xref: /openbsd-src/gnu/usr.bin/perl/pod/perlreapi.pod (revision 50b7afb2c2c0993b0894d4e34bf857cb13ed9c80)
1=head1 NAME
2
3perlreapi - Perl regular expression plugin interface
4
5=head1 DESCRIPTION
6
7As of Perl 5.9.5 there is a new interface for plugging and using
8regular expression engines other than the default one.
9
10Each engine is supposed to provide access to a constant structure of the
11following format:
12
13    typedef struct regexp_engine {
14        REGEXP* (*comp) (pTHX_
15                         const SV * const pattern, const U32 flags);
16        I32     (*exec) (pTHX_
17                         REGEXP * const rx,
18                         char* stringarg,
19                         char* strend, char* strbeg,
20                         I32 minend, SV* screamer,
21                         void* data, U32 flags);
22        char*   (*intuit) (pTHX_
23                           REGEXP * const rx, SV *sv,
24                           char *strpos, char *strend, U32 flags,
25                           struct re_scream_pos_data_s *data);
26        SV*     (*checkstr) (pTHX_ REGEXP * const rx);
27        void    (*free) (pTHX_ REGEXP * const rx);
28        void    (*numbered_buff_FETCH) (pTHX_
29                                        REGEXP * const rx,
30                                        const I32 paren,
31                                        SV * const sv);
32        void    (*numbered_buff_STORE) (pTHX_
33                                        REGEXP * const rx,
34                                        const I32 paren,
35                                        SV const * const value);
36        I32     (*numbered_buff_LENGTH) (pTHX_
37                                         REGEXP * const rx,
38                                         const SV * const sv,
39                                         const I32 paren);
40        SV*     (*named_buff) (pTHX_
41                               REGEXP * const rx,
42                               SV * const key,
43                               SV * const value,
44                               U32 flags);
45        SV*     (*named_buff_iter) (pTHX_
46                                    REGEXP * const rx,
47                                    const SV * const lastkey,
48                                    const U32 flags);
49        SV*     (*qr_package)(pTHX_ REGEXP * const rx);
50    #ifdef USE_ITHREADS
51        void*   (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
52    #endif
53        REGEXP* (*op_comp) (...);
54
55
56When a regexp is compiled, its C<engine> field is then set to point at
57the appropriate structure, so that when it needs to be used Perl can find
58the right routines to do so.
59
60In order to install a new regexp handler, C<$^H{regcomp}> is set
61to an integer which (when casted appropriately) resolves to one of these
62structures.  When compiling, the C<comp> method is executed, and the
63resulting C<regexp> structure's engine field is expected to point back at
64the same structure.
65
66The pTHX_ symbol in the definition is a macro used by Perl under threading
67to provide an extra argument to the routine holding a pointer back to
68the interpreter that is executing the regexp. So under threading all
69routines get an extra argument.
70
71=head1 Callbacks
72
73=head2 comp
74
75    REGEXP* comp(pTHX_ const SV * const pattern, const U32 flags);
76
77Compile the pattern stored in C<pattern> using the given C<flags> and
78return a pointer to a prepared C<REGEXP> structure that can perform
79the match.  See L</The REGEXP structure> below for an explanation of
80the individual fields in the REGEXP struct.
81
82The C<pattern> parameter is the scalar that was used as the
83pattern.  Previous versions of Perl would pass two C<char*> indicating
84the start and end of the stringified pattern; the following snippet can
85be used to get the old parameters:
86
87    STRLEN plen;
88    char*  exp = SvPV(pattern, plen);
89    char* xend = exp + plen;
90
91Since any scalar can be passed as a pattern, it's possible to implement
92an engine that does something with an array (C<< "ook" =~ [ qw/ eek
93hlagh / ] >>) or with the non-stringified form of a compiled regular
94expression (C<< "ook" =~ qr/eek/ >>).  Perl's own engine will always
95stringify everything using the snippet above, but that doesn't mean
96other engines have to.
97
98The C<flags> parameter is a bitfield which indicates which of the
99C<msixp> flags the regex was compiled with.  It also contains
100additional info, such as if C<use locale> is in effect.
101
102The C<eogc> flags are stripped out before being passed to the comp
103routine.  The regex engine does not need to know if any of these
104are set, as those flags should only affect what Perl does with the
105pattern and its match variables, not how it gets compiled and
106executed.
107
108By the time the comp callback is called, some of these flags have
109already had effect (noted below where applicable).  However most of
110their effect occurs after the comp callback has run, in routines that
111read the C<< rx->extflags >> field which it populates.
112
113In general the flags should be preserved in C<< rx->extflags >> after
114compilation, although the regex engine might want to add or delete
115some of them to invoke or disable some special behavior in Perl.  The
116flags along with any special behavior they cause are documented below:
117
118The pattern modifiers:
119
120=over 4
121
122=item C</m> - RXf_PMf_MULTILINE
123
124If this is in C<< rx->extflags >> it will be passed to
125C<Perl_fbm_instr> by C<pp_split> which will treat the subject string
126as a multi-line string.
127
128=item C</s> - RXf_PMf_SINGLELINE
129
130=item C</i> - RXf_PMf_FOLD
131
132=item C</x> - RXf_PMf_EXTENDED
133
134If present on a regex, C<"#"> comments will be handled differently by the
135tokenizer in some cases.
136
137TODO: Document those cases.
138
139=item C</p> - RXf_PMf_KEEPCOPY
140
141TODO: Document this
142
143=item Character set
144
145The character set semantics are determined by an enum that is contained
146in this field.  This is still experimental and subject to change, but
147the current interface returns the rules by use of the in-line function
148C<get_regex_charset(const U32 flags)>.  The only currently documented
149value returned from it is REGEX_LOCALE_CHARSET, which is set if
150C<use locale> is in effect. If present in C<< rx->extflags >>,
151C<split> will use the locale dependent definition of whitespace
152when RXf_SKIPWHITE or RXf_WHITE is in effect.  ASCII whitespace
153is defined as per L<isSPACE|perlapi/isSPACE>, and by the internal
154macros C<is_utf8_space> under UTF-8, and C<isSPACE_LC> under C<use
155locale>.
156
157=back
158
159Additional flags:
160
161=over 4
162
163=item RXf_SPLIT
164
165This flag was removed in perl 5.18.0.  C<split ' '> is now special-cased
166solely in the parser.  RXf_SPLIT is still #defined, so you can test for it.
167This is how it used to work:
168
169If C<split> is invoked as C<split ' '> or with no arguments (which
170really means C<split(' ', $_)>, see L<split|perlfunc/split>), Perl will
171set this flag.  The regex engine can then check for it and set the
172SKIPWHITE and WHITE extflags.  To do this, the Perl engine does:
173
174    if (flags & RXf_SPLIT && r->prelen == 1 && r->precomp[0] == ' ')
175        r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
176
177=back
178
179These flags can be set during compilation to enable optimizations in
180the C<split> operator.
181
182=over 4
183
184=item RXf_SKIPWHITE
185
186This flag was removed in perl 5.18.0.  It is still #defined, so you can
187set it, but doing so will have no effect.  This is how it used to work:
188
189If the flag is present in C<< rx->extflags >> C<split> will delete
190whitespace from the start of the subject string before it's operated
191on.  What is considered whitespace depends on if the subject is a
192UTF-8 string and if the C<RXf_PMf_LOCALE> flag is set.
193
194If RXf_WHITE is set in addition to this flag, C<split> will behave like
195C<split " "> under the Perl engine.
196
197=item RXf_START_ONLY
198
199Tells the split operator to split the target string on newlines
200(C<\n>) without invoking the regex engine.
201
202Perl's engine sets this if the pattern is C</^/> (C<plen == 1 && *exp
203== '^'>), even under C</^/s>; see L<split|perlfunc>.  Of course a
204different regex engine might want to use the same optimizations
205with a different syntax.
206
207=item RXf_WHITE
208
209Tells the split operator to split the target string on whitespace
210without invoking the regex engine.  The definition of whitespace varies
211depending on if the target string is a UTF-8 string and on
212if RXf_PMf_LOCALE is set.
213
214Perl's engine sets this flag if the pattern is C<\s+>.
215
216=item RXf_NULL
217
218Tells the split operator to split the target string on
219characters.  The definition of character varies depending on if
220the target string is a UTF-8 string.
221
222Perl's engine sets this flag on empty patterns, this optimization
223makes C<split //> much faster than it would otherwise be.  It's even
224faster than C<unpack>.
225
226=item RXf_NO_INPLACE_SUBST
227
228Added in perl 5.18.0, this flag indicates that a regular expression might
229perform an operation that would interfere with inplace substituion. For
230instance it might contain lookbehind, or assign to non-magical variables
231(such as $REGMARK and $REGERROR) during matching.  C<s///> will skip
232certain optimisations when this is set.
233
234=back
235
236=head2 exec
237
238    I32 exec(pTHX_ REGEXP * const rx,
239             char *stringarg, char* strend, char* strbeg,
240             I32 minend, SV* screamer,
241             void* data, U32 flags);
242
243Execute a regexp. The arguments are
244
245=over 4
246
247=item rx
248
249The regular expression to execute.
250
251=item screamer
252
253This strangely-named arg is the SV to be matched against.  Note that the
254actual char array to be matched against is supplied by the arguments
255described below; the SV is just used to determine UTF8ness, C<pos()> etc.
256
257=item strbeg
258
259Pointer to the physical start of the string.
260
261=item strend
262
263Pointer to the character following the physical end of the string (i.e.
264the C<\0>).
265
266=item stringarg
267
268Pointer to the position in the string where matching should start; it might
269not be equal to C<strbeg> (for example in a later iteration of C</.../g>).
270
271=item minend
272
273Minimum length of string (measured in bytes from C<stringarg>) that must
274match; if the engine reaches the end of the match but hasn't reached this
275position in the string, it should fail.
276
277=item data
278
279Optimisation data; subject to change.
280
281=item flags
282
283Optimisation flags; subject to change.
284
285=back
286
287=head2 intuit
288
289    char* intuit(pTHX_ REGEXP * const rx,
290                  SV *sv, char *strpos, char *strend,
291                  const U32 flags, struct re_scream_pos_data_s *data);
292
293Find the start position where a regex match should be attempted,
294or possibly if the regex engine should not be run because the
295pattern can't match.  This is called, as appropriate, by the core,
296depending on the values of the C<extflags> member of the C<regexp>
297structure.
298
299=head2 checkstr
300
301    SV*	checkstr(pTHX_ REGEXP * const rx);
302
303Return a SV containing a string that must appear in the pattern. Used
304by C<split> for optimising matches.
305
306=head2 free
307
308    void free(pTHX_ REGEXP * const rx);
309
310Called by Perl when it is freeing a regexp pattern so that the engine
311can release any resources pointed to by the C<pprivate> member of the
312C<regexp> structure.  This is only responsible for freeing private data;
313Perl will handle releasing anything else contained in the C<regexp> structure.
314
315=head2 Numbered capture callbacks
316
317Called to get/set the value of C<$`>, C<$'>, C<$&> and their named
318equivalents, ${^PREMATCH}, ${^POSTMATCH} and $^{MATCH}, as well as the
319numbered capture groups (C<$1>, C<$2>, ...).
320
321The C<paren> parameter will be C<1> for C<$1>, C<2> for C<$2> and so
322forth, and have these symbolic values for the special variables:
323
324    ${^PREMATCH}  RX_BUFF_IDX_CARET_PREMATCH
325    ${^POSTMATCH} RX_BUFF_IDX_CARET_POSTMATCH
326    ${^MATCH}     RX_BUFF_IDX_CARET_FULLMATCH
327    $`            RX_BUFF_IDX_PREMATCH
328    $'            RX_BUFF_IDX_POSTMATCH
329    $&            RX_BUFF_IDX_FULLMATCH
330
331Note that in Perl 5.17.3 and earlier, the last three constants were also
332used for the caret variants of the variables.
333
334
335The names have been chosen by analogy with L<Tie::Scalar> methods
336names with an additional B<LENGTH> callback for efficiency.  However
337named capture variables are currently not tied internally but
338implemented via magic.
339
340=head3 numbered_buff_FETCH
341
342    void numbered_buff_FETCH(pTHX_ REGEXP * const rx, const I32 paren,
343                             SV * const sv);
344
345Fetch a specified numbered capture.  C<sv> should be set to the scalar
346to return, the scalar is passed as an argument rather than being
347returned from the function because when it's called Perl already has a
348scalar to store the value, creating another one would be
349redundant.  The scalar can be set with C<sv_setsv>, C<sv_setpvn> and
350friends, see L<perlapi>.
351
352This callback is where Perl untaints its own capture variables under
353taint mode (see L<perlsec>).  See the C<Perl_reg_numbered_buff_fetch>
354function in F<regcomp.c> for how to untaint capture variables if
355that's something you'd like your engine to do as well.
356
357=head3 numbered_buff_STORE
358
359    void    (*numbered_buff_STORE) (pTHX_
360                                    REGEXP * const rx,
361                                    const I32 paren,
362                                    SV const * const value);
363
364Set the value of a numbered capture variable.  C<value> is the scalar
365that is to be used as the new value.  It's up to the engine to make
366sure this is used as the new value (or reject it).
367
368Example:
369
370    if ("ook" =~ /(o*)/) {
371        # 'paren' will be '1' and 'value' will be 'ee'
372        $1 =~ tr/o/e/;
373    }
374
375Perl's own engine will croak on any attempt to modify the capture
376variables, to do this in another engine use the following callback
377(copied from C<Perl_reg_numbered_buff_store>):
378
379    void
380    Example_reg_numbered_buff_store(pTHX_
381                                    REGEXP * const rx,
382                                    const I32 paren,
383                                    SV const * const value)
384    {
385        PERL_UNUSED_ARG(rx);
386        PERL_UNUSED_ARG(paren);
387        PERL_UNUSED_ARG(value);
388
389        if (!PL_localizing)
390            Perl_croak(aTHX_ PL_no_modify);
391    }
392
393Actually Perl will not I<always> croak in a statement that looks
394like it would modify a numbered capture variable.  This is because the
395STORE callback will not be called if Perl can determine that it
396doesn't have to modify the value.  This is exactly how tied variables
397behave in the same situation:
398
399    package CaptureVar;
400    use base 'Tie::Scalar';
401
402    sub TIESCALAR { bless [] }
403    sub FETCH { undef }
404    sub STORE { die "This doesn't get called" }
405
406    package main;
407
408    tie my $sv => "CaptureVar";
409    $sv =~ y/a/b/;
410
411Because C<$sv> is C<undef> when the C<y///> operator is applied to it,
412the transliteration won't actually execute and the program won't
413C<die>.  This is different to how 5.8 and earlier versions behaved
414since the capture variables were READONLY variables then; now they'll
415just die when assigned to in the default engine.
416
417=head3 numbered_buff_LENGTH
418
419    I32 numbered_buff_LENGTH (pTHX_
420                              REGEXP * const rx,
421                              const SV * const sv,
422                              const I32 paren);
423
424Get the C<length> of a capture variable.  There's a special callback
425for this so that Perl doesn't have to do a FETCH and run C<length> on
426the result, since the length is (in Perl's case) known from an offset
427stored in C<< rx->offs >>, this is much more efficient:
428
429    I32 s1  = rx->offs[paren].start;
430    I32 s2  = rx->offs[paren].end;
431    I32 len = t1 - s1;
432
433This is a little bit more complex in the case of UTF-8, see what
434C<Perl_reg_numbered_buff_length> does with
435L<is_utf8_string_loclen|perlapi/is_utf8_string_loclen>.
436
437=head2 Named capture callbacks
438
439Called to get/set the value of C<%+> and C<%->, as well as by some
440utility functions in L<re>.
441
442There are two callbacks, C<named_buff> is called in all the cases the
443FETCH, STORE, DELETE, CLEAR, EXISTS and SCALAR L<Tie::Hash> callbacks
444would be on changes to C<%+> and C<%-> and C<named_buff_iter> in the
445same cases as FIRSTKEY and NEXTKEY.
446
447The C<flags> parameter can be used to determine which of these
448operations the callbacks should respond to.  The following flags are
449currently defined:
450
451Which L<Tie::Hash> operation is being performed from the Perl level on
452C<%+> or C<%+>, if any:
453
454    RXapif_FETCH
455    RXapif_STORE
456    RXapif_DELETE
457    RXapif_CLEAR
458    RXapif_EXISTS
459    RXapif_SCALAR
460    RXapif_FIRSTKEY
461    RXapif_NEXTKEY
462
463If C<%+> or C<%-> is being operated on, if any.
464
465    RXapif_ONE /* %+ */
466    RXapif_ALL /* %- */
467
468If this is being called as C<re::regname>, C<re::regnames> or
469C<re::regnames_count>, if any.  The first two will be combined with
470C<RXapif_ONE> or C<RXapif_ALL>.
471
472    RXapif_REGNAME
473    RXapif_REGNAMES
474    RXapif_REGNAMES_COUNT
475
476Internally C<%+> and C<%-> are implemented with a real tied interface
477via L<Tie::Hash::NamedCapture>.  The methods in that package will call
478back into these functions.  However the usage of
479L<Tie::Hash::NamedCapture> for this purpose might change in future
480releases.  For instance this might be implemented by magic instead
481(would need an extension to mgvtbl).
482
483=head3 named_buff
484
485    SV*     (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
486                           SV * const value, U32 flags);
487
488=head3 named_buff_iter
489
490    SV*     (*named_buff_iter) (pTHX_
491                                REGEXP * const rx,
492                                const SV * const lastkey,
493                                const U32 flags);
494
495=head2 qr_package
496
497    SV* qr_package(pTHX_ REGEXP * const rx);
498
499The package the qr// magic object is blessed into (as seen by C<ref
500qr//>).  It is recommended that engines change this to their package
501name for identification regardless of if they implement methods
502on the object.
503
504The package this method returns should also have the internal
505C<Regexp> package in its C<@ISA>.  C<< qr//->isa("Regexp") >> should always
506be true regardless of what engine is being used.
507
508Example implementation might be:
509
510    SV*
511    Example_qr_package(pTHX_ REGEXP * const rx)
512    {
513    	PERL_UNUSED_ARG(rx);
514    	return newSVpvs("re::engine::Example");
515    }
516
517Any method calls on an object created with C<qr//> will be dispatched to the
518package as a normal object.
519
520    use re::engine::Example;
521    my $re = qr//;
522    $re->meth; # dispatched to re::engine::Example::meth()
523
524To retrieve the C<REGEXP> object from the scalar in an XS function use
525the C<SvRX> macro, see L<"REGEXP Functions" in perlapi|perlapi/REGEXP
526Functions>.
527
528    void meth(SV * rv)
529    PPCODE:
530        REGEXP * re = SvRX(sv);
531
532=head2 dupe
533
534    void* dupe(pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
535
536On threaded builds a regexp may need to be duplicated so that the pattern
537can be used by multiple threads.  This routine is expected to handle the
538duplication of any private data pointed to by the C<pprivate> member of
539the C<regexp> structure.  It will be called with the preconstructed new
540C<regexp> structure as an argument, the C<pprivate> member will point at
541the B<old> private structure, and it is this routine's responsibility to
542construct a copy and return a pointer to it (which Perl will then use to
543overwrite the field as passed to this routine.)
544
545This allows the engine to dupe its private data but also if necessary
546modify the final structure if it really must.
547
548On unthreaded builds this field doesn't exist.
549
550=head2 op_comp
551
552This is private to the Perl core and subject to change. Should be left
553null.
554
555=head1 The REGEXP structure
556
557The REGEXP struct is defined in F<regexp.h>.
558All regex engines must be able to
559correctly build such a structure in their L</comp> routine.
560
561The REGEXP structure contains all the data that Perl needs to be aware of
562to properly work with the regular expression.  It includes data about
563optimisations that Perl can use to determine if the regex engine should
564really be used, and various other control info that is needed to properly
565execute patterns in various contexts, such as if the pattern anchored in
566some way, or what flags were used during the compile, or if the
567program contains special constructs that Perl needs to be aware of.
568
569In addition it contains two fields that are intended for the private
570use of the regex engine that compiled the pattern.  These are the
571C<intflags> and C<pprivate> members.  C<pprivate> is a void pointer to
572an arbitrary structure, whose use and management is the responsibility
573of the compiling engine.  Perl will never modify either of these
574values.
575
576    typedef struct regexp {
577        /* what engine created this regexp? */
578        const struct regexp_engine* engine;
579
580        /* what re is this a lightweight copy of? */
581        struct regexp* mother_re;
582
583        /* Information about the match that the Perl core uses to manage
584         * things */
585        U32 extflags;   /* Flags used both externally and internally */
586	I32 minlen;	/* mininum possible number of chars in */
587                           string to match */
588	I32 minlenret;	/* mininum possible number of chars in $& */
589        U32 gofs;       /* chars left of pos that we search from */
590
591        /* substring data about strings that must appear
592           in the final match, used for optimisations */
593        struct reg_substr_data *substrs;
594
595        U32 nparens;  /* number of capture groups */
596
597        /* private engine specific data */
598        U32 intflags;   /* Engine Specific Internal flags */
599        void *pprivate; /* Data private to the regex engine which
600                           created this object. */
601
602        /* Data about the last/current match. These are modified during
603         * matching*/
604        U32 lastparen;            /* highest close paren matched ($+) */
605        U32 lastcloseparen;       /* last close paren matched ($^N) */
606        regexp_paren_pair *swap;  /* Swap copy of *offs */
607        regexp_paren_pair *offs;  /* Array of offsets for (@-) and
608                                     (@+) */
609
610        char *subbeg;  /* saved or original string so \digit works
611                          forever. */
612        SV_SAVED_COPY  /* If non-NULL, SV which is COW from original */
613        I32 sublen;    /* Length of string pointed by subbeg */
614        I32 suboffset;	/* byte offset of subbeg from logical start of
615                           str */
616	I32 subcoffset;	/* suboffset equiv, but in chars (for @-/@+) */
617
618        /* Information about the match that isn't often used */
619        I32 prelen;           /* length of precomp */
620        const char *precomp;  /* pre-compilation regular expression */
621
622        char *wrapped;  /* wrapped version of the pattern */
623        I32 wraplen;    /* length of wrapped */
624
625        I32 seen_evals;   /* number of eval groups in the pattern - for
626                             security checks */
627        HV *paren_names;  /* Optional hash of paren names */
628
629        /* Refcount of this regexp */
630        I32 refcnt;             /* Refcount of this regexp */
631    } regexp;
632
633The fields are discussed in more detail below:
634
635=head2 C<engine>
636
637This field points at a C<regexp_engine> structure which contains pointers
638to the subroutines that are to be used for performing a match.  It
639is the compiling routine's responsibility to populate this field before
640returning the regexp object.
641
642Internally this is set to C<NULL> unless a custom engine is specified in
643C<$^H{regcomp}>, Perl's own set of callbacks can be accessed in the struct
644pointed to by C<RE_ENGINE_PTR>.
645
646=head2 C<mother_re>
647
648TODO, see L<http://www.mail-archive.com/perl5-changes@perl.org/msg17328.html>
649
650=head2 C<extflags>
651
652This will be used by Perl to see what flags the regexp was compiled
653with, this will normally be set to the value of the flags parameter by
654the L<comp|/comp> callback.  See the L<comp|/comp> documentation for
655valid flags.
656
657=head2 C<minlen> C<minlenret>
658
659The minimum string length (in characters) required for the pattern to match.
660This is used to
661prune the search space by not bothering to match any closer to the end of a
662string than would allow a match.  For instance there is no point in even
663starting the regex engine if the minlen is 10 but the string is only 5
664characters long.  There is no way that the pattern can match.
665
666C<minlenret> is the minimum length (in characters) of the string that would
667be found in $& after a match.
668
669The difference between C<minlen> and C<minlenret> can be seen in the
670following pattern:
671
672    /ns(?=\d)/
673
674where the C<minlen> would be 3 but C<minlenret> would only be 2 as the \d is
675required to match but is not actually
676included in the matched content.  This
677distinction is particularly important as the substitution logic uses the
678C<minlenret> to tell if it can do in-place substitutions (these can
679result in considerable speed-up).
680
681=head2 C<gofs>
682
683Left offset from pos() to start match at.
684
685=head2 C<substrs>
686
687Substring data about strings that must appear in the final match.  This
688is currently only used internally by Perl's engine, but might be
689used in the future for all engines for optimisations.
690
691=head2 C<nparens>, C<lastparen>, and C<lastcloseparen>
692
693These fields are used to keep track of how many paren groups could be matched
694in the pattern, which was the last open paren to be entered, and which was
695the last close paren to be entered.
696
697=head2 C<intflags>
698
699The engine's private copy of the flags the pattern was compiled with. Usually
700this is the same as C<extflags> unless the engine chose to modify one of them.
701
702=head2 C<pprivate>
703
704A void* pointing to an engine-defined
705data structure.  The Perl engine uses the
706C<regexp_internal> structure (see L<perlreguts/Base Structures>) but a custom
707engine should use something else.
708
709=head2 C<swap>
710
711Unused.  Left in for compatibility with Perl 5.10.0.
712
713=head2 C<offs>
714
715A C<regexp_paren_pair> structure which defines offsets into the string being
716matched which correspond to the C<$&> and C<$1>, C<$2> etc. captures, the
717C<regexp_paren_pair> struct is defined as follows:
718
719    typedef struct regexp_paren_pair {
720        I32 start;
721        I32 end;
722    } regexp_paren_pair;
723
724If C<< ->offs[num].start >> or C<< ->offs[num].end >> is C<-1> then that
725capture group did not match.
726C<< ->offs[0].start/end >> represents C<$&> (or
727C<${^MATCH}> under C<//p>) and C<< ->offs[paren].end >> matches C<$$paren> where
728C<$paren >= 1>.
729
730=head2 C<precomp> C<prelen>
731
732Used for optimisations.  C<precomp> holds a copy of the pattern that
733was compiled and C<prelen> its length.  When a new pattern is to be
734compiled (such as inside a loop) the internal C<regcomp> operator
735checks if the last compiled C<REGEXP>'s C<precomp> and C<prelen>
736are equivalent to the new one, and if so uses the old pattern instead
737of compiling a new one.
738
739The relevant snippet from C<Perl_pp_regcomp>:
740
741	if (!re || !re->precomp || re->prelen != (I32)len ||
742	    memNE(re->precomp, t, len))
743        /* Compile a new pattern */
744
745=head2 C<paren_names>
746
747This is a hash used internally to track named capture groups and their
748offsets.  The keys are the names of the buffers the values are dualvars,
749with the IV slot holding the number of buffers with the given name and the
750pv being an embedded array of I32.  The values may also be contained
751independently in the data array in cases where named backreferences are
752used.
753
754=head2 C<substrs>
755
756Holds information on the longest string that must occur at a fixed
757offset from the start of the pattern, and the longest string that must
758occur at a floating offset from the start of the pattern.  Used to do
759Fast-Boyer-Moore searches on the string to find out if its worth using
760the regex engine at all, and if so where in the string to search.
761
762=head2 C<subbeg> C<sublen> C<saved_copy> C<suboffset> C<subcoffset>
763
764Used during the execution phase for managing search and replace patterns,
765and for providing the text for C<$&>, C<$1> etc. C<subbeg> points to a
766buffer (either the original string, or a copy in the case of
767C<RX_MATCH_COPIED(rx)>), and C<sublen> is the length of the buffer.  The
768C<RX_OFFS> start and end indices index into this buffer.
769
770In the presence of the C<REXEC_COPY_STR> flag, but with the addition of
771the C<REXEC_COPY_SKIP_PRE> or C<REXEC_COPY_SKIP_POST> flags, an engine
772can choose not to copy the full buffer (although it must still do so in
773the presence of C<RXf_PMf_KEEPCOPY> or the relevant bits being set in
774C<PL_sawampersand>).  In this case, it may set C<suboffset> to indicate the
775number of bytes from the logical start of the buffer to the physical start
776(i.e. C<subbeg>).  It should also set C<subcoffset>, the number of
777characters in the offset. The latter is needed to support C<@-> and C<@+>
778which work in characters, not bytes.
779
780=head2 C<wrapped> C<wraplen>
781
782Stores the string C<qr//> stringifies to. The Perl engine for example
783stores C<(?^:eek)> in the case of C<qr/eek/>.
784
785When using a custom engine that doesn't support the C<(?:)> construct
786for inline modifiers, it's probably best to have C<qr//> stringify to
787the supplied pattern, note that this will create undesired patterns in
788cases such as:
789
790    my $x = qr/a|b/;  # "a|b"
791    my $y = qr/c/i;   # "c"
792    my $z = qr/$x$y/; # "a|bc"
793
794There's no solution for this problem other than making the custom
795engine understand a construct like C<(?:)>.
796
797=head2 C<seen_evals>
798
799This stores the number of eval groups in
800the pattern.  This is used for security
801purposes when embedding compiled regexes into larger patterns with C<qr//>.
802
803=head2 C<refcnt>
804
805The number of times the structure is referenced.  When
806this falls to 0, the regexp is automatically freed
807by a call to pregfree.  This should be set to 1 in
808each engine's L</comp> routine.
809
810=head1 HISTORY
811
812Originally part of L<perlreguts>.
813
814=head1 AUTHORS
815
816Originally written by Yves Orton, expanded by E<AElig>var ArnfjE<ouml>rE<eth>
817Bjarmason.
818
819=head1 LICENSE
820
821Copyright 2006 Yves Orton and 2007 E<AElig>var ArnfjE<ouml>rE<eth> Bjarmason.
822
823This program is free software; you can redistribute it and/or modify it under
824the same terms as Perl itself.
825
826=cut
827