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 following table shows the mapping of POSIX character class
320 names, common escapes, literal escape sequences and their equivalent
321 Unicode style property names.
322 X<character class> X<\p> X<\p{}>
323 X<alpha> X<alnum> X<ascii> X<blank> X<cntrl> X<digit> X<graph>
324 X<lower> X<print> X<punct> X<space> X<upper> X<word> X<xdigit>
326 B<Note:> up to Perl 5.10 the property names used were shared with
327 standard Unicode properties, this was changed in Perl 5.11, see
328 L<perl5110delta> for details.
330 POSIX Esc Class Property Note
331 --------------------------------------------------------
332 alnum [0-9A-Za-z] IsPosixAlnum
333 alpha [A-Za-z] IsPosixAlpha
334 ascii [\000-\177] IsASCII
335 blank [\011 ] IsPosixBlank [1]
336 cntrl [\0-\37\177] IsPosixCntrl
337 digit \d [0-9] IsPosixDigit
338 graph [!-~] IsPosixGraph
339 lower [a-z] IsPosixLower
340 print [ -~] IsPosixPrint
341 punct [!-/:-@[-`{-~] IsPosixPunct
342 space [\11-\15 ] IsPosixSpace [2]
343 \s [\11\12\14\15 ] IsPerlSpace [2]
344 upper [A-Z] IsPosixUpper
345 word \w [0-9A-Z_a-z] IsPerlWord [3]
346 xdigit [0-9A-Fa-f] IsXDigit
352 A GNU extension equivalent to C<[ \t]>, "all horizontal whitespace".
356 Note that C<\s> and C<[[:space:]]> are B<not> equivalent as C<[[:space:]]>
357 includes also the (very rare) "vertical tabulator", "\cK" or chr(11) in
362 A Perl extension, see above.
366 For example use C<[:upper:]> to match all the uppercase characters.
367 Note that the C<[]> are part of the C<[::]> construct, not part of the
368 whole character class. For example:
372 matches zero, one, any alphabetic character, and the percent sign.
380 =item C<+> C<< < >> C<=> C<< > >> C<|> C<~>
386 Modifier symbols (accents)
391 The other named classes are:
398 Any control character. Usually characters that don't produce output as
399 such but instead control the terminal somehow: for example newline and
400 backspace are control characters. All characters with ord() less than
401 32 are usually classified as control characters (assuming ASCII,
402 the ISO Latin character sets, and Unicode), as is the character with
403 the ord() value of 127 (C<DEL>).
408 Any alphanumeric or punctuation (special) character.
413 Any alphanumeric or punctuation (special) character or the space character.
418 Any punctuation (special) character.
423 Any hexadecimal digit. Though this may feel silly ([0-9A-Fa-f] would
424 work just fine) it is included for completeness.
428 You can negate the [::] character classes by prefixing the class name
429 with a '^'. This is a Perl extension. For example:
430 X<character class, negation>
432 POSIX traditional Unicode
434 [[:^digit:]] \D \P{IsPosixDigit}
435 [[:^space:]] \S \P{IsPosixSpace}
436 [[:^word:]] \W \P{IsPerlWord}
438 Perl respects the POSIX standard in that POSIX character classes are
439 only supported within a character class. The POSIX character classes
440 [.cc.] and [=cc=] are recognized but B<not> supported and trying to
441 use them will cause an error.
445 Perl defines the following zero-width assertions:
446 X<zero-width assertion> X<assertion> X<regex, zero-width assertion>
447 X<regexp, zero-width assertion>
448 X<regular expression, zero-width assertion>
449 X<\b> X<\B> X<\A> X<\Z> X<\z> X<\G>
451 \b Match a word boundary
452 \B Match except at a word boundary
453 \A Match only at beginning of string
454 \Z Match only at end of string, or before newline at the end
455 \z Match only at end of string
456 \G Match only at pos() (e.g. at the end-of-match position
459 A word boundary (C<\b>) is a spot between two characters
460 that has a C<\w> on one side of it and a C<\W> on the other side
461 of it (in either order), counting the imaginary characters off the
462 beginning and end of the string as matching a C<\W>. (Within
463 character classes C<\b> represents backspace rather than a word
464 boundary, just as it normally does in any double-quoted string.)
465 The C<\A> and C<\Z> are just like "^" and "$", except that they
466 won't match multiple times when the C</m> modifier is used, while
467 "^" and "$" will match at every internal line boundary. To match
468 the actual end of the string and not ignore an optional trailing
470 X<\b> X<\A> X<\Z> X<\z> X</m>
472 The C<\G> assertion can be used to chain global matches (using
473 C<m//g>), as described in L<perlop/"Regexp Quote-Like Operators">.
474 It is also useful when writing C<lex>-like scanners, when you have
475 several patterns that you want to match against consequent substrings
476 of your string, see the previous reference. The actual location
477 where C<\G> will match can also be influenced by using C<pos()> as
478 an lvalue: see L<perlfunc/pos>. Note that the rule for zero-length
479 matches is modified somewhat, in that contents to the left of C<\G> is
480 not counted when determining the length of the match. Thus the following
481 will not match forever:
490 It will print 'A' and then terminate, as it considers the match to
491 be zero-width, and thus will not match at the same position twice in a
494 It is worth noting that C<\G> improperly used can result in an infinite
495 loop. Take care when using patterns that include C<\G> in an alternation.
497 =head3 Capture buffers
499 The bracketing construct C<( ... )> creates capture buffers. To refer
500 to the current contents of a buffer later on, within the same pattern,
501 use \1 for the first, \2 for the second, and so on.
502 Outside the match use "$" instead of "\". (The
503 \<digit> notation works in certain circumstances outside
504 the match. See the warning below about \1 vs $1 for details.)
505 Referring back to another part of the match is called a
507 X<regex, capture buffer> X<regexp, capture buffer>
508 X<regular expression, capture buffer> X<backreference>
510 There is no limit to the number of captured substrings that you may
511 use. However Perl also uses \10, \11, etc. as aliases for \010,
512 \011, etc. (Recall that 0 means octal, so \011 is the character at
513 number 9 in your coded character set; which would be the 10th character,
514 a horizontal tab under ASCII.) Perl resolves this
515 ambiguity by interpreting \10 as a backreference only if at least 10
516 left parentheses have opened before it. Likewise \11 is a
517 backreference only if at least 11 left parentheses have opened
518 before it. And so on. \1 through \9 are always interpreted as
521 X<\g{1}> X<\g{-1}> X<\g{name}> X<relative backreference> X<named backreference>
522 In order to provide a safer and easier way to construct patterns using
523 backreferences, Perl provides the C<\g{N}> notation (starting with perl
524 5.10.0). The curly brackets are optional, however omitting them is less
525 safe as the meaning of the pattern can be changed by text (such as digits)
526 following it. When N is a positive integer the C<\g{N}> notation is
527 exactly equivalent to using normal backreferences. When N is a negative
528 integer then it is a relative backreference referring to the previous N'th
529 capturing group. When the bracket form is used and N is not an integer, it
530 is treated as a reference to a named buffer.
532 Thus C<\g{-1}> refers to the last buffer, C<\g{-2}> refers to the
533 buffer before that. For example:
539 \g{-1} # backref to buffer 3
540 \g{-3} # backref to buffer 1
544 and would match the same as C</(Y) ( (X) \3 \1 )/x>.
546 Additionally, as of Perl 5.10.0 you may use named capture buffers and named
547 backreferences. The notation is C<< (?<name>...) >> to declare and C<< \k<name> >>
548 to reference. You may also use apostrophes instead of angle brackets to delimit the
549 name; and you may use the bracketed C<< \g{name} >> backreference syntax.
550 It's possible to refer to a named capture buffer by absolute and relative number as well.
551 Outside the pattern, a named capture buffer is available via the C<%+> hash.
552 When different buffers within the same pattern have the same name, C<$+{name}>
553 and C<< \k<name> >> refer to the leftmost defined group. (Thus it's possible
554 to do things with named capture buffers that would otherwise require C<(??{})>
556 X<named capture buffer> X<regular expression, named capture buffer>
557 X<%+> X<$+{name}> X<< \k<name> >>
561 s/^([^ ]*) *([^ ]*)/$2 $1/; # swap first two words
563 /(.)\1/ # find first doubled char
564 and print "'$1' is the first doubled character\n";
566 /(?<char>.)\k<char>/ # ... a different way
567 and print "'$+{char}' is the first doubled character\n";
569 /(?'char'.)\1/ # ... mix and match
570 and print "'$1' is the first doubled character\n";
572 if (/Time: (..):(..):(..)/) { # parse out values
578 Several special variables also refer back to portions of the previous
579 match. C<$+> returns whatever the last bracket match matched.
580 C<$&> returns the entire matched string. (At one point C<$0> did
581 also, but now it returns the name of the program.) C<$`> returns
582 everything before the matched string. C<$'> returns everything
583 after the matched string. And C<$^N> contains whatever was matched by
584 the most-recently closed group (submatch). C<$^N> can be used in
585 extended patterns (see below), for example to assign a submatch to a
587 X<$+> X<$^N> X<$&> X<$`> X<$'>
589 The numbered match variables ($1, $2, $3, etc.) and the related punctuation
590 set (C<$+>, C<$&>, C<$`>, C<$'>, and C<$^N>) are all dynamically scoped
591 until the end of the enclosing block or until the next successful
592 match, whichever comes first. (See L<perlsyn/"Compound Statements">.)
593 X<$+> X<$^N> X<$&> X<$`> X<$'>
594 X<$1> X<$2> X<$3> X<$4> X<$5> X<$6> X<$7> X<$8> X<$9>
597 B<NOTE>: Failed matches in Perl do not reset the match variables,
598 which makes it easier to write code that tests for a series of more
599 specific cases and remembers the best match.
601 B<WARNING>: Once Perl sees that you need one of C<$&>, C<$`>, or
602 C<$'> anywhere in the program, it has to provide them for every
603 pattern match. This may substantially slow your program. Perl
604 uses the same mechanism to produce $1, $2, etc, so you also pay a
605 price for each pattern that contains capturing parentheses. (To
606 avoid this cost while retaining the grouping behaviour, use the
607 extended regular expression C<(?: ... )> instead.) But if you never
608 use C<$&>, C<$`> or C<$'>, then patterns I<without> capturing
609 parentheses will not be penalized. So avoid C<$&>, C<$'>, and C<$`>
610 if you can, but if you can't (and some algorithms really appreciate
611 them), once you've used them once, use them at will, because you've
612 already paid the price. As of 5.005, C<$&> is not so costly as the
616 As a workaround for this problem, Perl 5.10.0 introduces C<${^PREMATCH}>,
617 C<${^MATCH}> and C<${^POSTMATCH}>, which are equivalent to C<$`>, C<$&>
618 and C<$'>, B<except> that they are only guaranteed to be defined after a
619 successful match that was executed with the C</p> (preserve) modifier.
620 The use of these variables incurs no global performance penalty, unlike
621 their punctuation char equivalents, however at the trade-off that you
622 have to tell perl when you want to use them.
625 Backslashed metacharacters in Perl are alphanumeric, such as C<\b>,
626 C<\w>, C<\n>. Unlike some other regular expression languages, there
627 are no backslashed symbols that aren't alphanumeric. So anything
628 that looks like \\, \(, \), \<, \>, \{, or \} is always
629 interpreted as a literal character, not a metacharacter. This was
630 once used in a common idiom to disable or quote the special meanings
631 of regular expression metacharacters in a string that you want to
632 use for a pattern. Simply quote all non-"word" characters:
634 $pattern =~ s/(\W)/\\$1/g;
636 (If C<use locale> is set, then this depends on the current locale.)
637 Today it is more common to use the quotemeta() function or the C<\Q>
638 metaquoting escape sequence to disable all metacharacters' special
641 /$unquoted\Q$quoted\E$unquoted/
643 Beware that if you put literal backslashes (those not inside
644 interpolated variables) between C<\Q> and C<\E>, double-quotish
645 backslash interpolation may lead to confusing results. If you
646 I<need> to use literal backslashes within C<\Q...\E>,
647 consult L<perlop/"Gory details of parsing quoted constructs">.
649 =head2 Extended Patterns
651 Perl also defines a consistent extension syntax for features not
652 found in standard tools like B<awk> and B<lex>. The syntax is a
653 pair of parentheses with a question mark as the first thing within
654 the parentheses. The character after the question mark indicates
657 The stability of these extensions varies widely. Some have been
658 part of the core language for many years. Others are experimental
659 and may change without warning or be completely removed. Check
660 the documentation on an individual feature to verify its current
663 A question mark was chosen for this and for the minimal-matching
664 construct because 1) question marks are rare in older regular
665 expressions, and 2) whenever you see one, you should stop and
666 "question" exactly what is going on. That's psychology...
673 A comment. The text is ignored. If the C</x> modifier enables
674 whitespace formatting, a simple C<#> will suffice. Note that Perl closes
675 the comment as soon as it sees a C<)>, so there is no way to put a literal
678 =item C<(?pimsx-imsx)>
681 One or more embedded pattern-match modifiers, to be turned on (or
682 turned off, if preceded by C<->) for the remainder of the pattern or
683 the remainder of the enclosing pattern group (if any). This is
684 particularly useful for dynamic patterns, such as those read in from a
685 configuration file, taken from an argument, or specified in a table
686 somewhere. Consider the case where some patterns want to be case
687 sensitive and some do not: The case insensitive ones merely need to
688 include C<(?i)> at the front of the pattern. For example:
691 if ( /$pattern/i ) { }
695 $pattern = "(?i)foobar";
696 if ( /$pattern/ ) { }
698 These modifiers are restored at the end of the enclosing group. For example,
702 will match C<blah> in any case, some spaces, and an exact (I<including the case>!)
703 repetition of the previous word, assuming the C</x> modifier, and no C</i>
704 modifier outside this group.
706 These modifiers do not carry over into named subpatterns called in the
707 enclosing group. In other words, a pattern such as C<((?i)(&NAME))> does not
708 change the case-sensitivity of the "NAME" pattern.
710 Note that the C<p> modifier is special in that it can only be enabled,
711 not disabled, and that its presence anywhere in a pattern has a global
712 effect. Thus C<(?-p)> and C<(?-p:...)> are meaningless and will warn
713 when executed under C<use warnings>.
718 =item C<(?imsx-imsx:pattern)>
720 This is for clustering, not capturing; it groups subexpressions like
721 "()", but doesn't make backreferences as "()" does. So
723 @fields = split(/\b(?:a|b|c)\b/)
727 @fields = split(/\b(a|b|c)\b/)
729 but doesn't spit out extra fields. It's also cheaper not to capture
730 characters if you don't need to.
732 Any letters between C<?> and C<:> act as flags modifiers as with
733 C<(?imsx-imsx)>. For example,
735 /(?s-i:more.*than).*million/i
737 is equivalent to the more verbose
739 /(?:(?s-i)more.*than).*million/i
742 X<(?|)> X<Branch reset>
744 This is the "branch reset" pattern, which has the special property
745 that the capture buffers are numbered from the same starting point
746 in each alternation branch. It is available starting from perl 5.10.0.
748 Capture buffers are numbered from left to right, but inside this
749 construct the numbering is restarted for each branch.
751 The numbering within each branch will be as normal, and any buffers
752 following this construct will be numbered as though the construct
753 contained only one branch, that being the one with the most capture
756 This construct will be useful when you want to capture one of a
757 number of alternative matches.
759 Consider the following pattern. The numbers underneath show in
760 which buffer the captured content will be stored.
763 # before ---------------branch-reset----------- after
764 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
767 Note: as of Perl 5.10.0, branch resets interfere with the contents of
768 the C<%+> hash, that holds named captures. Consider using C<%-> instead.
770 =item Look-Around Assertions
771 X<look-around assertion> X<lookaround assertion> X<look-around> X<lookaround>
773 Look-around assertions are zero width patterns which match a specific
774 pattern without including it in C<$&>. Positive assertions match when
775 their subpattern matches, negative assertions match when their subpattern
776 fails. Look-behind matches text up to the current match position,
777 look-ahead matches text following the current match position.
782 X<(?=)> X<look-ahead, positive> X<lookahead, positive>
784 A zero-width positive look-ahead assertion. For example, C</\w+(?=\t)/>
785 matches a word followed by a tab, without including the tab in C<$&>.
788 X<(?!)> X<look-ahead, negative> X<lookahead, negative>
790 A zero-width negative look-ahead assertion. For example C</foo(?!bar)/>
791 matches any occurrence of "foo" that isn't followed by "bar". Note
792 however that look-ahead and look-behind are NOT the same thing. You cannot
793 use this for look-behind.
795 If you are looking for a "bar" that isn't preceded by a "foo", C</(?!foo)bar/>
796 will not do what you want. That's because the C<(?!foo)> is just saying that
797 the next thing cannot be "foo"--and it's not, it's a "bar", so "foobar" will
798 match. You would have to do something like C</(?!foo)...bar/> for that. We
799 say "like" because there's the case of your "bar" not having three characters
800 before it. You could cover that this way: C</(?:(?!foo)...|^.{0,2})bar/>.
801 Sometimes it's still easier just to say:
803 if (/bar/ && $` !~ /foo$/)
805 For look-behind see below.
807 =item C<(?<=pattern)> C<\K>
808 X<(?<=)> X<look-behind, positive> X<lookbehind, positive> X<\K>
810 A zero-width positive look-behind assertion. For example, C</(?<=\t)\w+/>
811 matches a word that follows a tab, without including the tab in C<$&>.
812 Works only for fixed-width look-behind.
814 There is a special form of this construct, called C<\K>, which causes the
815 regex engine to "keep" everything it had matched prior to the C<\K> and
816 not include it in C<$&>. This effectively provides variable length
817 look-behind. The use of C<\K> inside of another look-around assertion
818 is allowed, but the behaviour is currently not well defined.
820 For various reasons C<\K> may be significantly more efficient than the
821 equivalent C<< (?<=...) >> construct, and it is especially useful in
822 situations where you want to efficiently remove something following
823 something else in a string. For instance
827 can be rewritten as the much more efficient
831 =item C<(?<!pattern)>
832 X<(?<!)> X<look-behind, negative> X<lookbehind, negative>
834 A zero-width negative look-behind assertion. For example C</(?<!bar)foo/>
835 matches any occurrence of "foo" that does not follow "bar". Works
836 only for fixed-width look-behind.
840 =item C<(?'NAME'pattern)>
842 =item C<< (?<NAME>pattern) >>
843 X<< (?<NAME>) >> X<(?'NAME')> X<named capture> X<capture>
845 A named capture buffer. Identical in every respect to normal capturing
846 parentheses C<()> but for the additional fact that C<%+> or C<%-> may be
847 used after a successful match to refer to a named buffer. See C<perlvar>
848 for more details on the C<%+> and C<%-> hashes.
850 If multiple distinct capture buffers have the same name then the
851 $+{NAME} will refer to the leftmost defined buffer in the match.
853 The forms C<(?'NAME'pattern)> and C<< (?<NAME>pattern) >> are equivalent.
855 B<NOTE:> While the notation of this construct is the same as the similar
856 function in .NET regexes, the behavior is not. In Perl the buffers are
857 numbered sequentially regardless of being named or not. Thus in the
862 $+{foo} will be the same as $2, and $3 will contain 'z' instead of
863 the opposite which is what a .NET regex hacker might expect.
865 Currently NAME is restricted to simple identifiers only.
866 In other words, it must match C</^[_A-Za-z][_A-Za-z0-9]*\z/> or
867 its Unicode extension (see L<utf8>),
868 though it isn't extended by the locale (see L<perllocale>).
870 B<NOTE:> In order to make things easier for programmers with experience
871 with the Python or PCRE regex engines, the pattern C<< (?PE<lt>NAMEE<gt>pattern) >>
872 may be used instead of C<< (?<NAME>pattern) >>; however this form does not
873 support the use of single quotes as a delimiter for the name.
875 =item C<< \k<NAME> >>
877 =item C<< \k'NAME' >>
879 Named backreference. Similar to numeric backreferences, except that
880 the group is designated by name and not number. If multiple groups
881 have the same name then it refers to the leftmost defined group in
884 It is an error to refer to a name not defined by a C<< (?<NAME>) >>
885 earlier in the pattern.
887 Both forms are equivalent.
889 B<NOTE:> In order to make things easier for programmers with experience
890 with the Python or PCRE regex engines, the pattern C<< (?P=NAME) >>
891 may be used instead of C<< \k<NAME> >>.
894 X<(?{})> X<regex, code in> X<regexp, code in> X<regular expression, code in>
896 B<WARNING>: This extended regular expression feature is considered
897 experimental, and may be changed without notice. Code executed that
898 has side effects may not perform identically from version to version
899 due to the effect of future optimisations in the regex engine.
901 This zero-width assertion evaluates any embedded Perl code. It
902 always succeeds, and its C<code> is not interpolated. Currently,
903 the rules to determine where the C<code> ends are somewhat convoluted.
905 This feature can be used together with the special variable C<$^N> to
906 capture the results of submatches in variables without having to keep
907 track of the number of nested parentheses. For example:
909 $_ = "The brown fox jumps over the lazy dog";
910 /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
911 print "color = $color, animal = $animal\n";
913 Inside the C<(?{...})> block, C<$_> refers to the string the regular
914 expression is matching against. You can also use C<pos()> to know what is
915 the current position of matching within this string.
917 The C<code> is properly scoped in the following sense: If the assertion
918 is backtracked (compare L<"Backtracking">), all changes introduced after
919 C<local>ization are undone, so that
923 (?{ $cnt = 0 }) # Initialize $cnt.
927 local $cnt = $cnt + 1; # Update $cnt, backtracking-safe.
931 (?{ $res = $cnt }) # On success copy to non-localized
935 will set C<$res = 4>. Note that after the match, C<$cnt> returns to the globally
936 introduced value, because the scopes that restrict C<local> operators
939 This assertion may be used as a C<(?(condition)yes-pattern|no-pattern)>
940 switch. If I<not> used in this way, the result of evaluation of
941 C<code> is put into the special variable C<$^R>. This happens
942 immediately, so C<$^R> can be used from other C<(?{ code })> assertions
943 inside the same regular expression.
945 The assignment to C<$^R> above is properly localized, so the old
946 value of C<$^R> is restored if the assertion is backtracked; compare
949 Due to an unfortunate implementation issue, the Perl code contained in these
950 blocks is treated as a compile time closure that can have seemingly bizarre
951 consequences when used with lexically scoped variables inside of subroutines
952 or loops. There are various workarounds for this, including simply using
953 global variables instead. If you are using this construct and strange results
954 occur then check for the use of lexically scoped variables.
956 For reasons of security, this construct is forbidden if the regular
957 expression involves run-time interpolation of variables, unless the
958 perilous C<use re 'eval'> pragma has been used (see L<re>), or the
959 variables contain results of C<qr//> operator (see
960 L<perlop/"qr/STRING/imosx">).
962 This restriction is due to the wide-spread and remarkably convenient
963 custom of using run-time determined strings as patterns. For example:
969 Before Perl knew how to execute interpolated code within a pattern,
970 this operation was completely safe from a security point of view,
971 although it could raise an exception from an illegal pattern. If
972 you turn on the C<use re 'eval'>, though, it is no longer secure,
973 so you should only do so if you are also using taint checking.
974 Better yet, use the carefully constrained evaluation within a Safe
975 compartment. See L<perlsec> for details about both these mechanisms.
977 Because Perl's regex engine is currently not re-entrant, interpolated
978 code may not invoke the regex engine either directly with C<m//> or C<s///>),
979 or indirectly with functions such as C<split>.
981 =item C<(??{ code })>
983 X<regex, postponed> X<regexp, postponed> X<regular expression, postponed>
985 B<WARNING>: This extended regular expression feature is considered
986 experimental, and may be changed without notice. Code executed that
987 has side effects may not perform identically from version to version
988 due to the effect of future optimisations in the regex engine.
990 This is a "postponed" regular subexpression. The C<code> is evaluated
991 at run time, at the moment this subexpression may match. The result
992 of evaluation is considered as a regular expression and matched as
993 if it were inserted instead of this construct. Note that this means
994 that the contents of capture buffers defined inside an eval'ed pattern
995 are not available outside of the pattern, and vice versa, there is no
996 way for the inner pattern to refer to a capture buffer defined outside.
999 ('a' x 100)=~/(??{'(.)' x 100})/
1001 B<will> match, it will B<not> set $1.
1003 The C<code> is not interpolated. As before, the rules to determine
1004 where the C<code> ends are currently somewhat convoluted.
1006 The following pattern matches a parenthesized group:
1011 (?> [^()]+ ) # Non-parens without backtracking
1013 (??{ $re }) # Group with matching parens
1018 See also C<(?PARNO)> for a different, more efficient way to accomplish
1021 Because perl's regex engine is not currently re-entrant, delayed
1022 code may not invoke the regex engine either directly with C<m//> or C<s///>),
1023 or indirectly with functions such as C<split>.
1025 Recursing deeper than 50 times without consuming any input string will
1026 result in a fatal error. The maximum depth is compiled into perl, so
1027 changing it requires a custom build.
1029 =item C<(?PARNO)> C<(?-PARNO)> C<(?+PARNO)> C<(?R)> C<(?0)>
1030 X<(?PARNO)> X<(?1)> X<(?R)> X<(?0)> X<(?-1)> X<(?+1)> X<(?-PARNO)> X<(?+PARNO)>
1031 X<regex, recursive> X<regexp, recursive> X<regular expression, recursive>
1032 X<regex, relative recursion>
1034 Similar to C<(??{ code })> except it does not involve compiling any code,
1035 instead it treats the contents of a capture buffer as an independent
1036 pattern that must match at the current position. Capture buffers
1037 contained by the pattern will have the value as determined by the
1038 outermost recursion.
1040 PARNO is a sequence of digits (not starting with 0) whose value reflects
1041 the paren-number of the capture buffer to recurse to. C<(?R)> recurses to
1042 the beginning of the whole pattern. C<(?0)> is an alternate syntax for
1043 C<(?R)>. If PARNO is preceded by a plus or minus sign then it is assumed
1044 to be relative, with negative numbers indicating preceding capture buffers
1045 and positive ones following. Thus C<(?-1)> refers to the most recently
1046 declared buffer, and C<(?+1)> indicates the next buffer to be declared.
1047 Note that the counting for relative recursion differs from that of
1048 relative backreferences, in that with recursion unclosed buffers B<are>
1051 The following pattern matches a function foo() which may contain
1052 balanced parentheses as the argument.
1054 $re = qr{ ( # paren group 1 (full function)
1056 ( # paren group 2 (parens)
1058 ( # paren group 3 (contents of parens)
1060 (?> [^()]+ ) # Non-parens without backtracking
1062 (?2) # Recurse to start of paren group 2
1070 If the pattern was used as follows
1072 'foo(bar(baz)+baz(bop))'=~/$re/
1073 and print "\$1 = $1\n",
1077 the output produced should be the following:
1079 $1 = foo(bar(baz)+baz(bop))
1080 $2 = (bar(baz)+baz(bop))
1081 $3 = bar(baz)+baz(bop)
1083 If there is no corresponding capture buffer defined, then it is a
1084 fatal error. Recursing deeper than 50 times without consuming any input
1085 string will also result in a fatal error. The maximum depth is compiled
1086 into perl, so changing it requires a custom build.
1088 The following shows how using negative indexing can make it
1089 easier to embed recursive patterns inside of a C<qr//> construct
1092 my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1093 if (/foo $parens \s+ + \s+ bar $parens/x) {
1094 # do something here...
1097 B<Note> that this pattern does not behave the same way as the equivalent
1098 PCRE or Python construct of the same form. In Perl you can backtrack into
1099 a recursed group, in PCRE and Python the recursed into group is treated
1100 as atomic. Also, modifiers are resolved at compile time, so constructs
1101 like (?i:(?1)) or (?:(?i)(?1)) do not affect how the sub-pattern will
1107 Recurse to a named subpattern. Identical to C<(?PARNO)> except that the
1108 parenthesis to recurse to is determined by name. If multiple parentheses have
1109 the same name, then it recurses to the leftmost.
1111 It is an error to refer to a name that is not declared somewhere in the
1114 B<NOTE:> In order to make things easier for programmers with experience
1115 with the Python or PCRE regex engines the pattern C<< (?P>NAME) >>
1116 may be used instead of C<< (?&NAME) >>.
1118 =item C<(?(condition)yes-pattern|no-pattern)>
1121 =item C<(?(condition)yes-pattern)>
1123 Conditional expression. C<(condition)> should be either an integer in
1124 parentheses (which is valid if the corresponding pair of parentheses
1125 matched), a look-ahead/look-behind/evaluate zero-width assertion, a
1126 name in angle brackets or single quotes (which is valid if a buffer
1127 with the given name matched), or the special symbol (R) (true when
1128 evaluated inside of recursion or eval). Additionally the R may be
1129 followed by a number, (which will be true when evaluated when recursing
1130 inside of the appropriate group), or by C<&NAME>, in which case it will
1131 be true only when evaluated during recursion in the named group.
1133 Here's a summary of the possible predicates:
1139 Checks if the numbered capturing buffer has matched something.
1141 =item (<NAME>) ('NAME')
1143 Checks if a buffer with the given name has matched something.
1147 Treats the code block as the condition.
1151 Checks if the expression has been evaluated inside of recursion.
1155 Checks if the expression has been evaluated while executing directly
1156 inside of the n-th capture group. This check is the regex equivalent of
1158 if ((caller(0))[3] eq 'subname') { ... }
1160 In other words, it does not check the full recursion stack.
1164 Similar to C<(R1)>, this predicate checks to see if we're executing
1165 directly inside of the leftmost group with a given name (this is the same
1166 logic used by C<(?&NAME)> to disambiguate). It does not check the full
1167 stack, but only the name of the innermost active recursion.
1171 In this case, the yes-pattern is never directly executed, and no
1172 no-pattern is allowed. Similar in spirit to C<(?{0})> but more efficient.
1173 See below for details.
1184 matches a chunk of non-parentheses, possibly included in parentheses
1187 A special form is the C<(DEFINE)> predicate, which never executes directly
1188 its yes-pattern, and does not allow a no-pattern. This allows to define
1189 subpatterns which will be executed only by using the recursion mechanism.
1190 This way, you can define a set of regular expression rules that can be
1191 bundled into any pattern you choose.
1193 It is recommended that for this usage you put the DEFINE block at the
1194 end of the pattern, and that you name any subpatterns defined within it.
1196 Also, it's worth noting that patterns defined this way probably will
1197 not be as efficient, as the optimiser is not very clever about
1200 An example of how this might be used is as follows:
1202 /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
1208 Note that capture buffers matched inside of recursion are not accessible
1209 after the recursion returns, so the extra layer of capturing buffers is
1210 necessary. Thus C<$+{NAME_PAT}> would not be defined even though
1211 C<$+{NAME}> would be.
1213 =item C<< (?>pattern) >>
1214 X<backtrack> X<backtracking> X<atomic> X<possessive>
1216 An "independent" subexpression, one which matches the substring
1217 that a I<standalone> C<pattern> would match if anchored at the given
1218 position, and it matches I<nothing other than this substring>. This
1219 construct is useful for optimizations of what would otherwise be
1220 "eternal" matches, because it will not backtrack (see L<"Backtracking">).
1221 It may also be useful in places where the "grab all you can, and do not
1222 give anything back" semantic is desirable.
1224 For example: C<< ^(?>a*)ab >> will never match, since C<< (?>a*) >>
1225 (anchored at the beginning of string, as above) will match I<all>
1226 characters C<a> at the beginning of string, leaving no C<a> for
1227 C<ab> to match. In contrast, C<a*ab> will match the same as C<a+b>,
1228 since the match of the subgroup C<a*> is influenced by the following
1229 group C<ab> (see L<"Backtracking">). In particular, C<a*> inside
1230 C<a*ab> will match fewer characters than a standalone C<a*>, since
1231 this makes the tail match.
1233 An effect similar to C<< (?>pattern) >> may be achieved by writing
1234 C<(?=(pattern))\1>. This matches the same substring as a standalone
1235 C<a+>, and the following C<\1> eats the matched string; it therefore
1236 makes a zero-length assertion into an analogue of C<< (?>...) >>.
1237 (The difference between these two constructs is that the second one
1238 uses a capturing group, thus shifting ordinals of backreferences
1239 in the rest of a regular expression.)
1241 Consider this pattern:
1252 That will efficiently match a nonempty group with matching parentheses
1253 two levels deep or less. However, if there is no such group, it
1254 will take virtually forever on a long string. That's because there
1255 are so many different ways to split a long string into several
1256 substrings. This is what C<(.+)+> is doing, and C<(.+)+> is similar
1257 to a subpattern of the above pattern. Consider how the pattern
1258 above detects no-match on C<((()aaaaaaaaaaaaaaaaaa> in several
1259 seconds, but that each extra letter doubles this time. This
1260 exponential performance will make it appear that your program has
1261 hung. However, a tiny change to this pattern
1265 (?> [^()]+ ) # change x+ above to (?> x+ )
1272 which uses C<< (?>...) >> matches exactly when the one above does (verifying
1273 this yourself would be a productive exercise), but finishes in a fourth
1274 the time when used on a similar string with 1000000 C<a>s. Be aware,
1275 however, that this pattern currently triggers a warning message under
1276 the C<use warnings> pragma or B<-w> switch saying it
1277 C<"matches null string many times in regex">.
1279 On simple groups, such as the pattern C<< (?> [^()]+ ) >>, a comparable
1280 effect may be achieved by negative look-ahead, as in C<[^()]+ (?! [^()] )>.
1281 This was only 4 times slower on a string with 1000000 C<a>s.
1283 The "grab all you can, and do not give anything back" semantic is desirable
1284 in many situations where on the first sight a simple C<()*> looks like
1285 the correct solution. Suppose we parse text with comments being delimited
1286 by C<#> followed by some optional (horizontal) whitespace. Contrary to
1287 its appearance, C<#[ \t]*> I<is not> the correct subexpression to match
1288 the comment delimiter, because it may "give up" some whitespace if
1289 the remainder of the pattern can be made to match that way. The correct
1290 answer is either one of these:
1295 For example, to grab non-empty comments into $1, one should use either
1298 / (?> \# [ \t]* ) ( .+ ) /x;
1299 / \# [ \t]* ( [^ \t] .* ) /x;
1301 Which one you pick depends on which of these expressions better reflects
1302 the above specification of comments.
1304 In some literature this construct is called "atomic matching" or
1305 "possessive matching".
1307 Possessive quantifiers are equivalent to putting the item they are applied
1308 to inside of one of these constructs. The following equivalences apply:
1310 Quantifier Form Bracketing Form
1311 --------------- ---------------
1315 PAT{min,max}+ (?>PAT{min,max})
1319 =head2 Special Backtracking Control Verbs
1321 B<WARNING:> These patterns are experimental and subject to change or
1322 removal in a future version of Perl. Their usage in production code should
1323 be noted to avoid problems during upgrades.
1325 These special patterns are generally of the form C<(*VERB:ARG)>. Unless
1326 otherwise stated the ARG argument is optional; in some cases, it is
1329 Any pattern containing a special backtracking verb that allows an argument
1330 has the special behaviour that when executed it sets the current packages'
1331 C<$REGERROR> and C<$REGMARK> variables. When doing so the following
1334 On failure, the C<$REGERROR> variable will be set to the ARG value of the
1335 verb pattern, if the verb was involved in the failure of the match. If the
1336 ARG part of the pattern was omitted, then C<$REGERROR> will be set to the
1337 name of the last C<(*MARK:NAME)> pattern executed, or to TRUE if there was
1338 none. Also, the C<$REGMARK> variable will be set to FALSE.
1340 On a successful match, the C<$REGERROR> variable will be set to FALSE, and
1341 the C<$REGMARK> variable will be set to the name of the last
1342 C<(*MARK:NAME)> pattern executed. See the explanation for the
1343 C<(*MARK:NAME)> verb below for more details.
1345 B<NOTE:> C<$REGERROR> and C<$REGMARK> are not magic variables like C<$1>
1346 and most other regex related variables. They are not local to a scope, nor
1347 readonly, but instead are volatile package variables similar to C<$AUTOLOAD>.
1348 Use C<local> to localize changes to them to a specific scope if necessary.
1350 If a pattern does not contain a special backtracking verb that allows an
1351 argument, then C<$REGERROR> and C<$REGMARK> are not touched at all.
1355 =item Verbs that take an argument
1359 =item C<(*PRUNE)> C<(*PRUNE:NAME)>
1360 X<(*PRUNE)> X<(*PRUNE:NAME)>
1362 This zero-width pattern prunes the backtracking tree at the current point
1363 when backtracked into on failure. Consider the pattern C<A (*PRUNE) B>,
1364 where A and B are complex patterns. Until the C<(*PRUNE)> verb is reached,
1365 A may backtrack as necessary to match. Once it is reached, matching
1366 continues in B, which may also backtrack as necessary; however, should B
1367 not match, then no further backtracking will take place, and the pattern
1368 will fail outright at the current starting position.
1370 The following example counts all the possible matching strings in a
1371 pattern (without actually matching any of them).
1373 'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
1374 print "Count=$count\n";
1389 If we add a C<(*PRUNE)> before the count like the following
1391 'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
1392 print "Count=$count\n";
1394 we prevent backtracking and find the count of the longest matching
1395 at each matching starting point like so:
1402 Any number of C<(*PRUNE)> assertions may be used in a pattern.
1404 See also C<< (?>pattern) >> and possessive quantifiers for other ways to
1405 control backtracking. In some cases, the use of C<(*PRUNE)> can be
1406 replaced with a C<< (?>pattern) >> with no functional difference; however,
1407 C<(*PRUNE)> can be used to handle cases that cannot be expressed using a
1408 C<< (?>pattern) >> alone.
1411 =item C<(*SKIP)> C<(*SKIP:NAME)>
1414 This zero-width pattern is similar to C<(*PRUNE)>, except that on
1415 failure it also signifies that whatever text that was matched leading up
1416 to the C<(*SKIP)> pattern being executed cannot be part of I<any> match
1417 of this pattern. This effectively means that the regex engine "skips" forward
1418 to this position on failure and tries to match again, (assuming that
1419 there is sufficient room to match).
1421 The name of the C<(*SKIP:NAME)> pattern has special significance. If a
1422 C<(*MARK:NAME)> was encountered while matching, then it is that position
1423 which is used as the "skip point". If no C<(*MARK)> of that name was
1424 encountered, then the C<(*SKIP)> operator has no effect. When used
1425 without a name the "skip point" is where the match point was when
1426 executing the (*SKIP) pattern.
1428 Compare the following to the examples in C<(*PRUNE)>, note the string
1431 'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
1432 print "Count=$count\n";
1440 Once the 'aaab' at the start of the string has matched, and the C<(*SKIP)>
1441 executed, the next starting point will be where the cursor was when the
1442 C<(*SKIP)> was executed.
1444 =item C<(*MARK:NAME)> C<(*:NAME)>
1445 X<(*MARK)> C<(*MARK:NAME)> C<(*:NAME)>
1447 This zero-width pattern can be used to mark the point reached in a string
1448 when a certain part of the pattern has been successfully matched. This
1449 mark may be given a name. A later C<(*SKIP)> pattern will then skip
1450 forward to that point if backtracked into on failure. Any number of
1451 C<(*MARK)> patterns are allowed, and the NAME portion is optional and may
1454 In addition to interacting with the C<(*SKIP)> pattern, C<(*MARK:NAME)>
1455 can be used to "label" a pattern branch, so that after matching, the
1456 program can determine which branches of the pattern were involved in the
1459 When a match is successful, the C<$REGMARK> variable will be set to the
1460 name of the most recently executed C<(*MARK:NAME)> that was involved
1463 This can be used to determine which branch of a pattern was matched
1464 without using a separate capture buffer for each branch, which in turn
1465 can result in a performance improvement, as perl cannot optimize
1466 C</(?:(x)|(y)|(z))/> as efficiently as something like
1467 C</(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/>.
1469 When a match has failed, and unless another verb has been involved in
1470 failing the match and has provided its own name to use, the C<$REGERROR>
1471 variable will be set to the name of the most recently executed
1474 See C<(*SKIP)> for more details.
1476 As a shortcut C<(*MARK:NAME)> can be written C<(*:NAME)>.
1478 =item C<(*THEN)> C<(*THEN:NAME)>
1480 This is similar to the "cut group" operator C<::> from Perl 6. Like
1481 C<(*PRUNE)>, this verb always matches, and when backtracked into on
1482 failure, it causes the regex engine to try the next alternation in the
1483 innermost enclosing group (capturing or otherwise).
1485 Its name comes from the observation that this operation combined with the
1486 alternation operator (C<|>) can be used to create what is essentially a
1487 pattern-based if/then/else block:
1489 ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
1491 Note that if this operator is used and NOT inside of an alternation then
1492 it acts exactly like the C<(*PRUNE)> operator.
1502 / ( A (*THEN) B | C (*THEN) D ) /
1506 / ( A (*PRUNE) B | C (*PRUNE) D ) /
1508 as after matching the A but failing on the B the C<(*THEN)> verb will
1509 backtrack and try C; but the C<(*PRUNE)> verb will simply fail.
1514 This is the Perl 6 "commit pattern" C<< <commit> >> or C<:::>. It's a
1515 zero-width pattern similar to C<(*SKIP)>, except that when backtracked
1516 into on failure it causes the match to fail outright. No further attempts
1517 to find a valid match by advancing the start pointer will occur again.
1520 'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
1521 print "Count=$count\n";
1528 In other words, once the C<(*COMMIT)> has been entered, and if the pattern
1529 does not match, the regex engine will not try any further matching on the
1534 =item Verbs without an argument
1538 =item C<(*FAIL)> C<(*F)>
1541 This pattern matches nothing and always fails. It can be used to force the
1542 engine to backtrack. It is equivalent to C<(?!)>, but easier to read. In
1543 fact, C<(?!)> gets optimised into C<(*FAIL)> internally.
1545 It is probably useful only when combined with C<(?{})> or C<(??{})>.
1550 B<WARNING:> This feature is highly experimental. It is not recommended
1551 for production code.
1553 This pattern matches nothing and causes the end of successful matching at
1554 the point at which the C<(*ACCEPT)> pattern was encountered, regardless of
1555 whether there is actually more to match in the string. When inside of a
1556 nested pattern, such as recursion, or in a subpattern dynamically generated
1557 via C<(??{})>, only the innermost pattern is ended immediately.
1559 If the C<(*ACCEPT)> is inside of capturing buffers then the buffers are
1560 marked as ended at the point at which the C<(*ACCEPT)> was encountered.
1563 'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
1565 will match, and C<$1> will be C<AB> and C<$2> will be C<B>, C<$3> will not
1566 be set. If another branch in the inner parentheses were matched, such as in the
1567 string 'ACDE', then the C<D> and C<E> would have to be matched as well.
1574 X<backtrack> X<backtracking>
1576 NOTE: This section presents an abstract approximation of regular
1577 expression behavior. For a more rigorous (and complicated) view of
1578 the rules involved in selecting a match among possible alternatives,
1579 see L<Combining RE Pieces>.
1581 A fundamental feature of regular expression matching involves the
1582 notion called I<backtracking>, which is currently used (when needed)
1583 by all regular non-possessive expression quantifiers, namely C<*>, C<*?>, C<+>,
1584 C<+?>, C<{n,m}>, and C<{n,m}?>. Backtracking is often optimized
1585 internally, but the general principle outlined here is valid.
1587 For a regular expression to match, the I<entire> regular expression must
1588 match, not just part of it. So if the beginning of a pattern containing a
1589 quantifier succeeds in a way that causes later parts in the pattern to
1590 fail, the matching engine backs up and recalculates the beginning
1591 part--that's why it's called backtracking.
1593 Here is an example of backtracking: Let's say you want to find the
1594 word following "foo" in the string "Food is on the foo table.":
1596 $_ = "Food is on the foo table.";
1597 if ( /\b(foo)\s+(\w+)/i ) {
1598 print "$2 follows $1.\n";
1601 When the match runs, the first part of the regular expression (C<\b(foo)>)
1602 finds a possible match right at the beginning of the string, and loads up
1603 $1 with "Foo". However, as soon as the matching engine sees that there's
1604 no whitespace following the "Foo" that it had saved in $1, it realizes its
1605 mistake and starts over again one character after where it had the
1606 tentative match. This time it goes all the way until the next occurrence
1607 of "foo". The complete regular expression matches this time, and you get
1608 the expected output of "table follows foo."
1610 Sometimes minimal matching can help a lot. Imagine you'd like to match
1611 everything between "foo" and "bar". Initially, you write something
1614 $_ = "The food is under the bar in the barn.";
1615 if ( /foo(.*)bar/ ) {
1619 Which perhaps unexpectedly yields:
1621 got <d is under the bar in the >
1623 That's because C<.*> was greedy, so you get everything between the
1624 I<first> "foo" and the I<last> "bar". Here it's more effective
1625 to use minimal matching to make sure you get the text between a "foo"
1626 and the first "bar" thereafter.
1628 if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
1629 got <d is under the >
1631 Here's another example. Let's say you'd like to match a number at the end
1632 of a string, and you also want to keep the preceding part of the match.
1635 $_ = "I have 2 numbers: 53147";
1636 if ( /(.*)(\d*)/ ) { # Wrong!
1637 print "Beginning is <$1>, number is <$2>.\n";
1640 That won't work at all, because C<.*> was greedy and gobbled up the
1641 whole string. As C<\d*> can match on an empty string the complete
1642 regular expression matched successfully.
1644 Beginning is <I have 2 numbers: 53147>, number is <>.
1646 Here are some variants, most of which don't work:
1648 $_ = "I have 2 numbers: 53147";
1661 printf "%-12s ", $pat;
1663 print "<$1> <$2>\n";
1669 That will print out:
1671 (.*)(\d*) <I have 2 numbers: 53147> <>
1672 (.*)(\d+) <I have 2 numbers: 5314> <7>
1674 (.*?)(\d+) <I have > <2>
1675 (.*)(\d+)$ <I have 2 numbers: 5314> <7>
1676 (.*?)(\d+)$ <I have 2 numbers: > <53147>
1677 (.*)\b(\d+)$ <I have 2 numbers: > <53147>
1678 (.*\D)(\d+)$ <I have 2 numbers: > <53147>
1680 As you see, this can be a bit tricky. It's important to realize that a
1681 regular expression is merely a set of assertions that gives a definition
1682 of success. There may be 0, 1, or several different ways that the
1683 definition might succeed against a particular string. And if there are
1684 multiple ways it might succeed, you need to understand backtracking to
1685 know which variety of success you will achieve.
1687 When using look-ahead assertions and negations, this can all get even
1688 trickier. Imagine you'd like to find a sequence of non-digits not
1689 followed by "123". You might try to write that as
1692 if ( /^\D*(?!123)/ ) { # Wrong!
1693 print "Yup, no 123 in $_\n";
1696 But that isn't going to match; at least, not the way you're hoping. It
1697 claims that there is no 123 in the string. Here's a clearer picture of
1698 why that pattern matches, contrary to popular expectations:
1703 print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
1704 print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
1706 print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
1707 print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
1715 You might have expected test 3 to fail because it seems to a more
1716 general purpose version of test 1. The important difference between
1717 them is that test 3 contains a quantifier (C<\D*>) and so can use
1718 backtracking, whereas test 1 will not. What's happening is
1719 that you've asked "Is it true that at the start of $x, following 0 or more
1720 non-digits, you have something that's not 123?" If the pattern matcher had
1721 let C<\D*> expand to "ABC", this would have caused the whole pattern to
1724 The search engine will initially match C<\D*> with "ABC". Then it will
1725 try to match C<(?!123> with "123", which fails. But because
1726 a quantifier (C<\D*>) has been used in the regular expression, the
1727 search engine can backtrack and retry the match differently
1728 in the hope of matching the complete regular expression.
1730 The pattern really, I<really> wants to succeed, so it uses the
1731 standard pattern back-off-and-retry and lets C<\D*> expand to just "AB" this
1732 time. Now there's indeed something following "AB" that is not
1733 "123". It's "C123", which suffices.
1735 We can deal with this by using both an assertion and a negation.
1736 We'll say that the first part in $1 must be followed both by a digit
1737 and by something that's not "123". Remember that the look-aheads
1738 are zero-width expressions--they only look, but don't consume any
1739 of the string in their match. So rewriting this way produces what
1740 you'd expect; that is, case 5 will fail, but case 6 succeeds:
1742 print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
1743 print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
1747 In other words, the two zero-width assertions next to each other work as though
1748 they're ANDed together, just as you'd use any built-in assertions: C</^$/>
1749 matches only if you're at the beginning of the line AND the end of the
1750 line simultaneously. The deeper underlying truth is that juxtaposition in
1751 regular expressions always means AND, except when you write an explicit OR
1752 using the vertical bar. C</ab/> means match "a" AND (then) match "b",
1753 although the attempted matches are made at different positions because "a"
1754 is not a zero-width assertion, but a one-width assertion.
1756 B<WARNING>: Particularly complicated regular expressions can take
1757 exponential time to solve because of the immense number of possible
1758 ways they can use backtracking to try for a match. For example, without
1759 internal optimizations done by the regular expression engine, this will
1760 take a painfully long time to run:
1762 'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
1764 And if you used C<*>'s in the internal groups instead of limiting them
1765 to 0 through 5 matches, then it would take forever--or until you ran
1766 out of stack space. Moreover, these internal optimizations are not
1767 always applicable. For example, if you put C<{0,5}> instead of C<*>
1768 on the external group, no current optimization is applicable, and the
1769 match takes a long time to finish.
1771 A powerful tool for optimizing such beasts is what is known as an
1772 "independent group",
1773 which does not backtrack (see L<C<< (?>pattern) >>>). Note also that
1774 zero-length look-ahead/look-behind assertions will not backtrack to make
1775 the tail match, since they are in "logical" context: only
1776 whether they match is considered relevant. For an example
1777 where side-effects of look-ahead I<might> have influenced the
1778 following match, see L<C<< (?>pattern) >>>.
1780 =head2 Version 8 Regular Expressions
1781 X<regular expression, version 8> X<regex, version 8> X<regexp, version 8>
1783 In case you're not familiar with the "regular" Version 8 regex
1784 routines, here are the pattern-matching rules not described above.
1786 Any single character matches itself, unless it is a I<metacharacter>
1787 with a special meaning described here or above. You can cause
1788 characters that normally function as metacharacters to be interpreted
1789 literally by prefixing them with a "\" (e.g., "\." matches a ".", not any
1790 character; "\\" matches a "\"). This escape mechanism is also required
1791 for the character used as the pattern delimiter.
1793 A series of characters matches that series of characters in the target
1794 string, so the pattern C<blurfl> would match "blurfl" in the target
1797 You can specify a character class, by enclosing a list of characters
1798 in C<[]>, which will match any character from the list. If the
1799 first character after the "[" is "^", the class matches any character not
1800 in the list. Within a list, the "-" character specifies a
1801 range, so that C<a-z> represents all characters between "a" and "z",
1802 inclusive. If you want either "-" or "]" itself to be a member of a
1803 class, put it at the start of the list (possibly after a "^"), or
1804 escape it with a backslash. "-" is also taken literally when it is
1805 at the end of the list, just before the closing "]". (The
1806 following all specify the same class of three characters: C<[-az]>,
1807 C<[az-]>, and C<[a\-z]>. All are different from C<[a-z]>, which
1808 specifies a class containing twenty-six characters, even on EBCDIC-based
1809 character sets.) Also, if you try to use the character
1810 classes C<\w>, C<\W>, C<\s>, C<\S>, C<\d>, or C<\D> as endpoints of
1811 a range, the "-" is understood literally.
1813 Note also that the whole range idea is rather unportable between
1814 character sets--and even within character sets they may cause results
1815 you probably didn't expect. A sound principle is to use only ranges
1816 that begin from and end at either alphabetics of equal case ([a-e],
1817 [A-E]), or digits ([0-9]). Anything else is unsafe. If in doubt,
1818 spell out the character sets in full.
1820 Characters may be specified using a metacharacter syntax much like that
1821 used in C: "\n" matches a newline, "\t" a tab, "\r" a carriage return,
1822 "\f" a form feed, etc. More generally, \I<nnn>, where I<nnn> is a string
1823 of octal digits, matches the character whose coded character set value
1824 is I<nnn>. Similarly, \xI<nn>, where I<nn> are hexadecimal digits,
1825 matches the character whose numeric value is I<nn>. The expression \cI<x>
1826 matches the character control-I<x>. Finally, the "." metacharacter
1827 matches any character except "\n" (unless you use C</s>).
1829 You can specify a series of alternatives for a pattern using "|" to
1830 separate them, so that C<fee|fie|foe> will match any of "fee", "fie",
1831 or "foe" in the target string (as would C<f(e|i|o)e>). The
1832 first alternative includes everything from the last pattern delimiter
1833 ("(", "[", or the beginning of the pattern) up to the first "|", and
1834 the last alternative contains everything from the last "|" to the next
1835 pattern delimiter. That's why it's common practice to include
1836 alternatives in parentheses: to minimize confusion about where they
1839 Alternatives are tried from left to right, so the first
1840 alternative found for which the entire expression matches, is the one that
1841 is chosen. This means that alternatives are not necessarily greedy. For
1842 example: when matching C<foo|foot> against "barefoot", only the "foo"
1843 part will match, as that is the first alternative tried, and it successfully
1844 matches the target string. (This might not seem important, but it is
1845 important when you are capturing matched text using parentheses.)
1847 Also remember that "|" is interpreted as a literal within square brackets,
1848 so if you write C<[fee|fie|foe]> you're really only matching C<[feio|]>.
1850 Within a pattern, you may designate subpatterns for later reference
1851 by enclosing them in parentheses, and you may refer back to the
1852 I<n>th subpattern later in the pattern using the metacharacter
1853 \I<n>. Subpatterns are numbered based on the left to right order
1854 of their opening parenthesis. A backreference matches whatever
1855 actually matched the subpattern in the string being examined, not
1856 the rules for that subpattern. Therefore, C<(0|0x)\d*\s\1\d*> will
1857 match "0x1234 0x4321", but not "0x1234 01234", because subpattern
1858 1 matched "0x", even though the rule C<0|0x> could potentially match
1859 the leading 0 in the second number.
1861 =head2 Warning on \1 Instead of $1
1863 Some people get too used to writing things like:
1865 $pattern =~ s/(\W)/\\\1/g;
1867 This is grandfathered for the RHS of a substitute to avoid shocking the
1868 B<sed> addicts, but it's a dirty habit to get into. That's because in
1869 PerlThink, the righthand side of an C<s///> is a double-quoted string. C<\1> in
1870 the usual double-quoted string means a control-A. The customary Unix
1871 meaning of C<\1> is kludged in for C<s///>. However, if you get into the habit
1872 of doing that, you get yourself into trouble if you then add an C</e>
1875 s/(\d+)/ \1 + 1 /eg; # causes warning under -w
1881 You can't disambiguate that by saying C<\{1}000>, whereas you can fix it with
1882 C<${1}000>. The operation of interpolation should not be confused
1883 with the operation of matching a backreference. Certainly they mean two
1884 different things on the I<left> side of the C<s///>.
1886 =head2 Repeated Patterns Matching a Zero-length Substring
1888 B<WARNING>: Difficult material (and prose) ahead. This section needs a rewrite.
1890 Regular expressions provide a terse and powerful programming language. As
1891 with most other power tools, power comes together with the ability
1894 A common abuse of this power stems from the ability to make infinite
1895 loops using regular expressions, with something as innocuous as:
1897 'foo' =~ m{ ( o? )* }x;
1899 The C<o?> matches at the beginning of C<'foo'>, and since the position
1900 in the string is not moved by the match, C<o?> would match again and again
1901 because of the C<*> quantifier. Another common way to create a similar cycle
1902 is with the looping modifier C<//g>:
1904 @matches = ( 'foo' =~ m{ o? }xg );
1908 print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
1910 or the loop implied by split().
1912 However, long experience has shown that many programming tasks may
1913 be significantly simplified by using repeated subexpressions that
1914 may match zero-length substrings. Here's a simple example being:
1916 @chars = split //, $string; # // is not magic in split
1917 ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
1919 Thus Perl allows such constructs, by I<forcefully breaking
1920 the infinite loop>. The rules for this are different for lower-level
1921 loops given by the greedy quantifiers C<*+{}>, and for higher-level
1922 ones like the C</g> modifier or split() operator.
1924 The lower-level loops are I<interrupted> (that is, the loop is
1925 broken) when Perl detects that a repeated expression matched a
1926 zero-length substring. Thus
1928 m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
1930 is made equivalent to
1932 m{ (?: NON_ZERO_LENGTH )*
1937 The higher level-loops preserve an additional state between iterations:
1938 whether the last match was zero-length. To break the loop, the following
1939 match after a zero-length match is prohibited to have a length of zero.
1940 This prohibition interacts with backtracking (see L<"Backtracking">),
1941 and so the I<second best> match is chosen if the I<best> match is of
1949 results in C<< <><b><><a><><r><> >>. At each position of the string the best
1950 match given by non-greedy C<??> is the zero-length match, and the I<second
1951 best> match is what is matched by C<\w>. Thus zero-length matches
1952 alternate with one-character-long matches.
1954 Similarly, for repeated C<m/()/g> the second-best match is the match at the
1955 position one notch further in the string.
1957 The additional state of being I<matched with zero-length> is associated with
1958 the matched string, and is reset by each assignment to pos().
1959 Zero-length matches at the end of the previous match are ignored
1962 =head2 Combining RE Pieces
1964 Each of the elementary pieces of regular expressions which were described
1965 before (such as C<ab> or C<\Z>) could match at most one substring
1966 at the given position of the input string. However, in a typical regular
1967 expression these elementary pieces are combined into more complicated
1968 patterns using combining operators C<ST>, C<S|T>, C<S*> etc
1969 (in these examples C<S> and C<T> are regular subexpressions).
1971 Such combinations can include alternatives, leading to a problem of choice:
1972 if we match a regular expression C<a|ab> against C<"abc">, will it match
1973 substring C<"a"> or C<"ab">? One way to describe which substring is
1974 actually matched is the concept of backtracking (see L<"Backtracking">).
1975 However, this description is too low-level and makes you think
1976 in terms of a particular implementation.
1978 Another description starts with notions of "better"/"worse". All the
1979 substrings which may be matched by the given regular expression can be
1980 sorted from the "best" match to the "worst" match, and it is the "best"
1981 match which is chosen. This substitutes the question of "what is chosen?"
1982 by the question of "which matches are better, and which are worse?".
1984 Again, for elementary pieces there is no such question, since at most
1985 one match at a given position is possible. This section describes the
1986 notion of better/worse for combining operators. In the description
1987 below C<S> and C<T> are regular subexpressions.
1993 Consider two possible matches, C<AB> and C<A'B'>, C<A> and C<A'> are
1994 substrings which can be matched by C<S>, C<B> and C<B'> are substrings
1995 which can be matched by C<T>.
1997 If C<A> is better match for C<S> than C<A'>, C<AB> is a better
2000 If C<A> and C<A'> coincide: C<AB> is a better match than C<AB'> if
2001 C<B> is better match for C<T> than C<B'>.
2005 When C<S> can match, it is a better match than when only C<T> can match.
2007 Ordering of two matches for C<S> is the same as for C<S>. Similar for
2008 two matches for C<T>.
2010 =item C<S{REPEAT_COUNT}>
2012 Matches as C<SSS...S> (repeated as many times as necessary).
2016 Matches as C<S{max}|S{max-1}|...|S{min+1}|S{min}>.
2018 =item C<S{min,max}?>
2020 Matches as C<S{min}|S{min+1}|...|S{max-1}|S{max}>.
2022 =item C<S?>, C<S*>, C<S+>
2024 Same as C<S{0,1}>, C<S{0,BIG_NUMBER}>, C<S{1,BIG_NUMBER}> respectively.
2026 =item C<S??>, C<S*?>, C<S+?>
2028 Same as C<S{0,1}?>, C<S{0,BIG_NUMBER}?>, C<S{1,BIG_NUMBER}?> respectively.
2032 Matches the best match for C<S> and only that.
2034 =item C<(?=S)>, C<(?<=S)>
2036 Only the best match for C<S> is considered. (This is important only if
2037 C<S> has capturing parentheses, and backreferences are used somewhere
2038 else in the whole regular expression.)
2040 =item C<(?!S)>, C<(?<!S)>
2042 For this grouping operator there is no need to describe the ordering, since
2043 only whether or not C<S> can match is important.
2045 =item C<(??{ EXPR })>, C<(?PARNO)>
2047 The ordering is the same as for the regular expression which is
2048 the result of EXPR, or the pattern contained by capture buffer PARNO.
2050 =item C<(?(condition)yes-pattern|no-pattern)>
2052 Recall that which of C<yes-pattern> or C<no-pattern> actually matches is
2053 already determined. The ordering of the matches is the same as for the
2054 chosen subexpression.
2058 The above recipes describe the ordering of matches I<at a given position>.
2059 One more rule is needed to understand how a match is determined for the
2060 whole regular expression: a match at an earlier position is always better
2061 than a match at a later position.
2063 =head2 Creating Custom RE Engines
2065 Overloaded constants (see L<overload>) provide a simple way to extend
2066 the functionality of the RE engine.
2068 Suppose that we want to enable a new RE escape-sequence C<\Y|> which
2069 matches at a boundary between whitespace characters and non-whitespace
2070 characters. Note that C<(?=\S)(?<!\S)|(?!\S)(?<=\S)> matches exactly
2071 at these positions, so we want to have each C<\Y|> in the place of the
2072 more complicated version. We can create a module C<customre> to do
2080 die "No argument to customre::import allowed" if @_;
2081 overload::constant 'qr' => \&convert;
2084 sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2086 # We must also take care of not escaping the legitimate \\Y|
2087 # sequence, hence the presence of '\\' in the conversion rules.
2088 my %rules = ( '\\' => '\\\\',
2089 'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2095 { $rules{$1} or invalid($re,$1) }sgex;
2099 Now C<use customre> enables the new escape in constant regular
2100 expressions, i.e., those without any runtime variable interpolations.
2101 As documented in L<overload>, this conversion will work only over
2102 literal parts of regular expressions. For C<\Y|$re\Y|> the variable
2103 part of this regular expression needs to be converted explicitly
2104 (but only if the special meaning of C<\Y|> should be enabled inside $re):
2109 $re = customre::convert $re;
2112 =head1 PCRE/Python Support
2114 As of Perl 5.10.0, Perl supports several Python/PCRE specific extensions
2115 to the regex syntax. While Perl programmers are encouraged to use the
2116 Perl specific syntax, the following are also accepted:
2120 =item C<< (?PE<lt>NAMEE<gt>pattern) >>
2122 Define a named capture buffer. Equivalent to C<< (?<NAME>pattern) >>.
2124 =item C<< (?P=NAME) >>
2126 Backreference to a named capture buffer. Equivalent to C<< \g{NAME} >>.
2128 =item C<< (?P>NAME) >>
2130 Subroutine call to a named capture buffer. Equivalent to C<< (?&NAME) >>.
2136 This document varies from difficult to understand to completely
2137 and utterly opaque. The wandering prose riddled with jargon is
2138 hard to fathom in several places.
2140 This document needs a rewrite that separates the tutorial content
2141 from the reference content.
2149 L<perlop/"Regexp Quote-Like Operators">.
2151 L<perlop/"Gory details of parsing quoted constructs">.
2161 I<Mastering Regular Expressions> by Jeffrey Friedl, published
2162 by O'Reilly and Associates.