3 perlreapi - perl regular expression plugin interface
7 As of Perl 5.9.5 there is a new interface for using other regexp engines than
8 the default one. Each engine is supposed to provide access to a constant
9 structure of the following format:
11 typedef struct regexp_engine {
12 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
13 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
14 char* strbeg, I32 minend, SV* screamer,
15 void* data, U32 flags);
16 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
17 char *strend, U32 flags,
18 struct re_scream_pos_data_s *data);
19 SV* (*checkstr) (pTHX_ regexp *prog);
20 void (*free) (pTHX_ struct regexp* r);
21 SV* (*numbered_buff_get) (pTHX_ const REGEXP * const rx, I32 paren, SV* usesv);
22 SV* (*named_buff_get)(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags);
23 SV* (*qr_pkg)(pTHX_ const REGEXP * const rx);
25 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
29 When a regexp is compiled, its C<engine> field is then set to point at
30 the appropriate structure so that when it needs to be used Perl can find
31 the right routines to do so.
33 In order to install a new regexp handler, C<$^H{regcomp}> is set
34 to an integer which (when casted appropriately) resolves to one of these
35 structures. When compiling, the C<comp> method is executed, and the
36 resulting regexp structure's engine field is expected to point back at
39 The pTHX_ symbol in the definition is a macro used by perl under threading
40 to provide an extra argument to the routine holding a pointer back to
41 the interpreter that is executing the regexp. So under threading all
42 routines get an extra argument.
44 The routines are as follows:
48 REGEXP* comp(pTHX_ const SV * const pattern, const U32 flags);
50 Compile the pattern stored in C<pattern> using the given C<flags> and
51 return a pointer to a prepared C<REGEXP> structure that can perform
52 the match. See L</The REGEXP structure> below for an explanation of
53 the individual fields in the REGEXP struct.
55 The C<pattern> parameter is the scalar that was used as the
56 pattern. previous versions of perl would pass two C<char*> indicating
57 the start and end of the stringifed pattern, the following snippet can
58 be used to get the old parameters:
61 char* exp = SvPV(pattern, plen);
62 char* xend = exp + plen;
64 Since any scalar can be passed as a pattern it's possible to implement
65 an engine that does something with an array (C<< "ook" =~ [ qw/ eek
66 hlagh / ] >>) or with the non-stringified form of a compiled regular
67 expression (C<< "ook" =~ qr/eek/ >>). perl's own engine will always
68 stringify everything using the snippet above but that doesn't mean
69 other engines have to.
71 The C<flags> paramater is a bitfield which indicates which of the
72 C<msixk> flags the regex was compiled with. In addition it contains
73 info about whether C<use locale> is in effect and optimization info
74 for C<split>. A regex engine might want to use the same split
75 optimizations with a different syntax, for instance a Perl6 engine
76 would treat C<split /^^/> equivalently to perl's C<split /^/>, see
77 L<split documentation|perlfunc> and the relevant code in C<pp_split>
78 in F<pp.c> to find out whether your engine should be setting these.
80 The C<eogc> flags are stripped out before being passed to the comp
81 routine. The regex engine does not need to know whether any of these
82 are set as those flags should only affect what perl does with the
83 pattern and its match variables, not how it gets compiled & executed.
89 C<split ' '> or C<split> with no arguments (which really means
90 C<split(' ', $_> see L<split|perlfunc>).
94 Set if the pattern is C</^/> (C<<r->prelen == 1 && r->precomp[0] ==
95 '^'>>). Will be used by the C<split> operator to split the given
96 string on C<\n> (even under C</^/s>, see L<split|perlfunc>).
100 Set if the pattern is exactly C</\s+/> and used by C<split>, the
101 definition of whitespace varies depending on whether RXf_UTF8 or
102 RXf_PMf_LOCALE is set.
106 Makes C<split> use the locale dependant definition of whitespace under C<use
107 locale> when RXf_SKIPWHITE or RXf_WHITE is in effect. Under ASCII whitespace is
108 defined as per L<isSPACE|perlapi/ISSPACE>, and by the internal macros
109 C<is_utf8_space> under UTF-8 and C<isSPACE_LC> under C<use locale>.
111 =item RXf_PMf_MULTILINE
113 The C</m> flag, this ends up being passed to C<Perl_fbm_instr> by
114 C<pp_split> regardless of the engine.
116 =item RXf_PMf_SINGLELINE
118 The C</s> flag. Guaranteed not to be used outside the regex engine.
122 The C</i> flag. Guaranteed not to be used outside the regex engine.
124 =item RXf_PMf_EXTENDED
126 The C</x> flag. Guaranteed not to be used outside the regex
127 engine. However if present on a regex C<#> comments will be stripped
128 by the tokenizer regardless of the engine currently in use.
130 =item RXf_PMf_KEEPCOPY
136 Set if the pattern is L<SvUTF8()|perlapi/SvUTF8>, set by Perl_pmruntime.
140 In general these flags should be preserved in regex->extflags after
141 compilation, although it is possible the regex includes constructs
142 that changes them. The perl engine for instance may upgrade non-utf8
143 strings to utf8 if the pattern includes constructs such as C<\x{...}>
144 that can only match unicode values. RXf_SKIPWHITE should always be
145 preserved verbatim in regex->extflags.
149 I32 exec(regexp* prog,
150 char *stringarg, char* strend, char* strbeg,
151 I32 minend, SV* screamer,
152 void* data, U32 flags);
158 char* intuit( regexp *prog,
159 SV *sv, char *strpos, char *strend,
160 U32 flags, struct re_scream_pos_data_s *data);
162 Find the start position where a regex match should be attempted,
163 or possibly whether the regex engine should not be run because the
164 pattern can't match. This is called as appropriate by the core
165 depending on the values of the extflags member of the regexp
170 SV* checkstr(regexp *prog);
172 Return a SV containing a string that must appear in the pattern. Used
173 by C<split> for optimising matches.
177 void free(regexp *prog);
179 Called by perl when it is freeing a regexp pattern so that the engine
180 can release any resources pointed to by the C<pprivate> member of the
181 regexp structure. This is only responsible for freeing private data;
182 perl will handle releasing anything else contained in the regexp structure.
184 =head2 numbered_buff_get
186 SV* numbered_buff_get(pTHX_ const REGEXP * const rx, I32 paren, SV* usesv);
190 =head2 named_buff_get
192 SV* named_buff_get(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags);
198 SV* qr_pkg(pTHX_ const REGEXP * const rx);
200 The package the qr// magic object is blessed into (as seen by C<ref
201 qr//>). It is recommended that engines change this to its package
205 Example_reg_qr_pkg(pTHX_ const REGEXP * const rx)
208 return newSVpvs("re::engine::Example");
211 Any method calls on an object created with C<qr//> will be dispatched to the
212 package as a normal object.
214 use re::engine::Example;
216 $re->meth; # dispatched to re::engine::Example::meth()
218 To retrieve the C<REGEXP> object from the scalar in an XS function use the
229 (sv = (SV*)SvRV(sv)) && /* assignment deliberate */
230 SvTYPE(sv) == SVt_PVMG &&
231 (mg = mg_find(sv, PERL_MAGIC_qr))) /* assignment deliberate */
233 re = (REGEXP *)mg->mg_obj;
236 Or use the (CURRENTLY UNDOCUMENETED!) C<Perl_get_re_arg> function:
240 const REGEXP * const re = (REGEXP *)Perl_get_re_arg( aTHX_ rv, 0, NULL );
244 void* dupe(const regexp *r, CLONE_PARAMS *param);
246 On threaded builds a regexp may need to be duplicated so that the pattern
247 can be used by mutiple threads. This routine is expected to handle the
248 duplication of any private data pointed to by the C<pprivate> member of
249 the regexp structure. It will be called with the preconstructed new
250 regexp structure as an argument, the C<pprivate> member will point at
251 the B<old> private structue, and it is this routine's responsibility to
252 construct a copy and return a pointer to it (which perl will then use to
253 overwrite the field as passed to this routine.)
255 This allows the engine to dupe its private data but also if necessary
256 modify the final structure if it really must.
258 On unthreaded builds this field doesn't exist.
260 =head1 The REGEXP structure
262 The REGEXP struct is defined in F<regexp.h>. All regex engines must be able to
263 correctly build such a structure in their L</comp> routine.
265 The REGEXP structure contains all the data that perl needs to be aware of
266 to properly work with the regular expression. It includes data about
267 optimisations that perl can use to determine if the regex engine should
268 really be used, and various other control info that is needed to properly
269 execute patterns in various contexts such as is the pattern anchored in
270 some way, or what flags were used during the compile, or whether the
271 program contains special constructs that perl needs to be aware of.
273 In addition it contains two fields that are intended for the private use
274 of the regex engine that compiled the pattern. These are the C<intflags>
275 and pprivate members. The C<pprivate> is a void pointer to an arbitrary
276 structure whose use and management is the responsibility of the compiling
277 engine. perl will never modify either of these values.
279 typedef struct regexp {
280 /* what engine created this regexp? */
281 const struct regexp_engine* engine;
283 /* what re is this a lightweight copy of? */
284 struct regexp* mother_re;
286 /* Information about the match that the perl core uses to manage things */
287 U32 extflags; /* Flags used both externally and internally */
288 I32 minlen; /* mininum possible length of string to match */
289 I32 minlenret; /* mininum possible length of $& */
290 U32 gofs; /* chars left of pos that we search from */
292 /* substring data about strings that must appear
293 in the final match, used for optimisations */
294 struct reg_substr_data *substrs;
296 U32 nparens; /* number of capture buffers */
298 /* private engine specific data */
299 U32 intflags; /* Engine Specific Internal flags */
300 void *pprivate; /* Data private to the regex engine which
301 created this object. */
303 /* Data about the last/current match. These are modified during matching*/
304 U32 lastparen; /* last open paren matched */
305 U32 lastcloseparen; /* last close paren matched */
306 regexp_paren_pair *swap; /* Swap copy of *offs */
307 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
309 char *subbeg; /* saved or original string so \digit works forever. */
310 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
311 I32 sublen; /* Length of string pointed by subbeg */
313 /* Information about the match that isn't often used */
314 I32 prelen; /* length of precomp */
315 const char *precomp; /* pre-compilation regular expression */
317 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
318 char *wrapped; /* wrapped version of the pattern */
319 I32 wraplen; /* length of wrapped */
321 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
322 HV *paren_names; /* Optional hash of paren names */
324 /* Refcount of this regexp */
325 I32 refcnt; /* Refcount of this regexp */
328 The fields are discussed in more detail below:
334 This field points at a regexp_engine structure which contains pointers
335 to the subroutines that are to be used for performing a match. It
336 is the compiling routine's responsibility to populate this field before
337 returning the regexp object.
339 Internally this is set to C<NULL> unless a custom engine is specified in
340 C<$^H{regcomp}>, perl's own set of callbacks can be accessed in the struct
341 pointed to by C<RE_ENGINE_PTR>.
345 TODO, see L<http://www.mail-archive.com/perl5-changes@perl.org/msg17328.html>
349 This will be used by perl to see what flags the regexp was compiled with, this
350 will normally be set to the value of the flags parameter on L</comp>.
352 =item C<minlen> C<minlenret>
354 The minimum string length required for the pattern to match. This is used to
355 prune the search space by not bothering to match any closer to the end of a
356 string than would allow a match. For instance there is no point in even
357 starting the regex engine if the minlen is 10 but the string is only 5
358 characters long. There is no way that the pattern can match.
360 C<minlenret> is the minimum length of the string that would be found
363 The difference between C<minlen> and C<minlenret> can be seen in the
368 where the C<minlen> would be 3 but C<minlenret> would only be 2 as the \d is
369 required to match but is not actually included in the matched content. This
370 distinction is particularly important as the substitution logic uses the
371 C<minlenret> to tell whether it can do in-place substition which can result in
372 considerable speedup.
376 Left offset from pos() to start match at.
382 =item C<nparens>, C<lasparen>, and C<lastcloseparen>
384 These fields are used to keep track of how many paren groups could be matched
385 in the pattern, which was the last open paren to be entered, and which was
386 the last close paren to be entered.
390 The engine's private copy of the flags the pattern was compiled with. Usually
391 this is the same as C<extflags> unless the engine chose to modify one of them
395 A void* pointing to an engine-defined data structure. The perl engine uses the
396 C<regexp_internal> structure (see L<perlreguts/Base Structures>) but a custom
397 engine should use something else.
405 A C<regexp_paren_pair> structure which defines offsets into the string being
406 matched which correspond to the C<$&> and C<$1>, C<$2> etc. captures, the
407 C<regexp_paren_pair> struct is defined as follows:
409 typedef struct regexp_paren_pair {
414 If C<< ->offs[num].start >> or C<< ->offs[num].end >> is C<-1> then that
415 capture buffer did not match. C<< ->offs[0].start/end >> represents C<$&> (or
416 C<${^MATCH> under C<//p>) and C<< ->offs[paren].end >> matches C<$$paren> where
419 =item C<precomp> C<prelen>
421 Used for debugging purposes. C<precomp> holds a copy of the pattern
422 that was compiled and C<prelen> its length.
426 This is a hash used internally to track named capture buffers and their
427 offsets. The keys are the names of the buffers the values are dualvars,
428 with the IV slot holding the number of buffers with the given name and the
429 pv being an embedded array of I32. The values may also be contained
430 independently in the data array in cases where named backreferences are
433 =item C<reg_substr_data>
435 Holds information on the longest string that must occur at a fixed
436 offset from the start of the pattern, and the longest string that must
437 occur at a floating offset from the start of the pattern. Used to do
438 Fast-Boyer-Moore searches on the string to find out if its worth using
439 the regex engine at all, and if so where in the string to search.
441 =item C<startp>, C<endp>
443 These fields store arrays that are used to hold the offsets of the begining
444 and end of each capture group that has matched. -1 is used to indicate no match.
446 These are the source for @- and @+.
448 =item C<subbeg> C<sublen> C<saved_copy>
450 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
451 if (RX_MATCH_COPIED(ret))
452 ret->subbeg = SAVEPVN(ret->subbeg, ret->sublen);
456 C<PL_sawampersand || rx->extflags & RXf_PMf_KEEPCOPY>
458 These are used during execution phase for managing search and replace
461 =item C<wrapped> C<wraplen>
463 Stores the string C<qr//> stringifies to, for example C<(?-xism:eek)>
464 in the case of C<qr/eek/>.
466 When using a custom engine that doesn't support the C<(?:)> construct for
467 inline modifiers it's best to have C<qr//> stringify to the supplied pattern,
468 note that this will create invalid patterns in cases such as:
470 my $x = qr/a|b/; # "a|b"
472 my $z = qr/$x$y/; # "a|bc"
474 There's no solution for such problems other than making the custom engine
475 understand some for of inline modifiers.
477 The C<Perl_reg_stringify> in F<regcomp.c> does the stringification work.
481 This stores the number of eval groups in the pattern. This is used for security
482 purposes when embedding compiled regexes into larger patterns with C<qr//>.
486 The number of times the structure is referenced. When this falls to 0 the
487 regexp is automatically freed by a call to pregfree. This should be set to 1 in
488 each engine's L</comp> routine.
492 =head2 De-allocation and Cloning
494 Any patch that adds data items to the REGEXP struct will need to include
495 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c> (C<pregfree()>). This
496 involves freeing or cloning items in the regexp's data array based on the data
501 Originally part of L<perlreguts>.
505 Originally written by Yves Orton, expanded by E<AElig>var ArnfjE<ouml>rE<eth>
510 Copyright 2006 Yves Orton and 2007 E<AElig>var ArnfjE<ouml>rE<eth> Bjarmason.
512 This program is free software; you can redistribute it and/or modify it under
513 the same terms as Perl itself.