3 perlreapi - perl regular expression plugin interface
7 As of Perl 5.9.5 there is a new interface for plugging and using other
8 regular expression engines than the default one.
10 Each engine is supposed to provide access to a constant structure of the
13 typedef struct regexp_engine {
14 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
15 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
16 char* strbeg, I32 minend, SV* screamer,
17 void* data, U32 flags);
18 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
19 char *strend, U32 flags,
20 struct re_scream_pos_data_s *data);
21 SV* (*checkstr) (pTHX_ REGEXP * const rx);
22 void (*free) (pTHX_ REGEXP * const rx);
23 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
25 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
26 SV const * const value);
27 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
29 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
30 SV * const value, U32 flags);
31 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
33 SV* (*qr_package)(pTHX_ REGEXP * const rx);
35 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
38 When a regexp is compiled, its C<engine> field is then set to point at
39 the appropriate structure, so that when it needs to be used Perl can find
40 the right routines to do so.
42 In order to install a new regexp handler, C<$^H{regcomp}> is set
43 to an integer which (when casted appropriately) resolves to one of these
44 structures. When compiling, the C<comp> method is executed, and the
45 resulting regexp structure's engine field is expected to point back at
48 The pTHX_ symbol in the definition is a macro used by perl under threading
49 to provide an extra argument to the routine holding a pointer back to
50 the interpreter that is executing the regexp. So under threading all
51 routines get an extra argument.
57 REGEXP* comp(pTHX_ const SV * const pattern, const U32 flags);
59 Compile the pattern stored in C<pattern> using the given C<flags> and
60 return a pointer to a prepared C<REGEXP> structure that can perform
61 the match. See L</The REGEXP structure> below for an explanation of
62 the individual fields in the REGEXP struct.
64 The C<pattern> parameter is the scalar that was used as the
65 pattern. previous versions of perl would pass two C<char*> indicating
66 the start and end of the stringified pattern, the following snippet can
67 be used to get the old parameters:
70 char* exp = SvPV(pattern, plen);
71 char* xend = exp + plen;
73 Since any scalar can be passed as a pattern it's possible to implement
74 an engine that does something with an array (C<< "ook" =~ [ qw/ eek
75 hlagh / ] >>) or with the non-stringified form of a compiled regular
76 expression (C<< "ook" =~ qr/eek/ >>). perl's own engine will always
77 stringify everything using the snippet above but that doesn't mean
78 other engines have to.
80 The C<flags> parameter is a bitfield which indicates which of the
81 C<msixp> flags the regex was compiled with. It also contains
82 additional info such as whether C<use locale> is in effect.
84 The C<eogc> flags are stripped out before being passed to the comp
85 routine. The regex engine does not need to know whether any of these
86 are set as those flags should only affect what perl does with the
87 pattern and its match variables, not how it gets compiled and
90 By the time the comp callback is called, some of these flags have
91 already had effect (noted below where applicable). However most of
92 their effect occurs after the comp callback has run in routines that
93 read the C<< rx->extflags >> field which it populates.
95 In general the flags should be preserved in C<< rx->extflags >> after
96 compilation, although the regex engine might want to add or delete
97 some of them to invoke or disable some special behavior in perl. The
98 flags along with any special behavior they cause are documented below:
100 The pattern modifiers:
104 =item C</m> - RXf_PMf_MULTILINE
106 If this is in C<< rx->extflags >> it will be passed to
107 C<Perl_fbm_instr> by C<pp_split> which will treat the subject string
108 as a multi-line string.
110 =item C</s> - RXf_PMf_SINGLELINE
112 =item C</i> - RXf_PMf_FOLD
114 =item C</x> - RXf_PMf_EXTENDED
116 If present on a regex C<#> comments will be handled differently by the
117 tokenizer in some cases.
119 TODO: Document those cases.
121 =item C</p> - RXf_PMf_KEEPCOPY
127 The character set semantics are determined by an enum that is contained
128 in this field. This is still experimental and subject to change, but
129 the current interface returns the rules by use of the in-line function
130 C<get_regex_charset(const U32 flags)>. The only currently documented
131 value returned from it is REGEX_LOCALE_CHARSET, which is set if
132 C<use locale> is in effect. If present in C<< rx->extflags >>,
133 C<split> will use the locale dependent definition of whitespace
134 when RXf_SKIPWHITE or RXf_WHITE is in effect. ASCII whitespace
135 is defined as per L<isSPACE|perlapi/isSPACE>, and by the internal
136 macros C<is_utf8_space> under UTF-8, and C<isSPACE_LC> under C<use
147 Set if the pattern is L<SvUTF8()|perlapi/SvUTF8>, set by Perl_pmruntime.
149 A regex engine may want to set or disable this flag during
150 compilation. The perl engine for instance may upgrade non-UTF-8
151 strings to UTF-8 if the pattern includes constructs such as C<\x{...}>
152 that can only match Unicode values.
156 If C<split> is invoked as C<split ' '> or with no arguments (which
157 really means C<split(' ', $_)>, see L<split|perlfunc/split>), perl will
158 set this flag. The regex engine can then check for it and set the
159 SKIPWHITE and WHITE extflags. To do this the perl engine does:
161 if (flags & RXf_SPLIT && r->prelen == 1 && r->precomp[0] == ' ')
162 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
166 These flags can be set during compilation to enable optimizations in
167 the C<split> operator.
173 If the flag is present in C<< rx->extflags >> C<split> will delete
174 whitespace from the start of the subject string before it's operated
175 on. What is considered whitespace depends on whether the subject is a
176 UTF-8 string and whether the C<RXf_PMf_LOCALE> flag is set.
178 If RXf_WHITE is set in addition to this flag C<split> will behave like
179 C<split " "> under the perl engine.
183 Tells the split operator to split the target string on newlines
184 (C<\n>) without invoking the regex engine.
186 Perl's engine sets this if the pattern is C</^/> (C<plen == 1 && *exp
187 == '^'>), even under C</^/s>, see L<split|perlfunc>. Of course a
188 different regex engine might want to use the same optimizations
189 with a different syntax.
193 Tells the split operator to split the target string on whitespace
194 without invoking the regex engine. The definition of whitespace varies
195 depending on whether the target string is a UTF-8 string and on
196 whether RXf_PMf_LOCALE is set.
198 Perl's engine sets this flag if the pattern is C<\s+>.
202 Tells the split operator to split the target string on
203 characters. The definition of character varies depending on whether
204 the target string is a UTF-8 string.
206 Perl's engine sets this flag on empty patterns, this optimization
207 makes C<split //> much faster than it would otherwise be. It's even
208 faster than C<unpack>.
214 I32 exec(pTHX_ REGEXP * const rx,
215 char *stringarg, char* strend, char* strbeg,
216 I32 minend, SV* screamer,
217 void* data, U32 flags);
223 char* intuit(pTHX_ REGEXP * const rx,
224 SV *sv, char *strpos, char *strend,
225 const U32 flags, struct re_scream_pos_data_s *data);
227 Find the start position where a regex match should be attempted,
228 or possibly whether the regex engine should not be run because the
229 pattern can't match. This is called as appropriate by the core
230 depending on the values of the extflags member of the regexp
235 SV* checkstr(pTHX_ REGEXP * const rx);
237 Return a SV containing a string that must appear in the pattern. Used
238 by C<split> for optimising matches.
242 void free(pTHX_ REGEXP * const rx);
244 Called by perl when it is freeing a regexp pattern so that the engine
245 can release any resources pointed to by the C<pprivate> member of the
246 regexp structure. This is only responsible for freeing private data;
247 perl will handle releasing anything else contained in the regexp structure.
249 =head2 Numbered capture callbacks
251 Called to get/set the value of C<$`>, C<$'>, C<$&> and their named
252 equivalents, ${^PREMATCH}, ${^POSTMATCH} and $^{MATCH}, as well as the
253 numbered capture groups (C<$1>, C<$2>, ...).
255 The C<paren> parameter will be C<-2> for C<$`>, C<-1> for C<$'>, C<0>
256 for C<$&>, C<1> for C<$1> and so forth.
258 The names have been chosen by analogy with L<Tie::Scalar> methods
259 names with an additional B<LENGTH> callback for efficiency. However
260 named capture variables are currently not tied internally but
261 implemented via magic.
263 =head3 numbered_buff_FETCH
265 void numbered_buff_FETCH(pTHX_ REGEXP * const rx, const I32 paren,
268 Fetch a specified numbered capture. C<sv> should be set to the scalar
269 to return, the scalar is passed as an argument rather than being
270 returned from the function because when it's called perl already has a
271 scalar to store the value, creating another one would be
272 redundant. The scalar can be set with C<sv_setsv>, C<sv_setpvn> and
273 friends, see L<perlapi>.
275 This callback is where perl untaints its own capture variables under
276 taint mode (see L<perlsec>). See the C<Perl_reg_numbered_buff_fetch>
277 function in F<regcomp.c> for how to untaint capture variables if
278 that's something you'd like your engine to do as well.
280 =head3 numbered_buff_STORE
282 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
283 SV const * const value);
285 Set the value of a numbered capture variable. C<value> is the scalar
286 that is to be used as the new value. It's up to the engine to make
287 sure this is used as the new value (or reject it).
291 if ("ook" =~ /(o*)/) {
292 # `paren' will be `1' and `value' will be `ee'
296 Perl's own engine will croak on any attempt to modify the capture
297 variables, to do this in another engine use the following callback
298 (copied from C<Perl_reg_numbered_buff_store>):
301 Example_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
302 SV const * const value)
305 PERL_UNUSED_ARG(paren);
306 PERL_UNUSED_ARG(value);
309 Perl_croak(aTHX_ PL_no_modify);
312 Actually perl will not I<always> croak in a statement that looks
313 like it would modify a numbered capture variable. This is because the
314 STORE callback will not be called if perl can determine that it
315 doesn't have to modify the value. This is exactly how tied variables
316 behave in the same situation:
319 use base 'Tie::Scalar';
321 sub TIESCALAR { bless [] }
323 sub STORE { die "This doesn't get called" }
327 tie my $sv => "CaptureVar";
330 Because C<$sv> is C<undef> when the C<y///> operator is applied to it
331 the transliteration won't actually execute and the program won't
332 C<die>. This is different to how 5.8 and earlier versions behaved
333 since the capture variables were READONLY variables then, now they'll
334 just die when assigned to in the default engine.
336 =head3 numbered_buff_LENGTH
338 I32 numbered_buff_LENGTH (pTHX_ REGEXP * const rx, const SV * const sv,
341 Get the C<length> of a capture variable. There's a special callback
342 for this so that perl doesn't have to do a FETCH and run C<length> on
343 the result, since the length is (in perl's case) known from an offset
344 stored in C<< rx->offs >> this is much more efficient:
346 I32 s1 = rx->offs[paren].start;
347 I32 s2 = rx->offs[paren].end;
350 This is a little bit more complex in the case of UTF-8, see what
351 C<Perl_reg_numbered_buff_length> does with
352 L<is_utf8_string_loclen|perlapi/is_utf8_string_loclen>.
354 =head2 Named capture callbacks
356 Called to get/set the value of C<%+> and C<%-> as well as by some
357 utility functions in L<re>.
359 There are two callbacks, C<named_buff> is called in all the cases the
360 FETCH, STORE, DELETE, CLEAR, EXISTS and SCALAR L<Tie::Hash> callbacks
361 would be on changes to C<%+> and C<%-> and C<named_buff_iter> in the
362 same cases as FIRSTKEY and NEXTKEY.
364 The C<flags> parameter can be used to determine which of these
365 operations the callbacks should respond to, the following flags are
368 Which L<Tie::Hash> operation is being performed from the Perl level on
369 C<%+> or C<%+>, if any:
380 Whether C<%+> or C<%-> is being operated on, if any.
385 Whether this is being called as C<re::regname>, C<re::regnames> or
386 C<re::regnames_count>, if any. The first two will be combined with
387 C<RXapif_ONE> or C<RXapif_ALL>.
391 RXapif_REGNAMES_COUNT
393 Internally C<%+> and C<%-> are implemented with a real tied interface
394 via L<Tie::Hash::NamedCapture>. The methods in that package will call
395 back into these functions. However the usage of
396 L<Tie::Hash::NamedCapture> for this purpose might change in future
397 releases. For instance this might be implemented by magic instead
398 (would need an extension to mgvtbl).
402 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
403 SV * const value, U32 flags);
405 =head3 named_buff_iter
407 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
412 SV* qr_package(pTHX_ REGEXP * const rx);
414 The package the qr// magic object is blessed into (as seen by C<ref
415 qr//>). It is recommended that engines change this to their package
416 name for identification regardless of whether they implement methods
419 The package this method returns should also have the internal
420 C<Regexp> package in its C<@ISA>. C<< qr//->isa("Regexp") >> should always
421 be true regardless of what engine is being used.
423 Example implementation might be:
426 Example_qr_package(pTHX_ REGEXP * const rx)
429 return newSVpvs("re::engine::Example");
432 Any method calls on an object created with C<qr//> will be dispatched to the
433 package as a normal object.
435 use re::engine::Example;
437 $re->meth; # dispatched to re::engine::Example::meth()
439 To retrieve the C<REGEXP> object from the scalar in an XS function use
440 the C<SvRX> macro, see L<"REGEXP Functions" in perlapi|perlapi/REGEXP
445 REGEXP * re = SvRX(sv);
449 void* dupe(pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
451 On threaded builds a regexp may need to be duplicated so that the pattern
452 can be used by multiple threads. This routine is expected to handle the
453 duplication of any private data pointed to by the C<pprivate> member of
454 the regexp structure. It will be called with the preconstructed new
455 regexp structure as an argument, the C<pprivate> member will point at
456 the B<old> private structure, and it is this routine's responsibility to
457 construct a copy and return a pointer to it (which perl will then use to
458 overwrite the field as passed to this routine.)
460 This allows the engine to dupe its private data but also if necessary
461 modify the final structure if it really must.
463 On unthreaded builds this field doesn't exist.
465 =head1 The REGEXP structure
467 The REGEXP struct is defined in F<regexp.h>. All regex engines must be able to
468 correctly build such a structure in their L</comp> routine.
470 The REGEXP structure contains all the data that perl needs to be aware of
471 to properly work with the regular expression. It includes data about
472 optimisations that perl can use to determine if the regex engine should
473 really be used, and various other control info that is needed to properly
474 execute patterns in various contexts such as is the pattern anchored in
475 some way, or what flags were used during the compile, or whether the
476 program contains special constructs that perl needs to be aware of.
478 In addition it contains two fields that are intended for the private
479 use of the regex engine that compiled the pattern. These are the
480 C<intflags> and C<pprivate> members. C<pprivate> is a void pointer to
481 an arbitrary structure whose use and management is the responsibility
482 of the compiling engine. perl will never modify either of these
485 typedef struct regexp {
486 /* what engine created this regexp? */
487 const struct regexp_engine* engine;
489 /* what re is this a lightweight copy of? */
490 struct regexp* mother_re;
492 /* Information about the match that the perl core uses to manage things */
493 U32 extflags; /* Flags used both externally and internally */
494 I32 minlen; /* mininum possible length of string to match */
495 I32 minlenret; /* mininum possible length of $& */
496 U32 gofs; /* chars left of pos that we search from */
498 /* substring data about strings that must appear
499 in the final match, used for optimisations */
500 struct reg_substr_data *substrs;
502 U32 nparens; /* number of capture groups */
504 /* private engine specific data */
505 U32 intflags; /* Engine Specific Internal flags */
506 void *pprivate; /* Data private to the regex engine which
507 created this object. */
509 /* Data about the last/current match. These are modified during matching*/
510 U32 lastparen; /* last open paren matched */
511 U32 lastcloseparen; /* last close paren matched */
512 regexp_paren_pair *swap; /* Swap copy of *offs */
513 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
515 char *subbeg; /* saved or original string so \digit works forever. */
516 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
517 I32 sublen; /* Length of string pointed by subbeg */
519 /* Information about the match that isn't often used */
520 I32 prelen; /* length of precomp */
521 const char *precomp; /* pre-compilation regular expression */
523 char *wrapped; /* wrapped version of the pattern */
524 I32 wraplen; /* length of wrapped */
526 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
527 HV *paren_names; /* Optional hash of paren names */
529 /* Refcount of this regexp */
530 I32 refcnt; /* Refcount of this regexp */
533 The fields are discussed in more detail below:
537 This field points at a regexp_engine structure which contains pointers
538 to the subroutines that are to be used for performing a match. It
539 is the compiling routine's responsibility to populate this field before
540 returning the regexp object.
542 Internally this is set to C<NULL> unless a custom engine is specified in
543 C<$^H{regcomp}>, perl's own set of callbacks can be accessed in the struct
544 pointed to by C<RE_ENGINE_PTR>.
548 TODO, see L<http://www.mail-archive.com/perl5-changes@perl.org/msg17328.html>
552 This will be used by perl to see what flags the regexp was compiled
553 with, this will normally be set to the value of the flags parameter by
554 the L<comp|/comp> callback. See the L<comp|/comp> documentation for
557 =head2 C<minlen> C<minlenret>
559 The minimum string length required for the pattern to match. This is used to
560 prune the search space by not bothering to match any closer to the end of a
561 string than would allow a match. For instance there is no point in even
562 starting the regex engine if the minlen is 10 but the string is only 5
563 characters long. There is no way that the pattern can match.
565 C<minlenret> is the minimum length of the string that would be found
568 The difference between C<minlen> and C<minlenret> can be seen in the
573 where the C<minlen> would be 3 but C<minlenret> would only be 2 as the \d is
574 required to match but is not actually included in the matched content. This
575 distinction is particularly important as the substitution logic uses the
576 C<minlenret> to tell whether it can do in-place substitution which can result in
577 considerable speedup.
581 Left offset from pos() to start match at.
585 Substring data about strings that must appear in the final match. This
586 is currently only used internally by perl's engine for but might be
587 used in the future for all engines for optimisations.
589 =head2 C<nparens>, C<lastparen>, and C<lastcloseparen>
591 These fields are used to keep track of how many paren groups could be matched
592 in the pattern, which was the last open paren to be entered, and which was
593 the last close paren to be entered.
597 The engine's private copy of the flags the pattern was compiled with. Usually
598 this is the same as C<extflags> unless the engine chose to modify one of them.
602 A void* pointing to an engine-defined data structure. The perl engine uses the
603 C<regexp_internal> structure (see L<perlreguts/Base Structures>) but a custom
604 engine should use something else.
608 Unused. Left in for compatibility with perl 5.10.0.
612 A C<regexp_paren_pair> structure which defines offsets into the string being
613 matched which correspond to the C<$&> and C<$1>, C<$2> etc. captures, the
614 C<regexp_paren_pair> struct is defined as follows:
616 typedef struct regexp_paren_pair {
621 If C<< ->offs[num].start >> or C<< ->offs[num].end >> is C<-1> then that
622 capture group did not match. C<< ->offs[0].start/end >> represents C<$&> (or
623 C<${^MATCH> under C<//p>) and C<< ->offs[paren].end >> matches C<$$paren> where
626 =head2 C<precomp> C<prelen>
628 Used for optimisations. C<precomp> holds a copy of the pattern that
629 was compiled and C<prelen> its length. When a new pattern is to be
630 compiled (such as inside a loop) the internal C<regcomp> operator
631 checks whether the last compiled C<REGEXP>'s C<precomp> and C<prelen>
632 are equivalent to the new one, and if so uses the old pattern instead
633 of compiling a new one.
635 The relevant snippet from C<Perl_pp_regcomp>:
637 if (!re || !re->precomp || re->prelen != (I32)len ||
638 memNE(re->precomp, t, len))
639 /* Compile a new pattern */
641 =head2 C<paren_names>
643 This is a hash used internally to track named capture groups and their
644 offsets. The keys are the names of the buffers the values are dualvars,
645 with the IV slot holding the number of buffers with the given name and the
646 pv being an embedded array of I32. The values may also be contained
647 independently in the data array in cases where named backreferences are
652 Holds information on the longest string that must occur at a fixed
653 offset from the start of the pattern, and the longest string that must
654 occur at a floating offset from the start of the pattern. Used to do
655 Fast-Boyer-Moore searches on the string to find out if its worth using
656 the regex engine at all, and if so where in the string to search.
658 =head2 C<subbeg> C<sublen> C<saved_copy>
660 Used during execution phase for managing search and replace patterns.
662 =head2 C<wrapped> C<wraplen>
664 Stores the string C<qr//> stringifies to. The perl engine for example
665 stores C<(?^:eek)> in the case of C<qr/eek/>.
667 When using a custom engine that doesn't support the C<(?:)> construct
668 for inline modifiers, it's probably best to have C<qr//> stringify to
669 the supplied pattern, note that this will create undesired patterns in
672 my $x = qr/a|b/; # "a|b"
673 my $y = qr/c/i; # "c"
674 my $z = qr/$x$y/; # "a|bc"
676 There's no solution for this problem other than making the custom
677 engine understand a construct like C<(?:)>.
681 This stores the number of eval groups in the pattern. This is used for security
682 purposes when embedding compiled regexes into larger patterns with C<qr//>.
686 The number of times the structure is referenced. When this falls to 0 the
687 regexp is automatically freed by a call to pregfree. This should be set to 1 in
688 each engine's L</comp> routine.
692 Originally part of L<perlreguts>.
696 Originally written by Yves Orton, expanded by E<AElig>var ArnfjE<ouml>rE<eth>
701 Copyright 2006 Yves Orton and 2007 E<AElig>var ArnfjE<ouml>rE<eth> Bjarmason.
703 This program is free software; you can redistribute it and/or modify it under
704 the same terms as Perl itself.