreferences, except that
918 the group is designated by name and not number. If multiple groups
919 have the same name then it refers to the leftmost defined group in
922 It is an error to refer to a name not defined by a C<< (?<NAME>) >>
923 earlier in the pattern.
925 Both forms are equivalent.
927 B<NOTE:> In order to make things easier for programmers with experience
928 with the Python or PCRE regex engines, the pattern C<< (?P=NAME) >>
929 may be used instead of C<< \k<NAME> >>.
932 X<(?{})> X<regex, code in> X<regexp, code in> X<regular expression, code in>
934 B<WARNING>: This extended regular expression feature is considered
935 experimental, and may be changed without notice. Code executed that
936 has side effects may not perform identically from version to version
937 due to the effect of future optimisations in the regex engine.
939 This zero-width assertion evaluates any embedded Perl code. It
940 always succeeds, and its C<code> is not interpolated. Currently,
941 the rules to determine where the C<code> ends are somewhat convoluted.
943 This feature can be used together with the special variable C<$^N> to
944 capture the results of submatches in variables without having to keep
945 track of the number of nested parentheses. For example:
947 $_ = "The brown fox jumps over the lazy dog";
948 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
949 print "color = $color, animal = $animal\n";
951 Inside the C<(?{...})> block, C<$_> refers to the string the regular
952 expression is matching against. You can also use C<pos()> to know what is
953 the current position of matching within this string.
955 The C<code> is properly scoped in the following sense: If the assertion
956 is backtracked (compare L<"Backtracking">), all changes introduced after
957 C<local>ization are undone, so that
961 (?{ $cnt = 0 }) # Initialize $cnt.
965 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
969 (?{ $res = $cnt }) # On success copy to non-localized
973 will set C<$res = 4>. Note that after the match, C<$cnt> returns to the globally
974 introduced value, because the scopes that restrict C<local> operators
977 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
978 switch. If I<not> used in this way, the result of evaluation of
979 C<code> is put into the special variable C<$^R>. This happens
980 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
981 inside the same regular expression.
983 The assignment to C<$^R> above is properly localized, so the old
984 value of C<$^R> is restored if the assertion is backtracked; compare
987 Due to an unfortunate implementation issue, the Perl code contained in these
988 blocks is treated as a compile time closure that can have seemingly bizarre
989 consequences when used with lexically scoped variables inside of subroutines
990 or loops. There are various workarounds for this, including simply using
991 global variables instead. If you are using this construct and strange results
992 occur then check for the use of lexically scoped variables.
994 For reasons of security, this construct is forbidden if the regular
995 expression involves run-time interpolation of variables, unless the
996 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
997 variables contain results of C<qr//> operator (see
998 L<perlop/"qr/STRING/imosx">).
1000 This restriction is due to the wide-spread and remarkably convenient
1001 custom of using run-time determined strings as patterns. For example:
1007 Before Perl knew how to execute interpolated code within a pattern,
1008 this operation was completely safe from a security point of view,
1009 although it could raise an exception from an illegal pattern. If
1010 you turn on the C<use re 'eval'>, though, it is no longer secure,
1011 so you should only do so if you are also using taint checking.
1012 Better yet, use the carefully constrained evaluation within a Safe
1013 compartment. See L<perlsec> for details about both these mechanisms.
1015 Because Perl's regex engine is currently not re-entrant, interpolated
1016 code may not invoke the regex engine either directly with C<m//> or C<s///>),
1017 or indirectly with functions such as C<split>.
1019 =item C<(??{ code })>
1021 X<regex, postponed> X<regexp, postponed> X<regular expression, postponed>
1023 B<WARNING>: This extended regular expression feature is considered
1024 experimental, and may be changed without notice. Code executed that
1025 has side effects may not perform identically from version to version
1026 due to the effect of future optimisations in the regex engine.
1028 This is a "postponed" regular subexpression. The C<code> is evaluated
1029 at run time, at the moment this subexpression may match. The result
1030 of evaluation is considered as a regular expression and matched as
1031 if it were inserted instead of this construct. Note that this means
1032 that the contents of capture buffers defined inside an eval'ed pattern
1033 are not available outside of the pattern, and vice versa, there is no
1034 way for the inner pattern to refer to a capture buffer defined outside.
1037 ('a' x 100)=~/(??{'(.)' x 100})/
1039 B<will> match, it will B<not> set $1.
1041 The C<code> is not interpolated. As before, the rules to determine
1042 where the C<code> ends are currently somewhat convoluted.
1044 The following pattern matches a parenthesized group:
1049 (?> [^()]+ ) # Non-parens without backtracking
1051 (??{ $re }) # Group with matching parens
1056 See also C<(?PARNO)> for a different, more efficient way to accomplish
1059 Because perl's regex engine is not currently re-entrant, delayed
1060 code may not invoke the regex engine either directly with C<m//> or C<s///>),
1061 or indirectly with functions such as C<split>.
1063 Recursing deeper than 50 times without consuming any input string will
1064 result in a fatal error. The maximum depth is compiled into perl, so
1065 changing it requires a custom build.
1067 =item C<(?PARNO)> C<(?-PARNO)> C<(?+PARNO)> C<(?R)> C<(?0)>
1068 X<(?PARNO)> X<(?1)> X<(?R)> X<(?0)> X<(?-1)> X<(?+1)> X<(?-PARNO)> X<(?+PARNO)>
1069 X<regex, recursive> X<regexp, recursive> X<regular expression, recursive>
1070 X<regex, relative recursion>
1072 Similar to C<(??{ code })> except it does not involve compiling any code,
1073 instead it treats the contents of a capture buffer as an independent
1074 pattern that must match at the current position. Capture buffers
1075 contained by the pattern will have the value as determined by the
1076 outermost recursion.
1078 PARNO is a sequence of digits (not starting with 0) whose value reflects
1079 the paren-number of the capture buffer to recurse to. C<(?R)> recurses to
1080 the beginning of the whole pattern. C<(?0)> is an alternate syntax for
1081 C<(?R)>. If PARNO is preceded by a plus or minus sign then it is assumed
1082 to be relative, with negative numbers indicating preceding capture buffers
1083 and positive ones following. Thus C<(?-1)> refers to the most recently
1084 declared buffer, and C<(?+1)> indicates the next buffer to be declared.
1085 Note that the counting for relative recursion differs from that of
1086 relative backreferences, in that with recursion unclosed buffers B<are>
1089 The following pattern matches a function foo() which may contain
1090 balanced parentheses as the argument.
1092 $re = qr{ ( # paren group 1 (full function)
1094 ( # paren group 2 (parens)
1096 ( # paren group 3 (contents of parens)
1098 (?> [^()]+ ) # Non-parens without backtracking
1100 (?2) # Recurse to start of paren group 2
1108 If the pattern was used as follows
1110 'foo(bar(baz)+baz(bop))'=~/$re/
1111 and print "\$1 = $1\n",
1115 the output produced should be the following:
1117 $1 = foo(bar(baz)+baz(bop))
1118 $2 = (bar(baz)+baz(bop))
1119 $3 = bar(baz)+baz(bop)
1121 If there is no corresponding capture buffer defined, then it is a
1122 fatal error. Recursing deeper than 50 times without consuming any input
1123 string will also result in a fatal error. The maximum depth is compiled
1124 into perl, so changing it requires a custom build.
1126 The following shows how using negative indexing can make it
1127 easier to embed recursive patterns inside of a C<qr//> construct
1130 my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1131 if (/foo $parens \s+ + \s+ bar $parens/x) {
1132 # do something here...
1135 B<Note> that this pattern does not behave the same way as the equivalent
1136 PCRE or Python construct of the same form. In Perl you can backtrack into
1137 a recursed group, in PCRE and Python the recursed into group is treated
1138 as atomic. Also, modifiers are resolved at compile time, so constructs
1139 like (?i:(?1)) or (?:(?i)(?1)) do not affect how the sub-pattern will
1145 Recurse to a named subpattern. Identical to C<(?PARNO)> except that the
1146 parenthesis to recurse to is determined by name. If multiple parentheses have
1147 the same name, then it recurses to the leftmost.
1149 It is an error to refer to a name that is not declared somewhere in the
1152 B<NOTE:> In order to make things easier for programmers with experience
1153 with the Python or PCRE regex engines the pattern C<< (?P>NAME) >>
1154 may be used instead of C<< (?&NAME) >>.
1156 =item C<(?(condition)yes-pattern|no-pattern)>
1159 =item C<(?(condition)yes-pattern)>
1161 Conditional expression. C<(condition)> should be either an integer in
1162 parentheses (which is valid if the corresponding pair of parentheses
1163 matched), a look-ahead/look-behind/evaluate zero-width assertion, a
1164 name in angle brackets or single quotes (which is valid if a buffer
1165 with the given name matched), or the special symbol (R) (true when
1166 evaluated inside of recursion or eval). Additionally the R may be
1167 followed by a number, (which will be true when evaluated when recursing
1168 inside of the appropriate group), or by C<&NAME>, in which case it will
1169 be true only when evaluated during recursion in the named group.
1171 Here's a summary of the possible predicates:
1177 Checks if the numbered capturing buffer has matched something.
1179 =item (<NAME>) ('NAME')
1181 Checks if a buffer with the given name has matched something.
1185 Treats the code block as the condition.
1189 Checks if the expression has been evaluated inside of recursion.
1193 Checks if the expression has been evaluated while executing directly
1194 inside of the n-th capture group. This check is the regex equivalent of
1196 if ((caller(0))[3] eq 'subname') { ... }
1198 In other words, it does not check the full recursion stack.
1202 Similar to C<(R1)>, this predicate checks to see if we're executing
1203 directly inside of the leftmost group with a given name (this is the same
1204 logic used by C<(?&NAME)> to disambiguate). It does not check the full
1205 stack, but only the name of the innermost active recursion.
1209 In this case, the yes-pattern is never directly executed, and no
1210 no-pattern is allowed. Similar in spirit to C<(?{0})> but more efficient.
1211 See below for details.
1222 matches a chunk of non-parentheses, possibly included in parentheses
1225 A special form is the C<(DEFINE)> predicate, which never executes directly
1226 its yes-pattern, and does not allow a no-pattern. This allows to define
1227 subpatterns which will be executed only by using the recursion mechanism.
1228 This way, you can define a set of regular expression rules that can be
1229 bundled into any pattern you choose.
1231 It is recommended that for this usage you put the DEFINE block at the
1232 end of the pattern, and that you name any subpatterns defined within it.
1234 Also, it's worth noting that patterns defined this way probably will
1235 not be as efficient, as the optimiser is not very clever about
1238 An example of how this might be used is as follows:
1240 /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
1246 Note that capture buffers matched inside of recursion are not accessible
1247 after the recursion returns, so the extra layer of capturing buffers is
1248 necessary. Thus C<$+{NAME_PAT}> would not be defined even though
1249 C<$+{NAME}> would be.
1251 =item C<< (?>pattern) >>
1252 X<backtrack> X<backtracking> X<atomic> X<possessive>
1254 An "independent" subexpression, one which matches the substring
1255 that a I<standalone> C<pattern> would match if anchored at the given
1256 position, and it matches I<nothing other than this substring>. This
1257 construct is useful for optimizations of what would otherwise be
1258 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
1259 It may also be useful in places where the "grab all you can, and do not
1260 give anything back" semantic is desirable.
1262 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
1263 (anchored at the beginning of string, as above) will match I<all>
1264 characters C<a> at the beginning of string, leaving no C<a> for
1265 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
1266 since the match of the subgroup C<a*> is influenced by the following
1267 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
1268 C<a*ab> will match fewer characters than a standalone C<a*>, since
1269 this makes the tail match.
1271 An effect similar to C<< (?>pattern) >> may be achieved by writing
1272 C<(?=(pattern))\1>. This matches the same substring as a standalone
1273 C<a+>, and the following C<\1> eats the matched string; it therefore
1274 makes a zero-length assertion into an analogue of C<< (?>...) >>.
1275 (The difference between these two constructs is that the second one
1276 uses a capturing group, thus shifting ordinals of backreferences
1277 in the rest of a regular expression.)
1279 Consider this pattern:
1290 That will efficiently match a nonempty group with matching parentheses
1291 two levels deep or less. However, if there is no such group, it
1292 will take virtually forever on a long string. That's because there
1293 are so many different ways to split a long string into several
1294 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
1295 to a subpattern of the above pattern. Consider how the pattern
1296 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
1297 seconds, but that each extra letter doubles this time. This
1298 exponential performance will make it appear that your program has
1299 hung. However, a tiny change to this pattern
1303 (?> [^()]+ ) # change x+ above to (?> x+ )
1310 which uses C<< (?>...) >> matches exactly when the one above does (verifying
1311 this yourself would be a productive exercise), but finishes in a fourth
1312 the time when used on a similar string with 1000000 C<a>s. Be aware,
1313 however, that this pattern currently triggers a warning message under
1314 the C<use warnings> pragma or B<-w> switch saying it
1315 C<"matches null string many times in regex">.
1317 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
1318 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
1319 This was only 4 times slower on a string with 1000000 C<a>s.
1321 The "grab all you can, and do not give anything back" semantic is desirable
1322 in many situations where on the first sight a simple C<()*> looks like
1323 the correct solution. Suppose we parse text with comments being delimited
1324 by C<#> followed by some optional (horizontal) whitespace. Contrary to
1325 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
1326 the comment delimiter, because it may "give up" some whitespace if
1327 the remainder of the pattern can be made to match that way. The correct
1328 answer is either one of these:
1333 For example, to grab non-empty comments into $1, one should use either
1336 / (?> \# [ \t]* ) ( .+ ) /x;
1337 / \# [ \t]* ( [^ \t] .* ) /x;
1339 Which one you pick depends on which of these expressions better reflects
1340 the above specification of comments.
1342 In some literature this construct is called "atomic matching" or
1343 "possessive matching".
1345 Possessive quantifiers are equivalent to putting the item they are applied
1346 to inside of one of these constructs. The following equivalences apply:
1348 Quantifier Form Bracketing Form
1349 --------------- ---------------
1353 PAT{min,max}+ (?>PAT{min,max})
1357 =head2 Special Backtracking Control Verbs
1359 B<WARNING:> These patterns are experimental and subject to change or
1360 removal in a future version of Perl. Their usage in production code should
1361 be noted to avoid problems during upgrades.
1363 These special patterns are generally of the form C<(*VERB:ARG)>. Unless
1364 otherwise stated the ARG argument is optional; in some cases, it is
1367 Any pattern containing a special backtracking verb that allows an argument
1368 has the special behaviour that when executed it sets the current packages'
1369 C<$REGERROR> and C<$REGMARK> variables. When doing so the following
1372 On failure, the C<$REGERROR> variable will be set to the ARG value of the
1373 verb pattern, if the verb was involved in the failure of the match. If the
1374 ARG part of the pattern was omitted, then C<$REGERROR> will be set to the
1375 name of the last C<(*MARK:NAME)> pattern executed, or to TRUE if there was
1376 none. Also, the C<$REGMARK> variable will be set to FALSE.
1378 On a successful match, the C<$REGERROR> variable will be set to FALSE, and
1379 the C<$REGMARK> variable will be set to the name of the last
1380 C<(*MARK:NAME)> pattern executed. See the explanation for the
1381 C<(*MARK:NAME)> verb below for more details.
1383 B<NOTE:> C<$REGERROR> and C<$REGMARK> are not magic variables like C<$1>
1384 and most other regex related variables. They are not local to a scope, nor
1385 readonly, but instead are volatile package variables similar to C<$AUTOLOAD>.
1386 Use C<local> to localize changes to them to a specific scope if necessary.
1388 If a pattern does not contain a special backtracking verb that allows an
1389 argument, then C<$REGERROR> and C<$REGMARK> are not touched at all.
1393 =item Verbs that take an argument
1397 =item C<(*PRUNE)> C<(*PRUNE:NAME)>
1398 X<(*PRUNE)> X<(*PRUNE:NAME)>
1400 This zero-width pattern prunes the backtracking tree at the current point
1401 when backtracked into on failure. Consider the pattern C<A (*PRUNE) B>,
1402 where A and B are complex patterns. Until the C<(*PRUNE)> verb is reached,
1403 A may backtrack as necessary to match. Once it is reached, matching
1404 continues in B, which may also backtrack as necessary; however, should B
1405 not match, then no further backtracking will take place, and the pattern
1406 will fail outright at the current starting position.
1408 The following example counts all the possible matching strings in a
1409 pattern (without actually matching any of them).
1411 'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
1412 print "Count=$count\n";
1427 If we add a C<(*PRUNE)> before the count like the following
1429 'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
1430 print "Count=$count\n";
1432 we prevent backtracking and find the count of the longest matching
1433 at each matching starting point like so:
1440 Any number of C<(*PRUNE)> assertions may be used in a pattern.
1442 See also C<< (?>pattern) >> and possessive quantifiers for other ways to
1443 control backtracking. In some cases, the use of C<(*PRUNE)> can be
1444 replaced with a C<< (?>pattern) >> with no functional difference; however,
1445 C<(*PRUNE)> can be used to handle cases that cannot be expressed using a
1446 C<< (?>pattern) >> alone.
1449 =item C<(*SKIP)> C<(*SKIP:NAME)>
1452 This zero-width pattern is similar to C<(*PRUNE)>, except that on
1453 failure it also signifies that whatever text that was matched leading up
1454 to the C<(*SKIP)> pattern being executed cannot be part of I<any> match
1455 of this pattern. This effectively means that the regex engine "skips" forward
1456 to this position on failure and tries to match again, (assuming that
1457 there is sufficient room to match).
1459 The name of the C<(*SKIP:NAME)> pattern has special significance. If a
1460 C<(*MARK:NAME)> was encountered while matching, then it is that position
1461 which is used as the "skip point". If no C<(*MARK)> of that name was
1462 encountered, then the C<(*SKIP)> operator has no effect. When used
1463 without a name the "skip point" is where the match point was when
1464 executing the (*SKIP) pattern.
1466 Compare the following to the examples in C<(*PRUNE)>, note the string
1469 'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
1470 print "Count=$count\n";
1478 Once the 'aaab' at the start of the string has matched, and the C<(*SKIP)>
1479 executed, the next starting point will be where the cursor was when the
1480 C<(*SKIP)> was executed.
1482 =item C<(*MARK:NAME)> C<(*:NAME)>
1483 X<(*MARK)> C<(*MARK:NAME)> C<(*:NAME)>
1485 This zero-width pattern can be used to mark the point reached in a string
1486 when a certain part of the pattern has been successfully matched. This
1487 mark may be given a name. A later C<(*SKIP)> pattern will then skip
1488 forward to that point if backtracked into on failure. Any number of
1489 C<(*MARK)> patterns are allowed, and the NAME portion is optional and may
1492 In addition to interacting with the C<(*SKIP)> pattern, C<(*MARK:NAME)>
1493 can be used to "label" a pattern branch, so that after matching, the
1494 program can determine which branches of the pattern were involved in the
1497 When a match is successful, the C<$REGMARK> variable will be set to the
1498 name of the most recently executed C<(*MARK:NAME)> that was involved
1501 This can be used to determine which branch of a pattern was matched
1502 without using a separate capture buffer for each branch, which in turn
1503 can result in a performance improvement, as perl cannot optimize
1504 C</(?:(x)|(y)|(z))/> as efficiently as something like
1505 C</(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/>.
1507 When a match has failed, and unless another verb has been involved in
1508 failing the match and has provided its own name to use, the C<$REGERROR>
1509 variable will be set to the name of the most recently executed
1512 See C<(*SKIP)> for more details.
1514 As a shortcut C<(*MARK:NAME)> can be written C<(*:NAME)>.
1516 =item C<(*THEN)> C<(*THEN:NAME)>
1518 This is similar to the "cut group" operator C<::> from Perl 6. Like
1519 C<(*PRUNE)>, this verb always matches, and when backtracked into on
1520 failure, it causes the regex engine to try the next alternation in the
1521 innermost enclosing group (capturing or otherwise).
1523 Its name comes from the observation that this operation combined with the
1524 alternation operator (C<|>) can be used to create what is essentially a
1525 pattern-based if/then/else block:
1527 ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
1529 Note that if this operator is used and NOT inside of an alternation then
1530 it acts exactly like the C<(*PRUNE)> operator.
1540 / ( A (*THEN) B | C (*THEN) D ) /
1544 / ( A (*PRUNE) B | C (*PRUNE) D ) /
1546 as after matching the A but failing on the B the C<(*THEN)> verb will
1547 backtrack and try C; but the C<(*PRUNE)> verb will simply fail.
1552 This is the Perl 6 "commit pattern" C<< <commit> >> or C<:::>. It's a
1553 zero-width pattern similar to C<(*SKIP)>, except that when backtracked
1554 into on failure it causes the match to fail outright. No further attempts
1555 to find a valid match by advancing the start pointer will occur again.
1558 'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
1559 print "Count=$count\n";
1566 In other words, once the C<(*COMMIT)> has been entered, and if the pattern
1567 does not match, the regex engine will not try any further matching on the
1572 =item Verbs without an argument
1576 =item C<(*FAIL)> C<(*F)>
1579 This pattern matches nothing and always fails. It can be used to force the
1580 engine to backtrack. It is equivalent to C<(?!)>, but easier to read. In
1581 fact, C<(?!)> gets optimised into C<(*FAIL)> internally.
1583 It is probably useful only when combined with C<(?{})> or C<(??{})>.
1588 B<WARNING:> This feature is highly experimental. It is not recommended
1589 for production code.
1591 This pattern matches nothing and causes the end of successful matching at
1592 the point at which the C<(*ACCEPT)> pattern was encountered, regardless of
1593 whether there is actually more to match in the string. When inside of a
1594 nested pattern, such as recursion, or in a subpattern dynamically generated
1595 via C<(??{})>, only the innermost pattern is ended immediately.
1597 If the C<(*ACCEPT)> is inside of capturing buffers then the buffers are
1598 marked as ended at the point at which the C<(*ACCEPT)> was encountered.
1601 'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
1603 will match, and C<$1> will be C<AB> and C<$2> will be C<B>, C<$3> will not
1604 be set. If another branch in the inner parentheses were matched, such as in the
1605 string 'ACDE', then the C<D> and C<E> would have to be matched as well.
1612 X<backtrack> X<backtracking>
1614 NOTE: This section presents an abstract approximation of regular
1615 expression behavior. For a more rigorous (and complicated) view of
1616 the rules involved in selecting a match among possible alternatives,
1617 see L<Combining RE Pieces>.
1619 A fundamental feature of regular expression matching involves the
1620 notion called I<backtracking>, which is currently used (when needed)
1621 by all regular non-possessive expression quantifiers, namely C<*>, C<*?>, C<+>,
1622 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
1623 internally, but the general principle outlined here is valid.
1625 For a regular expression to match, the I<entire> regular expression must
1626 match, not just part of it. So if the beginning of a pattern containing a
1627 quantifier succeeds in a way that causes later parts in the pattern to
1628 fail, the matching engine backs up and recalculates the beginning
1629 part--that's why it's called backtracking.
1631 Here is an example of backtracking: Let's say you want to find the
1632 word following "foo" in the string "Food is on the foo table.":
1634 $_ = "Food is on the foo table.";
1635 if ( /\b(foo)\s+(\w+)/i ) {
1636 print "$2 follows $1.\n";
1639 When the match runs, the first part of the regular expression (C<\b(foo)>)
1640 finds a possible match right at the beginning of the string, and loads up
1641 $1 with "Foo". However, as soon as the matching engine sees that there's
1642 no whitespace following the "Foo" that it had saved in $1, it realizes its
1643 mistake and starts over again one character after where it had the
1644 tentative match. This time it goes all the way until the next occurrence
1645 of "foo". The complete regular expression matches this time, and you get
1646 the expected output of "table follows foo."
1648 Sometimes minimal matching can help a lot. Imagine you'd like to match
1649 everything between "foo" and "bar". Initially, you write something
1652 $_ = "The food is under the bar in the barn.";
1653 if ( /foo(.*)bar/ ) {
1657 Which perhaps unexpectedly yields:
1659 got <d is under the bar in the >
1661 That's because C<.*> was greedy, so you get everything between the
1662 I<first> "foo" and the I<last> "bar". Here it's more effective
1663 to use minimal matching to make sure you get the text between a "foo"
1664 and the first "bar" thereafter.
1666 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
1667 got <d is under the >
1669 Here's another example. Let's say you'd like to match a number at the end
1670 of a string, and you also want to keep the preceding part of the match.
1673 $_ = "I have 2 numbers: 53147";
1674 if ( /(.*)(\d*)/ ) { # Wrong!
1675 print "Beginning is <$1>, number is <$2>.\n";
1678 That won't work at all, because C<.*> was greedy and gobbled up the
1679 whole string. As C<\d*> can match on an empty string the complete
1680 regular expression matched successfully.
1682 Beginning is <I have 2 numbers: 53147>, number is <>.
1684 Here are some variants, most of which don't work:
1686 $_ = "I have 2 numbers: 53147";
1699 printf "%-12s ", $pat;
1701 print "<$1> <$2>\n";
1707 That will print out:
1709 (.*)(\d*) <I have 2 numbers: 53147> <>
1710 (.*)(\d+) <I have 2 numbers: 5314> <7>
1712 (.*?)(\d+) <I have > <2>
1713 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
1714 (.*?)(\d+)$ <I have 2 numbers: > <53147>
1715 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
1716 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
1718 As you see, this can be a bit tricky. It's important to realize that a
1719 regular expression is merely a set of assertions that gives a definition
1720 of success. There may be 0, 1, or several different ways that the
1721 definition might succeed against a particular string. And if there are
1722 multiple ways it might succeed, you need to understand backtracking to
1723 know which variety of success you will achieve.
1725 When using look-ahead assertions and negations, this can all get even
1726 trickier. Imagine you'd like to find a sequence of non-digits not
1727 followed by "123". You might try to write that as
1730 if ( /^\D*(?!123)/ ) { # Wrong!
1731 print "Yup, no 123 in $_\n";
1734 But that isn't going to match; at least, not the way you're hoping. It
1735 claims that there is no 123 in the string. Here's a clearer picture of
1736 why that pattern matches, contrary to popular expectations:
1741 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
1742 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
1744 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
1745 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
1753 You might have expected test 3 to fail because it seems to a more
1754 general purpose version of test 1. The important difference between
1755 them is that test 3 contains a quantifier (C<\D*>) and so can use
1756 backtracking, whereas test 1 will not. What's happening is
1757 that you've asked "Is it true that at the start of $x, following 0 or more
1758 non-digits, you have something that's not 123?" If the pattern matcher had
1759 let C<\D*> expand to "ABC", this would have caused the whole pattern to
1762 The search engine will initially match C<\D*> with "ABC". Then it will
1763 try to match C<(?!123> with "123", which fails. But because
1764 a quantifier (C<\D*>) has been used in the regular expression, the
1765 search engine can backtrack and retry the match differently
1766 in the hope of matching the complete regular expression.
1768 The pattern really, I<really> wants to succeed, so it uses the
1769 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
1770 time. Now there's indeed something following "AB" that is not
1771 "123". It's "C123", which suffices.
1773 We can deal with this by using both an assertion and a negation.
1774 We'll say that the first part in $1 must be followed both by a digit
1775 and by something that's not "123". Remember that the look-aheads
1776 are zero-width expressions--they only look, but don't consume any
1777 of the string in their match. So rewriting this way produces what
1778 you'd expect; that is, case 5 will fail, but case 6 succeeds:
1780 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
1781 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
1785 In other words, the two zero-width assertions next to each other work as though
1786 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
1787 matches only if you're at the beginning of the line AND the end of the
1788 line simultaneously. The deeper underlying truth is that juxtaposition in
1789 regular expressions always means AND, except when you write an explicit OR
1790 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
1791 although the attempted matches are made at different positions because "a"
1792 is not a zero-width assertion, but a one-width assertion.
1794 B<WARNING>: Particularly complicated regular expressions can take
1795 exponential time to solve because of the immense number of possible
1796 ways they can use backtracking to try for a match. For example, without
1797 internal optimizations done by the regular expression engine, this will
1798 take a painfully long time to run:
1800 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
1802 And if you used C<*>'s in the internal groups instead of limiting them
1803 to 0 through 5 matches, then it would take forever--or until you ran
1804 out of stack space. Moreover, these internal optimizations are not
1805 always applicable. For example, if you put C<{0,5}> instead of C<*>
1806 on the external group, no current optimization is applicable, and the
1807 match takes a long time to finish.
1809 A powerful tool for optimizing such beasts is what is known as an
1810 "independent group",
1811 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
1812 zero-length look-ahead/look-behind assertions will not backtrack to make
1813 the tail match, since they are in "logical" context: only
1814 whether they match is considered relevant. For an example
1815 where side-effects of look-ahead I<might> have influenced the
1816 following match, see L<C<< (?>pattern) >>>.
1818 =head2 Version 8 Regular Expressions
1819 X<regular expression, version 8> X<regex, version 8> X<regexp, version 8>
1821 In case you're not familiar with the "regular" Version 8 regex
1822 routines, here are the pattern-matching rules not described above.
1824 Any single character matches itself, unless it is a I<metacharacter>
1825 with a special meaning described here or above. You can cause
1826 characters that normally function as metacharacters to be interpreted
1827 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
1828 character; "\\" matches a "\"). This escape mechanism is also required
1829 for the character used as the pattern delimiter.
1831 A series of characters matches that series of characters in the target
1832 string, so the pattern C<blurfl> would match "blurfl" in the target
1835 You can specify a character class, by enclosing a list of characters
1836 in C<[]>, which will match any character from the list. If the
1837 first character after the "[" is "^", the class matches any character not
1838 in the list. Within a list, the "-" character specifies a
1839 range, so that C<a-z> represents all characters between "a" and "z",
1840 inclusive. If you want either "-" or "]" itself to be a member of a
1841 class, put it at the start of the list (possibly after a "^"), or
1842 escape it with a backslash. "-" is also taken literally when it is
1843 at the end of the list, just before the closing "]". (The
1844 following all specify the same class of three characters: C<[-az]>,
1845 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1846 specifies a class containing twenty-six characters, even on EBCDIC-based
1847 character sets.) Also, if you try to use the character
1848 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1849 a range, the "-" is understood literally.
1851 Note also that the whole range idea is rather unportable between
1852 character sets--and even within character sets they may cause results
1853 you probably didn't expect. A sound principle is to use only ranges
1854 that begin from and end at either alphabetics of equal case ([a-e],
1855 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1856 spell out the character sets in full.
1858 Characters may be specified using a metacharacter syntax much like that
1859 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1860 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1861 of octal digits, matches the character whose coded character set value
1862 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1863 matches the character whose numeric value is I<nn>. The expression \cI<x>
1864 matches the character control-I<x>. Finally, the "." metacharacter
1865 matches any character except "\n" (unless you use C</s>).
1867 You can specify a series of alternatives for a pattern using "|" to
1868 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1869 or "foe" in the target string (as would C<f(e|i|o)e>). The
1870 first alternative includes everything from the last pattern delimiter
1871 ("(", "[", or the beginning of the pattern) up to the first "|", and
1872 the last alternative contains everything from the last "|" to the next
1873 pattern delimiter. That's why it's common practice to include
1874 alternatives in parentheses: to minimize confusion about where they
1877 Alternatives are tried from left to right, so the first
1878 alternative found for which the entire expression matches, is the one that
1879 is chosen. This means that alternatives are not necessarily greedy. For
1880 example: when matching C<foo|foot> against "barefoot", only the "foo"
1881 part will match, as that is the first alternative tried, and it successfully
1882 matches the target string. (This might not seem important, but it is
1883 important when you are capturing matched text using parentheses.)
1885 Also remember that "|" is interpreted as a literal within square brackets,
1886 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1888 Within a pattern, you may designate subpatterns for later reference
1889 by enclosing them in parentheses, and you may refer back to the
1890 I<n>th subpattern later in the pattern using the metacharacter
1891 \I<n>. Subpatterns are numbered based on the left to right order
1892 of their opening parenthesis. A backreference matches whatever
1893 actually matched the subpattern in the string being examined, not
1894 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1895 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1896 1 matched "0x", even though the rule C<0|0x> could potentially match
1897 the leading 0 in the second number.
1899 =head2 Warning on \1 Instead of $1
1901 Some people get too used to writing things like:
1903 $pattern =~ s/(\W)/\\\1/g;
1905 This is grandfathered for the RHS of a substitute to avoid shocking the
1906 B<sed> addicts, but it's a dirty habit to get into. That's because in
1907 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1908 the usual double-quoted string means a control-A. The customary Unix
1909 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1910 of doing that, you get yourself into trouble if you then add an C</e>
1913 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1919 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1920 C<${1}000>. The operation of interpolation should not be confused
1921 with the operation of matching a backreference. Certainly they mean two
1922 different things on the I<left> side of the C<s///>.
1924 =head2 Repeated Patterns Matching a Zero-length Substring
1926 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1928 Regular expressions provide a terse and powerful programming language. As
1929 with most other power tools, power comes together with the ability
1932 A common abuse of this power stems from the ability to make infinite
1933 loops using regular expressions, with something as innocuous as:
1935 'foo' =~ m{ ( o? )* }x;
1937 The C<o?> matches at the beginning of C<'foo'>, and since the position
1938 in the string is not moved by the match, C<o?> would match again and again
1939 because of the C<*> quantifier. Another common way to create a similar cycle
1940 is with the looping modifier C<//g>:
1942 @matches = ( 'foo' =~ m{ o? }xg );
1946 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1948 or the loop implied by split().
1950 However, long experience has shown that many programming tasks may
1951 be significantly simplified by using repeated subexpressions that
1952 may match zero-length substrings. Here's a simple example being:
1954 @chars = split //, $string; # // is not magic in split
1955 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1957 Thus Perl allows such constructs, by I<forcefully breaking
1958 the infinite loop>. The rules for this are different for lower-level
1959 loops given by the greedy quantifiers C<*+{}>, and for higher-level
1960 ones like the C</g> modifier or split() operator.
1962 The lower-level loops are I<interrupted> (that is, the loop is
1963 broken) when Perl detects that a repeated expression matched a
1964 zero-length substring. Thus
1966 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1968 is made equivalent to
1970 m{ (?: NON_ZERO_LENGTH )*
1975 The higher level-loops preserve an additional state between iterations:
1976 whether the last match was zero-length. To break the loop, the following
1977 match after a zero-length match is prohibited to have a length of zero.
1978 This prohibition interacts with backtracking (see L<"Backtracking">),
1979 and so the I<second best> match is chosen if the I<best> match is of
1987 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1988 match given by non-greedy C<??> is the zero-length match, and the I<second
1989 best> match is what is matched by C<\w>. Thus zero-length matches
1990 alternate with one-character-long matches.
1992 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1993 position one notch further in the string.
1995 The additional state of being I<matched with zero-length> is associated with
1996 the matched string, and is reset by each assignment to pos().
1997 Zero-length matches at the end of the previous match are ignored
2000 =head2 Combining RE Pieces
2002 Each of the elementary pieces of regular expressions which were described
2003 before (such as C<ab> or C<\Z>) could match at most one substring
2004 at the given position of the input string. However, in a typical regular
2005 expression these elementary pieces are combined into more complicated
2006 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
2007 (in these examples C<S> and C<T> are regular subexpressions).
2009 Such combinations can include alternatives, leading to a problem of choice:
2010 if we match a regular expression C<a|ab> against C<"abc">, will it match
2011 substring C<"a"> or C<"ab">? One way to describe which substring is
2012 actually matched is the concept of backtracking (see L<"Backtracking">).
2013 However, this description is too low-level and makes you think
2014 in terms of a particular implementation.
2016 Another description starts with notions of "better"/"worse". All the
2017 substrings which may be matched by the given regular expression can be
2018 sorted from the "best" match to the "worst" match, and it is the "best"
2019 match which is chosen. This substitutes the question of "what is chosen?"
2020 by the question of "which matches are better, and which are worse?".
2022 Again, for elementary pieces there is no such question, since at most
2023 one match at a given position is possible. This section describes the
2024 notion of better/worse for combining operators. In the description
2025 below C<S> and C<T> are regular subexpressions.
2031 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
2032 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
2033 which can be matched by C<T>.
2035 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
2038 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
2039 C<B> is better match for C<T> than C<B'>.
2043 When C<S> can match, it is a better match than when only C<T> can match.
2045 Ordering of two matches for C<S> is the same as for C<S>. Similar for
2046 two matches for C<T>.
2048 =item C<S{REPEAT_COUNT}>
2050 Matches as C<SSS...S> (repeated as many times as necessary).
2054 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
2056 =item C<S{min,max}?>
2058 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
2060 =item C<S?>, C<S*>, C<S+>
2062 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
2064 =item C<S??>, C<S*?>, C<S+?>
2066 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
2070 Matches the best match for C<S> and only that.
2072 =item C<(?=S)>, C<(?<=S)>
2074 Only the best match for C<S> is considered. (This is important only if
2075 C<S> has capturing parentheses, and backreferences are used somewhere
2076 else in the whole regular expression.)
2078 =item C<(?!S)>, C<(?<!S)>
2080 For this grouping operator there is no need to describe the ordering, since
2081 only whether or not C<S> can match is important.
2083 =item C<(??{ EXPR })>, C<(?PARNO)>
2085 The ordering is the same as for the regular expression which is
2086 the result of EXPR, or the pattern contained by capture buffer PARNO.
2088 =item C<(?(condition)yes-pattern|no-pattern)>
2090 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
2091 already determined. The ordering of the matches is the same as for the
2092 chosen subexpression.
2096 The above recipes describe the ordering of matches I<at a given position>.
2097 One more rule is needed to understand how a match is determined for the
2098 whole regular expression: a match at an earlier position is always better
2099 than a match at a later position.
2101 =head2 Creating Custom RE Engines
2103 Overloaded constants (see L<overload>) provide a simple way to extend
2104 the functionality of the RE engine.
2106 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
2107 matches at a boundary between whitespace characters and non-whitespace
2108 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
2109 at these positions, so we want to have each C<\Y|> in the place of the
2110 more complicated version. We can create a module C<customre> to do
2118 die "No argument to customre::import allowed" if @_;
2119 overload::constant 'qr' => \&convert;
2122 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2124 # We must also take care of not escaping the legitimate \\Y|
2125 # sequence, hence the presence of '\\' in the conversion rules.
2126 my %rules = ( '\\' => '\\\\',
2127 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2133 { $rules{$1} or invalid($re,$1) }sgex;
2137 Now C<use customre> enables the new escape in constant regular
2138 expressions, i.e., those without any runtime variable interpolations.
2139 As documented in L<overload>, this conversion will work only over
2140 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
2141 part of this regular expression needs to be converted explicitly
2142 (but only if the special meaning of C<\Y|> should be enabled inside $re):
2147 $re = customre::convert $re;
2150 =head1 PCRE/Python Support
2152 As of Perl 5.10.0, Perl supports several Python/PCRE specific extensions
2153 to the regex syntax. While Perl programmers are encouraged to use the
2154 Perl specific syntax, the following are also accepted:
2158 =item C<< (?PE<lt>NAMEE<gt>pattern) >>
2160 Define a named capture buffer. Equivalent to C<< (?<NAME>pattern) >>.
2162 =item C<< (?P=NAME) >>
2164 Backreference to a named capture buffer. Equivalent to C<< \g{NAME} >>.
2166 =item C<< (?P>NAME) >>
2168 Subroutine call to a named capture buffer. Equivalent to C<< (?&NAME) >>.
2174 This document varies from difficult to understand to completely
2175 and utterly opaque. The wandering prose riddled with jargon is
2176 hard to fathom in several places.
2178 This document needs a rewrite that separates the tutorial content
2179 from the reference content.
2187 L<perlop/"Regexp Quote-Like Operators">.
2189 L<perlop/"Gory details of parsing quoted constructs">.
2199 I<Mastering Regular Expressions> by Jeffrey Friedl, published
2200 by O'Reilly and Associates.