1PERLRE(1)              Perl Programmers Reference Guide              PERLRE(1)
2
3
4

NAME

6       perlre - Perl regular expressions
7

DESCRIPTION

9       This page describes the syntax of regular expressions in Perl.
10
11       If you haven't used regular expressions before, a tutorial introduction
12       is available in perlretut.  If you know just a little about them, a
13       quick-start introduction is available in perlrequick.
14
15       Except for "The Basics" section, this page assumes you are familiar
16       with regular expression basics, like what is a "pattern", what does it
17       look like, and how it is basically used.  For a reference on how they
18       are used, plus various examples of the same, see discussions of "m//",
19       "s///", "qr//" and "??" in "Regexp Quote-Like Operators" in perlop.
20
21       New in v5.22, "use re 'strict'" applies stricter rules than otherwise
22       when compiling regular expression patterns.  It can find things that,
23       while legal, may not be what you intended.
24
25   The Basics
26       Regular expressions are strings with the very particular syntax and
27       meaning described in this document and auxiliary documents referred to
28       by this one.  The strings are called "patterns".  Patterns are used to
29       determine if some other string, called the "target", has (or doesn't
30       have) the characteristics specified by the pattern.  We call this
31       "matching" the target string against the pattern.  Usually the match is
32       done by having the target be the first operand, and the pattern be the
33       second operand, of one of the two binary operators "=~" and "!~",
34       listed in "Binding Operators" in perlop; and the pattern will have been
35       converted from an ordinary string by one of the operators in "Regexp
36       Quote-Like Operators" in perlop, like so:
37
38        $foo =~ m/abc/
39
40       This evaluates to true if and only if the string in the variable $foo
41       contains somewhere in it, the sequence of characters "a", "b", then
42       "c".  (The "=~ m", or match operator, is described in
43       "m/PATTERN/msixpodualngc" in perlop.)
44
45       Patterns that aren't already stored in some variable must be
46       delimitted, at both ends, by delimitter characters.  These are often,
47       as in the example above, forward slashes, and the typical way a pattern
48       is written in documentation is with those slashes.  In most cases, the
49       delimitter is the same character, fore and aft, but there are a few
50       cases where a character looks like it has a mirror-image mate, where
51       the opening version is the beginning delimiter, and the closing one is
52       the ending delimiter, like
53
54        $foo =~ m<abc>
55
56       Most times, the pattern is evaluated in double-quotish context, but it
57       is possible to choose delimiters to force single-quotish, like
58
59        $foo =~ m'abc'
60
61       If the pattern contains its delimiter within it, that delimiter must be
62       escaped.  Prefixing it with a backslash (e.g., "/foo\/bar/") serves
63       this purpose.
64
65       Any single character in a pattern matches that same character in the
66       target string, unless the character is a metacharacter with a special
67       meaning described in this document.  A sequence of non-metacharacters
68       matches the same sequence in the target string, as we saw above with
69       "m/abc/".
70
71       Only a few characters (all of them being ASCII punctuation characters)
72       are metacharacters.  The most commonly used one is a dot ".", which
73       normally matches almost any character (including a dot itself).
74
75       You can cause characters that normally function as metacharacters to be
76       interpreted literally by prefixing them with a "\", just like the
77       pattern's delimiter must be escaped if it also occurs within the
78       pattern.  Thus, "\." matches just a literal dot, "." instead of its
79       normal meaning.  This means that the backslash is also a metacharacter,
80       so "\\" matches a single "\".  And a sequence that contains an escaped
81       metacharacter matches the same sequence (but without the escape) in the
82       target string.  So, the pattern "/blur\\fl/" would match any target
83       string that contains the sequence "blur\fl".
84
85       The metacharacter "|" is used to match one thing or another.  Thus
86
87        $foo =~ m/this|that/
88
89       is TRUE if and only if $foo contains either the sequence "this" or the
90       sequence "that".  Like all metacharacters, prefixing the "|" with a
91       backslash makes it match the plain punctuation character; in its case,
92       the VERTICAL LINE.
93
94        $foo =~ m/this\|that/
95
96       is TRUE if and only if $foo contains the sequence "this|that".
97
98       You aren't limited to just a single "|".
99
100        $foo =~ m/fee|fie|foe|fum/
101
102       is TRUE if and only if $foo contains any of those 4 sequences from the
103       children's story "Jack and the Beanstalk".
104
105       As you can see, the "|" binds less tightly than a sequence of ordinary
106       characters.  We can override this by using the grouping metacharacters,
107       the parentheses "(" and ")".
108
109        $foo =~ m/th(is|at) thing/
110
111       is TRUE if and only if $foo contains either the sequence "this thing"
112       or the sequence "that thing".  The portions of the string that match
113       the portions of the pattern enclosed in parentheses are normally made
114       available separately for use later in the pattern, substitution, or
115       program.  This is called "capturing", and it can get complicated.  See
116       "Capture groups".
117
118       The first alternative includes everything from the last pattern
119       delimiter ("(", "(?:" (described later), etc. or the beginning of the
120       pattern) up to the first "|", and the last alternative contains
121       everything from the last "|" to the next closing pattern delimiter.
122       That's why it's common practice to include alternatives in parentheses:
123       to minimize confusion about where they start and end.
124
125       Alternatives are tried from left to right, so the first alternative
126       found for which the entire expression matches, is the one that is
127       chosen. This means that alternatives are not necessarily greedy. For
128       example: when matching "foo|foot" against "barefoot", only the "foo"
129       part will match, as that is the first alternative tried, and it
130       successfully matches the target string. (This might not seem important,
131       but it is important when you are capturing matched text using
132       parentheses.)
133
134       Besides taking away the special meaning of a metacharacter, a prefixed
135       backslash changes some letter and digit characters away from matching
136       just themselves to instead have special meaning.  These are called
137       "escape sequences", and all such are described in perlrebackslash.  A
138       backslash sequence (of a letter or digit) that doesn't currently have
139       special meaning to Perl will raise a warning if warnings are enabled,
140       as those are reserved for potential future use.
141
142       One such sequence is "\b", which matches a boundary of some sort.
143       "\b{wb}" and a few others give specialized types of boundaries.  (They
144       are all described in detail starting at "\b{}, \b, \B{}, \B" in
145       perlrebackslash.)  Note that these don't match characters, but the
146       zero-width spaces between characters.  They are an example of a zero-
147       width assertion.  Consider again,
148
149        $foo =~ m/fee|fie|foe|fum/
150
151       It evaluates to TRUE if, besides those 4 words, any of the sequences
152       "feed", "field", "Defoe", "fume", and many others are in $foo.  By
153       judicious use of "\b" (or better (because it is designed to handle
154       natural language) "\b{wb}"), we can make sure that only the Giant's
155       words are matched:
156
157        $foo =~ m/\b(fee|fie|foe|fum)\b/
158        $foo =~ m/\b{wb}(fee|fie|foe|fum)\b{wb}/
159
160       The final example shows that the characters "{" and "}" are
161       metacharacters.
162
163       Another use for escape sequences is to specify characters that cannot
164       (or which you prefer not to) be written literally.  These are described
165       in detail in "Character Escapes" in perlrebackslash, but the next three
166       paragraphs briefly describe some of them.
167
168       Various control characters can be written in C language style: "\n"
169       matches a newline, "\t" a tab, "\r" a carriage return, "\f" a form
170       feed, etc.
171
172       More generally, "\nnn", where nnn is a string of three octal digits,
173       matches the character whose native code point is nnn.  You can easily
174       run into trouble if you don't have exactly three digits.  So always use
175       three, or since Perl 5.14, you can use "\o{...}" to specify any number
176       of octal digits.
177
178       Similarly, "\xnn", where nn are hexadecimal digits, matches the
179       character whose native ordinal is nn.  Again, not using exactly two
180       digits is a recipe for disaster, but you can use "\x{...}" to specify
181       any number of hex digits.
182
183       Besides being a metacharacter, the "." is an example of a "character
184       class", something that can match any single character of a given set of
185       them.  In its case, the set is just about all possible characters.
186       Perl predefines several character classes besides the "."; there is a
187       separate reference page about just these, perlrecharclass.
188
189       You can define your own custom character classes, by putting into your
190       pattern in the appropriate place(s), a list of all the characters you
191       want in the set.  You do this by enclosing the list within "[]" bracket
192       characters.  These are called "bracketed character classes" when we are
193       being precise, but often the word "bracketed" is dropped.  (Dropping it
194       usually doesn't cause confusion.)  This means that the "[" character is
195       another metacharacter.  It doesn't match anything just by itself; it is
196       used only to tell Perl that what follows it is a bracketed character
197       class.  If you want to match a literal left square bracket, you must
198       escape it, like "\[".  The matching "]" is also a metacharacter; again
199       it doesn't match anything by itself, but just marks the end of your
200       custom class to Perl.  It is an example of a "sometimes metacharacter".
201       It isn't a metacharacter if there is no corresponding "[", and matches
202       its literal self:
203
204        print "]" =~ /]/;  # prints 1
205
206       The list of characters within the character class gives the set of
207       characters matched by the class.  "[abc]" matches a single "a" or "b"
208       or "c".  But if the first character after the "[" is "^", the class
209       instead matches any character not in the list.  Within a list, the "-"
210       character specifies a range of characters, so that "a-z" represents all
211       characters between "a" and "z", inclusive.  If you want either "-" or
212       "]" itself to be a member of a class, put it at the start of the list
213       (possibly after a "^"), or escape it with a backslash.  "-" is also
214       taken literally when it is at the end of the list, just before the
215       closing "]".  (The following all specify the same class of three
216       characters: "[-az]", "[az-]", and "[a\-z]".  All are different from
217       "[a-z]", which specifies a class containing twenty-six characters, even
218       on EBCDIC-based character sets.)
219
220       There is lots more to bracketed character classes; full details are in
221       "Bracketed Character Classes" in perlrecharclass.
222
223       Metacharacters
224
225       "The Basics" introduced some of the metacharacters.  This section gives
226       them all.  Most of them have the same meaning as in the egrep command.
227
228       Only the "\" is always a metacharacter.  The others are metacharacters
229       just sometimes.  The following tables lists all of them, summarizes
230       their use, and gives the contexts where they are metacharacters.
231       Outside those contexts or if prefixed by a "\", they match their
232       corresponding punctuation character.  In some cases, their meaning
233       varies depending on various pattern modifiers that alter the default
234       behaviors.  See "Modifiers".
235
236                   PURPOSE                                  WHERE
237        \   Escape the next character                    Always, except when
238                                                         escaped by another \
239        ^   Match the beginning of the string            Not in []
240              (or line, if /m is used)
241        ^   Complement the [] class                      At the beginning of []
242        .   Match any single character except newline    Not in []
243              (under /s, includes newline)
244        $   Match the end of the string                  Not in [], but can
245              (or before newline at the end of the       mean interpolate a
246              string; or before any newline if /m is     scalar
247              used)
248        |   Alternation                                  Not in []
249        ()  Grouping                                     Not in []
250        [   Start Bracketed Character class              Not in []
251        ]   End Bracketed Character class                Only in [], and
252                                                           not first
253        *   Matches the preceding element 0 or more      Not in []
254              times
255        +   Matches the preceding element 1 or more      Not in []
256              times
257        ?   Matches the preceding element 0 or 1         Not in []
258              times
259        {   Starts a sequence that gives number(s)       Not in []
260              of times the preceding element can be
261              matched
262        {   when following certain escape sequences
263              starts a modifier to the meaning of the
264              sequence
265        }   End sequence started by {
266        -   Indicates a range                            Only in [] interior
267        #   Beginning of comment, extends to line end    Only with /x modifier
268
269       Notice that most of the metacharacters lose their special meaning when
270       they occur in a bracketed character class, except "^" has a different
271       meaning when it is at the beginning of such a class.  And "-" and "]"
272       are metacharacters only at restricted positions within bracketed
273       character classes; while "}" is a metacharacter only when closing a
274       special construct started by "{".
275
276       In double-quotish context, as is usually the case,  you need to be
277       careful about "$" and the non-metacharacter "@".  Those could
278       interpolate variables, which may or may not be what you intended.
279
280       These rules were designed for compactness of expression, rather than
281       legibility and maintainability.  The "/x and /xx" pattern modifiers
282       allow you to insert white space to improve readability.  And use of
283       "re 'strict'" adds extra checking to catch some typos that might
284       silently compile into something unintended.
285
286       By default, the "^" character is guaranteed to match only the beginning
287       of the string, the "$" character only the end (or before the newline at
288       the end), and Perl does certain optimizations with the assumption that
289       the string contains only one line.  Embedded newlines will not be
290       matched by "^" or "$".  You may, however, wish to treat a string as a
291       multi-line buffer, such that the "^" will match after any newline
292       within the string (except if the newline is the last character in the
293       string), and "$" will match before any newline.  At the cost of a
294       little more overhead, you can do this by using the ""/m"" modifier on
295       the pattern match operator.  (Older programs did this by setting $*,
296       but this option was removed in perl 5.10.)
297
298       To simplify multi-line substitutions, the "." character never matches a
299       newline unless you use the "/s" modifier, which in effect tells Perl to
300       pretend the string is a single line--even if it isn't.
301
302   Modifiers
303       Overview
304
305       The default behavior for matching can be changed, using various
306       modifiers.  Modifiers that relate to the interpretation of the pattern
307       are listed just below.  Modifiers that alter the way a pattern is used
308       by Perl are detailed in "Regexp Quote-Like Operators" in perlop and
309       "Gory details of parsing quoted constructs" in perlop.
310
311       "m" Treat the string being matched against as multiple lines.  That is,
312           change "^" and "$" from matching the start of the string's first
313           line and the end of its last line to matching the start and end of
314           each line within the string.
315
316       "s" Treat the string as single line.  That is, change "." to match any
317           character whatsoever, even a newline, which normally it would not
318           match.
319
320           Used together, as "/ms", they let the "." match any character
321           whatsoever, while still allowing "^" and "$" to match,
322           respectively, just after and just before newlines within the
323           string.
324
325       "i" Do case-insensitive pattern matching.  For example, "A" will match
326           "a" under "/i".
327
328           If locale matching rules are in effect, the case map is taken from
329           the current locale for code points less than 255, and from Unicode
330           rules for larger code points.  However, matches that would cross
331           the Unicode rules/non-Unicode rules boundary (ords 255/256) will
332           not succeed, unless the locale is a UTF-8 one.  See perllocale.
333
334           There are a number of Unicode characters that match a sequence of
335           multiple characters under "/i".  For example, "LATIN SMALL LIGATURE
336           FI" should match the sequence "fi".  Perl is not currently able to
337           do this when the multiple characters are in the pattern and are
338           split between groupings, or when one or more are quantified.  Thus
339
340            "\N{LATIN SMALL LIGATURE FI}" =~ /fi/i;          # Matches
341            "\N{LATIN SMALL LIGATURE FI}" =~ /[fi][fi]/i;    # Doesn't match!
342            "\N{LATIN SMALL LIGATURE FI}" =~ /fi*/i;         # Doesn't match!
343
344            # The below doesn't match, and it isn't clear what $1 and $2 would
345            # be even if it did!!
346            "\N{LATIN SMALL LIGATURE FI}" =~ /(f)(i)/i;      # Doesn't match!
347
348           Perl doesn't match multiple characters in a bracketed character
349           class unless the character that maps to them is explicitly
350           mentioned, and it doesn't match them at all if the character class
351           is inverted, which otherwise could be highly confusing.  See
352           "Bracketed Character Classes" in perlrecharclass, and "Negation" in
353           perlrecharclass.
354
355       "x" and "xx"
356           Extend your pattern's legibility by permitting whitespace and
357           comments.  Details in "/x and  /xx"
358
359       "p" Preserve the string matched such that "${^PREMATCH}", "${^MATCH}",
360           and "${^POSTMATCH}" are available for use after matching.
361
362           In Perl 5.20 and higher this is ignored. Due to a new copy-on-write
363           mechanism, "${^PREMATCH}", "${^MATCH}", and "${^POSTMATCH}" will be
364           available after the match regardless of the modifier.
365
366       "a", "d", "l", and "u"
367           These modifiers, all new in 5.14, affect which character-set rules
368           (Unicode, etc.) are used, as described below in "Character set
369           modifiers".
370
371       "n" Prevent the grouping metacharacters "()" from capturing. This
372           modifier, new in 5.22, will stop $1, $2, etc... from being filled
373           in.
374
375             "hello" =~ /(hi|hello)/;   # $1 is "hello"
376             "hello" =~ /(hi|hello)/n;  # $1 is undef
377
378           This is equivalent to putting "?:" at the beginning of every
379           capturing group:
380
381             "hello" =~ /(?:hi|hello)/; # $1 is undef
382
383           "/n" can be negated on a per-group basis. Alternatively, named
384           captures may still be used.
385
386             "hello" =~ /(?-n:(hi|hello))/n;   # $1 is "hello"
387             "hello" =~ /(?<greet>hi|hello)/n; # $1 is "hello", $+{greet} is
388                                               # "hello"
389
390       Other Modifiers
391           There are a number of flags that can be found at the end of regular
392           expression constructs that are not generic regular expression
393           flags, but apply to the operation being performed, like matching or
394           substitution ("m//" or "s///" respectively).
395
396           Flags described further in "Using regular expressions in Perl" in
397           perlretut are:
398
399             c  - keep the current position during repeated matching
400             g  - globally match the pattern repeatedly in the string
401
402           Substitution-specific modifiers described in
403           "s/PATTERN/REPLACEMENT/msixpodualngcer" in perlop are:
404
405             e  - evaluate the right-hand side as an expression
406             ee - evaluate the right side as a string then eval the result
407             o  - pretend to optimize your code, but actually introduce bugs
408             r  - perform non-destructive substitution and return the new value
409
410       Regular expression modifiers are usually written in documentation as
411       e.g., "the "/x" modifier", even though the delimiter in question might
412       not really be a slash.  The modifiers "/imnsxadlup" may also be
413       embedded within the regular expression itself using the "(?...)"
414       construct, see "Extended Patterns" below.
415
416       Details on some modifiers
417
418       Some of the modifiers require more explanation than given in the
419       "Overview" above.
420
421       "/x" and  "/xx"
422
423       A single "/x" tells the regular expression parser to ignore most
424       whitespace that is neither backslashed nor within a bracketed character
425       class.  You can use this to break up your regular expression into more
426       readable parts.  Also, the "#" character is treated as a metacharacter
427       introducing a comment that runs up to the pattern's closing delimiter,
428       or to the end of the current line if the pattern extends onto the next
429       line.  Hence, this is very much like an ordinary Perl code comment.
430       (You can include the closing delimiter within the comment only if you
431       precede it with a backslash, so be careful!)
432
433       Use of "/x" means that if you want real whitespace or "#" characters in
434       the pattern (outside a bracketed character class, which is unaffected
435       by "/x"), then you'll either have to escape them (using backslashes or
436       "\Q...\E") or encode them using octal, hex, or "\N{}" escapes.  It is
437       ineffective to try to continue a comment onto the next line by escaping
438       the "\n" with a backslash or "\Q".
439
440       You can use "(?#text)" to create a comment that ends earlier than the
441       end of the current line, but "text" also can't contain the closing
442       delimiter unless escaped with a backslash.
443
444       A common pitfall is to forget that "#" characters begin a comment under
445       "/x" and are not matched literally.  Just keep that in mind when trying
446       to puzzle out why a particular "/x" pattern isn't working as expected.
447
448       Starting in Perl v5.26, if the modifier has a second "x" within it, it
449       does everything that a single "/x" does, but additionally non-
450       backslashed SPACE and TAB characters within bracketed character classes
451       are also generally ignored, and hence can be added to make the classes
452       more readable.
453
454           / [d-e g-i 3-7]/xx
455           /[ ! @ " # $ % ^ & * () = ? <> ' ]/xx
456
457       may be easier to grasp than the squashed equivalents
458
459           /[d-eg-i3-7]/
460           /[!@"#$%^&*()=?<>']/
461
462       Taken together, these features go a long way towards making Perl's
463       regular expressions more readable.  Here's an example:
464
465           # Delete (most) C comments.
466           $program =~ s {
467               /\*     # Match the opening delimiter.
468               .*?     # Match a minimal number of characters.
469               \*/     # Match the closing delimiter.
470           } []gsx;
471
472       Note that anything inside a "\Q...\E" stays unaffected by "/x".  And
473       note that "/x" doesn't affect space interpretation within a single
474       multi-character construct.  For example in "\x{...}", regardless of the
475       "/x" modifier, there can be no spaces.  Same for a quantifier such as
476       "{3}" or "{5,}".  Similarly, "(?:...)" can't have a space between the
477       "(", "?", and ":".  Within any delimiters for such a construct, allowed
478       spaces are not affected by "/x", and depend on the construct.  For
479       example, "\x{...}" can't have spaces because hexadecimal numbers don't
480       have spaces in them.  But, Unicode properties can have spaces, so in
481       "\p{...}" there can be spaces that follow the Unicode rules, for which
482       see "Properties accessible through \p{} and \P{}" in perluniprops.
483
484       The set of characters that are deemed whitespace are those that Unicode
485       calls "Pattern White Space", namely:
486
487        U+0009 CHARACTER TABULATION
488        U+000A LINE FEED
489        U+000B LINE TABULATION
490        U+000C FORM FEED
491        U+000D CARRIAGE RETURN
492        U+0020 SPACE
493        U+0085 NEXT LINE
494        U+200E LEFT-TO-RIGHT MARK
495        U+200F RIGHT-TO-LEFT MARK
496        U+2028 LINE SEPARATOR
497        U+2029 PARAGRAPH SEPARATOR
498
499       Character set modifiers
500
501       "/d", "/u", "/a", and "/l", available starting in 5.14, are called the
502       character set modifiers; they affect the character set rules used for
503       the regular expression.
504
505       The "/d", "/u", and "/l" modifiers are not likely to be of much use to
506       you, and so you need not worry about them very much.  They exist for
507       Perl's internal use, so that complex regular expression data structures
508       can be automatically serialized and later exactly reconstituted,
509       including all their nuances.  But, since Perl can't keep a secret, and
510       there may be rare instances where they are useful, they are documented
511       here.
512
513       The "/a" modifier, on the other hand, may be useful.  Its purpose is to
514       allow code that is to work mostly on ASCII data to not have to concern
515       itself with Unicode.
516
517       Briefly, "/l" sets the character set to that of whatever Locale is in
518       effect at the time of the execution of the pattern match.
519
520       "/u" sets the character set to Unicode.
521
522       "/a" also sets the character set to Unicode, BUT adds several
523       restrictions for ASCII-safe matching.
524
525       "/d" is the old, problematic, pre-5.14 Default character set behavior.
526       Its only use is to force that old behavior.
527
528       At any given time, exactly one of these modifiers is in effect.  Their
529       existence allows Perl to keep the originally compiled behavior of a
530       regular expression, regardless of what rules are in effect when it is
531       actually executed.  And if it is interpolated into a larger regex, the
532       original's rules continue to apply to it, and don't affect the other
533       parts.
534
535       The "/l" and "/u" modifiers are automatically selected for regular
536       expressions compiled within the scope of various pragmas, and we
537       recommend that in general, you use those pragmas instead of specifying
538       these modifiers explicitly.  For one thing, the modifiers affect only
539       pattern matching, and do not extend to even any replacement done,
540       whereas using the pragmas gives consistent results for all appropriate
541       operations within their scopes.  For example,
542
543        s/foo/\Ubar/il
544
545       will match "foo" using the locale's rules for case-insensitive
546       matching, but the "/l" does not affect how the "\U" operates.  Most
547       likely you want both of them to use locale rules.  To do this, instead
548       compile the regular expression within the scope of "use locale".  This
549       both implicitly adds the "/l", and applies locale rules to the "\U".
550       The lesson is to "use locale", and not "/l" explicitly.
551
552       Similarly, it would be better to use "use feature 'unicode_strings'"
553       instead of,
554
555        s/foo/\Lbar/iu
556
557       to get Unicode rules, as the "\L" in the former (but not necessarily
558       the latter) would also use Unicode rules.
559
560       More detail on each of the modifiers follows.  Most likely you don't
561       need to know this detail for "/l", "/u", and "/d", and can skip ahead
562       to /a.
563
564       /l
565
566       means to use the current locale's rules (see perllocale) when pattern
567       matching.  For example, "\w" will match the "word" characters of that
568       locale, and "/i" case-insensitive matching will match according to the
569       locale's case folding rules.  The locale used will be the one in effect
570       at the time of execution of the pattern match.  This may not be the
571       same as the compilation-time locale, and can differ from one match to
572       another if there is an intervening call of the setlocale() function.
573
574       Prior to v5.20, Perl did not support multi-byte locales.  Starting
575       then, UTF-8 locales are supported.  No other multi byte locales are
576       ever likely to be supported.  However, in all locales, one can have
577       code points above 255 and these will always be treated as Unicode no
578       matter what locale is in effect.
579
580       Under Unicode rules, there are a few case-insensitive matches that
581       cross the 255/256 boundary.  Except for UTF-8 locales in Perls v5.20
582       and later, these are disallowed under "/l".  For example, 0xFF (on
583       ASCII platforms) does not caselessly match the character at 0x178,
584       "LATIN CAPITAL LETTER Y WITH DIAERESIS", because 0xFF may not be "LATIN
585       SMALL LETTER Y WITH DIAERESIS" in the current locale, and Perl has no
586       way of knowing if that character even exists in the locale, much less
587       what code point it is.
588
589       In a UTF-8 locale in v5.20 and later, the only visible difference
590       between locale and non-locale in regular expressions should be tainting
591       (see perlsec).
592
593       This modifier may be specified to be the default by "use locale", but
594       see "Which character set modifier is in effect?".
595
596       /u
597
598       means to use Unicode rules when pattern matching.  On ASCII platforms,
599       this means that the code points between 128 and 255 take on their
600       Latin-1 (ISO-8859-1) meanings (which are the same as Unicode's).
601       (Otherwise Perl considers their meanings to be undefined.)  Thus, under
602       this modifier, the ASCII platform effectively becomes a Unicode
603       platform; and hence, for example, "\w" will match any of the more than
604       100_000 word characters in Unicode.
605
606       Unlike most locales, which are specific to a language and country pair,
607       Unicode classifies all the characters that are letters somewhere in the
608       world as "\w".  For example, your locale might not think that "LATIN
609       SMALL LETTER ETH" is a letter (unless you happen to speak Icelandic),
610       but Unicode does.  Similarly, all the characters that are decimal
611       digits somewhere in the world will match "\d"; this is hundreds, not
612       10, possible matches.  And some of those digits look like some of the
613       10 ASCII digits, but mean a different number, so a human could easily
614       think a number is a different quantity than it really is.  For example,
615       "BENGALI DIGIT FOUR" (U+09EA) looks very much like an "ASCII DIGIT
616       EIGHT" (U+0038), and "LEPCHA DIGIT SIX" (U+1C46) looks very much like
617       an "ASCII DIGIT FIVE" (U+0035).  And, "\d+", may match strings of
618       digits that are a mixture from different writing systems, creating a
619       security issue.  A fraudulent website, for example, could display the
620       price of something using U+1C46, and it would appear to the user that
621       something cost 500 units, but it really costs 600.  A browser that
622       enforced script runs ("Script Runs") would prevent that fraudulent
623       display.  "num()" in Unicode::UCD can also be used to sort this out.
624       Or the "/a" modifier can be used to force "\d" to match just the ASCII
625       0 through 9.
626
627       Also, under this modifier, case-insensitive matching works on the full
628       set of Unicode characters.  The "KELVIN SIGN", for example matches the
629       letters "k" and "K"; and "LATIN SMALL LIGATURE FF" matches the sequence
630       "ff", which, if you're not prepared, might make it look like a
631       hexadecimal constant, presenting another potential security issue.  See
632       <http://unicode.org/reports/tr36> for a detailed discussion of Unicode
633       security issues.
634
635       This modifier may be specified to be the default by "use feature
636       'unicode_strings", "use locale ':not_characters'", or "use 5.012" (or
637       higher), but see "Which character set modifier is in effect?".
638
639       /d
640
641       This modifier means to use the "Default" native rules of the platform
642       except when there is cause to use Unicode rules instead, as follows:
643
644       1.  the target string is encoded in UTF-8; or
645
646       2.  the pattern is encoded in UTF-8; or
647
648       3.  the pattern explicitly mentions a code point that is above 255 (say
649           by "\x{100}"); or
650
651       4.  the pattern uses a Unicode name ("\N{...}");  or
652
653       5.  the pattern uses a Unicode property ("\p{...}" or "\P{...}"); or
654
655       6.  the pattern uses a Unicode break ("\b{...}" or "\B{...}"); or
656
657       7.  the pattern uses ""(?[ ])""
658
659       8.  the pattern uses "(*script_run: ...)"
660
661       Another mnemonic for this modifier is "Depends", as the rules actually
662       used depend on various things, and as a result you can get unexpected
663       results.  See "The "Unicode Bug"" in perlunicode.  The Unicode Bug has
664       become rather infamous, leading to yet another (without swearing) name
665       for this modifier, "Dodgy".
666
667       Unless the pattern or string are encoded in UTF-8, only ASCII
668       characters can match positively.
669
670       Here are some examples of how that works on an ASCII platform:
671
672        $str =  "\xDF";      # $str is not in UTF-8 format.
673        $str =~ /^\w/;       # No match, as $str isn't in UTF-8 format.
674        $str .= "\x{0e0b}";  # Now $str is in UTF-8 format.
675        $str =~ /^\w/;       # Match! $str is now in UTF-8 format.
676        chop $str;
677        $str =~ /^\w/;       # Still a match! $str remains in UTF-8 format.
678
679       This modifier is automatically selected by default when none of the
680       others are, so yet another name for it is "Default".
681
682       Because of the unexpected behaviors associated with this modifier, you
683       probably should only explicitly use it to maintain weird backward
684       compatibilities.
685
686       /a (and /aa)
687
688       This modifier stands for ASCII-restrict (or ASCII-safe).  This modifier
689       may be doubled-up to increase its effect.
690
691       When it appears singly, it causes the sequences "\d", "\s", "\w", and
692       the Posix character classes to match only in the ASCII range.  They
693       thus revert to their pre-5.6, pre-Unicode meanings.  Under "/a",  "\d"
694       always means precisely the digits "0" to "9"; "\s" means the five
695       characters "[ \f\n\r\t]", and starting in Perl v5.18, the vertical tab;
696       "\w" means the 63 characters "[A-Za-z0-9_]"; and likewise, all the
697       Posix classes such as "[[:print:]]" match only the appropriate ASCII-
698       range characters.
699
700       This modifier is useful for people who only incidentally use Unicode,
701       and who do not wish to be burdened with its complexities and security
702       concerns.
703
704       With "/a", one can write "\d" with confidence that it will only match
705       ASCII characters, and should the need arise to match beyond ASCII, you
706       can instead use "\p{Digit}" (or "\p{Word}" for "\w").  There are
707       similar "\p{...}" constructs that can match beyond ASCII both white
708       space (see "Whitespace" in perlrecharclass), and Posix classes (see
709       "POSIX Character Classes" in perlrecharclass).  Thus, this modifier
710       doesn't mean you can't use Unicode, it means that to get Unicode
711       matching you must explicitly use a construct ("\p{}", "\P{}") that
712       signals Unicode.
713
714       As you would expect, this modifier causes, for example, "\D" to mean
715       the same thing as "[^0-9]"; in fact, all non-ASCII characters match
716       "\D", "\S", and "\W".  "\b" still means to match at the boundary
717       between "\w" and "\W", using the "/a" definitions of them (similarly
718       for "\B").
719
720       Otherwise, "/a" behaves like the "/u" modifier, in that case-
721       insensitive matching uses Unicode rules; for example, "k" will match
722       the Unicode "\N{KELVIN SIGN}" under "/i" matching, and code points in
723       the Latin1 range, above ASCII will have Unicode rules when it comes to
724       case-insensitive matching.
725
726       To forbid ASCII/non-ASCII matches (like "k" with "\N{KELVIN SIGN}"),
727       specify the "a" twice, for example "/aai" or "/aia".  (The first
728       occurrence of "a" restricts the "\d", etc., and the second occurrence
729       adds the "/i" restrictions.)  But, note that code points outside the
730       ASCII range will use Unicode rules for "/i" matching, so the modifier
731       doesn't really restrict things to just ASCII; it just forbids the
732       intermixing of ASCII and non-ASCII.
733
734       To summarize, this modifier provides protection for applications that
735       don't wish to be exposed to all of Unicode.  Specifying it twice gives
736       added protection.
737
738       This modifier may be specified to be the default by "use re '/a'" or
739       "use re '/aa'".  If you do so, you may actually have occasion to use
740       the "/u" modifier explicitly if there are a few regular expressions
741       where you do want full Unicode rules (but even here, it's best if
742       everything were under feature "unicode_strings", along with the "use re
743       '/aa'").  Also see "Which character set modifier is in effect?".
744
745       Which character set modifier is in effect?
746
747       Which of these modifiers is in effect at any given point in a regular
748       expression depends on a fairly complex set of interactions.  These have
749       been designed so that in general you don't have to worry about it, but
750       this section gives the gory details.  As explained below in "Extended
751       Patterns" it is possible to explicitly specify modifiers that apply
752       only to portions of a regular expression.  The innermost always has
753       priority over any outer ones, and one applying to the whole expression
754       has priority over any of the default settings that are described in the
755       remainder of this section.
756
757       The "use re '/foo'" pragma can be used to set default modifiers
758       (including these) for regular expressions compiled within its scope.
759       This pragma has precedence over the other pragmas listed below that
760       also change the defaults.
761
762       Otherwise, "use locale" sets the default modifier to "/l"; and "use
763       feature 'unicode_strings", or "use 5.012" (or higher) set the default
764       to "/u" when not in the same scope as either "use locale" or "use
765       bytes".  ("use locale ':not_characters'" also sets the default to "/u",
766       overriding any plain "use locale".)  Unlike the mechanisms mentioned
767       above, these affect operations besides regular expressions pattern
768       matching, and so give more consistent results with other operators,
769       including using "\U", "\l", etc. in substitution replacements.
770
771       If none of the above apply, for backwards compatibility reasons, the
772       "/d" modifier is the one in effect by default.  As this can lead to
773       unexpected results, it is best to specify which other rule set should
774       be used.
775
776       Character set modifier behavior prior to Perl 5.14
777
778       Prior to 5.14, there were no explicit modifiers, but "/l" was implied
779       for regexes compiled within the scope of "use locale", and "/d" was
780       implied otherwise.  However, interpolating a regex into a larger regex
781       would ignore the original compilation in favor of whatever was in
782       effect at the time of the second compilation.  There were a number of
783       inconsistencies (bugs) with the "/d" modifier, where Unicode rules
784       would be used when inappropriate, and vice versa.  "\p{}" did not imply
785       Unicode rules, and neither did all occurrences of "\N{}", until 5.12.
786
787   Regular Expressions
788       Quantifiers
789
790       Quantifiers are used when a particular portion of a pattern needs to
791       match a certain number (or numbers) of times.  If there isn't a
792       quantifier the number of times to match is exactly one.  The following
793       standard quantifiers are recognized:
794
795           *           Match 0 or more times
796           +           Match 1 or more times
797           ?           Match 1 or 0 times
798           {n}         Match exactly n times
799           {n,}        Match at least n times
800           {n,m}       Match at least n but not more than m times
801
802       (If a non-escaped curly bracket occurs in a context other than one of
803       the quantifiers listed above, where it does not form part of a
804       backslashed sequence like "\x{...}", it is either a fatal syntax error,
805       or treated as a regular character, generally with a deprecation warning
806       raised.  To escape it, you can precede it with a backslash ("\{") or
807       enclose it within square brackets  ("[{]").  This change will allow for
808       future syntax extensions (like making the lower bound of a quantifier
809       optional), and better error checking of quantifiers).
810
811       The "*" quantifier is equivalent to "{0,}", the "+" quantifier to
812       "{1,}", and the "?" quantifier to "{0,1}".  n and m are limited to non-
813       negative integral values less than a preset limit defined when perl is
814       built.  This is usually 32766 on the most common platforms.  The actual
815       limit can be seen in the error message generated by code such as this:
816
817           $_ **= $_ , / {$_} / for 2 .. 42;
818
819       By default, a quantified subpattern is "greedy", that is, it will match
820       as many times as possible (given a particular starting location) while
821       still allowing the rest of the pattern to match.  If you want it to
822       match the minimum number of times possible, follow the quantifier with
823       a "?".  Note that the meanings don't change, just the "greediness":
824
825           *?        Match 0 or more times, not greedily
826           +?        Match 1 or more times, not greedily
827           ??        Match 0 or 1 time, not greedily
828           {n}?      Match exactly n times, not greedily (redundant)
829           {n,}?     Match at least n times, not greedily
830           {n,m}?    Match at least n but not more than m times, not greedily
831
832       Normally when a quantified subpattern does not allow the rest of the
833       overall pattern to match, Perl will backtrack. However, this behaviour
834       is sometimes undesirable. Thus Perl provides the "possessive"
835       quantifier form as well.
836
837        *+     Match 0 or more times and give nothing back
838        ++     Match 1 or more times and give nothing back
839        ?+     Match 0 or 1 time and give nothing back
840        {n}+   Match exactly n times and give nothing back (redundant)
841        {n,}+  Match at least n times and give nothing back
842        {n,m}+ Match at least n but not more than m times and give nothing back
843
844       For instance,
845
846          'aaaa' =~ /a++a/
847
848       will never match, as the "a++" will gobble up all the "a"'s in the
849       string and won't leave any for the remaining part of the pattern. This
850       feature can be extremely useful to give perl hints about where it
851       shouldn't backtrack. For instance, the typical "match a double-quoted
852       string" problem can be most efficiently performed when written as:
853
854          /"(?:[^"\\]++|\\.)*+"/
855
856       as we know that if the final quote does not match, backtracking will
857       not help. See the independent subexpression ""(?>pattern)"" for more
858       details; possessive quantifiers are just syntactic sugar for that
859       construct. For instance the above example could also be written as
860       follows:
861
862          /"(?>(?:(?>[^"\\]+)|\\.)*)"/
863
864       Note that the possessive quantifier modifier can not be combined with
865       the non-greedy modifier. This is because it would make no sense.
866       Consider the follow equivalency table:
867
868           Illegal         Legal
869           ------------    ------
870           X??+            X{0}
871           X+?+            X{1}
872           X{min,max}?+    X{min}
873
874       Escape sequences
875
876       Because patterns are processed as double-quoted strings, the following
877       also work:
878
879        \t          tab                   (HT, TAB)
880        \n          newline               (LF, NL)
881        \r          return                (CR)
882        \f          form feed             (FF)
883        \a          alarm (bell)          (BEL)
884        \e          escape (think troff)  (ESC)
885        \cK         control char          (example: VT)
886        \x{}, \x00  character whose ordinal is the given hexadecimal number
887        \N{name}    named Unicode character or character sequence
888        \N{U+263D}  Unicode character     (example: FIRST QUARTER MOON)
889        \o{}, \000  character whose ordinal is the given octal number
890        \l          lowercase next char (think vi)
891        \u          uppercase next char (think vi)
892        \L          lowercase until \E (think vi)
893        \U          uppercase until \E (think vi)
894        \Q          quote (disable) pattern metacharacters until \E
895        \E          end either case modification or quoted section, think vi
896
897       Details are in "Quote and Quote-like Operators" in perlop.
898
899       Character Classes and other Special Escapes
900
901       In addition, Perl defines the following:
902
903        Sequence   Note    Description
904         [...]     [1]  Match a character according to the rules of the
905                          bracketed character class defined by the "...".
906                          Example: [a-z] matches "a" or "b" or "c" ... or "z"
907         [[:...:]] [2]  Match a character according to the rules of the POSIX
908                          character class "..." within the outer bracketed
909                          character class.  Example: [[:upper:]] matches any
910                          uppercase character.
911         (?[...])  [8]  Extended bracketed character class
912         \w        [3]  Match a "word" character (alphanumeric plus "_", plus
913                          other connector punctuation chars plus Unicode
914                          marks)
915         \W        [3]  Match a non-"word" character
916         \s        [3]  Match a whitespace character
917         \S        [3]  Match a non-whitespace character
918         \d        [3]  Match a decimal digit character
919         \D        [3]  Match a non-digit character
920         \pP       [3]  Match P, named property.  Use \p{Prop} for longer names
921         \PP       [3]  Match non-P
922         \X        [4]  Match Unicode "eXtended grapheme cluster"
923         \1        [5]  Backreference to a specific capture group or buffer.
924                          '1' may actually be any positive integer.
925         \g1       [5]  Backreference to a specific or previous group,
926         \g{-1}    [5]  The number may be negative indicating a relative
927                          previous group and may optionally be wrapped in
928                          curly brackets for safer parsing.
929         \g{name}  [5]  Named backreference
930         \k<name>  [5]  Named backreference
931         \K        [6]  Keep the stuff left of the \K, don't include it in $&
932         \N        [7]  Any character but \n.  Not affected by /s modifier
933         \v        [3]  Vertical whitespace
934         \V        [3]  Not vertical whitespace
935         \h        [3]  Horizontal whitespace
936         \H        [3]  Not horizontal whitespace
937         \R        [4]  Linebreak
938
939       [1] See "Bracketed Character Classes" in perlrecharclass for details.
940
941       [2] See "POSIX Character Classes" in perlrecharclass for details.
942
943       [3] See "Unicode Character Properties" in perlunicode for details
944
945       [4] See "Misc" in perlrebackslash for details.
946
947       [5] See "Capture groups" below for details.
948
949       [6] See "Extended Patterns" below for details.
950
951       [7] Note that "\N" has two meanings.  When of the form "\N{NAME}", it
952           matches the character or character sequence whose name is NAME; and
953           similarly when of the form "\N{U+hex}", it matches the character
954           whose Unicode code point is hex.  Otherwise it matches any
955           character but "\n".
956
957       [8] See "Extended Bracketed Character Classes" in perlrecharclass for
958           details.
959
960       Assertions
961
962       Besides "^" and "$", Perl defines the following zero-width assertions:
963
964        \b{}   Match at Unicode boundary of specified type
965        \B{}   Match where corresponding \b{} doesn't match
966        \b     Match a \w\W or \W\w boundary
967        \B     Match except at a \w\W or \W\w boundary
968        \A     Match only at beginning of string
969        \Z     Match only at end of string, or before newline at the end
970        \z     Match only at end of string
971        \G     Match only at pos() (e.g. at the end-of-match position
972               of prior m//g)
973
974       A Unicode boundary ("\b{}"), available starting in v5.22, is a spot
975       between two characters, or before the first character in the string, or
976       after the final character in the string where certain criteria defined
977       by Unicode are met.  See "\b{}, \b, \B{}, \B" in perlrebackslash for
978       details.
979
980       A word boundary ("\b") is a spot between two characters that has a "\w"
981       on one side of it and a "\W" on the other side of it (in either order),
982       counting the imaginary characters off the beginning and end of the
983       string as matching a "\W".  (Within character classes "\b" represents
984       backspace rather than a word boundary, just as it normally does in any
985       double-quoted string.)  The "\A" and "\Z" are just like "^" and "$",
986       except that they won't match multiple times when the "/m" modifier is
987       used, while "^" and "$" will match at every internal line boundary.  To
988       match the actual end of the string and not ignore an optional trailing
989       newline, use "\z".
990
991       The "\G" assertion can be used to chain global matches (using "m//g"),
992       as described in "Regexp Quote-Like Operators" in perlop.  It is also
993       useful when writing "lex"-like scanners, when you have several patterns
994       that you want to match against consequent substrings of your string;
995       see the previous reference.  The actual location where "\G" will match
996       can also be influenced by using "pos()" as an lvalue: see "pos" in
997       perlfunc. Note that the rule for zero-length matches (see "Repeated
998       Patterns Matching a Zero-length Substring") is modified somewhat, in
999       that contents to the left of "\G" are not counted when determining the
1000       length of the match. Thus the following will not match forever:
1001
1002            my $string = 'ABC';
1003            pos($string) = 1;
1004            while ($string =~ /(.\G)/g) {
1005                print $1;
1006            }
1007
1008       It will print 'A' and then terminate, as it considers the match to be
1009       zero-width, and thus will not match at the same position twice in a
1010       row.
1011
1012       It is worth noting that "\G" improperly used can result in an infinite
1013       loop. Take care when using patterns that include "\G" in an
1014       alternation.
1015
1016       Note also that "s///" will refuse to overwrite part of a substitution
1017       that has already been replaced; so for example this will stop after the
1018       first iteration, rather than iterating its way backwards through the
1019       string:
1020
1021           $_ = "123456789";
1022           pos = 6;
1023           s/.(?=.\G)/X/g;
1024           print;      # prints 1234X6789, not XXXXX6789
1025
1026       Capture groups
1027
1028       The grouping construct "( ... )" creates capture groups (also referred
1029       to as capture buffers). To refer to the current contents of a group
1030       later on, within the same pattern, use "\g1" (or "\g{1}") for the
1031       first, "\g2" (or "\g{2}") for the second, and so on.  This is called a
1032       backreference.
1033
1034
1035
1036
1037
1038
1039
1040
1041       There is no limit to the number of captured substrings that you may
1042       use.  Groups are numbered with the leftmost open parenthesis being
1043       number 1, etc.  If a group did not match, the associated backreference
1044       won't match either. (This can happen if the group is optional, or in a
1045       different branch of an alternation.)  You can omit the "g", and write
1046       "\1", etc, but there are some issues with this form, described below.
1047
1048       You can also refer to capture groups relatively, by using a negative
1049       number, so that "\g-1" and "\g{-1}" both refer to the immediately
1050       preceding capture group, and "\g-2" and "\g{-2}" both refer to the
1051       group before it.  For example:
1052
1053               /
1054                (Y)            # group 1
1055                (              # group 2
1056                   (X)         # group 3
1057                   \g{-1}      # backref to group 3
1058                   \g{-3}      # backref to group 1
1059                )
1060               /x
1061
1062       would match the same as "/(Y) ( (X) \g3 \g1 )/x".  This allows you to
1063       interpolate regexes into larger regexes and not have to worry about the
1064       capture groups being renumbered.
1065
1066       You can dispense with numbers altogether and create named capture
1067       groups.  The notation is "(?<name>...)" to declare and "\g{name}" to
1068       reference.  (To be compatible with .Net regular expressions, "\g{name}"
1069       may also be written as "\k{name}", "\k<name>" or "\k'name'".)  name
1070       must not begin with a number, nor contain hyphens.  When different
1071       groups within the same pattern have the same name, any reference to
1072       that name assumes the leftmost defined group.  Named groups count in
1073       absolute and relative numbering, and so can also be referred to by
1074       those numbers.  (It's possible to do things with named capture groups
1075       that would otherwise require "(??{})".)
1076
1077       Capture group contents are dynamically scoped and available to you
1078       outside the pattern until the end of the enclosing block or until the
1079       next successful match, whichever comes first.  (See "Compound
1080       Statements" in perlsyn.)  You can refer to them by absolute number
1081       (using "$1" instead of "\g1", etc); or by name via the "%+" hash, using
1082       "$+{name}".
1083
1084       Braces are required in referring to named capture groups, but are
1085       optional for absolute or relative numbered ones.  Braces are safer when
1086       creating a regex by concatenating smaller strings.  For example if you
1087       have "qr/$a$b/", and $a contained "\g1", and $b contained "37", you
1088       would get "/\g137/" which is probably not what you intended.
1089
1090       The "\g" and "\k" notations were introduced in Perl 5.10.0.  Prior to
1091       that there were no named nor relative numbered capture groups.
1092       Absolute numbered groups were referred to using "\1", "\2", etc., and
1093       this notation is still accepted (and likely always will be).  But it
1094       leads to some ambiguities if there are more than 9 capture groups, as
1095       "\10" could mean either the tenth capture group, or the character whose
1096       ordinal in octal is 010 (a backspace in ASCII).  Perl resolves this
1097       ambiguity by interpreting "\10" as a backreference only if at least 10
1098       left parentheses have opened before it.  Likewise "\11" is a
1099       backreference only if at least 11 left parentheses have opened before
1100       it.  And so on.  "\1" through "\9" are always interpreted as
1101       backreferences.  There are several examples below that illustrate these
1102       perils.  You can avoid the ambiguity by always using "\g{}" or "\g" if
1103       you mean capturing groups; and for octal constants always using "\o{}",
1104       or for "\077" and below, using 3 digits padded with leading zeros,
1105       since a leading zero implies an octal constant.
1106
1107       The "\digit" notation also works in certain circumstances outside the
1108       pattern.  See "Warning on \1 Instead of $1" below for details.
1109
1110       Examples:
1111
1112           s/^([^ ]*) *([^ ]*)/$2 $1/;     # swap first two words
1113
1114           /(.)\g1/                        # find first doubled char
1115                and print "'$1' is the first doubled character\n";
1116
1117           /(?<char>.)\k<char>/            # ... a different way
1118                and print "'$+{char}' is the first doubled character\n";
1119
1120           /(?'char'.)\g1/                 # ... mix and match
1121                and print "'$1' is the first doubled character\n";
1122
1123           if (/Time: (..):(..):(..)/) {   # parse out values
1124               $hours = $1;
1125               $minutes = $2;
1126               $seconds = $3;
1127           }
1128
1129           /(.)(.)(.)(.)(.)(.)(.)(.)(.)\g10/   # \g10 is a backreference
1130           /(.)(.)(.)(.)(.)(.)(.)(.)(.)\10/    # \10 is octal
1131           /((.)(.)(.)(.)(.)(.)(.)(.)(.))\10/  # \10 is a backreference
1132           /((.)(.)(.)(.)(.)(.)(.)(.)(.))\010/ # \010 is octal
1133
1134           $a = '(.)\1';        # Creates problems when concatenated.
1135           $b = '(.)\g{1}';     # Avoids the problems.
1136           "aa" =~ /${a}/;      # True
1137           "aa" =~ /${b}/;      # True
1138           "aa0" =~ /${a}0/;    # False!
1139           "aa0" =~ /${b}0/;    # True
1140           "aa\x08" =~ /${a}0/;  # True!
1141           "aa\x08" =~ /${b}0/;  # False
1142
1143       Several special variables also refer back to portions of the previous
1144       match.  $+ returns whatever the last bracket match matched.  $& returns
1145       the entire matched string.  (At one point $0 did also, but now it
1146       returns the name of the program.)  "$`" returns everything before the
1147       matched string.  "$'" returns everything after the matched string. And
1148       $^N contains whatever was matched by the most-recently closed group
1149       (submatch). $^N can be used in extended patterns (see below), for
1150       example to assign a submatch to a variable.
1151
1152       These special variables, like the "%+" hash and the numbered match
1153       variables ($1, $2, $3, etc.) are dynamically scoped until the end of
1154       the enclosing block or until the next successful match, whichever comes
1155       first.  (See "Compound Statements" in perlsyn.)
1156
1157       NOTE: Failed matches in Perl do not reset the match variables, which
1158       makes it easier to write code that tests for a series of more specific
1159       cases and remembers the best match.
1160
1161       WARNING: If your code is to run on Perl 5.16 or earlier, beware that
1162       once Perl sees that you need one of $&, "$`", or "$'" anywhere in the
1163       program, it has to provide them for every pattern match.  This may
1164       substantially slow your program.
1165
1166       Perl uses the same mechanism to produce $1, $2, etc, so you also pay a
1167       price for each pattern that contains capturing parentheses.  (To avoid
1168       this cost while retaining the grouping behaviour, use the extended
1169       regular expression "(?: ... )" instead.)  But if you never use $&, "$`"
1170       or "$'", then patterns without capturing parentheses will not be
1171       penalized.  So avoid $&, "$'", and "$`" if you can, but if you can't
1172       (and some algorithms really appreciate them), once you've used them
1173       once, use them at will, because you've already paid the price.
1174
1175       Perl 5.16 introduced a slightly more efficient mechanism that notes
1176       separately whether each of "$`", $&, and "$'" have been seen, and thus
1177       may only need to copy part of the string.  Perl 5.20 introduced a much
1178       more efficient copy-on-write mechanism which eliminates any slowdown.
1179
1180       As another workaround for this problem, Perl 5.10.0 introduced
1181       "${^PREMATCH}", "${^MATCH}" and "${^POSTMATCH}", which are equivalent
1182       to "$`", $& and "$'", except that they are only guaranteed to be
1183       defined after a successful match that was executed with the "/p"
1184       (preserve) modifier.  The use of these variables incurs no global
1185       performance penalty, unlike their punctuation character equivalents,
1186       however at the trade-off that you have to tell perl when you want to
1187       use them.  As of Perl 5.20, these three variables are equivalent to
1188       "$`", $& and "$'", and "/p" is ignored.
1189
1190   Quoting metacharacters
1191       Backslashed metacharacters in Perl are alphanumeric, such as "\b",
1192       "\w", "\n".  Unlike some other regular expression languages, there are
1193       no backslashed symbols that aren't alphanumeric.  So anything that
1194       looks like "\\", "\(", "\)", "\[", "\]", "\{", or "\}" is always
1195       interpreted as a literal character, not a metacharacter.  This was once
1196       used in a common idiom to disable or quote the special meanings of
1197       regular expression metacharacters in a string that you want to use for
1198       a pattern. Simply quote all non-"word" characters:
1199
1200           $pattern =~ s/(\W)/\\$1/g;
1201
1202       (If "use locale" is set, then this depends on the current locale.)
1203       Today it is more common to use the "quotemeta()" function or the "\Q"
1204       metaquoting escape sequence to disable all metacharacters' special
1205       meanings like this:
1206
1207           /$unquoted\Q$quoted\E$unquoted/
1208
1209       Beware that if you put literal backslashes (those not inside
1210       interpolated variables) between "\Q" and "\E", double-quotish backslash
1211       interpolation may lead to confusing results.  If you need to use
1212       literal backslashes within "\Q...\E", consult "Gory details of parsing
1213       quoted constructs" in perlop.
1214
1215       "quotemeta()" and "\Q" are fully described in "quotemeta" in perlfunc.
1216
1217   Extended Patterns
1218       Perl also defines a consistent extension syntax for features not found
1219       in standard tools like awk and lex.  The syntax for most of these is a
1220       pair of parentheses with a question mark as the first thing within the
1221       parentheses.  The character after the question mark indicates the
1222       extension.
1223
1224       A question mark was chosen for this and for the minimal-matching
1225       construct because 1) question marks are rare in older regular
1226       expressions, and 2) whenever you see one, you should stop and
1227       "question" exactly what is going on.  That's psychology....
1228
1229       "(?#text)"
1230           A comment.  The text is ignored.  Note that Perl closes the comment
1231           as soon as it sees a ")", so there is no way to put a literal ")"
1232           in the comment.  The pattern's closing delimiter must be escaped by
1233           a backslash if it appears in the comment.
1234
1235           See "/x" for another way to have comments in patterns.
1236
1237           Note that a comment can go just about anywhere, except in the
1238           middle of an escape sequence.   Examples:
1239
1240            qr/foo(?#comment)bar/'  # Matches 'foobar'
1241
1242            # The pattern below matches 'abcd', 'abccd', or 'abcccd'
1243            qr/abc(?#comment between literal and its quantifier){1,3}d/
1244
1245            # The pattern below generates a syntax error, because the '\p' must
1246            # be followed immediately by a '{'.
1247            qr/\p(?#comment between \p and its property name){Any}/
1248
1249            # The pattern below generates a syntax error, because the initial
1250            # '\(' is a literal opening parenthesis, and so there is nothing
1251            # for the  closing ')' to match
1252            qr/\(?#the backslash means this isn't a comment)p{Any}/
1253
1254            # Comments can be used to fold long patterns into multiple lines
1255            qr/First part of a long regex(?#
1256              )remaining part/
1257
1258       "(?adlupimnsx-imnsx)"
1259       "(?^alupimnsx)"
1260           Zero or more embedded pattern-match modifiers, to be turned on (or
1261           turned off if preceded by "-") for the remainder of the pattern or
1262           the remainder of the enclosing pattern group (if any).
1263
1264           This is particularly useful for dynamically-generated patterns,
1265           such as those read in from a configuration file, taken from an
1266           argument, or specified in a table somewhere.  Consider the case
1267           where some patterns want to be case-sensitive and some do not:  The
1268           case-insensitive ones merely need to include "(?i)" at the front of
1269           the pattern.  For example:
1270
1271               $pattern = "foobar";
1272               if ( /$pattern/i ) { }
1273
1274               # more flexible:
1275
1276               $pattern = "(?i)foobar";
1277               if ( /$pattern/ ) { }
1278
1279           These modifiers are restored at the end of the enclosing group. For
1280           example,
1281
1282               ( (?i) blah ) \s+ \g1
1283
1284           will match "blah" in any case, some spaces, and an exact (including
1285           the case!)  repetition of the previous word, assuming the "/x"
1286           modifier, and no "/i" modifier outside this group.
1287
1288           These modifiers do not carry over into named subpatterns called in
1289           the enclosing group. In other words, a pattern such as
1290           "((?i)(?&NAME))" does not change the case-sensitivity of the NAME
1291           pattern.
1292
1293           A modifier is overridden by later occurrences of this construct in
1294           the same scope containing the same modifier, so that
1295
1296               /((?im)foo(?-m)bar)/
1297
1298           matches all of "foobar" case insensitively, but uses "/m" rules for
1299           only the "foo" portion.  The "a" flag overrides "aa" as well;
1300           likewise "aa" overrides "a".  The same goes for "x" and "xx".
1301           Hence, in
1302
1303               /(?-x)foo/xx
1304
1305           both "/x" and "/xx" are turned off during matching "foo".  And in
1306
1307               /(?x)foo/x
1308
1309           "/x" but NOT "/xx" is turned on for matching "foo".  (One might
1310           mistakenly think that since the inner "(?x)" is already in the
1311           scope of "/x", that the result would effectively be the sum of
1312           them, yielding "/xx".  It doesn't work that way.)  Similarly, doing
1313           something like "(?xx-x)foo" turns off all "x" behavior for matching
1314           "foo", it is not that you subtract 1 "x" from 2 to get 1 "x"
1315           remaining.
1316
1317           Any of these modifiers can be set to apply globally to all regular
1318           expressions compiled within the scope of a "use re".  See "'/flags'
1319           mode" in re.
1320
1321           Starting in Perl 5.14, a "^" (caret or circumflex accent)
1322           immediately after the "?" is a shorthand equivalent to "d-imnsx".
1323           Flags (except "d") may follow the caret to override it.  But a
1324           minus sign is not legal with it.
1325
1326           Note that the "a", "d", "l", "p", and "u" modifiers are special in
1327           that they can only be enabled, not disabled, and the "a", "d", "l",
1328           and "u" modifiers are mutually exclusive: specifying one de-
1329           specifies the others, and a maximum of one (or two "a"'s) may
1330           appear in the construct.  Thus, for example, "(?-p)" will warn when
1331           compiled under "use warnings"; "(?-d:...)" and "(?dl:...)" are
1332           fatal errors.
1333
1334           Note also that the "p" modifier is special in that its presence
1335           anywhere in a pattern has a global effect.
1336
1337           Having zero modifiers makes this a no-op (so why did you specify
1338           it, unless it's generated code), and starting in v5.30, warns under
1339           "use re 'strict'".
1340
1341       "(?:pattern)"
1342       "(?adluimnsx-imnsx:pattern)"
1343       "(?^aluimnsx:pattern)"
1344           This is for clustering, not capturing; it groups subexpressions
1345           like "()", but doesn't make backreferences as "()" does.  So
1346
1347               @fields = split(/\b(?:a|b|c)\b/)
1348
1349           matches the same field delimiters as
1350
1351               @fields = split(/\b(a|b|c)\b/)
1352
1353           but doesn't spit out the delimiters themselves as extra fields
1354           (even though that's the behaviour of "split" in perlfunc when its
1355           pattern contains capturing groups).  It's also cheaper not to
1356           capture characters if you don't need to.
1357
1358           Any letters between "?" and ":" act as flags modifiers as with
1359           "(?adluimnsx-imnsx)".  For example,
1360
1361               /(?s-i:more.*than).*million/i
1362
1363           is equivalent to the more verbose
1364
1365               /(?:(?s-i)more.*than).*million/i
1366
1367           Note that any "()" constructs enclosed within this one will still
1368           capture unless the "/n" modifier is in effect.
1369
1370           Like the "(?adlupimnsx-imnsx)" construct, "aa" and "a" override
1371           each other, as do "xx" and "x".  They are not additive.  So, doing
1372           something like "(?xx-x:foo)" turns off all "x" behavior for
1373           matching "foo".
1374
1375           Starting in Perl 5.14, a "^" (caret or circumflex accent)
1376           immediately after the "?" is a shorthand equivalent to "d-imnsx".
1377           Any positive flags (except "d") may follow the caret, so
1378
1379               (?^x:foo)
1380
1381           is equivalent to
1382
1383               (?x-imns:foo)
1384
1385           The caret tells Perl that this cluster doesn't inherit the flags of
1386           any surrounding pattern, but uses the system defaults ("d-imnsx"),
1387           modified by any flags specified.
1388
1389           The caret allows for simpler stringification of compiled regular
1390           expressions.  These look like
1391
1392               (?^:pattern)
1393
1394           with any non-default flags appearing between the caret and the
1395           colon.  A test that looks at such stringification thus doesn't need
1396           to have the system default flags hard-coded in it, just the caret.
1397           If new flags are added to Perl, the meaning of the caret's
1398           expansion will change to include the default for those flags, so
1399           the test will still work, unchanged.
1400
1401           Specifying a negative flag after the caret is an error, as the flag
1402           is redundant.
1403
1404           Mnemonic for "(?^...)":  A fresh beginning since the usual use of a
1405           caret is to match at the beginning.
1406
1407       "(?|pattern)"
1408           This is the "branch reset" pattern, which has the special property
1409           that the capture groups are numbered from the same starting point
1410           in each alternation branch. It is available starting from perl
1411           5.10.0.
1412
1413           Capture groups are numbered from left to right, but inside this
1414           construct the numbering is restarted for each branch.
1415
1416           The numbering within each branch will be as normal, and any groups
1417           following this construct will be numbered as though the construct
1418           contained only one branch, that being the one with the most capture
1419           groups in it.
1420
1421           This construct is useful when you want to capture one of a number
1422           of alternative matches.
1423
1424           Consider the following pattern.  The numbers underneath show in
1425           which group the captured content will be stored.
1426
1427               # before  ---------------branch-reset----------- after
1428               / ( a )  (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1429               # 1            2         2  3        2     3     4
1430
1431           Be careful when using the branch reset pattern in combination with
1432           named captures. Named captures are implemented as being aliases to
1433           numbered groups holding the captures, and that interferes with the
1434           implementation of the branch reset pattern. If you are using named
1435           captures in a branch reset pattern, it's best to use the same
1436           names, in the same order, in each of the alternations:
1437
1438              /(?|  (?<a> x ) (?<b> y )
1439                 |  (?<a> z ) (?<b> w )) /x
1440
1441           Not doing so may lead to surprises:
1442
1443             "12" =~ /(?| (?<a> \d+ ) | (?<b> \D+))/x;
1444             say $+{a};    # Prints '12'
1445             say $+{b};    # *Also* prints '12'.
1446
1447           The problem here is that both the group named "a" and the group
1448           named "b" are aliases for the group belonging to $1.
1449
1450       Lookaround Assertions
1451           Lookaround assertions are zero-width patterns which match a
1452           specific pattern without including it in $&. Positive assertions
1453           match when their subpattern matches, negative assertions match when
1454           their subpattern fails. Lookbehind matches text up to the current
1455           match position, lookahead matches text following the current match
1456           position.
1457
1458           "(?=pattern)"
1459           "(*pla:pattern)"
1460           "(*positive_lookahead:pattern)"
1461               A zero-width positive lookahead assertion.  For example,
1462               "/\w+(?=\t)/" matches a word followed by a tab, without
1463               including the tab in $&.
1464
1465               The alphabetic forms are experimental; using them yields a
1466               warning in the "experimental::alpha_assertions" category.
1467
1468           "(?!pattern)"
1469           "(*nla:pattern)"
1470           "(*negative_lookahead:pattern)"
1471               A zero-width negative lookahead assertion.  For example
1472               "/foo(?!bar)/" matches any occurrence of "foo" that isn't
1473               followed by "bar".  Note however that lookahead and lookbehind
1474               are NOT the same thing.  You cannot use this for lookbehind.
1475
1476               If you are looking for a "bar" that isn't preceded by a "foo",
1477               "/(?!foo)bar/" will not do what you want.  That's because the
1478               "(?!foo)" is just saying that the next thing cannot be
1479               "foo"--and it's not, it's a "bar", so "foobar" will match.  Use
1480               lookbehind instead (see below).
1481
1482               The alphabetic forms are experimental; using them yields a
1483               warning in the "experimental::alpha_assertions" category.
1484
1485           "(?<=pattern)"
1486           "\K"
1487           "(*plb:pattern)"
1488           "(*positive_lookbehind:pattern)"
1489               A zero-width positive lookbehind assertion.  For example,
1490               "/(?<=\t)\w+/" matches a word that follows a tab, without
1491               including the tab in $&.
1492
1493               Prior to Perl 5.30, it worked only for fixed-width lookbehind,
1494               but starting in that release, it can handle variable lengths
1495               from 1 to 255 characters as an experimental feature.  The
1496               feature is enabled automatically if you use a variable length
1497               lookbehind assertion, but will raise a warning at pattern
1498               compilation time, unless turned off, in the "experimental::vlb"
1499               category.  This is to warn you that the exact behavior is
1500               subject to change should feedback from actual use in the field
1501               indicate to do so; or even complete removal if the problems
1502               found are not practically surmountable.  You can achieve close
1503               to pre-5.30 behavior by fatalizing warnings in this category.
1504
1505               There is a special form of this construct, called "\K"
1506               (available since Perl 5.10.0), which causes the regex engine to
1507               "keep" everything it had matched prior to the "\K" and not
1508               include it in $&. This effectively provides non-experimental
1509               variable-length lookbehind of any length.
1510
1511               And, there is a technique that can be used to handle variable
1512               length lookbehinds on earlier releases, and longer than 255
1513               characters.  It is described in
1514               <http://www.drregex.com/2019/02/variable-length-lookbehinds-actually.html>.
1515
1516               Note that under "/i", a few single characters match two or
1517               three other characters.  This makes them variable length, and
1518               the 255 length applies to the maximum number of characters in
1519               the match.  For example "qr/\N{LATIN SMALL LETTER SHARP S}/i"
1520               matches the sequence "ss".  Your lookbehind assertion could
1521               contain 127 Sharp S characters under "/i", but adding a 128th
1522               would generate a compilation error, as that could match 256 "s"
1523               characters in a row.
1524
1525               The use of "\K" inside of another lookaround assertion is
1526               allowed, but the behaviour is currently not well defined.
1527
1528               For various reasons "\K" may be significantly more efficient
1529               than the equivalent "(?<=...)" construct, and it is especially
1530               useful in situations where you want to efficiently remove
1531               something following something else in a string. For instance
1532
1533                 s/(foo)bar/$1/g;
1534
1535               can be rewritten as the much more efficient
1536
1537                 s/foo\Kbar//g;
1538
1539               Use of the non-greedy modifier "?" may not give you the
1540               expected results if it is within a capturing group within the
1541               construct.
1542
1543               The alphabetic forms (not including "\K" are experimental;
1544               using them yields a warning in the
1545               "experimental::alpha_assertions" category.
1546
1547           "(?<!pattern)"
1548           "(*nlb:pattern)"
1549           "(*negative_lookbehind:pattern)"
1550               A zero-width negative lookbehind assertion.  For example
1551               "/(?<!bar)foo/" matches any occurrence of "foo" that does not
1552               follow "bar".
1553
1554               Prior to Perl 5.30, it worked only for fixed-width lookbehind,
1555               but starting in that release, it can handle variable lengths
1556               from 1 to 255 characters as an experimental feature.  The
1557               feature is enabled automatically if you use a variable length
1558               lookbehind assertion, but will raise a warning at pattern
1559               compilation time, unless turned off, in the "experimental::vlb"
1560               category.  This is to warn you that the exact behavior is
1561               subject to change should feedback from actual use in the field
1562               indicate to do so; or even complete removal if the problems
1563               found are not practically surmountable.  You can achieve close
1564               to pre-5.30 behavior by fatalizing warnings in this category.
1565
1566               There is a technique that can be used to handle variable length
1567               lookbehinds on earlier releases, and longer than 255
1568               characters.  It is described in
1569               <http://www.drregex.com/2019/02/variable-length-lookbehinds-actually.html>.
1570
1571               Note that under "/i", a few single characters match two or
1572               three other characters.  This makes them variable length, and
1573               the 255 length applies to the maximum number of characters in
1574               the match.  For example "qr/\N{LATIN SMALL LETTER SHARP S}/i"
1575               matches the sequence "ss".  Your lookbehind assertion could
1576               contain 127 Sharp S characters under "/i", but adding a 128th
1577               would generate a compilation error, as that could match 256 "s"
1578               characters in a row.
1579
1580               Use of the non-greedy modifier "?" may not give you the
1581               expected results if it is within a capturing group within the
1582               construct.
1583
1584               The alphabetic forms are experimental; using them yields a
1585               warning in the "experimental::alpha_assertions" category.
1586
1587       "(?<NAME>pattern)"
1588       "(?'NAME'pattern)"
1589           A named capture group. Identical in every respect to normal
1590           capturing parentheses "()" but for the additional fact that the
1591           group can be referred to by name in various regular expression
1592           constructs (like "\g{NAME}") and can be accessed by name after a
1593           successful match via "%+" or "%-". See perlvar for more details on
1594           the "%+" and "%-" hashes.
1595
1596           If multiple distinct capture groups have the same name, then
1597           $+{NAME} will refer to the leftmost defined group in the match.
1598
1599           The forms "(?'NAME'pattern)" and "(?<NAME>pattern)" are equivalent.
1600
1601           NOTE: While the notation of this construct is the same as the
1602           similar function in .NET regexes, the behavior is not. In Perl the
1603           groups are numbered sequentially regardless of being named or not.
1604           Thus in the pattern
1605
1606             /(x)(?<foo>y)(z)/
1607
1608           $+{foo} will be the same as $2, and $3 will contain 'z' instead of
1609           the opposite which is what a .NET regex hacker might expect.
1610
1611           Currently NAME is restricted to simple identifiers only.  In other
1612           words, it must match "/^[_A-Za-z][_A-Za-z0-9]*\z/" or its Unicode
1613           extension (see utf8), though it isn't extended by the locale (see
1614           perllocale).
1615
1616           NOTE: In order to make things easier for programmers with
1617           experience with the Python or PCRE regex engines, the pattern
1618           "(?P<NAME>pattern)" may be used instead of "(?<NAME>pattern)";
1619           however this form does not support the use of single quotes as a
1620           delimiter for the name.
1621
1622       "\k<NAME>"
1623       "\k'NAME'"
1624           Named backreference. Similar to numeric backreferences, except that
1625           the group is designated by name and not number. If multiple groups
1626           have the same name then it refers to the leftmost defined group in
1627           the current match.
1628
1629           It is an error to refer to a name not defined by a "(?<NAME>)"
1630           earlier in the pattern.
1631
1632           Both forms are equivalent.
1633
1634           NOTE: In order to make things easier for programmers with
1635           experience with the Python or PCRE regex engines, the pattern
1636           "(?P=NAME)" may be used instead of "\k<NAME>".
1637
1638       "(?{ code })"
1639           WARNING: Using this feature safely requires that you understand its
1640           limitations.  Code executed that has side effects may not perform
1641           identically from version to version due to the effect of future
1642           optimisations in the regex engine.  For more information on this,
1643           see "Embedded Code Execution Frequency".
1644
1645           This zero-width assertion executes any embedded Perl code.  It
1646           always succeeds, and its return value is set as $^R.
1647
1648           In literal patterns, the code is parsed at the same time as the
1649           surrounding code. While within the pattern, control is passed
1650           temporarily back to the perl parser, until the logically-balancing
1651           closing brace is encountered. This is similar to the way that an
1652           array index expression in a literal string is handled, for example
1653
1654               "abc$array[ 1 + f('[') + g()]def"
1655
1656           In particular, braces do not need to be balanced:
1657
1658               s/abc(?{ f('{'); })/def/
1659
1660           Even in a pattern that is interpolated and compiled at run-time,
1661           literal code blocks will be compiled once, at perl compile time;
1662           the following prints "ABCD":
1663
1664               print "D";
1665               my $qr = qr/(?{ BEGIN { print "A" } })/;
1666               my $foo = "foo";
1667               /$foo$qr(?{ BEGIN { print "B" } })/;
1668               BEGIN { print "C" }
1669
1670           In patterns where the text of the code is derived from run-time
1671           information rather than appearing literally in a source code
1672           /pattern/, the code is compiled at the same time that the pattern
1673           is compiled, and for reasons of security, "use re 'eval'" must be
1674           in scope. This is to stop user-supplied patterns containing code
1675           snippets from being executable.
1676
1677           In situations where you need to enable this with "use re 'eval'",
1678           you should also have taint checking enabled.  Better yet, use the
1679           carefully constrained evaluation within a Safe compartment.  See
1680           perlsec for details about both these mechanisms.
1681
1682           From the viewpoint of parsing, lexical variable scope and closures,
1683
1684               /AAA(?{ BBB })CCC/
1685
1686           behaves approximately like
1687
1688               /AAA/ && do { BBB } && /CCC/
1689
1690           Similarly,
1691
1692               qr/AAA(?{ BBB })CCC/
1693
1694           behaves approximately like
1695
1696               sub { /AAA/ && do { BBB } && /CCC/ }
1697
1698           In particular:
1699
1700               { my $i = 1; $r = qr/(?{ print $i })/ }
1701               my $i = 2;
1702               /$r/; # prints "1"
1703
1704           Inside a "(?{...})" block, $_ refers to the string the regular
1705           expression is matching against. You can also use "pos()" to know
1706           what is the current position of matching within this string.
1707
1708           The code block introduces a new scope from the perspective of
1709           lexical variable declarations, but not from the perspective of
1710           "local" and similar localizing behaviours. So later code blocks
1711           within the same pattern will still see the values which were
1712           localized in earlier blocks.  These accumulated localizations are
1713           undone either at the end of a successful match, or if the assertion
1714           is backtracked (compare "Backtracking"). For example,
1715
1716             $_ = 'a' x 8;
1717             m<
1718                (?{ $cnt = 0 })               # Initialize $cnt.
1719                (
1720                  a
1721                  (?{
1722                      local $cnt = $cnt + 1;  # Update $cnt,
1723                                              # backtracking-safe.
1724                  })
1725                )*
1726                aaaa
1727                (?{ $res = $cnt })            # On success copy to
1728                                              # non-localized location.
1729              >x;
1730
1731           will initially increment $cnt up to 8; then during backtracking,
1732           its value will be unwound back to 4, which is the value assigned to
1733           $res.  At the end of the regex execution, $cnt will be wound back
1734           to its initial value of 0.
1735
1736           This assertion may be used as the condition in a
1737
1738               (?(condition)yes-pattern|no-pattern)
1739
1740           switch.  If not used in this way, the result of evaluation of code
1741           is put into the special variable $^R.  This happens immediately, so
1742           $^R can be used from other "(?{ code })" assertions inside the same
1743           regular expression.
1744
1745           The assignment to $^R above is properly localized, so the old value
1746           of $^R is restored if the assertion is backtracked; compare
1747           "Backtracking".
1748
1749           Note that the special variable $^N  is particularly useful with
1750           code blocks to capture the results of submatches in variables
1751           without having to keep track of the number of nested parentheses.
1752           For example:
1753
1754             $_ = "The brown fox jumps over the lazy dog";
1755             /the (\S+)(?{ $color = $^N }) (\S+)(?{ $animal = $^N })/i;
1756             print "color = $color, animal = $animal\n";
1757
1758       "(??{ code })"
1759           WARNING: Using this feature safely requires that you understand its
1760           limitations.  Code executed that has side effects may not perform
1761           identically from version to version due to the effect of future
1762           optimisations in the regex engine.  For more information on this,
1763           see "Embedded Code Execution Frequency".
1764
1765           This is a "postponed" regular subexpression.  It behaves in exactly
1766           the same way as a "(?{ code })" code block as described above,
1767           except that its return value, rather than being assigned to $^R, is
1768           treated as a pattern, compiled if it's a string (or used as-is if
1769           its a qr// object), then matched as if it were inserted instead of
1770           this construct.
1771
1772           During the matching of this sub-pattern, it has its own set of
1773           captures which are valid during the sub-match, but are discarded
1774           once control returns to the main pattern. For example, the
1775           following matches, with the inner pattern capturing "B" and
1776           matching "BB", while the outer pattern captures "A";
1777
1778               my $inner = '(.)\1';
1779               "ABBA" =~ /^(.)(??{ $inner })\1/;
1780               print $1; # prints "A";
1781
1782           Note that this means that  there is no way for the inner pattern to
1783           refer to a capture group defined outside.  (The code block itself
1784           can use $1, etc., to refer to the enclosing pattern's capture
1785           groups.)  Thus, although
1786
1787               ('a' x 100)=~/(??{'(.)' x 100})/
1788
1789           will match, it will not set $1 on exit.
1790
1791           The following pattern matches a parenthesized group:
1792
1793            $re = qr{
1794                       \(
1795                       (?:
1796                          (?> [^()]+ )  # Non-parens without backtracking
1797                        |
1798                          (??{ $re })   # Group with matching parens
1799                       )*
1800                       \)
1801                    }x;
1802
1803           See also "(?PARNO)" for a different, more efficient way to
1804           accomplish the same task.
1805
1806           Executing a postponed regular expression too many times without
1807           consuming any input string will also result in a fatal error.  The
1808           depth at which that happens is compiled into perl, so it can be
1809           changed with a custom build.
1810
1811       "(?PARNO)" "(?-PARNO)" "(?+PARNO)" "(?R)" "(?0)"
1812           Recursive subpattern. Treat the contents of a given capture buffer
1813           in the current pattern as an independent subpattern and attempt to
1814           match it at the current position in the string. Information about
1815           capture state from the caller for things like backreferences is
1816           available to the subpattern, but capture buffers set by the
1817           subpattern are not visible to the caller.
1818
1819           Similar to "(??{ code })" except that it does not involve executing
1820           any code or potentially compiling a returned pattern string;
1821           instead it treats the part of the current pattern contained within
1822           a specified capture group as an independent pattern that must match
1823           at the current position. Also different is the treatment of capture
1824           buffers, unlike "(??{ code })" recursive patterns have access to
1825           their caller's match state, so one can use backreferences safely.
1826
1827           PARNO is a sequence of digits (not starting with 0) whose value
1828           reflects the paren-number of the capture group to recurse to.
1829           "(?R)" recurses to the beginning of the whole pattern. "(?0)" is an
1830           alternate syntax for "(?R)". If PARNO is preceded by a plus or
1831           minus sign then it is assumed to be relative, with negative numbers
1832           indicating preceding capture groups and positive ones following.
1833           Thus "(?-1)" refers to the most recently declared group, and
1834           "(?+1)" indicates the next group to be declared.  Note that the
1835           counting for relative recursion differs from that of relative
1836           backreferences, in that with recursion unclosed groups are
1837           included.
1838
1839           The following pattern matches a function "foo()" which may contain
1840           balanced parentheses as the argument.
1841
1842             $re = qr{ (                   # paren group 1 (full function)
1843                         foo
1844                         (                 # paren group 2 (parens)
1845                           \(
1846                             (             # paren group 3 (contents of parens)
1847                             (?:
1848                              (?> [^()]+ ) # Non-parens without backtracking
1849                             |
1850                              (?2)         # Recurse to start of paren group 2
1851                             )*
1852                             )
1853                           \)
1854                         )
1855                       )
1856                     }x;
1857
1858           If the pattern was used as follows
1859
1860               'foo(bar(baz)+baz(bop))'=~/$re/
1861                   and print "\$1 = $1\n",
1862                             "\$2 = $2\n",
1863                             "\$3 = $3\n";
1864
1865           the output produced should be the following:
1866
1867               $1 = foo(bar(baz)+baz(bop))
1868               $2 = (bar(baz)+baz(bop))
1869               $3 = bar(baz)+baz(bop)
1870
1871           If there is no corresponding capture group defined, then it is a
1872           fatal error.  Recursing deeply without consuming any input string
1873           will also result in a fatal error.  The depth at which that happens
1874           is compiled into perl, so it can be changed with a custom build.
1875
1876           The following shows how using negative indexing can make it easier
1877           to embed recursive patterns inside of a "qr//" construct for later
1878           use:
1879
1880               my $parens = qr/(\((?:[^()]++|(?-1))*+\))/;
1881               if (/foo $parens \s+ \+ \s+ bar $parens/x) {
1882                  # do something here...
1883               }
1884
1885           Note that this pattern does not behave the same way as the
1886           equivalent PCRE or Python construct of the same form. In Perl you
1887           can backtrack into a recursed group, in PCRE and Python the
1888           recursed into group is treated as atomic. Also, modifiers are
1889           resolved at compile time, so constructs like "(?i:(?1))" or
1890           "(?:(?i)(?1))" do not affect how the sub-pattern will be processed.
1891
1892       "(?&NAME)"
1893           Recurse to a named subpattern. Identical to "(?PARNO)" except that
1894           the parenthesis to recurse to is determined by name. If multiple
1895           parentheses have the same name, then it recurses to the leftmost.
1896
1897           It is an error to refer to a name that is not declared somewhere in
1898           the pattern.
1899
1900           NOTE: In order to make things easier for programmers with
1901           experience with the Python or PCRE regex engines the pattern
1902           "(?P>NAME)" may be used instead of "(?&NAME)".
1903
1904       "(?(condition)yes-pattern|no-pattern)"
1905       "(?(condition)yes-pattern)"
1906           Conditional expression. Matches yes-pattern if condition yields a
1907           true value, matches no-pattern otherwise. A missing pattern always
1908           matches.
1909
1910           "(condition)" should be one of:
1911
1912           an integer in parentheses
1913               (which is valid if the corresponding pair of parentheses
1914               matched);
1915
1916           a lookahead/lookbehind/evaluate zero-width assertion;
1917           a name in angle brackets or single quotes
1918               (which is valid if a group with the given name matched);
1919
1920           the special symbol "(R)"
1921               (true when evaluated inside of recursion or eval).
1922               Additionally the "R" may be followed by a number, (which will
1923               be true when evaluated when recursing inside of the appropriate
1924               group), or by "&NAME", in which case it will be true only when
1925               evaluated during recursion in the named group.
1926
1927           Here's a summary of the possible predicates:
1928
1929           "(1)" "(2)" ...
1930               Checks if the numbered capturing group has matched something.
1931               Full syntax: "(?(1)then|else)"
1932
1933           "(<NAME>)" "('NAME')"
1934               Checks if a group with the given name has matched something.
1935               Full syntax: "(?(<name>)then|else)"
1936
1937           "(?=...)" "(?!...)" "(?<=...)" "(?<!...)"
1938               Checks whether the pattern matches (or does not match, for the
1939               "!"  variants).  Full syntax: "(?(?=lookahead)then|else)"
1940
1941           "(?{ CODE })"
1942               Treats the return value of the code block as the condition.
1943               Full syntax: "(?(?{ code })then|else)"
1944
1945           "(R)"
1946               Checks if the expression has been evaluated inside of
1947               recursion.  Full syntax: "(?(R)then|else)"
1948
1949           "(R1)" "(R2)" ...
1950               Checks if the expression has been evaluated while executing
1951               directly inside of the n-th capture group. This check is the
1952               regex equivalent of
1953
1954                 if ((caller(0))[3] eq 'subname') { ... }
1955
1956               In other words, it does not check the full recursion stack.
1957
1958               Full syntax: "(?(R1)then|else)"
1959
1960           "(R&NAME)"
1961               Similar to "(R1)", this predicate checks to see if we're
1962               executing directly inside of the leftmost group with a given
1963               name (this is the same logic used by "(?&NAME)" to
1964               disambiguate). It does not check the full stack, but only the
1965               name of the innermost active recursion.  Full syntax:
1966               "(?(R&name)then|else)"
1967
1968           "(DEFINE)"
1969               In this case, the yes-pattern is never directly executed, and
1970               no no-pattern is allowed. Similar in spirit to "(?{0})" but
1971               more efficient.  See below for details.  Full syntax:
1972               "(?(DEFINE)definitions...)"
1973
1974           For example:
1975
1976               m{ ( \( )?
1977                  [^()]+
1978                  (?(1) \) )
1979                }x
1980
1981           matches a chunk of non-parentheses, possibly included in
1982           parentheses themselves.
1983
1984           A special form is the "(DEFINE)" predicate, which never executes
1985           its yes-pattern directly, and does not allow a no-pattern. This
1986           allows one to define subpatterns which will be executed only by the
1987           recursion mechanism.  This way, you can define a set of regular
1988           expression rules that can be bundled into any pattern you choose.
1989
1990           It is recommended that for this usage you put the DEFINE block at
1991           the end of the pattern, and that you name any subpatterns defined
1992           within it.
1993
1994           Also, it's worth noting that patterns defined this way probably
1995           will not be as efficient, as the optimizer is not very clever about
1996           handling them.
1997
1998           An example of how this might be used is as follows:
1999
2000             /(?<NAME>(?&NAME_PAT))(?<ADDR>(?&ADDRESS_PAT))
2001              (?(DEFINE)
2002                (?<NAME_PAT>....)
2003                (?<ADDRESS_PAT>....)
2004              )/x
2005
2006           Note that capture groups matched inside of recursion are not
2007           accessible after the recursion returns, so the extra layer of
2008           capturing groups is necessary. Thus $+{NAME_PAT} would not be
2009           defined even though $+{NAME} would be.
2010
2011           Finally, keep in mind that subpatterns created inside a DEFINE
2012           block count towards the absolute and relative number of captures,
2013           so this:
2014
2015               my @captures = "a" =~ /(.)                  # First capture
2016                                      (?(DEFINE)
2017                                          (?<EXAMPLE> 1 )  # Second capture
2018                                      )/x;
2019               say scalar @captures;
2020
2021           Will output 2, not 1. This is particularly important if you intend
2022           to compile the definitions with the "qr//" operator, and later
2023           interpolate them in another pattern.
2024
2025       "(?>pattern)"
2026       "(*atomic:pattern)"
2027           An "independent" subexpression, one which matches the substring
2028           that a standalone pattern would match if anchored at the given
2029           position, and it matches nothing other than this substring.  This
2030           construct is useful for optimizations of what would otherwise be
2031           "eternal" matches, because it will not backtrack (see
2032           "Backtracking").  It may also be useful in places where the "grab
2033           all you can, and do not give anything back" semantic is desirable.
2034
2035           For example: "^(?>a*)ab" will never match, since "(?>a*)" (anchored
2036           at the beginning of string, as above) will match all characters "a"
2037           at the beginning of string, leaving no "a" for "ab" to match.  In
2038           contrast, "a*ab" will match the same as "a+b", since the match of
2039           the subgroup "a*" is influenced by the following group "ab" (see
2040           "Backtracking").  In particular, "a*" inside "a*ab" will match
2041           fewer characters than a standalone "a*", since this makes the tail
2042           match.
2043
2044           "(?>pattern)" does not disable backtracking altogether once it has
2045           matched. It is still possible to backtrack past the construct, but
2046           not into it. So "((?>a*)|(?>b*))ar" will still match "bar".
2047
2048           An effect similar to "(?>pattern)" may be achieved by writing
2049           "(?=(pattern))\g{-1}".  This matches the same substring as a
2050           standalone "a+", and the following "\g{-1}" eats the matched
2051           string; it therefore makes a zero-length assertion into an analogue
2052           of "(?>...)".  (The difference between these two constructs is that
2053           the second one uses a capturing group, thus shifting ordinals of
2054           backreferences in the rest of a regular expression.)
2055
2056           Consider this pattern:
2057
2058               m{ \(
2059                     (
2060                       [^()]+           # x+
2061                     |
2062                       \( [^()]* \)
2063                     )+
2064                  \)
2065                }x
2066
2067           That will efficiently match a nonempty group with matching
2068           parentheses two levels deep or less.  However, if there is no such
2069           group, it will take virtually forever on a long string.  That's
2070           because there are so many different ways to split a long string
2071           into several substrings.  This is what "(.+)+" is doing, and
2072           "(.+)+" is similar to a subpattern of the above pattern.  Consider
2073           how the pattern above detects no-match on "((()aaaaaaaaaaaaaaaaaa"
2074           in several seconds, but that each extra letter doubles this time.
2075           This exponential performance will make it appear that your program
2076           has hung.  However, a tiny change to this pattern
2077
2078               m{ \(
2079                     (
2080                       (?> [^()]+ )        # change x+ above to (?> x+ )
2081                     |
2082                       \( [^()]* \)
2083                     )+
2084                  \)
2085                }x
2086
2087           which uses "(?>...)" matches exactly when the one above does
2088           (verifying this yourself would be a productive exercise), but
2089           finishes in a fourth the time when used on a similar string with
2090           1000000 "a"s.  Be aware, however, that, when this construct is
2091           followed by a quantifier, it currently triggers a warning message
2092           under the "use warnings" pragma or -w switch saying it "matches
2093           null string many times in regex".
2094
2095           On simple groups, such as the pattern "(?> [^()]+ )", a comparable
2096           effect may be achieved by negative lookahead, as in "[^()]+ (?!
2097           [^()] )".  This was only 4 times slower on a string with 1000000
2098           "a"s.
2099
2100           The "grab all you can, and do not give anything back" semantic is
2101           desirable in many situations where on the first sight a simple
2102           "()*" looks like the correct solution.  Suppose we parse text with
2103           comments being delimited by "#" followed by some optional
2104           (horizontal) whitespace.  Contrary to its appearance, "#[ \t]*" is
2105           not the correct subexpression to match the comment delimiter,
2106           because it may "give up" some whitespace if the remainder of the
2107           pattern can be made to match that way.  The correct answer is
2108           either one of these:
2109
2110               (?>#[ \t]*)
2111               #[ \t]*(?![ \t])
2112
2113           For example, to grab non-empty comments into $1, one should use
2114           either one of these:
2115
2116               / (?> \# [ \t]* ) (        .+ ) /x;
2117               /     \# [ \t]*   ( [^ \t] .* ) /x;
2118
2119           Which one you pick depends on which of these expressions better
2120           reflects the above specification of comments.
2121
2122           In some literature this construct is called "atomic matching" or
2123           "possessive matching".
2124
2125           Possessive quantifiers are equivalent to putting the item they are
2126           applied to inside of one of these constructs. The following
2127           equivalences apply:
2128
2129               Quantifier Form     Bracketing Form
2130               ---------------     ---------------
2131               PAT*+               (?>PAT*)
2132               PAT++               (?>PAT+)
2133               PAT?+               (?>PAT?)
2134               PAT{min,max}+       (?>PAT{min,max})
2135
2136           Nested "(?>...)" constructs are not no-ops, even if at first glance
2137           they might seem to be.  This is because the nested "(?>...)" can
2138           restrict internal backtracking that otherwise might occur.  For
2139           example,
2140
2141            "abc" =~ /(?>a[bc]*c)/
2142
2143           matches, but
2144
2145            "abc" =~ /(?>a(?>[bc]*)c)/
2146
2147           does not.
2148
2149           The alphabetic form ("(*atomic:...)") is experimental; using it
2150           yields a warning in the "experimental::alpha_assertions" category.
2151
2152       "(?[ ])"
2153           See "Extended Bracketed Character Classes" in perlrecharclass.
2154
2155           Note that this feature is currently experimental; using it yields a
2156           warning in the "experimental::regex_sets" category.
2157
2158   Backtracking
2159       NOTE: This section presents an abstract approximation of regular
2160       expression behavior.  For a more rigorous (and complicated) view of the
2161       rules involved in selecting a match among possible alternatives, see
2162       "Combining RE Pieces".
2163
2164       A fundamental feature of regular expression matching involves the
2165       notion called backtracking, which is currently used (when needed) by
2166       all regular non-possessive expression quantifiers, namely "*", "*?",
2167       "+", "+?", "{n,m}", and "{n,m}?".  Backtracking is often optimized
2168       internally, but the general principle outlined here is valid.
2169
2170       For a regular expression to match, the entire regular expression must
2171       match, not just part of it.  So if the beginning of a pattern
2172       containing a quantifier succeeds in a way that causes later parts in
2173       the pattern to fail, the matching engine backs up and recalculates the
2174       beginning part--that's why it's called backtracking.
2175
2176       Here is an example of backtracking:  Let's say you want to find the
2177       word following "foo" in the string "Food is on the foo table.":
2178
2179           $_ = "Food is on the foo table.";
2180           if ( /\b(foo)\s+(\w+)/i ) {
2181               print "$2 follows $1.\n";
2182           }
2183
2184       When the match runs, the first part of the regular expression
2185       ("\b(foo)") finds a possible match right at the beginning of the
2186       string, and loads up $1 with "Foo".  However, as soon as the matching
2187       engine sees that there's no whitespace following the "Foo" that it had
2188       saved in $1, it realizes its mistake and starts over again one
2189       character after where it had the tentative match.  This time it goes
2190       all the way until the next occurrence of "foo". The complete regular
2191       expression matches this time, and you get the expected output of "table
2192       follows foo."
2193
2194       Sometimes minimal matching can help a lot.  Imagine you'd like to match
2195       everything between "foo" and "bar".  Initially, you write something
2196       like this:
2197
2198           $_ =  "The food is under the bar in the barn.";
2199           if ( /foo(.*)bar/ ) {
2200               print "got <$1>\n";
2201           }
2202
2203       Which perhaps unexpectedly yields:
2204
2205         got <d is under the bar in the >
2206
2207       That's because ".*" was greedy, so you get everything between the first
2208       "foo" and the last "bar".  Here it's more effective to use minimal
2209       matching to make sure you get the text between a "foo" and the first
2210       "bar" thereafter.
2211
2212           if ( /foo(.*?)bar/ ) { print "got <$1>\n" }
2213         got <d is under the >
2214
2215       Here's another example. Let's say you'd like to match a number at the
2216       end of a string, and you also want to keep the preceding part of the
2217       match.  So you write this:
2218
2219           $_ = "I have 2 numbers: 53147";
2220           if ( /(.*)(\d*)/ ) {                                # Wrong!
2221               print "Beginning is <$1>, number is <$2>.\n";
2222           }
2223
2224       That won't work at all, because ".*" was greedy and gobbled up the
2225       whole string. As "\d*" can match on an empty string the complete
2226       regular expression matched successfully.
2227
2228           Beginning is <I have 2 numbers: 53147>, number is <>.
2229
2230       Here are some variants, most of which don't work:
2231
2232           $_ = "I have 2 numbers: 53147";
2233           @pats = qw{
2234               (.*)(\d*)
2235               (.*)(\d+)
2236               (.*?)(\d*)
2237               (.*?)(\d+)
2238               (.*)(\d+)$
2239               (.*?)(\d+)$
2240               (.*)\b(\d+)$
2241               (.*\D)(\d+)$
2242           };
2243
2244           for $pat (@pats) {
2245               printf "%-12s ", $pat;
2246               if ( /$pat/ ) {
2247                   print "<$1> <$2>\n";
2248               } else {
2249                   print "FAIL\n";
2250               }
2251           }
2252
2253       That will print out:
2254
2255           (.*)(\d*)    <I have 2 numbers: 53147> <>
2256           (.*)(\d+)    <I have 2 numbers: 5314> <7>
2257           (.*?)(\d*)   <> <>
2258           (.*?)(\d+)   <I have > <2>
2259           (.*)(\d+)$   <I have 2 numbers: 5314> <7>
2260           (.*?)(\d+)$  <I have 2 numbers: > <53147>
2261           (.*)\b(\d+)$ <I have 2 numbers: > <53147>
2262           (.*\D)(\d+)$ <I have 2 numbers: > <53147>
2263
2264       As you see, this can be a bit tricky.  It's important to realize that a
2265       regular expression is merely a set of assertions that gives a
2266       definition of success.  There may be 0, 1, or several different ways
2267       that the definition might succeed against a particular string.  And if
2268       there are multiple ways it might succeed, you need to understand
2269       backtracking to know which variety of success you will achieve.
2270
2271       When using lookahead assertions and negations, this can all get even
2272       trickier.  Imagine you'd like to find a sequence of non-digits not
2273       followed by "123".  You might try to write that as
2274
2275           $_ = "ABC123";
2276           if ( /^\D*(?!123)/ ) {                # Wrong!
2277               print "Yup, no 123 in $_\n";
2278           }
2279
2280       But that isn't going to match; at least, not the way you're hoping.  It
2281       claims that there is no 123 in the string.  Here's a clearer picture of
2282       why that pattern matches, contrary to popular expectations:
2283
2284           $x = 'ABC123';
2285           $y = 'ABC445';
2286
2287           print "1: got $1\n" if $x =~ /^(ABC)(?!123)/;
2288           print "2: got $1\n" if $y =~ /^(ABC)(?!123)/;
2289
2290           print "3: got $1\n" if $x =~ /^(\D*)(?!123)/;
2291           print "4: got $1\n" if $y =~ /^(\D*)(?!123)/;
2292
2293       This prints
2294
2295           2: got ABC
2296           3: got AB
2297           4: got ABC
2298
2299       You might have expected test 3 to fail because it seems to a more
2300       general purpose version of test 1.  The important difference between
2301       them is that test 3 contains a quantifier ("\D*") and so can use
2302       backtracking, whereas test 1 will not.  What's happening is that you've
2303       asked "Is it true that at the start of $x, following 0 or more non-
2304       digits, you have something that's not 123?"  If the pattern matcher had
2305       let "\D*" expand to "ABC", this would have caused the whole pattern to
2306       fail.
2307
2308       The search engine will initially match "\D*" with "ABC".  Then it will
2309       try to match "(?!123)" with "123", which fails.  But because a
2310       quantifier ("\D*") has been used in the regular expression, the search
2311       engine can backtrack and retry the match differently in the hope of
2312       matching the complete regular expression.
2313
2314       The pattern really, really wants to succeed, so it uses the standard
2315       pattern back-off-and-retry and lets "\D*" expand to just "AB" this
2316       time.  Now there's indeed something following "AB" that is not "123".
2317       It's "C123", which suffices.
2318
2319       We can deal with this by using both an assertion and a negation.  We'll
2320       say that the first part in $1 must be followed both by a digit and by
2321       something that's not "123".  Remember that the lookaheads are zero-
2322       width expressions--they only look, but don't consume any of the string
2323       in their match.  So rewriting this way produces what you'd expect; that
2324       is, case 5 will fail, but case 6 succeeds:
2325
2326           print "5: got $1\n" if $x =~ /^(\D*)(?=\d)(?!123)/;
2327           print "6: got $1\n" if $y =~ /^(\D*)(?=\d)(?!123)/;
2328
2329           6: got ABC
2330
2331       In other words, the two zero-width assertions next to each other work
2332       as though they're ANDed together, just as you'd use any built-in
2333       assertions:  "/^$/" matches only if you're at the beginning of the line
2334       AND the end of the line simultaneously.  The deeper underlying truth is
2335       that juxtaposition in regular expressions always means AND, except when
2336       you write an explicit OR using the vertical bar.  "/ab/" means match
2337       "a" AND (then) match "b", although the attempted matches are made at
2338       different positions because "a" is not a zero-width assertion, but a
2339       one-width assertion.
2340
2341       WARNING: Particularly complicated regular expressions can take
2342       exponential time to solve because of the immense number of possible
2343       ways they can use backtracking to try for a match.  For example,
2344       without internal optimizations done by the regular expression engine,
2345       this will take a painfully long time to run:
2346
2347           'aaaaaaaaaaaa' =~ /((a{0,5}){0,5})*[c]/
2348
2349       And if you used "*"'s in the internal groups instead of limiting them
2350       to 0 through 5 matches, then it would take forever--or until you ran
2351       out of stack space.  Moreover, these internal optimizations are not
2352       always applicable.  For example, if you put "{0,5}" instead of "*" on
2353       the external group, no current optimization is applicable, and the
2354       match takes a long time to finish.
2355
2356       A powerful tool for optimizing such beasts is what is known as an
2357       "independent group", which does not backtrack (see ""(?>pattern)"").
2358       Note also that zero-length lookahead/lookbehind assertions will not
2359       backtrack to make the tail match, since they are in "logical" context:
2360       only whether they match is considered relevant.  For an example where
2361       side-effects of lookahead might have influenced the following match,
2362       see ""(?>pattern)"".
2363
2364   Script Runs
2365       A script run is basically a sequence of characters, all from the same
2366       Unicode script (see "Scripts" in perlunicode), such as Latin or Greek.
2367       In most places a single word would never be written in multiple
2368       scripts, unless it is a spoofing attack.  An infamous example, is
2369
2370        paypal.com
2371
2372       Those letters could all be Latin (as in the example just above), or
2373       they could be all Cyrillic (except for the dot), or they could be a
2374       mixture of the two.  In the case of an internet address the ".com"
2375       would be in Latin, And any Cyrillic ones would cause it to be a
2376       mixture, not a script run.  Someone clicking on such a link would not
2377       be directed to the real Paypal website, but an attacker would craft a
2378       look-alike one to attempt to gather sensitive information from the
2379       person.
2380
2381       Starting in Perl 5.28, it is now easy to detect strings that aren't
2382       script runs.  Simply enclose just about any pattern like either of
2383       these:
2384
2385        (*script_run:pattern)
2386        (*sr:pattern)
2387
2388       What happens is that after pattern succeeds in matching, it is
2389       subjected to the additional criterion that every character in it must
2390       be from the same script (see exceptions below).  If this isn't true,
2391       backtracking occurs until something all in the same script is found
2392       that matches, or all possibilities are exhausted.  This can cause a lot
2393       of backtracking, but generally, only malicious input will result in
2394       this, though the slow down could cause a denial of service attack.  If
2395       your needs permit, it is best to make the pattern atomic to cut down on
2396       the amount of backtracking.  This is so likely to be what you want,
2397       that instead of writing this:
2398
2399        (*script_run:(?>pattern))
2400
2401       you can write either of these:
2402
2403        (*atomic_script_run:pattern)
2404        (*asr:pattern)
2405
2406       (See ""(?>pattern)"".)
2407
2408       In Taiwan, Japan, and Korea, it is common for text to have a mixture of
2409       characters from their native scripts and base Chinese.  Perl follows
2410       Unicode's UTS 39 (<http://unicode.org/reports/tr39/>) Unicode Security
2411       Mechanisms in allowing such mixtures.  For example, the Japanese
2412       scripts Katakana and Hiragana are commonly mixed together in practice,
2413       along with some Chinese characters, and hence are treated as being in a
2414       single script run by Perl.
2415
2416       The rules used for matching decimal digits are slightly stricter.  Many
2417       scripts have their own sets of digits equivalent to the Western 0
2418       through 9 ones.  A few, such as Arabic, have more than one set.  For a
2419       string to be considered a script run, all digits in it must come from
2420       the same set of ten, as determined by the first digit encountered.  As
2421       an example,
2422
2423        qr/(*script_run: \d+ \b )/x
2424
2425       guarantees that the digits matched will all be from the same set of 10.
2426       You won't get a look-alike digit from a different script that has a
2427       different value than what it appears to be.
2428
2429       Unicode has three pseudo scripts that are handled specially.
2430
2431       "Unknown" is applied to code points whose meaning has yet to be
2432       determined.  Perl currently will match as a script run, any single
2433       character string consisting of one of these code points.  But any
2434       string longer than one code point containing one of these will not be
2435       considered a script run.
2436
2437       "Inherited" is applied to characters that modify another, such as an
2438       accent of some type.  These are considered to be in the script of the
2439       master character, and so never cause a script run to not match.
2440
2441       The other one is "Common".  This consists of mostly punctuation, emoji,
2442       and characters used in mathematics and music, the ASCII digits 0
2443       through 9, and full-width forms of these digits.  These characters can
2444       appear intermixed in text in many of the world's scripts.  These also
2445       don't cause a script run to not match.  But like other scripts, all
2446       digits in a run must come from the same set of 10.
2447
2448       This construct is non-capturing.  You can add parentheses to pattern to
2449       capture, if desired.  You will have to do this if you plan to use
2450       "(*ACCEPT) (*ACCEPT:arg)" and not have it bypass the script run
2451       checking.
2452
2453       This feature is experimental, and the exact syntax and details of
2454       operation are subject to change; using it yields a warning in the
2455       "experimental::script_run" category.
2456
2457       The "Script_Extensions" property as modified by UTS 39
2458       (<http://unicode.org/reports/tr39/>) is used as the basis for this
2459       feature.
2460
2461       To summarize,
2462
2463       ·   All length 0 or length 1 sequences are script runs.
2464
2465       ·   A longer sequence is a script run if and only if all of the
2466           following conditions are met:
2467
2468
2469
2470           1.  No code point in the sequence has the "Script_Extension"
2471               property of "Unknown".
2472
2473               This currently means that all code points in the sequence have
2474               been assigned by Unicode to be characters that aren't private
2475               use nor surrogate code points.
2476
2477           2.  All characters in the sequence come from the Common script
2478               and/or the Inherited script and/or a single other script.
2479
2480               The script of a character is determined by the
2481               "Script_Extensions" property as modified by UTS 39
2482               (<http://unicode.org/reports/tr39/>), as described above.
2483
2484           3.  All decimal digits in the sequence come from the same block of
2485               10 consecutive digits.
2486
2487   Special Backtracking Control Verbs
2488       These special patterns are generally of the form "(*VERB:arg)". Unless
2489       otherwise stated the arg argument is optional; in some cases, it is
2490       mandatory.
2491
2492       Any pattern containing a special backtracking verb that allows an
2493       argument has the special behaviour that when executed it sets the
2494       current package's $REGERROR and $REGMARK variables. When doing so the
2495       following rules apply:
2496
2497       On failure, the $REGERROR variable will be set to the arg value of the
2498       verb pattern, if the verb was involved in the failure of the match. If
2499       the arg part of the pattern was omitted, then $REGERROR will be set to
2500       the name of the last "(*MARK:NAME)" pattern executed, or to TRUE if
2501       there was none. Also, the $REGMARK variable will be set to FALSE.
2502
2503       On a successful match, the $REGERROR variable will be set to FALSE, and
2504       the $REGMARK variable will be set to the name of the last
2505       "(*MARK:NAME)" pattern executed.  See the explanation for the
2506       "(*MARK:NAME)" verb below for more details.
2507
2508       NOTE: $REGERROR and $REGMARK are not magic variables like $1 and most
2509       other regex-related variables. They are not local to a scope, nor
2510       readonly, but instead are volatile package variables similar to
2511       $AUTOLOAD.  They are set in the package containing the code that
2512       executed the regex (rather than the one that compiled it, where those
2513       differ).  If necessary, you can use "local" to localize changes to
2514       these variables to a specific scope before executing a regex.
2515
2516       If a pattern does not contain a special backtracking verb that allows
2517       an argument, then $REGERROR and $REGMARK are not touched at all.
2518
2519       Verbs
2520          "(*PRUNE)" "(*PRUNE:NAME)"
2521              This zero-width pattern prunes the backtracking tree at the
2522              current point when backtracked into on failure. Consider the
2523              pattern "/A (*PRUNE) B/", where A and B are complex patterns.
2524              Until the "(*PRUNE)" verb is reached, A may backtrack as
2525              necessary to match. Once it is reached, matching continues in B,
2526              which may also backtrack as necessary; however, should B not
2527              match, then no further backtracking will take place, and the
2528              pattern will fail outright at the current starting position.
2529
2530              The following example counts all the possible matching strings
2531              in a pattern (without actually matching any of them).
2532
2533                  'aaab' =~ /a+b?(?{print "$&\n"; $count++})(*FAIL)/;
2534                  print "Count=$count\n";
2535
2536              which produces:
2537
2538                  aaab
2539                  aaa
2540                  aa
2541                  a
2542                  aab
2543                  aa
2544                  a
2545                  ab
2546                  a
2547                  Count=9
2548
2549              If we add a "(*PRUNE)" before the count like the following
2550
2551                  'aaab' =~ /a+b?(*PRUNE)(?{print "$&\n"; $count++})(*FAIL)/;
2552                  print "Count=$count\n";
2553
2554              we prevent backtracking and find the count of the longest
2555              matching string at each matching starting point like so:
2556
2557                  aaab
2558                  aab
2559                  ab
2560                  Count=3
2561
2562              Any number of "(*PRUNE)" assertions may be used in a pattern.
2563
2564              See also "(?>pattern)" and possessive quantifiers for other ways
2565              to control backtracking. In some cases, the use of "(*PRUNE)"
2566              can be replaced with a "(?>pattern)" with no functional
2567              difference; however, "(*PRUNE)" can be used to handle cases that
2568              cannot be expressed using a "(?>pattern)" alone.
2569
2570          "(*SKIP)" "(*SKIP:NAME)"
2571              This zero-width pattern is similar to "(*PRUNE)", except that on
2572              failure it also signifies that whatever text that was matched
2573              leading up to the "(*SKIP)" pattern being executed cannot be
2574              part of any match of this pattern. This effectively means that
2575              the regex engine "skips" forward to this position on failure and
2576              tries to match again, (assuming that there is sufficient room to
2577              match).
2578
2579              The name of the "(*SKIP:NAME)" pattern has special significance.
2580              If a "(*MARK:NAME)" was encountered while matching, then it is
2581              that position which is used as the "skip point". If no "(*MARK)"
2582              of that name was encountered, then the "(*SKIP)" operator has no
2583              effect. When used without a name the "skip point" is where the
2584              match point was when executing the "(*SKIP)" pattern.
2585
2586              Compare the following to the examples in "(*PRUNE)"; note the
2587              string is twice as long:
2588
2589               'aaabaaab' =~ /a+b?(*SKIP)(?{print "$&\n"; $count++})(*FAIL)/;
2590               print "Count=$count\n";
2591
2592              outputs
2593
2594                  aaab
2595                  aaab
2596                  Count=2
2597
2598              Once the 'aaab' at the start of the string has matched, and the
2599              "(*SKIP)" executed, the next starting point will be where the
2600              cursor was when the "(*SKIP)" was executed.
2601
2602          "(*MARK:NAME)" "(*:NAME)"
2603              This zero-width pattern can be used to mark the point reached in
2604              a string when a certain part of the pattern has been
2605              successfully matched. This mark may be given a name. A later
2606              "(*SKIP)" pattern will then skip forward to that point if
2607              backtracked into on failure. Any number of "(*MARK)" patterns
2608              are allowed, and the NAME portion may be duplicated.
2609
2610              In addition to interacting with the "(*SKIP)" pattern,
2611              "(*MARK:NAME)" can be used to "label" a pattern branch, so that
2612              after matching, the program can determine which branches of the
2613              pattern were involved in the match.
2614
2615              When a match is successful, the $REGMARK variable will be set to
2616              the name of the most recently executed "(*MARK:NAME)" that was
2617              involved in the match.
2618
2619              This can be used to determine which branch of a pattern was
2620              matched without using a separate capture group for each branch,
2621              which in turn can result in a performance improvement, as perl
2622              cannot optimize "/(?:(x)|(y)|(z))/" as efficiently as something
2623              like "/(?:x(*MARK:x)|y(*MARK:y)|z(*MARK:z))/".
2624
2625              When a match has failed, and unless another verb has been
2626              involved in failing the match and has provided its own name to
2627              use, the $REGERROR variable will be set to the name of the most
2628              recently executed "(*MARK:NAME)".
2629
2630              See "(*SKIP)" for more details.
2631
2632              As a shortcut "(*MARK:NAME)" can be written "(*:NAME)".
2633
2634          "(*THEN)" "(*THEN:NAME)"
2635              This is similar to the "cut group" operator "::" from Perl 6.
2636              Like "(*PRUNE)", this verb always matches, and when backtracked
2637              into on failure, it causes the regex engine to try the next
2638              alternation in the innermost enclosing group (capturing or
2639              otherwise) that has alternations.  The two branches of a
2640              "(?(condition)yes-pattern|no-pattern)" do not count as an
2641              alternation, as far as "(*THEN)" is concerned.
2642
2643              Its name comes from the observation that this operation combined
2644              with the alternation operator ("|") can be used to create what
2645              is essentially a pattern-based if/then/else block:
2646
2647                ( COND (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ )
2648
2649              Note that if this operator is used and NOT inside of an
2650              alternation then it acts exactly like the "(*PRUNE)" operator.
2651
2652                / A (*PRUNE) B /
2653
2654              is the same as
2655
2656                / A (*THEN) B /
2657
2658              but
2659
2660                / ( A (*THEN) B | C ) /
2661
2662              is not the same as
2663
2664                / ( A (*PRUNE) B | C ) /
2665
2666              as after matching the A but failing on the B the "(*THEN)" verb
2667              will backtrack and try C; but the "(*PRUNE)" verb will simply
2668              fail.
2669
2670          "(*COMMIT)" "(*COMMIT:arg)"
2671              This is the Perl 6 "commit pattern" "<commit>" or ":::". It's a
2672              zero-width pattern similar to "(*SKIP)", except that when
2673              backtracked into on failure it causes the match to fail
2674              outright. No further attempts to find a valid match by advancing
2675              the start pointer will occur again.  For example,
2676
2677               'aaabaaab' =~ /a+b?(*COMMIT)(?{print "$&\n"; $count++})(*FAIL)/;
2678               print "Count=$count\n";
2679
2680              outputs
2681
2682                  aaab
2683                  Count=1
2684
2685              In other words, once the "(*COMMIT)" has been entered, and if
2686              the pattern does not match, the regex engine will not try any
2687              further matching on the rest of the string.
2688
2689          "(*FAIL)" "(*F)" "(*FAIL:arg)"
2690              This pattern matches nothing and always fails. It can be used to
2691              force the engine to backtrack. It is equivalent to "(?!)", but
2692              easier to read. In fact, "(?!)" gets optimised into "(*FAIL)"
2693              internally. You can provide an argument so that if the match
2694              fails because of this "FAIL" directive the argument can be
2695              obtained from $REGERROR.
2696
2697              It is probably useful only when combined with "(?{})" or
2698              "(??{})".
2699
2700          "(*ACCEPT)" "(*ACCEPT:arg)"
2701              This pattern matches nothing and causes the end of successful
2702              matching at the point at which the "(*ACCEPT)" pattern was
2703              encountered, regardless of whether there is actually more to
2704              match in the string. When inside of a nested pattern, such as
2705              recursion, or in a subpattern dynamically generated via
2706              "(??{})", only the innermost pattern is ended immediately.
2707
2708              If the "(*ACCEPT)" is inside of capturing groups then the groups
2709              are marked as ended at the point at which the "(*ACCEPT)" was
2710              encountered.  For instance:
2711
2712                'AB' =~ /(A (A|B(*ACCEPT)|C) D)(E)/x;
2713
2714              will match, and $1 will be "AB" and $2 will be "B", $3 will not
2715              be set. If another branch in the inner parentheses was matched,
2716              such as in the string 'ACDE', then the "D" and "E" would have to
2717              be matched as well.
2718
2719              You can provide an argument, which will be available in the var
2720              $REGMARK after the match completes.
2721
2722   Warning on "\1" Instead of $1
2723       Some people get too used to writing things like:
2724
2725           $pattern =~ s/(\W)/\\\1/g;
2726
2727       This is grandfathered (for \1 to \9) for the RHS of a substitute to
2728       avoid shocking the sed addicts, but it's a dirty habit to get into.
2729       That's because in PerlThink, the righthand side of an "s///" is a
2730       double-quoted string.  "\1" in the usual double-quoted string means a
2731       control-A.  The customary Unix meaning of "\1" is kludged in for
2732       "s///".  However, if you get into the habit of doing that, you get
2733       yourself into trouble if you then add an "/e" modifier.
2734
2735           s/(\d+)/ \1 + 1 /eg;            # causes warning under -w
2736
2737       Or if you try to do
2738
2739           s/(\d+)/\1000/;
2740
2741       You can't disambiguate that by saying "\{1}000", whereas you can fix it
2742       with "${1}000".  The operation of interpolation should not be confused
2743       with the operation of matching a backreference.  Certainly they mean
2744       two different things on the left side of the "s///".
2745
2746   Repeated Patterns Matching a Zero-length Substring
2747       WARNING: Difficult material (and prose) ahead.  This section needs a
2748       rewrite.
2749
2750       Regular expressions provide a terse and powerful programming language.
2751       As with most other power tools, power comes together with the ability
2752       to wreak havoc.
2753
2754       A common abuse of this power stems from the ability to make infinite
2755       loops using regular expressions, with something as innocuous as:
2756
2757           'foo' =~ m{ ( o? )* }x;
2758
2759       The "o?" matches at the beginning of ""foo"", and since the position in
2760       the string is not moved by the match, "o?" would match again and again
2761       because of the "*" quantifier.  Another common way to create a similar
2762       cycle is with the looping modifier "/g":
2763
2764           @matches = ( 'foo' =~ m{ o? }xg );
2765
2766       or
2767
2768           print "match: <$&>\n" while 'foo' =~ m{ o? }xg;
2769
2770       or the loop implied by "split()".
2771
2772       However, long experience has shown that many programming tasks may be
2773       significantly simplified by using repeated subexpressions that may
2774       match zero-length substrings.  Here's a simple example being:
2775
2776           @chars = split //, $string;           # // is not magic in split
2777           ($whitewashed = $string) =~ s/()/ /g; # parens avoid magic s// /
2778
2779       Thus Perl allows such constructs, by forcefully breaking the infinite
2780       loop.  The rules for this are different for lower-level loops given by
2781       the greedy quantifiers "*+{}", and for higher-level ones like the "/g"
2782       modifier or "split()" operator.
2783
2784       The lower-level loops are interrupted (that is, the loop is broken)
2785       when Perl detects that a repeated expression matched a zero-length
2786       substring.   Thus
2787
2788          m{ (?: NON_ZERO_LENGTH | ZERO_LENGTH )* }x;
2789
2790       is made equivalent to
2791
2792          m{ (?: NON_ZERO_LENGTH )* (?: ZERO_LENGTH )? }x;
2793
2794       For example, this program
2795
2796          #!perl -l
2797          "aaaaab" =~ /
2798            (?:
2799               a                 # non-zero
2800               |                 # or
2801              (?{print "hello"}) # print hello whenever this
2802                                 #    branch is tried
2803              (?=(b))            # zero-width assertion
2804            )*  # any number of times
2805           /x;
2806          print $&;
2807          print $1;
2808
2809       prints
2810
2811          hello
2812          aaaaa
2813          b
2814
2815       Notice that "hello" is only printed once, as when Perl sees that the
2816       sixth iteration of the outermost "(?:)*" matches a zero-length string,
2817       it stops the "*".
2818
2819       The higher-level loops preserve an additional state between iterations:
2820       whether the last match was zero-length.  To break the loop, the
2821       following match after a zero-length match is prohibited to have a
2822       length of zero.  This prohibition interacts with backtracking (see
2823       "Backtracking"), and so the second best match is chosen if the best
2824       match is of zero length.
2825
2826       For example:
2827
2828           $_ = 'bar';
2829           s/\w??/<$&>/g;
2830
2831       results in "<><b><><a><><r><>".  At each position of the string the
2832       best match given by non-greedy "??" is the zero-length match, and the
2833       second best match is what is matched by "\w".  Thus zero-length matches
2834       alternate with one-character-long matches.
2835
2836       Similarly, for repeated "m/()/g" the second-best match is the match at
2837       the position one notch further in the string.
2838
2839       The additional state of being matched with zero-length is associated
2840       with the matched string, and is reset by each assignment to "pos()".
2841       Zero-length matches at the end of the previous match are ignored during
2842       "split".
2843
2844   Combining RE Pieces
2845       Each of the elementary pieces of regular expressions which were
2846       described before (such as "ab" or "\Z") could match at most one
2847       substring at the given position of the input string.  However, in a
2848       typical regular expression these elementary pieces are combined into
2849       more complicated patterns using combining operators "ST", "S|T", "S*"
2850       etc.  (in these examples "S" and "T" are regular subexpressions).
2851
2852       Such combinations can include alternatives, leading to a problem of
2853       choice: if we match a regular expression "a|ab" against "abc", will it
2854       match substring "a" or "ab"?  One way to describe which substring is
2855       actually matched is the concept of backtracking (see "Backtracking").
2856       However, this description is too low-level and makes you think in terms
2857       of a particular implementation.
2858
2859       Another description starts with notions of "better"/"worse".  All the
2860       substrings which may be matched by the given regular expression can be
2861       sorted from the "best" match to the "worst" match, and it is the "best"
2862       match which is chosen.  This substitutes the question of "what is
2863       chosen?"  by the question of "which matches are better, and which are
2864       worse?".
2865
2866       Again, for elementary pieces there is no such question, since at most
2867       one match at a given position is possible.  This section describes the
2868       notion of better/worse for combining operators.  In the description
2869       below "S" and "T" are regular subexpressions.
2870
2871       "ST"
2872           Consider two possible matches, "AB" and "A'B'", "A" and "A'" are
2873           substrings which can be matched by "S", "B" and "B'" are substrings
2874           which can be matched by "T".
2875
2876           If "A" is a better match for "S" than "A'", "AB" is a better match
2877           than "A'B'".
2878
2879           If "A" and "A'" coincide: "AB" is a better match than "AB'" if "B"
2880           is a better match for "T" than "B'".
2881
2882       "S|T"
2883           When "S" can match, it is a better match than when only "T" can
2884           match.
2885
2886           Ordering of two matches for "S" is the same as for "S".  Similar
2887           for two matches for "T".
2888
2889       "S{REPEAT_COUNT}"
2890           Matches as "SSS...S" (repeated as many times as necessary).
2891
2892       "S{min,max}"
2893           Matches as "S{max}|S{max-1}|...|S{min+1}|S{min}".
2894
2895       "S{min,max}?"
2896           Matches as "S{min}|S{min+1}|...|S{max-1}|S{max}".
2897
2898       "S?", "S*", "S+"
2899           Same as "S{0,1}", "S{0,BIG_NUMBER}", "S{1,BIG_NUMBER}"
2900           respectively.
2901
2902       "S??", "S*?", "S+?"
2903           Same as "S{0,1}?", "S{0,BIG_NUMBER}?", "S{1,BIG_NUMBER}?"
2904           respectively.
2905
2906       "(?>S)"
2907           Matches the best match for "S" and only that.
2908
2909       "(?=S)", "(?<=S)"
2910           Only the best match for "S" is considered.  (This is important only
2911           if "S" has capturing parentheses, and backreferences are used
2912           somewhere else in the whole regular expression.)
2913
2914       "(?!S)", "(?<!S)"
2915           For this grouping operator there is no need to describe the
2916           ordering, since only whether or not "S" can match is important.
2917
2918       "(??{ EXPR })", "(?PARNO)"
2919           The ordering is the same as for the regular expression which is the
2920           result of EXPR, or the pattern contained by capture group PARNO.
2921
2922       "(?(condition)yes-pattern|no-pattern)"
2923           Recall that which of yes-pattern or no-pattern actually matches is
2924           already determined.  The ordering of the matches is the same as for
2925           the chosen subexpression.
2926
2927       The above recipes describe the ordering of matches at a given position.
2928       One more rule is needed to understand how a match is determined for the
2929       whole regular expression: a match at an earlier position is always
2930       better than a match at a later position.
2931
2932   Creating Custom RE Engines
2933       As of Perl 5.10.0, one can create custom regular expression engines.
2934       This is not for the faint of heart, as they have to plug in at the C
2935       level.  See perlreapi for more details.
2936
2937       As an alternative, overloaded constants (see overload) provide a simple
2938       way to extend the functionality of the RE engine, by substituting one
2939       pattern for another.
2940
2941       Suppose that we want to enable a new RE escape-sequence "\Y|" which
2942       matches at a boundary between whitespace characters and non-whitespace
2943       characters.  Note that "(?=\S)(?<!\S)|(?!\S)(?<=\S)" matches exactly at
2944       these positions, so we want to have each "\Y|" in the place of the more
2945       complicated version.  We can create a module "customre" to do this:
2946
2947           package customre;
2948           use overload;
2949
2950           sub import {
2951             shift;
2952             die "No argument to customre::import allowed" if @_;
2953             overload::constant 'qr' => \&convert;
2954           }
2955
2956           sub invalid { die "/$_[0]/: invalid escape '\\$_[1]'"}
2957
2958           # We must also take care of not escaping the legitimate \\Y|
2959           # sequence, hence the presence of '\\' in the conversion rules.
2960           my %rules = ( '\\' => '\\\\',
2961                         'Y|' => qr/(?=\S)(?<!\S)|(?!\S)(?<=\S)/ );
2962           sub convert {
2963             my $re = shift;
2964             $re =~ s{
2965                       \\ ( \\ | Y . )
2966                     }
2967                     { $rules{$1} or invalid($re,$1) }sgex;
2968             return $re;
2969           }
2970
2971       Now "use customre" enables the new escape in constant regular
2972       expressions, i.e., those without any runtime variable interpolations.
2973       As documented in overload, this conversion will work only over literal
2974       parts of regular expressions.  For "\Y|$re\Y|" the variable part of
2975       this regular expression needs to be converted explicitly (but only if
2976       the special meaning of "\Y|" should be enabled inside $re):
2977
2978           use customre;
2979           $re = <>;
2980           chomp $re;
2981           $re = customre::convert $re;
2982           /\Y|$re\Y|/;
2983
2984   Embedded Code Execution Frequency
2985       The exact rules for how often "(??{})" and "(?{})" are executed in a
2986       pattern are unspecified.  In the case of a successful match you can
2987       assume that they DWIM and will be executed in left to right order the
2988       appropriate number of times in the accepting path of the pattern as
2989       would any other meta-pattern.  How non-accepting pathways and match
2990       failures affect the number of times a pattern is executed is
2991       specifically unspecified and may vary depending on what optimizations
2992       can be applied to the pattern and is likely to change from version to
2993       version.
2994
2995       For instance in
2996
2997         "aaabcdeeeee"=~/a(?{print "a"})b(?{print "b"})cde/;
2998
2999       the exact number of times "a" or "b" are printed out is unspecified for
3000       failure, but you may assume they will be printed at least once during a
3001       successful match, additionally you may assume that if "b" is printed,
3002       it will be preceded by at least one "a".
3003
3004       In the case of branching constructs like the following:
3005
3006         /a(b|(?{ print "a" }))c(?{ print "c" })/;
3007
3008       you can assume that the input "ac" will output "ac", and that "abc"
3009       will output only "c".
3010
3011       When embedded code is quantified, successful matches will call the code
3012       once for each matched iteration of the quantifier.  For example:
3013
3014         "good" =~ /g(?:o(?{print "o"}))*d/;
3015
3016       will output "o" twice.
3017
3018   PCRE/Python Support
3019       As of Perl 5.10.0, Perl supports several Python/PCRE-specific
3020       extensions to the regex syntax. While Perl programmers are encouraged
3021       to use the Perl-specific syntax, the following are also accepted:
3022
3023       "(?P<NAME>pattern)"
3024           Define a named capture group. Equivalent to "(?<NAME>pattern)".
3025
3026       "(?P=NAME)"
3027           Backreference to a named capture group. Equivalent to "\g{NAME}".
3028
3029       "(?P>NAME)"
3030           Subroutine call to a named capture group. Equivalent to "(?&NAME)".
3031

BUGS

3033       There are a number of issues with regard to case-insensitive matching
3034       in Unicode rules.  See "i" under "Modifiers" above.
3035
3036       This document varies from difficult to understand to completely and
3037       utterly opaque.  The wandering prose riddled with jargon is hard to
3038       fathom in several places.
3039
3040       This document needs a rewrite that separates the tutorial content from
3041       the reference content.
3042

SEE ALSO

3044       The syntax of patterns used in Perl pattern matching evolved from those
3045       supplied in the Bell Labs Research Unix 8th Edition (Version 8) regex
3046       routines.  (The code is actually derived (distantly) from Henry
3047       Spencer's freely redistributable reimplementation of those V8
3048       routines.)
3049
3050       perlrequick.
3051
3052       perlretut.
3053
3054       "Regexp Quote-Like Operators" in perlop.
3055
3056       "Gory details of parsing quoted constructs" in perlop.
3057
3058       perlfaq6.
3059
3060       "pos" in perlfunc.
3061
3062       perllocale.
3063
3064       perlebcdic.
3065
3066       Mastering Regular Expressions by Jeffrey Friedl, published by O'Reilly
3067       and Associates.
3068
3069
3070
3071perl v5.30.1                      2019-11-29                         PERLRE(1)
Impressum