This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Retract #10874 (the hack should be unnecessary by now)
[perl5.git] / regcomp.c
1 /*    regcomp.c
2  */
3
4 /*
5  * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
6  */
7
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9  * confused with the original package (see point 3 below).  Thanks, Henry!
10  */
11
12 /* Additional note: this code is very heavily munged from Henry's version
13  * in places.  In some spots I've traded clarity for efficiency, so don't
14  * blame Henry for some of the lack of readability.
15  */
16
17 /* The names of the functions have been changed from regcomp and
18  * regexec to  pregcomp and pregexec in order to avoid conflicts
19  * with the POSIX routines of the same names.
20 */
21
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 #  ifndef PERL_IN_XSUB_RE
25 #    define PERL_IN_XSUB_RE
26 #  endif
27 /* need access to debugger hooks */
28 #  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
29 #    define DEBUGGING
30 #  endif
31 #endif
32
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 #  define Perl_pregcomp my_regcomp
36 #  define Perl_regdump my_regdump
37 #  define Perl_regprop my_regprop
38 #  define Perl_pregfree my_regfree
39 #  define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 #  define Perl_regnext my_regnext
42 #  define Perl_save_re_context my_save_re_context
43 #  define Perl_reginitcolors my_reginitcolors
44
45 #  define PERL_NO_GET_CONTEXT
46 #endif
47
48 /*SUPPRESS 112*/
49 /*
50  * pregcomp and pregexec -- regsub and regerror are not used in perl
51  *
52  *      Copyright (c) 1986 by University of Toronto.
53  *      Written by Henry Spencer.  Not derived from licensed software.
54  *
55  *      Permission is granted to anyone to use this software for any
56  *      purpose on any computer system, and to redistribute it freely,
57  *      subject to the following restrictions:
58  *
59  *      1. The author is not responsible for the consequences of use of
60  *              this software, no matter how awful, even if they arise
61  *              from defects in it.
62  *
63  *      2. The origin of this software must not be misrepresented, either
64  *              by explicit claim or by omission.
65  *
66  *      3. Altered versions must be plainly marked as such, and must not
67  *              be misrepresented as being the original software.
68  *
69  *
70  ****    Alterations to Henry's code are...
71  ****
72  ****    Copyright (c) 1991-2001, Larry Wall
73  ****
74  ****    You may distribute under the terms of either the GNU General Public
75  ****    License or the Artistic License, as specified in the README file.
76
77  *
78  * Beware that some of this code is subtly aware of the way operator
79  * precedence is structured in regular expressions.  Serious changes in
80  * regular-expression syntax might require a total rethink.
81  */
82 #include "EXTERN.h"
83 #define PERL_IN_REGCOMP_C
84 #include "perl.h"
85
86 #ifdef PERL_IN_XSUB_RE
87 #  if defined(PERL_CAPI) || defined(PERL_OBJECT)
88 #    include "XSUB.h"
89 #  endif
90 #else
91 #  include "INTERN.h"
92 #endif
93
94 #define REG_COMP_C
95 #include "regcomp.h"
96
97 #ifdef op
98 #undef op
99 #endif /* op */
100
101 #ifdef MSDOS
102 # if defined(BUGGY_MSC6)
103  /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
104  # pragma optimize("a",off)
105  /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
106  # pragma optimize("w",on )
107 # endif /* BUGGY_MSC6 */
108 #endif /* MSDOS */
109
110 #ifndef STATIC
111 #define STATIC  static
112 #endif
113
114 typedef struct RExC_state_t {
115     U16         flags16;                /* are we folding, multilining? */
116     char        *precomp;               /* uncompiled string. */
117     regexp      *rx;
118     char        *start;                 /* Start of input for compile */
119     char        *end;                   /* End of input for compile */
120     char        *parse;                 /* Input-scan pointer. */
121     I32         whilem_seen;            /* number of WHILEM in this expr */
122     regnode     *emit_start;            /* Start of emitted-code area */
123     regnode     *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
124     I32         naughty;                /* How bad is this pattern? */
125     I32         sawback;                /* Did we see \1, ...? */
126     U32         seen;
127     I32         size;                   /* Code size. */
128     I32         npar;                   /* () count. */
129     I32         extralen;
130     I32         seen_zerolen;
131     I32         seen_evals;
132     I32         utf8;
133 #if ADD_TO_REGEXEC
134     char        *starttry;              /* -Dr: where regtry was called. */
135 #define RExC_starttry   (pRExC_state->starttry)
136 #endif
137 } RExC_state_t;
138
139 #define RExC_flags16    (pRExC_state->flags16)
140 #define RExC_precomp    (pRExC_state->precomp)
141 #define RExC_rx         (pRExC_state->rx)
142 #define RExC_start      (pRExC_state->start)
143 #define RExC_end        (pRExC_state->end)
144 #define RExC_parse      (pRExC_state->parse)
145 #define RExC_whilem_seen        (pRExC_state->whilem_seen)
146 #define RExC_offsets    (pRExC_state->rx->offsets) /* I am not like the others */
147 #define RExC_emit       (pRExC_state->emit)
148 #define RExC_emit_start (pRExC_state->emit_start)
149 #define RExC_naughty    (pRExC_state->naughty)
150 #define RExC_sawback    (pRExC_state->sawback)
151 #define RExC_seen       (pRExC_state->seen)
152 #define RExC_size       (pRExC_state->size)
153 #define RExC_npar       (pRExC_state->npar)
154 #define RExC_extralen   (pRExC_state->extralen)
155 #define RExC_seen_zerolen       (pRExC_state->seen_zerolen)
156 #define RExC_seen_evals (pRExC_state->seen_evals)
157 #define RExC_utf8       (pRExC_state->utf8)
158
159 #define ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
160 #define ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
161         ((*s) == '{' && regcurly(s)))
162
163 #ifdef SPSTART
164 #undef SPSTART          /* dratted cpp namespace... */
165 #endif
166 /*
167  * Flags to be passed up and down.
168  */
169 #define WORST           0       /* Worst case. */
170 #define HASWIDTH        0x1     /* Known to match non-null strings. */
171 #define SIMPLE          0x2     /* Simple enough to be STAR/PLUS operand. */
172 #define SPSTART         0x4     /* Starts with * or +. */
173 #define TRYAGAIN        0x8     /* Weeded out a declaration. */
174
175 /* Length of a variant. */
176
177 typedef struct scan_data_t {
178     I32 len_min;
179     I32 len_delta;
180     I32 pos_min;
181     I32 pos_delta;
182     SV *last_found;
183     I32 last_end;                       /* min value, <0 unless valid. */
184     I32 last_start_min;
185     I32 last_start_max;
186     SV **longest;                       /* Either &l_fixed, or &l_float. */
187     SV *longest_fixed;
188     I32 offset_fixed;
189     SV *longest_float;
190     I32 offset_float_min;
191     I32 offset_float_max;
192     I32 flags;
193     I32 whilem_c;
194     I32 *last_closep;
195     struct regnode_charclass_class *start_class;
196 } scan_data_t;
197
198 /*
199  * Forward declarations for pregcomp()'s friends.
200  */
201
202 static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
203                                       0, 0, 0, 0, 0, 0};
204
205 #define SF_BEFORE_EOL           (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
206 #define SF_BEFORE_SEOL          0x1
207 #define SF_BEFORE_MEOL          0x2
208 #define SF_FIX_BEFORE_EOL       (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
209 #define SF_FL_BEFORE_EOL        (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
210
211 #ifdef NO_UNARY_PLUS
212 #  define SF_FIX_SHIFT_EOL      (0+2)
213 #  define SF_FL_SHIFT_EOL               (0+4)
214 #else
215 #  define SF_FIX_SHIFT_EOL      (+2)
216 #  define SF_FL_SHIFT_EOL               (+4)
217 #endif
218
219 #define SF_FIX_BEFORE_SEOL      (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
220 #define SF_FIX_BEFORE_MEOL      (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
221
222 #define SF_FL_BEFORE_SEOL       (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
223 #define SF_FL_BEFORE_MEOL       (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
224 #define SF_IS_INF               0x40
225 #define SF_HAS_PAR              0x80
226 #define SF_IN_PAR               0x100
227 #define SF_HAS_EVAL             0x200
228 #define SCF_DO_SUBSTR           0x400
229 #define SCF_DO_STCLASS_AND      0x0800
230 #define SCF_DO_STCLASS_OR       0x1000
231 #define SCF_DO_STCLASS          (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
232 #define SCF_WHILEM_VISITED_POS  0x2000
233
234 #define UTF RExC_utf8
235 #define LOC (RExC_flags16 & PMf_LOCALE)
236 #define FOLD (RExC_flags16 & PMf_FOLD)
237
238 #define OOB_UNICODE             12345678
239 #define OOB_NAMEDCLASS          -1
240
241 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
242 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
243
244
245 /* length of regex to show in messages that don't mark a position within */
246 #define RegexLengthToShowInErrorMessages 127
247
248 /*
249  * If MARKER[12] are adjusted, be sure to adjust the constants at the top
250  * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
251  * op/pragma/warn/regcomp.
252  */
253 #define MARKER1 "<-- HERE"    /* marker as it appears in the description */
254 #define MARKER2 " <-- HERE "  /* marker as it appears within the regex */
255
256 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
257
258 /*
259  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
260  * arg. Show regex, up to a maximum length. If it's too long, chop and add
261  * "...".
262  */
263 #define FAIL(msg)                                                             \
264     STMT_START {                                                             \
265         char *ellipses = "";                                                 \
266         unsigned len = strlen(RExC_precomp);                                \
267                                                                              \
268         if (!SIZE_ONLY)                                                      \
269             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
270                                                                              \
271         if (len > RegexLengthToShowInErrorMessages) {                        \
272             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
273             len = RegexLengthToShowInErrorMessages - 10;                     \
274             ellipses = "...";                                                \
275         }                                                                    \
276         Perl_croak(aTHX_ "%s in regex m/%.*s%s/",                            \
277                    msg, (int)len, RExC_precomp, ellipses);                  \
278     } STMT_END
279
280 /*
281  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
282  * args. Show regex, up to a maximum length. If it's too long, chop and add
283  * "...".
284  */
285 #define FAIL2(pat,msg)                                                        \
286     STMT_START {                                                             \
287         char *ellipses = "";                                                 \
288         unsigned len = strlen(RExC_precomp);                                \
289                                                                              \
290         if (!SIZE_ONLY)                                                      \
291             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
292                                                                              \
293         if (len > RegexLengthToShowInErrorMessages) {                        \
294             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
295             len = RegexLengthToShowInErrorMessages - 10;                     \
296             ellipses = "...";                                                \
297         }                                                                    \
298         S_re_croak2(aTHX_ pat, " in regex m/%.*s%s/",                        \
299                     msg, (int)len, RExC_precomp, ellipses);                \
300     } STMT_END
301
302
303 /*
304  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
305  */
306 #define Simple_vFAIL(m)                                                      \
307     STMT_START {                                                             \
308       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
309                                                                              \
310       Perl_croak(aTHX_ "%s" REPORT_LOCATION,               \
311                  m, (int)offset, RExC_precomp, RExC_precomp + offset);     \
312     } STMT_END
313
314 /*
315  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
316  */
317 #define vFAIL(m)                                                             \
318     STMT_START {                                                             \
319       if (!SIZE_ONLY)                                                        \
320             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
321       Simple_vFAIL(m);                                                       \
322     } STMT_END
323
324 /*
325  * Like Simple_vFAIL(), but accepts two arguments.
326  */
327 #define Simple_vFAIL2(m,a1)                                                  \
328     STMT_START {                                                             \
329       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
330                                                                              \
331       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1,       \
332                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
333     } STMT_END
334
335 /*
336  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
337  */
338 #define vFAIL2(m,a1)                                                         \
339     STMT_START {                                                             \
340       if (!SIZE_ONLY)                                                        \
341             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
342       Simple_vFAIL2(m, a1);                                                  \
343     } STMT_END
344
345
346 /*
347  * Like Simple_vFAIL(), but accepts three arguments.
348  */
349 #define Simple_vFAIL3(m, a1, a2)                                             \
350     STMT_START {                                                             \
351       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
352                                                                              \
353       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2,   \
354                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
355     } STMT_END
356
357 /*
358  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
359  */
360 #define vFAIL3(m,a1,a2)                                                      \
361     STMT_START {                                                             \
362       if (!SIZE_ONLY)                                                        \
363             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
364       Simple_vFAIL3(m, a1, a2);                                              \
365     } STMT_END
366
367 /*
368  * Like Simple_vFAIL(), but accepts four arguments.
369  */
370 #define Simple_vFAIL4(m, a1, a2, a3)                                         \
371     STMT_START {                                                             \
372       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
373                                                                              \
374       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3,\
375                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
376     } STMT_END
377
378 /*
379  * Like Simple_vFAIL(), but accepts five arguments.
380  */
381 #define Simple_vFAIL5(m, a1, a2, a3, a4)                                     \
382     STMT_START {                                                             \
383       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
384       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, a4,\
385                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
386     } STMT_END
387
388
389 #define vWARN(loc,m)                                                         \
390     STMT_START {                                                             \
391         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
392         Perl_warner(aTHX_ WARN_REGEXP, "%s" REPORT_LOCATION,\
393                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
394     } STMT_END                                                               \
395
396 #define vWARNdep(loc,m)                                                         \
397     STMT_START {                                                             \
398         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
399         int warn_cat = ckWARN(WARN_REGEXP) ? WARN_REGEXP : WARN_DEPRECATED;  \
400         Perl_warner(aTHX_ warn_cat, "%s" REPORT_LOCATION,\
401                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
402     } STMT_END                                                               \
403
404
405 #define vWARN2(loc, m, a1)                                                   \
406     STMT_START {                                                             \
407         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
408         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
409                  a1,                                                         \
410                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
411     } STMT_END
412
413 #define vWARN3(loc, m, a1, a2)                                               \
414     STMT_START {                                                             \
415       unsigned offset = strlen(RExC_precomp) - (RExC_end - (loc));        \
416         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,                    \
417                  a1, a2,                                                     \
418                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
419     } STMT_END
420
421 #define vWARN4(loc, m, a1, a2, a3)                                           \
422     STMT_START {                                                             \
423       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));            \
424         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
425                  a1, a2, a3,                                                 \
426                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
427     } STMT_END
428
429 /* used for the parse_flags section for (?c) -- japhy */
430 #define vWARN5(loc, m, a1, a2, a3, a4)                                       \
431   STMT_START {                                                   \
432       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));   \
433         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,      \
434                  a1, a2, a3, a4,                                 \
435                  (int)offset, RExC_precomp, RExC_precomp + offset);  \
436     } STMT_END
437
438
439 /* Allow for side effects in s */
440 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (void)(s);} STMT_END
441
442 /* Macros for recording node offsets.   20001227 mjd@plover.com 
443  * Nodes are numbered 1, 2, 3, 4.  Node #n's position is recorded in
444  * element 2*n-1 of the array.  Element #2n holds the byte length node #n.
445  * Element 0 holds the number n.
446  */
447
448 #define MJD_OFFSET_DEBUG(x)
449 /* #define MJD_OFFSET_DEBUG(x) fprintf x */
450
451
452 #  define Set_Node_Offset_To_R(node,byte)                           \
453    STMT_START {                                        \
454      if (! SIZE_ONLY) {                                  \
455        if((node) < 0) {                   \
456          Perl_croak(aTHX_ "value of node is %d in Offset macro", node); \
457        } else {                                                        \
458          RExC_offsets[2*(node)-1] = (byte);                               \
459        }                                                               \
460      }                                                                 \
461    } STMT_END
462
463 #  define Set_Node_Offset(node,byte) Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
464 #  define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
465
466 #  define Set_Node_Length_To_R(node,len)                            \
467    STMT_START {                                        \
468      if (! SIZE_ONLY) {                                  \
469        MJD_OFFSET_DEBUG((stderr, "** (%d) size of node %d is %d.\n", __LINE__, (node), (len))); \
470        if((node) < 0) {                   \
471          Perl_croak(aTHX_ "value of node is %d in Length macro", node); \
472        } else {                                                        \
473          RExC_offsets[2*(node)] = (len);                               \
474        }                                                               \
475      }                                                                 \
476    } STMT_END
477
478 #  define Set_Node_Length(node,len) Set_Node_Length_To_R((node)-RExC_emit_start, len)
479 #  define Set_Cur_Node_Length(len)  Set_Node_Length(RExC_emit, len)
480 #  define Set_Node_Cur_Length(node)   Set_Node_Length(node, RExC_parse - parse_start)
481
482 /* Get offsets and lengths */
483 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
484 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
485
486 static void clear_re(pTHXo_ void *r);
487
488 /* Mark that we cannot extend a found fixed substring at this point.
489    Updata the longest found anchored substring and the longest found
490    floating substrings if needed. */
491
492 STATIC void
493 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
494 {
495     STRLEN l = CHR_SVLEN(data->last_found);
496     STRLEN old_l = CHR_SVLEN(*data->longest);
497
498     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
499         sv_setsv(*data->longest, data->last_found);
500         if (*data->longest == data->longest_fixed) {
501             data->offset_fixed = l ? data->last_start_min : data->pos_min;
502             if (data->flags & SF_BEFORE_EOL)
503                 data->flags
504                     |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
505             else
506                 data->flags &= ~SF_FIX_BEFORE_EOL;
507         }
508         else {
509             data->offset_float_min = l ? data->last_start_min : data->pos_min;
510             data->offset_float_max = (l
511                                       ? data->last_start_max
512                                       : data->pos_min + data->pos_delta);
513             if (data->flags & SF_BEFORE_EOL)
514                 data->flags
515                     |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
516             else
517                 data->flags &= ~SF_FL_BEFORE_EOL;
518         }
519     }
520     SvCUR_set(data->last_found, 0);
521     data->last_end = -1;
522     data->flags &= ~SF_BEFORE_EOL;
523 }
524
525 /* Can match anything (initialization) */
526 STATIC void
527 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
528 {
529     int value;
530
531     ANYOF_CLASS_ZERO(cl);
532     for (value = 0; value < 256; ++value)
533         ANYOF_BITMAP_SET(cl, value);
534     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
535     if (LOC)
536         cl->flags |= ANYOF_LOCALE;
537 }
538
539 /* Can match anything (initialization) */
540 STATIC int
541 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
542 {
543     int value;
544
545     for (value = 0; value <= ANYOF_MAX; value += 2)
546         if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
547             return 1;
548     if (!(cl->flags & ANYOF_UNICODE_ALL))
549         return 0;
550     for (value = 0; value < 256; ++value)
551         if (!ANYOF_BITMAP_TEST(cl, value))
552             return 0;
553     return 1;
554 }
555
556 /* Can match anything (initialization) */
557 STATIC void
558 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
559 {
560     Zero(cl, 1, struct regnode_charclass_class);
561     cl->type = ANYOF;
562     cl_anything(pRExC_state, cl);
563 }
564
565 STATIC void
566 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
567 {
568     Zero(cl, 1, struct regnode_charclass_class);
569     cl->type = ANYOF;
570     cl_anything(pRExC_state, cl);
571     if (LOC)
572         cl->flags |= ANYOF_LOCALE;
573 }
574
575 /* 'And' a given class with another one.  Can create false positives */
576 /* We assume that cl is not inverted */
577 STATIC void
578 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
579          struct regnode_charclass_class *and_with)
580 {
581     if (!(and_with->flags & ANYOF_CLASS)
582         && !(cl->flags & ANYOF_CLASS)
583         && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
584         && !(and_with->flags & ANYOF_FOLD)
585         && !(cl->flags & ANYOF_FOLD)) {
586         int i;
587
588         if (and_with->flags & ANYOF_INVERT)
589             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
590                 cl->bitmap[i] &= ~and_with->bitmap[i];
591         else
592             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
593                 cl->bitmap[i] &= and_with->bitmap[i];
594     } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
595     if (!(and_with->flags & ANYOF_EOS))
596         cl->flags &= ~ANYOF_EOS;
597
598     if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
599         cl->flags &= ~ANYOF_UNICODE_ALL;
600         cl->flags |= ANYOF_UNICODE;
601         ARG_SET(cl, ARG(and_with));
602     }
603     if (!(and_with->flags & ANYOF_UNICODE_ALL))
604         cl->flags &= ~ANYOF_UNICODE_ALL;
605     if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
606         cl->flags &= ~ANYOF_UNICODE;
607 }
608
609 /* 'OR' a given class with another one.  Can create false positives */
610 /* We assume that cl is not inverted */
611 STATIC void
612 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
613 {
614     if (or_with->flags & ANYOF_INVERT) {
615         /* We do not use
616          * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
617          *   <= (B1 | !B2) | (CL1 | !CL2)
618          * which is wasteful if CL2 is small, but we ignore CL2:
619          *   (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
620          * XXXX Can we handle case-fold?  Unclear:
621          *   (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
622          *   (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
623          */
624         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
625              && !(or_with->flags & ANYOF_FOLD)
626              && !(cl->flags & ANYOF_FOLD) ) {
627             int i;
628
629             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
630                 cl->bitmap[i] |= ~or_with->bitmap[i];
631         } /* XXXX: logic is complicated otherwise */
632         else {
633             cl_anything(pRExC_state, cl);
634         }
635     } else {
636         /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
637         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
638              && (!(or_with->flags & ANYOF_FOLD)
639                  || (cl->flags & ANYOF_FOLD)) ) {
640             int i;
641
642             /* OR char bitmap and class bitmap separately */
643             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
644                 cl->bitmap[i] |= or_with->bitmap[i];
645             if (or_with->flags & ANYOF_CLASS) {
646                 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
647                     cl->classflags[i] |= or_with->classflags[i];
648                 cl->flags |= ANYOF_CLASS;
649             }
650         }
651         else { /* XXXX: logic is complicated, leave it along for a moment. */
652             cl_anything(pRExC_state, cl);
653         }
654     }
655     if (or_with->flags & ANYOF_EOS)
656         cl->flags |= ANYOF_EOS;
657
658     if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
659         ARG(cl) != ARG(or_with)) {
660         cl->flags |= ANYOF_UNICODE_ALL;
661         cl->flags &= ~ANYOF_UNICODE;
662     }
663     if (or_with->flags & ANYOF_UNICODE_ALL) {
664         cl->flags |= ANYOF_UNICODE_ALL;
665         cl->flags &= ~ANYOF_UNICODE;
666     }
667 }
668
669 /* REx optimizer.  Converts nodes into quickier variants "in place".
670    Finds fixed substrings.  */
671
672 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
673    to the position after last scanned or to NULL. */
674
675 STATIC I32
676 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
677                         /* scanp: Start here (read-write). */
678                         /* deltap: Write maxlen-minlen here. */
679                         /* last: Stop before this one. */
680 {
681     I32 min = 0, pars = 0, code;
682     regnode *scan = *scanp, *next;
683     I32 delta = 0;
684     int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
685     int is_inf_internal = 0;            /* The studied chunk is infinite */
686     I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
687     scan_data_t data_fake;
688     struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
689
690     while (scan && OP(scan) != END && scan < last) {
691         /* Peephole optimizer: */
692
693         if (PL_regkind[(U8)OP(scan)] == EXACT) {
694             /* Merge several consecutive EXACTish nodes into one. */
695             regnode *n = regnext(scan);
696             U32 stringok = 1;
697 #ifdef DEBUGGING
698             regnode *stop = scan;
699 #endif
700
701             next = scan + NODE_SZ_STR(scan);
702             /* Skip NOTHING, merge EXACT*. */
703             while (n &&
704                    ( PL_regkind[(U8)OP(n)] == NOTHING ||
705                      (stringok && (OP(n) == OP(scan))))
706                    && NEXT_OFF(n)
707                    && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
708                 if (OP(n) == TAIL || n > next)
709                     stringok = 0;
710                 if (PL_regkind[(U8)OP(n)] == NOTHING) {
711                     NEXT_OFF(scan) += NEXT_OFF(n);
712                     next = n + NODE_STEP_REGNODE;
713 #ifdef DEBUGGING
714                     if (stringok)
715                         stop = n;
716 #endif
717                     n = regnext(n);
718                 }
719                 else if (stringok) {
720                     int oldl = STR_LEN(scan);
721                     regnode *nnext = regnext(n);
722
723                     if (oldl + STR_LEN(n) > U8_MAX)
724                         break;
725                     NEXT_OFF(scan) += NEXT_OFF(n);
726                     STR_LEN(scan) += STR_LEN(n);
727                     next = n + NODE_SZ_STR(n);
728                     /* Now we can overwrite *n : */
729                     Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
730 #ifdef DEBUGGING
731                     stop = next - 1;
732 #endif
733                     n = nnext;
734                 }
735             }
736 #ifdef DEBUGGING
737             /* Allow dumping */
738             n = scan + NODE_SZ_STR(scan);
739             while (n <= stop) {
740                 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
741                     OP(n) = OPTIMIZED;
742                     NEXT_OFF(n) = 0;
743                 }
744                 n++;
745             }
746 #endif
747         }
748         /* Follow the next-chain of the current node and optimize
749            away all the NOTHINGs from it.  */
750         if (OP(scan) != CURLYX) {
751             int max = (reg_off_by_arg[OP(scan)]
752                        ? I32_MAX
753                        /* I32 may be smaller than U16 on CRAYs! */
754                        : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
755             int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
756             int noff;
757             regnode *n = scan;
758         
759             /* Skip NOTHING and LONGJMP. */
760             while ((n = regnext(n))
761                    && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
762                        || ((OP(n) == LONGJMP) && (noff = ARG(n))))
763                    && off + noff < max)
764                 off += noff;
765             if (reg_off_by_arg[OP(scan)])
766                 ARG(scan) = off;
767             else
768                 NEXT_OFF(scan) = off;
769         }
770         /* The principal pseudo-switch.  Cannot be a switch, since we
771            look into several different things.  */
772         if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
773                    || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
774             next = regnext(scan);
775             code = OP(scan);
776         
777             if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
778                 I32 max1 = 0, min1 = I32_MAX, num = 0;
779                 struct regnode_charclass_class accum;
780                 
781                 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
782                     scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
783                 if (flags & SCF_DO_STCLASS)
784                     cl_init_zero(pRExC_state, &accum);
785                 while (OP(scan) == code) {
786                     I32 deltanext, minnext, f = 0, fake;
787                     struct regnode_charclass_class this_class;
788
789                     num++;
790                     data_fake.flags = 0;
791                     if (data) {         
792                         data_fake.whilem_c = data->whilem_c;
793                         data_fake.last_closep = data->last_closep;
794                     }
795                     else
796                         data_fake.last_closep = &fake;
797                     next = regnext(scan);
798                     scan = NEXTOPER(scan);
799                     if (code != BRANCH)
800                         scan = NEXTOPER(scan);
801                     if (flags & SCF_DO_STCLASS) {
802                         cl_init(pRExC_state, &this_class);
803                         data_fake.start_class = &this_class;
804                         f = SCF_DO_STCLASS_AND;
805                     }           
806                     if (flags & SCF_WHILEM_VISITED_POS)
807                         f |= SCF_WHILEM_VISITED_POS;
808                     /* we suppose the run is continuous, last=next...*/
809                     minnext = study_chunk(pRExC_state, &scan, &deltanext,
810                                           next, &data_fake, f);
811                     if (min1 > minnext)
812                         min1 = minnext;
813                     if (max1 < minnext + deltanext)
814                         max1 = minnext + deltanext;
815                     if (deltanext == I32_MAX)
816                         is_inf = is_inf_internal = 1;
817                     scan = next;
818                     if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
819                         pars++;
820                     if (data && (data_fake.flags & SF_HAS_EVAL))
821                         data->flags |= SF_HAS_EVAL;
822                     if (data)
823                         data->whilem_c = data_fake.whilem_c;
824                     if (flags & SCF_DO_STCLASS)
825                         cl_or(pRExC_state, &accum, &this_class);
826                     if (code == SUSPEND)
827                         break;
828                 }
829                 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
830                     min1 = 0;
831                 if (flags & SCF_DO_SUBSTR) {
832                     data->pos_min += min1;
833                     data->pos_delta += max1 - min1;
834                     if (max1 != min1 || is_inf)
835                         data->longest = &(data->longest_float);
836                 }
837                 min += min1;
838                 delta += max1 - min1;
839                 if (flags & SCF_DO_STCLASS_OR) {
840                     cl_or(pRExC_state, data->start_class, &accum);
841                     if (min1) {
842                         cl_and(data->start_class, &and_with);
843                         flags &= ~SCF_DO_STCLASS;
844                     }
845                 }
846                 else if (flags & SCF_DO_STCLASS_AND) {
847                     if (min1) {
848                         cl_and(data->start_class, &accum);
849                         flags &= ~SCF_DO_STCLASS;
850                     }
851                     else {
852                         /* Switch to OR mode: cache the old value of
853                          * data->start_class */
854                         StructCopy(data->start_class, &and_with,
855                                    struct regnode_charclass_class);
856                         flags &= ~SCF_DO_STCLASS_AND;
857                         StructCopy(&accum, data->start_class,
858                                    struct regnode_charclass_class);
859                         flags |= SCF_DO_STCLASS_OR;
860                         data->start_class->flags |= ANYOF_EOS;
861                     }
862                 }
863             }
864             else if (code == BRANCHJ)   /* single branch is optimized. */
865                 scan = NEXTOPER(NEXTOPER(scan));
866             else                        /* single branch is optimized. */
867                 scan = NEXTOPER(scan);
868             continue;
869         }
870         else if (OP(scan) == EXACT) {
871             I32 l = STR_LEN(scan);
872             UV uc = *((U8*)STRING(scan));
873             if (UTF) {
874                 U8 *s = (U8*)STRING(scan);
875                 l = utf8_length(s, s + l);
876                 uc = utf8_to_uvchr(s, NULL);
877             }
878             min += l;
879             if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
880                 /* The code below prefers earlier match for fixed
881                    offset, later match for variable offset.  */
882                 if (data->last_end == -1) { /* Update the start info. */
883                     data->last_start_min = data->pos_min;
884                     data->last_start_max = is_inf
885                         ? I32_MAX : data->pos_min + data->pos_delta;
886                 }
887                 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
888                 data->last_end = data->pos_min + l;
889                 data->pos_min += l; /* As in the first entry. */
890                 data->flags &= ~SF_BEFORE_EOL;
891             }
892             if (flags & SCF_DO_STCLASS_AND) {
893                 /* Check whether it is compatible with what we know already! */
894                 int compat = 1;
895
896                 if (uc >= 0x100 ||
897                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
898                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
899                     && (!(data->start_class->flags & ANYOF_FOLD)
900                         || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
901                     )
902                     compat = 0;
903                 ANYOF_CLASS_ZERO(data->start_class);
904                 ANYOF_BITMAP_ZERO(data->start_class);
905                 if (compat)
906                     ANYOF_BITMAP_SET(data->start_class, uc);
907                 data->start_class->flags &= ~ANYOF_EOS;
908                 if (uc < 0x100)
909                   data->start_class->flags &= ~ANYOF_UNICODE_ALL;
910             }
911             else if (flags & SCF_DO_STCLASS_OR) {
912                 /* false positive possible if the class is case-folded */
913                 if (uc < 0x100)
914                     ANYOF_BITMAP_SET(data->start_class, uc);
915                 else
916                     data->start_class->flags |= ANYOF_UNICODE_ALL;
917                 data->start_class->flags &= ~ANYOF_EOS;
918                 cl_and(data->start_class, &and_with);
919             }
920             flags &= ~SCF_DO_STCLASS;
921         }
922         else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
923             I32 l = STR_LEN(scan);
924             UV uc = *((U8*)STRING(scan));
925
926             /* Search for fixed substrings supports EXACT only. */
927             if (flags & SCF_DO_SUBSTR)
928                 scan_commit(pRExC_state, data);
929             if (UTF) {
930                 U8 *s = (U8 *)STRING(scan);
931                 l = utf8_length(s, s + l);
932                 uc = utf8_to_uvchr(s, NULL);
933             }
934             min += l;
935             if (data && (flags & SCF_DO_SUBSTR))
936                 data->pos_min += l;
937             if (flags & SCF_DO_STCLASS_AND) {
938                 /* Check whether it is compatible with what we know already! */
939                 int compat = 1;
940
941                 if (uc >= 0x100 ||
942                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
943                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
944                      && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
945                     compat = 0;
946                 ANYOF_CLASS_ZERO(data->start_class);
947                 ANYOF_BITMAP_ZERO(data->start_class);
948                 if (compat) {
949                     ANYOF_BITMAP_SET(data->start_class, uc);
950                     data->start_class->flags &= ~ANYOF_EOS;
951                     data->start_class->flags |= ANYOF_FOLD;
952                     if (OP(scan) == EXACTFL)
953                         data->start_class->flags |= ANYOF_LOCALE;
954                 }
955             }
956             else if (flags & SCF_DO_STCLASS_OR) {
957                 if (data->start_class->flags & ANYOF_FOLD) {
958                     /* false positive possible if the class is case-folded.
959                        Assume that the locale settings are the same... */
960                     if (uc < 0x100)
961                         ANYOF_BITMAP_SET(data->start_class, uc);
962                     data->start_class->flags &= ~ANYOF_EOS;
963                 }
964                 cl_and(data->start_class, &and_with);
965             }
966             flags &= ~SCF_DO_STCLASS;
967         }
968         else if (strchr((char*)PL_varies,OP(scan))) {
969             I32 mincount, maxcount, minnext, deltanext, fl = 0;
970             I32 f = flags, pos_before = 0;
971             regnode *oscan = scan;
972             struct regnode_charclass_class this_class;
973             struct regnode_charclass_class *oclass = NULL;
974
975             switch (PL_regkind[(U8)OP(scan)]) {
976             case WHILEM:                /* End of (?:...)* . */
977                 scan = NEXTOPER(scan);
978                 goto finish;
979             case PLUS:
980                 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
981                     next = NEXTOPER(scan);
982                     if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
983                         mincount = 1;
984                         maxcount = REG_INFTY;
985                         next = regnext(scan);
986                         scan = NEXTOPER(scan);
987                         goto do_curly;
988                     }
989                 }
990                 if (flags & SCF_DO_SUBSTR)
991                     data->pos_min++;
992                 min++;
993                 /* Fall through. */
994             case STAR:
995                 if (flags & SCF_DO_STCLASS) {
996                     mincount = 0;
997                     maxcount = REG_INFTY;
998                     next = regnext(scan);
999                     scan = NEXTOPER(scan);
1000                     goto do_curly;
1001                 }
1002                 is_inf = is_inf_internal = 1;
1003                 scan = regnext(scan);
1004                 if (flags & SCF_DO_SUBSTR) {
1005                     scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
1006                     data->longest = &(data->longest_float);
1007                 }
1008                 goto optimize_curly_tail;
1009             case CURLY:
1010                 mincount = ARG1(scan);
1011                 maxcount = ARG2(scan);
1012                 next = regnext(scan);
1013                 if (OP(scan) == CURLYX) {
1014                     I32 lp = (data ? *(data->last_closep) : 0);
1015
1016                     scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
1017                 }
1018                 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
1019               do_curly:
1020                 if (flags & SCF_DO_SUBSTR) {
1021                     if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
1022                     pos_before = data->pos_min;
1023                 }
1024                 if (data) {
1025                     fl = data->flags;
1026                     data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
1027                     if (is_inf)
1028                         data->flags |= SF_IS_INF;
1029                 }
1030                 if (flags & SCF_DO_STCLASS) {
1031                     cl_init(pRExC_state, &this_class);
1032                     oclass = data->start_class;
1033                     data->start_class = &this_class;
1034                     f |= SCF_DO_STCLASS_AND;
1035                     f &= ~SCF_DO_STCLASS_OR;
1036                 }
1037                 /* These are the cases when once a subexpression
1038                    fails at a particular position, it cannot succeed
1039                    even after backtracking at the enclosing scope.
1040                 
1041                    XXXX what if minimal match and we are at the
1042                         initial run of {n,m}? */
1043                 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
1044                     f &= ~SCF_WHILEM_VISITED_POS;
1045
1046                 /* This will finish on WHILEM, setting scan, or on NULL: */
1047                 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data,
1048                                       mincount == 0
1049                                         ? (f & ~SCF_DO_SUBSTR) : f);
1050
1051                 if (flags & SCF_DO_STCLASS)
1052                     data->start_class = oclass;
1053                 if (mincount == 0 || minnext == 0) {
1054                     if (flags & SCF_DO_STCLASS_OR) {
1055                         cl_or(pRExC_state, data->start_class, &this_class);
1056                     }
1057                     else if (flags & SCF_DO_STCLASS_AND) {
1058                         /* Switch to OR mode: cache the old value of
1059                          * data->start_class */
1060                         StructCopy(data->start_class, &and_with,
1061                                    struct regnode_charclass_class);
1062                         flags &= ~SCF_DO_STCLASS_AND;
1063                         StructCopy(&this_class, data->start_class,
1064                                    struct regnode_charclass_class);
1065                         flags |= SCF_DO_STCLASS_OR;
1066                         data->start_class->flags |= ANYOF_EOS;
1067                     }
1068                 } else {                /* Non-zero len */
1069                     if (flags & SCF_DO_STCLASS_OR) {
1070                         cl_or(pRExC_state, data->start_class, &this_class);
1071                         cl_and(data->start_class, &and_with);
1072                     }
1073                     else if (flags & SCF_DO_STCLASS_AND)
1074                         cl_and(data->start_class, &this_class);
1075                     flags &= ~SCF_DO_STCLASS;
1076                 }
1077                 if (!scan)              /* It was not CURLYX, but CURLY. */
1078                     scan = next;
1079                 if (ckWARN(WARN_REGEXP)
1080                     && (minnext == 0) && (deltanext == 0)
1081                     && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1082                     && maxcount <= REG_INFTY/3) /* Complement check for big count */
1083                 {
1084                     vWARN(RExC_parse,
1085                           "Quantifier unexpected on zero-length expression");
1086                 }
1087
1088                 min += minnext * mincount;
1089                 is_inf_internal |= ((maxcount == REG_INFTY
1090                                      && (minnext + deltanext) > 0)
1091                                     || deltanext == I32_MAX);
1092                 is_inf |= is_inf_internal;
1093                 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1094
1095                 /* Try powerful optimization CURLYX => CURLYN. */
1096                 if (  OP(oscan) == CURLYX && data
1097                       && data->flags & SF_IN_PAR
1098                       && !(data->flags & SF_HAS_EVAL)
1099                       && !deltanext && minnext == 1 ) {
1100                     /* Try to optimize to CURLYN.  */
1101                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1102                     regnode *nxt1 = nxt;
1103 #ifdef DEBUGGING
1104                     regnode *nxt2;
1105 #endif
1106
1107                     /* Skip open. */
1108                     nxt = regnext(nxt);
1109                     if (!strchr((char*)PL_simple,OP(nxt))
1110                         && !(PL_regkind[(U8)OP(nxt)] == EXACT
1111                              && STR_LEN(nxt) == 1))
1112                         goto nogo;
1113 #ifdef DEBUGGING
1114                     nxt2 = nxt;
1115 #endif
1116                     nxt = regnext(nxt);
1117                     if (OP(nxt) != CLOSE)
1118                         goto nogo;
1119                     /* Now we know that nxt2 is the only contents: */
1120                     oscan->flags = ARG(nxt);
1121                     OP(oscan) = CURLYN;
1122                     OP(nxt1) = NOTHING; /* was OPEN. */
1123 #ifdef DEBUGGING
1124                     OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1125                     NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1126                     NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1127                     OP(nxt) = OPTIMIZED;        /* was CLOSE. */
1128                     OP(nxt + 1) = OPTIMIZED; /* was count. */
1129                     NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1130 #endif
1131                 }
1132               nogo:
1133
1134                 /* Try optimization CURLYX => CURLYM. */
1135                 if (  OP(oscan) == CURLYX && data
1136                       && !(data->flags & SF_HAS_PAR)
1137                       && !(data->flags & SF_HAS_EVAL)
1138                       && !deltanext  ) {
1139                     /* XXXX How to optimize if data == 0? */
1140                     /* Optimize to a simpler form.  */
1141                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1142                     regnode *nxt2;
1143
1144                     OP(oscan) = CURLYM;
1145                     while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1146                             && (OP(nxt2) != WHILEM))
1147                         nxt = nxt2;
1148                     OP(nxt2)  = SUCCEED; /* Whas WHILEM */
1149                     /* Need to optimize away parenths. */
1150                     if (data->flags & SF_IN_PAR) {
1151                         /* Set the parenth number.  */
1152                         regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1153
1154                         if (OP(nxt) != CLOSE)
1155                             FAIL("Panic opt close");
1156                         oscan->flags = ARG(nxt);
1157                         OP(nxt1) = OPTIMIZED;   /* was OPEN. */
1158                         OP(nxt) = OPTIMIZED;    /* was CLOSE. */
1159 #ifdef DEBUGGING
1160                         OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1161                         OP(nxt + 1) = OPTIMIZED; /* was count. */
1162                         NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1163                         NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1164 #endif
1165 #if 0
1166                         while ( nxt1 && (OP(nxt1) != WHILEM)) {
1167                             regnode *nnxt = regnext(nxt1);
1168                         
1169                             if (nnxt == nxt) {
1170                                 if (reg_off_by_arg[OP(nxt1)])
1171                                     ARG_SET(nxt1, nxt2 - nxt1);
1172                                 else if (nxt2 - nxt1 < U16_MAX)
1173                                     NEXT_OFF(nxt1) = nxt2 - nxt1;
1174                                 else
1175                                     OP(nxt) = NOTHING;  /* Cannot beautify */
1176                             }
1177                             nxt1 = nnxt;
1178                         }
1179 #endif
1180                         /* Optimize again: */
1181                         study_chunk(pRExC_state, &nxt1, &deltanext, nxt,
1182                                     NULL, 0);
1183                     }
1184                     else
1185                         oscan->flags = 0;
1186                 }
1187                 else if ((OP(oscan) == CURLYX)
1188                          && (flags & SCF_WHILEM_VISITED_POS)
1189                          /* See the comment on a similar expression above.
1190                             However, this time it not a subexpression
1191                             we care about, but the expression itself. */
1192                          && (maxcount == REG_INFTY)
1193                          && data && ++data->whilem_c < 16) {
1194                     /* This stays as CURLYX, we can put the count/of pair. */
1195                     /* Find WHILEM (as in regexec.c) */
1196                     regnode *nxt = oscan + NEXT_OFF(oscan);
1197
1198                     if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1199                         nxt += ARG(nxt);
1200                     PREVOPER(nxt)->flags = data->whilem_c
1201                         | (RExC_whilem_seen << 4); /* On WHILEM */
1202                 }
1203                 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
1204                     pars++;
1205                 if (flags & SCF_DO_SUBSTR) {
1206                     SV *last_str = Nullsv;
1207                     int counted = mincount != 0;
1208
1209                     if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1210                         I32 b = pos_before >= data->last_start_min
1211                             ? pos_before : data->last_start_min;
1212                         STRLEN l;
1213                         char *s = SvPV(data->last_found, l);
1214                         I32 old = b - data->last_start_min;
1215
1216                         if (UTF)
1217                             old = utf8_hop((U8*)s, old) - (U8*)s;
1218                         
1219                         l -= old;
1220                         /* Get the added string: */
1221                         last_str = newSVpvn(s  + old, l);
1222                         if (deltanext == 0 && pos_before == b) {
1223                             /* What was added is a constant string */
1224                             if (mincount > 1) {
1225                                 SvGROW(last_str, (mincount * l) + 1);
1226                                 repeatcpy(SvPVX(last_str) + l,
1227                                           SvPVX(last_str), l, mincount - 1);
1228                                 SvCUR(last_str) *= mincount;
1229                                 /* Add additional parts. */
1230                                 SvCUR_set(data->last_found,
1231                                           SvCUR(data->last_found) - l);
1232                                 sv_catsv(data->last_found, last_str);
1233                                 data->last_end += l * (mincount - 1);
1234                             }
1235                         } else {
1236                             /* start offset must point into the last copy */
1237                             data->last_start_min += minnext * (mincount - 1);
1238                             data->last_start_max += is_inf ? 0 : (maxcount - 1)
1239                                 * (minnext + data->pos_delta);
1240                         }
1241                     }
1242                     /* It is counted once already... */
1243                     data->pos_min += minnext * (mincount - counted);
1244                     data->pos_delta += - counted * deltanext +
1245                         (minnext + deltanext) * maxcount - minnext * mincount;
1246                     if (mincount != maxcount) {
1247                          /* Cannot extend fixed substrings found inside
1248                             the group.  */
1249                         scan_commit(pRExC_state,data);
1250                         if (mincount && last_str) {
1251                             sv_setsv(data->last_found, last_str);
1252                             data->last_end = data->pos_min;
1253                             data->last_start_min =
1254                                 data->pos_min - CHR_SVLEN(last_str);
1255                             data->last_start_max = is_inf
1256                                 ? I32_MAX
1257                                 : data->pos_min + data->pos_delta
1258                                 - CHR_SVLEN(last_str);
1259                         }
1260                         data->longest = &(data->longest_float);
1261                     }
1262                     SvREFCNT_dec(last_str);
1263                 }
1264                 if (data && (fl & SF_HAS_EVAL))
1265                     data->flags |= SF_HAS_EVAL;
1266               optimize_curly_tail:
1267                 if (OP(oscan) != CURLYX) {
1268                     while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1269                            && NEXT_OFF(next))
1270                         NEXT_OFF(oscan) += NEXT_OFF(next);
1271                 }
1272                 continue;
1273             default:                    /* REF and CLUMP only? */
1274                 if (flags & SCF_DO_SUBSTR) {
1275                     scan_commit(pRExC_state,data);      /* Cannot expect anything... */
1276                     data->longest = &(data->longest_float);
1277                 }
1278                 is_inf = is_inf_internal = 1;
1279                 if (flags & SCF_DO_STCLASS_OR)
1280                     cl_anything(pRExC_state, data->start_class);
1281                 flags &= ~SCF_DO_STCLASS;
1282                 break;
1283             }
1284         }
1285         else if (strchr((char*)PL_simple,OP(scan))) {
1286             int value = 0;
1287
1288             if (flags & SCF_DO_SUBSTR) {
1289                 scan_commit(pRExC_state,data);
1290                 data->pos_min++;
1291             }
1292             min++;
1293             if (flags & SCF_DO_STCLASS) {
1294                 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1295
1296                 /* Some of the logic below assumes that switching
1297                    locale on will only add false positives. */
1298                 switch (PL_regkind[(U8)OP(scan)]) {
1299                 case SANY:
1300                 default:
1301                   do_default:
1302                     /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1303                     if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1304                         cl_anything(pRExC_state, data->start_class);
1305                     break;
1306                 case REG_ANY:
1307                     if (OP(scan) == SANY)
1308                         goto do_default;
1309                     if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1310                         value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1311                                  || (data->start_class->flags & ANYOF_CLASS));
1312                         cl_anything(pRExC_state, data->start_class);
1313                     }
1314                     if (flags & SCF_DO_STCLASS_AND || !value)
1315                         ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1316                     break;
1317                 case ANYOF:
1318                     if (flags & SCF_DO_STCLASS_AND)
1319                         cl_and(data->start_class,
1320                                (struct regnode_charclass_class*)scan);
1321                     else
1322                         cl_or(pRExC_state, data->start_class,
1323                               (struct regnode_charclass_class*)scan);
1324                     break;
1325                 case ALNUM:
1326                     if (flags & SCF_DO_STCLASS_AND) {
1327                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1328                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1329                             for (value = 0; value < 256; value++)
1330                                 if (!isALNUM(value))
1331                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1332                         }
1333                     }
1334                     else {
1335                         if (data->start_class->flags & ANYOF_LOCALE)
1336                             ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1337                         else {
1338                             for (value = 0; value < 256; value++)
1339                                 if (isALNUM(value))
1340                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1341                         }
1342                     }
1343                     break;
1344                 case ALNUML:
1345                     if (flags & SCF_DO_STCLASS_AND) {
1346                         if (data->start_class->flags & ANYOF_LOCALE)
1347                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1348                     }
1349                     else {
1350                         ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1351                         data->start_class->flags |= ANYOF_LOCALE;
1352                     }
1353                     break;
1354                 case NALNUM:
1355                     if (flags & SCF_DO_STCLASS_AND) {
1356                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1357                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1358                             for (value = 0; value < 256; value++)
1359                                 if (isALNUM(value))
1360                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1361                         }
1362                     }
1363                     else {
1364                         if (data->start_class->flags & ANYOF_LOCALE)
1365                             ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1366                         else {
1367                             for (value = 0; value < 256; value++)
1368                                 if (!isALNUM(value))
1369                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1370                         }
1371                     }
1372                     break;
1373                 case NALNUML:
1374                     if (flags & SCF_DO_STCLASS_AND) {
1375                         if (data->start_class->flags & ANYOF_LOCALE)
1376                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1377                     }
1378                     else {
1379                         data->start_class->flags |= ANYOF_LOCALE;
1380                         ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1381                     }
1382                     break;
1383                 case SPACE:
1384                     if (flags & SCF_DO_STCLASS_AND) {
1385                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1386                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1387                             for (value = 0; value < 256; value++)
1388                                 if (!isSPACE(value))
1389                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1390                         }
1391                     }
1392                     else {
1393                         if (data->start_class->flags & ANYOF_LOCALE)
1394                             ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1395                         else {
1396                             for (value = 0; value < 256; value++)
1397                                 if (isSPACE(value))
1398                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1399                         }
1400                     }
1401                     break;
1402                 case SPACEL:
1403                     if (flags & SCF_DO_STCLASS_AND) {
1404                         if (data->start_class->flags & ANYOF_LOCALE)
1405                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1406                     }
1407                     else {
1408                         data->start_class->flags |= ANYOF_LOCALE;
1409                         ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1410                     }
1411                     break;
1412                 case NSPACE:
1413                     if (flags & SCF_DO_STCLASS_AND) {
1414                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1415                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1416                             for (value = 0; value < 256; value++)
1417                                 if (isSPACE(value))
1418                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1419                         }
1420                     }
1421                     else {
1422                         if (data->start_class->flags & ANYOF_LOCALE)
1423                             ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1424                         else {
1425                             for (value = 0; value < 256; value++)
1426                                 if (!isSPACE(value))
1427                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1428                         }
1429                     }
1430                     break;
1431                 case NSPACEL:
1432                     if (flags & SCF_DO_STCLASS_AND) {
1433                         if (data->start_class->flags & ANYOF_LOCALE) {
1434                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1435                             for (value = 0; value < 256; value++)
1436                                 if (!isSPACE(value))
1437                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1438                         }
1439                     }
1440                     else {
1441                         data->start_class->flags |= ANYOF_LOCALE;
1442                         ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1443                     }
1444                     break;
1445                 case DIGIT:
1446                     if (flags & SCF_DO_STCLASS_AND) {
1447                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1448                         for (value = 0; value < 256; value++)
1449                             if (!isDIGIT(value))
1450                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1451                     }
1452                     else {
1453                         if (data->start_class->flags & ANYOF_LOCALE)
1454                             ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1455                         else {
1456                             for (value = 0; value < 256; value++)
1457                                 if (isDIGIT(value))
1458                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1459                         }
1460                     }
1461                     break;
1462                 case NDIGIT:
1463                     if (flags & SCF_DO_STCLASS_AND) {
1464                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1465                         for (value = 0; value < 256; value++)
1466                             if (isDIGIT(value))
1467                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1468                     }
1469                     else {
1470                         if (data->start_class->flags & ANYOF_LOCALE)
1471                             ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1472                         else {
1473                             for (value = 0; value < 256; value++)
1474                                 if (!isDIGIT(value))
1475                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1476                         }
1477                     }
1478                     break;
1479                 }
1480                 if (flags & SCF_DO_STCLASS_OR)
1481                     cl_and(data->start_class, &and_with);
1482                 flags &= ~SCF_DO_STCLASS;
1483             }
1484         }
1485         else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1486             data->flags |= (OP(scan) == MEOL
1487                             ? SF_BEFORE_MEOL
1488                             : SF_BEFORE_SEOL);
1489         }
1490         else if (  PL_regkind[(U8)OP(scan)] == BRANCHJ
1491                  /* Lookbehind, or need to calculate parens/evals/stclass: */
1492                    && (scan->flags || data || (flags & SCF_DO_STCLASS))
1493                    && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1494             /* Lookahead/lookbehind */
1495             I32 deltanext, minnext, fake = 0;
1496             regnode *nscan;
1497             struct regnode_charclass_class intrnl;
1498             int f = 0;
1499
1500             data_fake.flags = 0;
1501             if (data) {         
1502                 data_fake.whilem_c = data->whilem_c;
1503                 data_fake.last_closep = data->last_closep;
1504             }
1505             else
1506                 data_fake.last_closep = &fake;
1507             if ( flags & SCF_DO_STCLASS && !scan->flags
1508                  && OP(scan) == IFMATCH ) { /* Lookahead */
1509                 cl_init(pRExC_state, &intrnl);
1510                 data_fake.start_class = &intrnl;
1511                 f |= SCF_DO_STCLASS_AND;
1512             }
1513             if (flags & SCF_WHILEM_VISITED_POS)
1514                 f |= SCF_WHILEM_VISITED_POS;
1515             next = regnext(scan);
1516             nscan = NEXTOPER(NEXTOPER(scan));
1517             minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1518             if (scan->flags) {
1519                 if (deltanext) {
1520                     vFAIL("Variable length lookbehind not implemented");
1521                 }
1522                 else if (minnext > U8_MAX) {
1523                     vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1524                 }
1525                 scan->flags = minnext;
1526             }
1527             if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1528                 pars++;
1529             if (data && (data_fake.flags & SF_HAS_EVAL))
1530                 data->flags |= SF_HAS_EVAL;
1531             if (data)
1532                 data->whilem_c = data_fake.whilem_c;
1533             if (f & SCF_DO_STCLASS_AND) {
1534                 int was = (data->start_class->flags & ANYOF_EOS);
1535
1536                 cl_and(data->start_class, &intrnl);
1537                 if (was)
1538                     data->start_class->flags |= ANYOF_EOS;
1539             }
1540         }
1541         else if (OP(scan) == OPEN) {
1542             pars++;
1543         }
1544         else if (OP(scan) == CLOSE) {
1545             if (ARG(scan) == is_par) {
1546                 next = regnext(scan);
1547
1548                 if ( next && (OP(next) != WHILEM) && next < last)
1549                     is_par = 0;         /* Disable optimization */
1550             }
1551             if (data)
1552                 *(data->last_closep) = ARG(scan);
1553         }
1554         else if (OP(scan) == EVAL) {
1555                 if (data)
1556                     data->flags |= SF_HAS_EVAL;
1557         }
1558         else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1559                 if (flags & SCF_DO_SUBSTR) {
1560                     scan_commit(pRExC_state,data);
1561                     data->longest = &(data->longest_float);
1562                 }
1563                 is_inf = is_inf_internal = 1;
1564                 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1565                     cl_anything(pRExC_state, data->start_class);
1566                 flags &= ~SCF_DO_STCLASS;
1567         }
1568         /* Else: zero-length, ignore. */
1569         scan = regnext(scan);
1570     }
1571
1572   finish:
1573     *scanp = scan;
1574     *deltap = is_inf_internal ? I32_MAX : delta;
1575     if (flags & SCF_DO_SUBSTR && is_inf)
1576         data->pos_delta = I32_MAX - data->pos_min;
1577     if (is_par > U8_MAX)
1578         is_par = 0;
1579     if (is_par && pars==1 && data) {
1580         data->flags |= SF_IN_PAR;
1581         data->flags &= ~SF_HAS_PAR;
1582     }
1583     else if (pars && data) {
1584         data->flags |= SF_HAS_PAR;
1585         data->flags &= ~SF_IN_PAR;
1586     }
1587     if (flags & SCF_DO_STCLASS_OR)
1588         cl_and(data->start_class, &and_with);
1589     return min;
1590 }
1591
1592 STATIC I32
1593 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1594 {
1595     if (RExC_rx->data) {
1596         Renewc(RExC_rx->data,
1597                sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1),
1598                char, struct reg_data);
1599         Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1600         RExC_rx->data->count += n;
1601     }
1602     else {
1603         Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1604              char, struct reg_data);
1605         New(1208, RExC_rx->data->what, n, U8);
1606         RExC_rx->data->count = n;
1607     }
1608     Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1609     return RExC_rx->data->count - n;
1610 }
1611
1612 void
1613 Perl_reginitcolors(pTHX)
1614 {
1615     int i = 0;
1616     char *s = PerlEnv_getenv("PERL_RE_COLORS");
1617         
1618     if (s) {
1619         PL_colors[0] = s = savepv(s);
1620         while (++i < 6) {
1621             s = strchr(s, '\t');
1622             if (s) {
1623                 *s = '\0';
1624                 PL_colors[i] = ++s;
1625             }
1626             else
1627                 PL_colors[i] = s = "";
1628         }
1629     } else {
1630         while (i < 6)
1631             PL_colors[i++] = "";
1632     }
1633     PL_colorset = 1;
1634 }
1635
1636
1637 /*
1638  - pregcomp - compile a regular expression into internal code
1639  *
1640  * We can't allocate space until we know how big the compiled form will be,
1641  * but we can't compile it (and thus know how big it is) until we've got a
1642  * place to put the code.  So we cheat:  we compile it twice, once with code
1643  * generation turned off and size counting turned on, and once "for real".
1644  * This also means that we don't allocate space until we are sure that the
1645  * thing really will compile successfully, and we never have to move the
1646  * code and thus invalidate pointers into it.  (Note that it has to be in
1647  * one piece because free() must be able to free it all.) [NB: not true in perl]
1648  *
1649  * Beware that the optimization-preparation code in here knows about some
1650  * of the structure of the compiled regexp.  [I'll say.]
1651  */
1652 regexp *
1653 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1654 {
1655     register regexp *r;
1656     regnode *scan;
1657     regnode *first;
1658     I32 flags;
1659     I32 minlen = 0;
1660     I32 sawplus = 0;
1661     I32 sawopen = 0;
1662     scan_data_t data;
1663     RExC_state_t RExC_state;
1664     RExC_state_t *pRExC_state = &RExC_state;
1665
1666     if (exp == NULL)
1667         FAIL("NULL regexp argument");
1668
1669     /* XXXX This looks very suspicious... */
1670     if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1671         RExC_utf8 = 1;
1672     else
1673         RExC_utf8 = 0;
1674
1675     RExC_precomp = exp;
1676     DEBUG_r(if (!PL_colorset) reginitcolors());
1677     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1678                       PL_colors[4],PL_colors[5],PL_colors[0],
1679                       (int)(xend - exp), RExC_precomp, PL_colors[1]));
1680     RExC_flags16 = pm->op_pmflags;
1681     RExC_sawback = 0;
1682
1683     RExC_seen = 0;
1684     RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1685     RExC_seen_evals = 0;
1686     RExC_extralen = 0;
1687
1688     /* First pass: determine size, legality. */
1689     RExC_parse = exp;
1690     RExC_start = exp;
1691     RExC_end = xend;
1692     RExC_naughty = 0;
1693     RExC_npar = 1;
1694     RExC_size = 0L;
1695     RExC_emit = &PL_regdummy;
1696     RExC_whilem_seen = 0;
1697 #if 0 /* REGC() is (currently) a NOP at the first pass.
1698        * Clever compilers notice this and complain. --jhi */
1699     REGC((U8)REG_MAGIC, (char*)RExC_emit);
1700 #endif
1701     if (reg(pRExC_state, 0, &flags) == NULL) {
1702         RExC_precomp = Nullch;
1703         return(NULL);
1704     }
1705     DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1706
1707     /* Small enough for pointer-storage convention?
1708        If extralen==0, this means that we will not need long jumps. */
1709     if (RExC_size >= 0x10000L && RExC_extralen)
1710         RExC_size += RExC_extralen;
1711     else
1712         RExC_extralen = 0;
1713     if (RExC_whilem_seen > 15)
1714         RExC_whilem_seen = 15;
1715
1716     /* Allocate space and initialize. */
1717     Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1718          char, regexp);
1719     if (r == NULL)
1720         FAIL("Regexp out of space");
1721
1722 #ifdef DEBUGGING
1723     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1724     Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1725 #endif
1726     r->refcnt = 1;
1727     r->prelen = xend - exp;
1728     r->precomp = savepvn(RExC_precomp, r->prelen);
1729     r->subbeg = NULL;
1730     r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1731     r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1732
1733     r->substrs = 0;                     /* Useful during FAIL. */
1734     r->startp = 0;                      /* Useful during FAIL. */
1735     r->endp = 0;                        /* Useful during FAIL. */
1736
1737     Newz(1304, r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
1738     if (r->offsets) {
1739       r->offsets[0] = RExC_size; 
1740     }
1741     DEBUG_r(PerlIO_printf(Perl_debug_log, 
1742                           "%s %"UVuf" bytes for offset annotations.\n", 
1743                           r->offsets ? "Got" : "Couldn't get", 
1744                           (UV)((2*RExC_size+1) * sizeof(U32))));
1745
1746     RExC_rx = r;
1747
1748     /* Second pass: emit code. */
1749     RExC_parse = exp;
1750     RExC_end = xend;
1751     RExC_naughty = 0;
1752     RExC_npar = 1;
1753     RExC_emit_start = r->program;
1754     RExC_emit = r->program;
1755     /* Store the count of eval-groups for security checks: */
1756     RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1757     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1758     r->data = 0;
1759     if (reg(pRExC_state, 0, &flags) == NULL)
1760         return(NULL);
1761
1762     /* Dig out information for optimizations. */
1763     r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1764     pm->op_pmflags = RExC_flags16;
1765     if (UTF)
1766         r->reganch |= ROPT_UTF8;
1767     r->regstclass = NULL;
1768     if (RExC_naughty >= 10)     /* Probably an expensive pattern. */
1769         r->reganch |= ROPT_NAUGHTY;
1770     scan = r->program + 1;              /* First BRANCH. */
1771
1772     /* XXXX To minimize changes to RE engine we always allocate
1773        3-units-long substrs field. */
1774     Newz(1004, r->substrs, 1, struct reg_substr_data);
1775
1776     StructCopy(&zero_scan_data, &data, scan_data_t);
1777     /* XXXX Should not we check for something else?  Usually it is OPEN1... */
1778     if (OP(scan) != BRANCH) {   /* Only one top-level choice. */
1779         I32 fake;
1780         STRLEN longest_float_length, longest_fixed_length;
1781         struct regnode_charclass_class ch_class;
1782         int stclass_flag;
1783         I32 last_close = 0;
1784
1785         first = scan;
1786         /* Skip introductions and multiplicators >= 1. */
1787         while ((OP(first) == OPEN && (sawopen = 1)) ||
1788                /* An OR of *one* alternative - should not happen now. */
1789             (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1790             (OP(first) == PLUS) ||
1791             (OP(first) == MINMOD) ||
1792                /* An {n,m} with n>0 */
1793             (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1794                 if (OP(first) == PLUS)
1795                     sawplus = 1;
1796                 else
1797                     first += regarglen[(U8)OP(first)];
1798                 first = NEXTOPER(first);
1799         }
1800
1801         /* Starting-point info. */
1802       again:
1803         if (PL_regkind[(U8)OP(first)] == EXACT) {
1804             if (OP(first) == EXACT)
1805                 ;       /* Empty, get anchored substr later. */
1806             else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1807                 r->regstclass = first;
1808         }
1809         else if (strchr((char*)PL_simple,OP(first)))
1810             r->regstclass = first;
1811         else if (PL_regkind[(U8)OP(first)] == BOUND ||
1812                  PL_regkind[(U8)OP(first)] == NBOUND)
1813             r->regstclass = first;
1814         else if (PL_regkind[(U8)OP(first)] == BOL) {
1815             r->reganch |= (OP(first) == MBOL
1816                            ? ROPT_ANCH_MBOL
1817                            : (OP(first) == SBOL
1818                               ? ROPT_ANCH_SBOL
1819                               : ROPT_ANCH_BOL));
1820             first = NEXTOPER(first);
1821             goto again;
1822         }
1823         else if (OP(first) == GPOS) {
1824             r->reganch |= ROPT_ANCH_GPOS;
1825             first = NEXTOPER(first);
1826             goto again;
1827         }
1828         else if ((OP(first) == STAR &&
1829             PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1830             !(r->reganch & ROPT_ANCH) )
1831         {
1832             /* turn .* into ^.* with an implied $*=1 */
1833             int type = OP(NEXTOPER(first));
1834
1835             if (type == REG_ANY)
1836                 type = ROPT_ANCH_MBOL;
1837             else
1838                 type = ROPT_ANCH_SBOL;
1839
1840             r->reganch |= type | ROPT_IMPLICIT;
1841             first = NEXTOPER(first);
1842             goto again;
1843         }
1844         if (sawplus && (!sawopen || !RExC_sawback)
1845             && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1846             /* x+ must match at the 1st pos of run of x's */
1847             r->reganch |= ROPT_SKIP;
1848
1849         /* Scan is after the zeroth branch, first is atomic matcher. */
1850         DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
1851                               (IV)(first - scan + 1)));
1852         /*
1853         * If there's something expensive in the r.e., find the
1854         * longest literal string that must appear and make it the
1855         * regmust.  Resolve ties in favor of later strings, since
1856         * the regstart check works with the beginning of the r.e.
1857         * and avoiding duplication strengthens checking.  Not a
1858         * strong reason, but sufficient in the absence of others.
1859         * [Now we resolve ties in favor of the earlier string if
1860         * it happens that c_offset_min has been invalidated, since the
1861         * earlier string may buy us something the later one won't.]
1862         */
1863         minlen = 0;
1864
1865         data.longest_fixed = newSVpvn("",0);
1866         data.longest_float = newSVpvn("",0);
1867         data.last_found = newSVpvn("",0);
1868         data.longest = &(data.longest_fixed);
1869         first = scan;
1870         if (!r->regstclass) {
1871             cl_init(pRExC_state, &ch_class);
1872             data.start_class = &ch_class;
1873             stclass_flag = SCF_DO_STCLASS_AND;
1874         } else                          /* XXXX Check for BOUND? */
1875             stclass_flag = 0;
1876         data.last_closep = &last_close;
1877
1878         minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1879                              &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1880         if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1881              && data.last_start_min == 0 && data.last_end > 0
1882              && !RExC_seen_zerolen
1883              && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1884             r->reganch |= ROPT_CHECK_ALL;
1885         scan_commit(pRExC_state, &data);
1886         SvREFCNT_dec(data.last_found);
1887
1888         longest_float_length = CHR_SVLEN(data.longest_float);
1889         if (longest_float_length
1890             || (data.flags & SF_FL_BEFORE_EOL
1891                 && (!(data.flags & SF_FL_BEFORE_MEOL)
1892                     || (RExC_flags16 & PMf_MULTILINE)))) {
1893             int t;
1894
1895             if (SvCUR(data.longest_fixed)                       /* ok to leave SvCUR */
1896                 && data.offset_fixed == data.offset_float_min
1897                 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1898                     goto remove_float;          /* As in (a)+. */
1899
1900             r->float_substr = data.longest_float;
1901             r->float_min_offset = data.offset_float_min;
1902             r->float_max_offset = data.offset_float_max;
1903             t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1904                        && (!(data.flags & SF_FL_BEFORE_MEOL)
1905                            || (RExC_flags16 & PMf_MULTILINE)));
1906             fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1907         }
1908         else {
1909           remove_float:
1910             r->float_substr = Nullsv;
1911             SvREFCNT_dec(data.longest_float);
1912             longest_float_length = 0;
1913         }
1914
1915         longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1916         if (longest_fixed_length
1917             || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1918                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1919                     || (RExC_flags16 & PMf_MULTILINE)))) {
1920             int t;
1921
1922             r->anchored_substr = data.longest_fixed;
1923             r->anchored_offset = data.offset_fixed;
1924             t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1925                  && (!(data.flags & SF_FIX_BEFORE_MEOL)
1926                      || (RExC_flags16 & PMf_MULTILINE)));
1927             fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1928         }
1929         else {
1930             r->anchored_substr = Nullsv;
1931             SvREFCNT_dec(data.longest_fixed);
1932             longest_fixed_length = 0;
1933         }
1934         if (r->regstclass
1935             && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1936             r->regstclass = NULL;
1937         if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1938             && !(data.start_class->flags & ANYOF_EOS)
1939             && !cl_is_anything(data.start_class)) {
1940             I32 n = add_data(pRExC_state, 1, "f");
1941
1942             New(1006, RExC_rx->data->data[n], 1,
1943                 struct regnode_charclass_class);
1944             StructCopy(data.start_class,
1945                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1946                        struct regnode_charclass_class);
1947             r->regstclass = (regnode*)RExC_rx->data->data[n];
1948             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1949             PL_regdata = r->data; /* for regprop() */
1950             DEBUG_r({ SV *sv = sv_newmortal();
1951                       regprop(sv, (regnode*)data.start_class);
1952                       PerlIO_printf(Perl_debug_log,
1953                                     "synthetic stclass `%s'.\n",
1954                                     SvPVX(sv));});
1955         }
1956
1957         /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1958         if (longest_fixed_length > longest_float_length) {
1959             r->check_substr = r->anchored_substr;
1960             r->check_offset_min = r->check_offset_max = r->anchored_offset;
1961             if (r->reganch & ROPT_ANCH_SINGLE)
1962                 r->reganch |= ROPT_NOSCAN;
1963         }
1964         else {
1965             r->check_substr = r->float_substr;
1966             r->check_offset_min = data.offset_float_min;
1967             r->check_offset_max = data.offset_float_max;
1968         }
1969         /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1970            This should be changed ASAP!  */
1971         if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1972             r->reganch |= RE_USE_INTUIT;
1973             if (SvTAIL(r->check_substr))
1974                 r->reganch |= RE_INTUIT_TAIL;
1975         }
1976     }
1977     else {
1978         /* Several toplevels. Best we can is to set minlen. */
1979         I32 fake;
1980         struct regnode_charclass_class ch_class;
1981         I32 last_close = 0;
1982         
1983         DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1984         scan = r->program + 1;
1985         cl_init(pRExC_state, &ch_class);
1986         data.start_class = &ch_class;
1987         data.last_closep = &last_close;
1988         minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1989         r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1990         if (!(data.start_class->flags & ANYOF_EOS)
1991             && !cl_is_anything(data.start_class)) {
1992             I32 n = add_data(pRExC_state, 1, "f");
1993
1994             New(1006, RExC_rx->data->data[n], 1,
1995                 struct regnode_charclass_class);
1996             StructCopy(data.start_class,
1997                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1998                        struct regnode_charclass_class);
1999             r->regstclass = (regnode*)RExC_rx->data->data[n];
2000             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
2001             DEBUG_r({ SV* sv = sv_newmortal();
2002                       regprop(sv, (regnode*)data.start_class);
2003                       PerlIO_printf(Perl_debug_log,
2004                                     "synthetic stclass `%s'.\n",
2005                                     SvPVX(sv));});
2006         }
2007     }
2008
2009     r->minlen = minlen;
2010     if (RExC_seen & REG_SEEN_GPOS)
2011         r->reganch |= ROPT_GPOS_SEEN;
2012     if (RExC_seen & REG_SEEN_LOOKBEHIND)
2013         r->reganch |= ROPT_LOOKBEHIND_SEEN;
2014     if (RExC_seen & REG_SEEN_EVAL)
2015         r->reganch |= ROPT_EVAL_SEEN;
2016     if (RExC_seen & REG_SEEN_CANY)
2017         r->reganch |= ROPT_CANY_SEEN;
2018     Newz(1002, r->startp, RExC_npar, I32);
2019     Newz(1002, r->endp, RExC_npar, I32);
2020     PL_regdata = r->data; /* for regprop() */
2021     DEBUG_r(regdump(r));
2022     return(r);
2023 }
2024
2025 /*
2026  - reg - regular expression, i.e. main body or parenthesized thing
2027  *
2028  * Caller must absorb opening parenthesis.
2029  *
2030  * Combining parenthesis handling with the base level of regular expression
2031  * is a trifle forced, but the need to tie the tails of the branches to what
2032  * follows makes it hard to avoid.
2033  */
2034 STATIC regnode *
2035 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2036     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2037 {
2038     register regnode *ret;              /* Will be the head of the group. */
2039     register regnode *br;
2040     register regnode *lastbr;
2041     register regnode *ender = 0;
2042     register I32 parno = 0;
2043     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2044
2045     /* for (?g), (?gc), and (?o) warnings; warning
2046        about (?c) will warn about (?g) -- japhy    */
2047
2048     I32 wastedflags = 0x00,
2049         wasted_o    = 0x01,
2050         wasted_g    = 0x02,
2051         wasted_gc   = 0x02 | 0x04,
2052         wasted_c    = 0x04;
2053
2054     char * parse_start = RExC_parse; /* MJD */
2055     char *oregcomp_parse = RExC_parse;
2056     char c;
2057
2058     *flagp = 0;                         /* Tentatively. */
2059
2060
2061     /* Make an OPEN node, if parenthesized. */
2062     if (paren) {
2063         if (*RExC_parse == '?') { /* (?...) */
2064             U16 posflags = 0, negflags = 0;
2065             U16 *flagsp = &posflags;
2066             int logical = 0;
2067             char *seqstart = RExC_parse;
2068
2069             RExC_parse++;
2070             paren = *RExC_parse++;
2071             ret = NULL;                 /* For look-ahead/behind. */
2072             switch (paren) {
2073             case '<':           /* (?<...) */
2074                 RExC_seen |= REG_SEEN_LOOKBEHIND;
2075                 if (*RExC_parse == '!')
2076                     paren = ',';
2077                 if (*RExC_parse != '=' && *RExC_parse != '!')
2078                     goto unknown;
2079                 RExC_parse++;
2080             case '=':           /* (?=...) */
2081             case '!':           /* (?!...) */
2082                 RExC_seen_zerolen++;
2083             case ':':           /* (?:...) */
2084             case '>':           /* (?>...) */
2085                 break;
2086             case '$':           /* (?$...) */
2087             case '@':           /* (?@...) */
2088                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2089                 break;
2090             case '#':           /* (?#...) */
2091                 while (*RExC_parse && *RExC_parse != ')')
2092                     RExC_parse++;
2093                 if (*RExC_parse != ')')
2094                     FAIL("Sequence (?#... not terminated");
2095                 nextchar(pRExC_state);
2096                 *flagp = TRYAGAIN;
2097                 return NULL;
2098             case 'p':           /* (?p...) */
2099                 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
2100                     vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
2101                 /* FALL THROUGH*/
2102             case '?':           /* (??...) */
2103                 logical = 1;
2104                 paren = *RExC_parse++;
2105                 /* FALL THROUGH */
2106             case '{':           /* (?{...}) */
2107             {
2108                 I32 count = 1, n = 0;
2109                 char c;
2110                 char *s = RExC_parse;
2111                 SV *sv;
2112                 OP_4tree *sop, *rop;
2113
2114                 RExC_seen_zerolen++;
2115                 RExC_seen |= REG_SEEN_EVAL;
2116                 while (count && (c = *RExC_parse)) {
2117                     if (c == '\\' && RExC_parse[1])
2118                         RExC_parse++;
2119                     else if (c == '{')
2120                         count++;
2121                     else if (c == '}')
2122                         count--;
2123                     RExC_parse++;
2124                 }
2125                 if (*RExC_parse != ')')
2126                 {
2127                     RExC_parse = s;             
2128                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2129                 }
2130                 if (!SIZE_ONLY) {
2131                     AV *av;
2132                 
2133                     if (RExC_parse - 1 - s)
2134                         sv = newSVpvn(s, RExC_parse - 1 - s);
2135                     else
2136                         sv = newSVpvn("", 0);
2137
2138                     ENTER;
2139                     Perl_save_re_context(aTHX);
2140                     rop = sv_compile_2op(sv, &sop, "re", &av);
2141                     sop->op_private |= OPpREFCOUNTED;
2142                     /* re_dup will OpREFCNT_inc */
2143                     OpREFCNT_set(sop, 1);
2144                     LEAVE;
2145
2146                     n = add_data(pRExC_state, 3, "nop");
2147                     RExC_rx->data->data[n] = (void*)rop;
2148                     RExC_rx->data->data[n+1] = (void*)sop;
2149                     RExC_rx->data->data[n+2] = (void*)av;
2150                     SvREFCNT_dec(sv);
2151                 }
2152                 else {                                          /* First pass */
2153                     if (PL_reginterp_cnt < ++RExC_seen_evals
2154                         && PL_curcop != &PL_compiling)
2155                         /* No compiled RE interpolated, has runtime
2156                            components ===> unsafe.  */
2157                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2158                     if (PL_tainting && PL_tainted)
2159                         FAIL("Eval-group in insecure regular expression");
2160                 }
2161                 
2162                 nextchar(pRExC_state);
2163                 if (logical) {
2164                     ret = reg_node(pRExC_state, LOGICAL);
2165                     if (!SIZE_ONLY)
2166                         ret->flags = 2;
2167                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2168                     /* deal with the length of this later - MJD */
2169                     return ret;
2170                 }
2171                 return reganode(pRExC_state, EVAL, n);
2172             }
2173             case '(':           /* (?(?{...})...) and (?(?=...)...) */
2174             {
2175                 if (RExC_parse[0] == '?') {        /* (?(?...)) */
2176                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2177                         || RExC_parse[1] == '<'
2178                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2179                         I32 flag;
2180                         
2181                         ret = reg_node(pRExC_state, LOGICAL);
2182                         if (!SIZE_ONLY)
2183                             ret->flags = 1;
2184                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2185                         goto insert_if;
2186                     }
2187                 }
2188                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2189                     /* (?(1)...) */
2190                     parno = atoi(RExC_parse++);
2191
2192                     while (isDIGIT(*RExC_parse))
2193                         RExC_parse++;
2194                     ret = reganode(pRExC_state, GROUPP, parno);
2195                     
2196                     if ((c = *nextchar(pRExC_state)) != ')')
2197                         vFAIL("Switch condition not recognized");
2198                   insert_if:
2199                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2200                     br = regbranch(pRExC_state, &flags, 1);
2201                     if (br == NULL)
2202                         br = reganode(pRExC_state, LONGJMP, 0);
2203                     else
2204                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2205                     c = *nextchar(pRExC_state);
2206                     if (flags&HASWIDTH)
2207                         *flagp |= HASWIDTH;
2208                     if (c == '|') {
2209                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2210                         regbranch(pRExC_state, &flags, 1);
2211                         regtail(pRExC_state, ret, lastbr);
2212                         if (flags&HASWIDTH)
2213                             *flagp |= HASWIDTH;
2214                         c = *nextchar(pRExC_state);
2215                     }
2216                     else
2217                         lastbr = NULL;
2218                     if (c != ')')
2219                         vFAIL("Switch (?(condition)... contains too many branches");
2220                     ender = reg_node(pRExC_state, TAIL);
2221                     regtail(pRExC_state, br, ender);
2222                     if (lastbr) {
2223                         regtail(pRExC_state, lastbr, ender);
2224                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2225                     }
2226                     else
2227                         regtail(pRExC_state, ret, ender);
2228                     return ret;
2229                 }
2230                 else {
2231                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2232                 }
2233             }
2234             case 0:
2235                 RExC_parse--; /* for vFAIL to print correctly */
2236                 vFAIL("Sequence (? incomplete");
2237                 break;
2238             default:
2239                 --RExC_parse;
2240               parse_flags:      /* (?i) */
2241                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2242                     /* (?g), (?gc) and (?o) are useless here
2243                        and must be globally applied -- japhy */
2244
2245                     if (*RExC_parse == 'o' || *RExC_parse == 'g') {
2246                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2247                             I32 wflagbit = *RExC_parse == 'o' ? wasted_o : wasted_g;
2248                             if (! (wastedflags & wflagbit) ) {
2249                                 wastedflags |= wflagbit;
2250                                 vWARN5(
2251                                     RExC_parse + 1,
2252                                     "Useless (%s%c) - %suse /%c modifier",
2253                                     flagsp == &negflags ? "?-" : "?",
2254                                     *RExC_parse,
2255                                     flagsp == &negflags ? "don't " : "",
2256                                     *RExC_parse
2257                                 );
2258                             }
2259                         }
2260                     }
2261                     else if (*RExC_parse == 'c') {
2262                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2263                             if (! (wastedflags & wasted_c) ) {
2264                                 wastedflags |= wasted_gc;
2265                                 vWARN3(
2266                                     RExC_parse + 1,
2267                                     "Useless (%sc) - %suse /gc modifier",
2268                                     flagsp == &negflags ? "?-" : "?",
2269                                     flagsp == &negflags ? "don't " : ""
2270                                 );
2271                             }
2272                         }
2273                     }
2274                     else { pmflag(flagsp, *RExC_parse); }
2275
2276                     ++RExC_parse;
2277                 }
2278                 if (*RExC_parse == '-') {
2279                     flagsp = &negflags;
2280                     wastedflags = 0;  /* reset so (?g-c) warns twice */
2281                     ++RExC_parse;
2282                     goto parse_flags;
2283                 }
2284                 RExC_flags16 |= posflags;
2285                 RExC_flags16 &= ~negflags;
2286                 if (*RExC_parse == ':') {
2287                     RExC_parse++;
2288                     paren = ':';
2289                     break;
2290                 }               
2291               unknown:
2292                 if (*RExC_parse != ')') {
2293                     RExC_parse++;
2294                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2295                 }
2296                 nextchar(pRExC_state);
2297                 *flagp = TRYAGAIN;
2298                 return NULL;
2299             }
2300         }
2301         else {                  /* (...) */
2302             parno = RExC_npar;
2303             RExC_npar++;
2304             ret = reganode(pRExC_state, OPEN, parno);
2305             Set_Node_Length(ret, 1); /* MJD */
2306             Set_Node_Offset(ret, RExC_parse); /* MJD */
2307             open = 1;
2308         }
2309     }
2310     else                        /* ! paren */
2311         ret = NULL;
2312
2313     /* Pick up the branches, linking them together. */
2314     parse_start = RExC_parse;   /* MJD */
2315     br = regbranch(pRExC_state, &flags, 1);
2316     /*     branch_len = (paren != 0); */
2317     
2318     if (br == NULL)
2319         return(NULL);
2320     if (*RExC_parse == '|') {
2321         if (!SIZE_ONLY && RExC_extralen) {
2322             reginsert(pRExC_state, BRANCHJ, br);
2323         }
2324         else {                  /* MJD */
2325             reginsert(pRExC_state, BRANCH, br);
2326             Set_Node_Length(br, paren != 0);
2327             Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2328         }
2329         have_branch = 1;
2330         if (SIZE_ONLY)
2331             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2332     }
2333     else if (paren == ':') {
2334         *flagp |= flags&SIMPLE;
2335     }
2336     if (open) {                         /* Starts with OPEN. */
2337         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2338     }
2339     else if (paren != '?')              /* Not Conditional */
2340         ret = br;
2341     if (flags&HASWIDTH)
2342         *flagp |= HASWIDTH;
2343     *flagp |= flags&SPSTART;
2344     lastbr = br;
2345     while (*RExC_parse == '|') {
2346         if (!SIZE_ONLY && RExC_extralen) {
2347             ender = reganode(pRExC_state, LONGJMP,0);
2348             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2349         }
2350         if (SIZE_ONLY)
2351             RExC_extralen += 2;         /* Account for LONGJMP. */
2352         nextchar(pRExC_state);
2353         br = regbranch(pRExC_state, &flags, 0);
2354         
2355         if (br == NULL)
2356             return(NULL);
2357         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2358         lastbr = br;
2359         if (flags&HASWIDTH)
2360             *flagp |= HASWIDTH;
2361         *flagp |= flags&SPSTART;
2362     }
2363
2364     if (have_branch || paren != ':') {
2365         /* Make a closing node, and hook it on the end. */
2366         switch (paren) {
2367         case ':':
2368             ender = reg_node(pRExC_state, TAIL);
2369             break;
2370         case 1:
2371             ender = reganode(pRExC_state, CLOSE, parno);
2372             Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2373             Set_Node_Length(ender,1); /* MJD */
2374             break;
2375         case '<':
2376         case ',':
2377         case '=':
2378         case '!':
2379             *flagp &= ~HASWIDTH;
2380             /* FALL THROUGH */
2381         case '>':
2382             ender = reg_node(pRExC_state, SUCCEED);
2383             break;
2384         case 0:
2385             ender = reg_node(pRExC_state, END);
2386             break;
2387         }
2388         regtail(pRExC_state, lastbr, ender);
2389
2390         if (have_branch) {
2391             /* Hook the tails of the branches to the closing node. */
2392             for (br = ret; br != NULL; br = regnext(br)) {
2393                 regoptail(pRExC_state, br, ender);
2394             }
2395         }
2396     }
2397
2398     {
2399         char *p;
2400         static char parens[] = "=!<,>";
2401
2402         if (paren && (p = strchr(parens, paren))) {
2403             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2404             int flag = (p - parens) > 1;
2405
2406             if (paren == '>')
2407                 node = SUSPEND, flag = 0;
2408             reginsert(pRExC_state, node,ret);
2409             ret->flags = flag;
2410             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2411         }
2412     }
2413
2414     /* Check for proper termination. */
2415     if (paren) {
2416         RExC_flags16 = oregflags;
2417         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2418             RExC_parse = oregcomp_parse;
2419             vFAIL("Unmatched (");
2420         }
2421     }
2422     else if (!paren && RExC_parse < RExC_end) {
2423         if (*RExC_parse == ')') {
2424             RExC_parse++;
2425             vFAIL("Unmatched )");
2426         }
2427         else
2428             FAIL("Junk on end of regexp");      /* "Can't happen". */
2429         /* NOTREACHED */
2430     }
2431
2432     return(ret);
2433 }
2434
2435 /*
2436  - regbranch - one alternative of an | operator
2437  *
2438  * Implements the concatenation operator.
2439  */
2440 STATIC regnode *
2441 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2442 {
2443     register regnode *ret;
2444     register regnode *chain = NULL;
2445     register regnode *latest;
2446     I32 flags = 0, c = 0;
2447
2448     if (first)
2449         ret = NULL;
2450     else {
2451         if (!SIZE_ONLY && RExC_extralen)
2452             ret = reganode(pRExC_state, BRANCHJ,0);
2453         else {
2454             ret = reg_node(pRExC_state, BRANCH);
2455             Set_Node_Length(ret, 1);
2456         }
2457     }
2458         
2459     if (!first && SIZE_ONLY)
2460         RExC_extralen += 1;                     /* BRANCHJ */
2461
2462     *flagp = WORST;                     /* Tentatively. */
2463
2464     RExC_parse--;
2465     nextchar(pRExC_state);
2466     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2467         flags &= ~TRYAGAIN;
2468         latest = regpiece(pRExC_state, &flags);
2469         if (latest == NULL) {
2470             if (flags & TRYAGAIN)
2471                 continue;
2472             return(NULL);
2473         }
2474         else if (ret == NULL)
2475             ret = latest;
2476         *flagp |= flags&HASWIDTH;
2477         if (chain == NULL)      /* First piece. */
2478             *flagp |= flags&SPSTART;
2479         else {
2480             RExC_naughty++;
2481             regtail(pRExC_state, chain, latest);
2482         }
2483         chain = latest;
2484         c++;
2485     }
2486     if (chain == NULL) {        /* Loop ran zero times. */
2487         chain = reg_node(pRExC_state, NOTHING);
2488         if (ret == NULL)
2489             ret = chain;
2490     }
2491     if (c == 1) {
2492         *flagp |= flags&SIMPLE;
2493     }
2494
2495     return(ret);
2496 }
2497
2498 /*
2499  - regpiece - something followed by possible [*+?]
2500  *
2501  * Note that the branching code sequences used for ? and the general cases
2502  * of * and + are somewhat optimized:  they use the same NOTHING node as
2503  * both the endmarker for their branch list and the body of the last branch.
2504  * It might seem that this node could be dispensed with entirely, but the
2505  * endmarker role is not redundant.
2506  */
2507 STATIC regnode *
2508 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2509 {
2510     register regnode *ret;
2511     register char op;
2512     register char *next;
2513     I32 flags;
2514     char *origparse = RExC_parse;
2515     char *maxpos;
2516     I32 min;
2517     I32 max = REG_INFTY;
2518     char *parse_start;
2519
2520     ret = regatom(pRExC_state, &flags);
2521     if (ret == NULL) {
2522         if (flags & TRYAGAIN)
2523             *flagp |= TRYAGAIN;
2524         return(NULL);
2525     }
2526
2527     op = *RExC_parse;
2528
2529     if (op == '{' && regcurly(RExC_parse)) {
2530         parse_start = RExC_parse; /* MJD */
2531         next = RExC_parse + 1;
2532         maxpos = Nullch;
2533         while (isDIGIT(*next) || *next == ',') {
2534             if (*next == ',') {
2535                 if (maxpos)
2536                     break;
2537                 else
2538                     maxpos = next;
2539             }
2540             next++;
2541         }
2542         if (*next == '}') {             /* got one */
2543             if (!maxpos)
2544                 maxpos = next;
2545             RExC_parse++;
2546             min = atoi(RExC_parse);
2547             if (*maxpos == ',')
2548                 maxpos++;
2549             else
2550                 maxpos = RExC_parse;
2551             max = atoi(maxpos);
2552             if (!max && *maxpos != '0')
2553                 max = REG_INFTY;                /* meaning "infinity" */
2554             else if (max >= REG_INFTY)
2555                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2556             RExC_parse = next;
2557             nextchar(pRExC_state);
2558
2559         do_curly:
2560             if ((flags&SIMPLE)) {
2561                 RExC_naughty += 2 + RExC_naughty / 2;
2562                 reginsert(pRExC_state, CURLY, ret);
2563                 Set_Node_Offset(ret, parse_start+1); /* MJD */
2564                 Set_Node_Cur_Length(ret);
2565             }
2566             else {
2567                 regnode *w = reg_node(pRExC_state, WHILEM);
2568
2569                 w->flags = 0;
2570                 regtail(pRExC_state, ret, w);
2571                 if (!SIZE_ONLY && RExC_extralen) {
2572                     reginsert(pRExC_state, LONGJMP,ret);
2573                     reginsert(pRExC_state, NOTHING,ret);
2574                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2575                 }
2576                 reginsert(pRExC_state, CURLYX,ret);
2577                                 /* MJD hk */
2578                 Set_Node_Offset(ret, parse_start+1);
2579                 Set_Node_Length(ret, 
2580                                 op == '{' ? (RExC_parse - parse_start) : 1);
2581                 
2582                 if (!SIZE_ONLY && RExC_extralen)
2583                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2584                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2585                 if (SIZE_ONLY)
2586                     RExC_whilem_seen++, RExC_extralen += 3;
2587                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2588             }
2589             ret->flags = 0;
2590
2591             if (min > 0)
2592                 *flagp = WORST;
2593             if (max > 0)
2594                 *flagp |= HASWIDTH;
2595             if (max && max < min)
2596                 vFAIL("Can't do {n,m} with n > m");
2597             if (!SIZE_ONLY) {
2598                 ARG1_SET(ret, min);
2599                 ARG2_SET(ret, max);
2600             }
2601
2602             goto nest_check;
2603         }
2604     }
2605
2606     if (!ISMULT1(op)) {
2607         *flagp = flags;
2608         return(ret);
2609     }
2610
2611 #if 0                           /* Now runtime fix should be reliable. */
2612
2613     /* if this is reinstated, don't forget to put this back into perldiag:
2614
2615             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2616
2617            (F) The part of the regexp subject to either the * or + quantifier
2618            could match an empty string. The {#} shows in the regular
2619            expression about where the problem was discovered.
2620
2621     */
2622
2623     if (!(flags&HASWIDTH) && op != '?')
2624       vFAIL("Regexp *+ operand could be empty");
2625 #endif
2626
2627     parse_start = RExC_parse;
2628     nextchar(pRExC_state);
2629
2630     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2631
2632     if (op == '*' && (flags&SIMPLE)) {
2633         reginsert(pRExC_state, STAR, ret);
2634         ret->flags = 0;
2635         RExC_naughty += 4;
2636     }
2637     else if (op == '*') {
2638         min = 0;
2639         goto do_curly;
2640     }
2641     else if (op == '+' && (flags&SIMPLE)) {
2642         reginsert(pRExC_state, PLUS, ret);
2643         ret->flags = 0;
2644         RExC_naughty += 3;
2645     }
2646     else if (op == '+') {
2647         min = 1;
2648         goto do_curly;
2649     }
2650     else if (op == '?') {
2651         min = 0; max = 1;
2652         goto do_curly;
2653     }
2654   nest_check:
2655     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2656         vWARN3(RExC_parse,
2657                "%.*s matches null string many times",
2658                RExC_parse - origparse,
2659                origparse);
2660     }
2661
2662     if (*RExC_parse == '?') {
2663         nextchar(pRExC_state);
2664         reginsert(pRExC_state, MINMOD, ret);
2665         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2666     }
2667     if (ISMULT2(RExC_parse)) {
2668         RExC_parse++;
2669         vFAIL("Nested quantifiers");
2670     }
2671
2672     return(ret);
2673 }
2674
2675 /*
2676  - regatom - the lowest level
2677  *
2678  * Optimization:  gobbles an entire sequence of ordinary characters so that
2679  * it can turn them into a single node, which is smaller to store and
2680  * faster to run.  Backslashed characters are exceptions, each becoming a
2681  * separate node; the code is simpler that way and it's not worth fixing.
2682  *
2683  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2684 STATIC regnode *
2685 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2686 {
2687     register regnode *ret = 0;
2688     I32 flags;
2689     char *parse_start = 0;
2690
2691     *flagp = WORST;             /* Tentatively. */
2692
2693 tryagain:
2694     switch (*RExC_parse) {
2695     case '^':
2696         RExC_seen_zerolen++;
2697         nextchar(pRExC_state);
2698         if (RExC_flags16 & PMf_MULTILINE)
2699             ret = reg_node(pRExC_state, MBOL);
2700         else if (RExC_flags16 & PMf_SINGLELINE)
2701             ret = reg_node(pRExC_state, SBOL);
2702         else
2703             ret = reg_node(pRExC_state, BOL);
2704         Set_Node_Length(ret, 1); /* MJD */
2705         break;
2706     case '$':
2707         nextchar(pRExC_state);
2708         if (*RExC_parse)
2709             RExC_seen_zerolen++;
2710         if (RExC_flags16 & PMf_MULTILINE)
2711             ret = reg_node(pRExC_state, MEOL);
2712         else if (RExC_flags16 & PMf_SINGLELINE)
2713             ret = reg_node(pRExC_state, SEOL);
2714         else
2715             ret = reg_node(pRExC_state, EOL);
2716         Set_Node_Length(ret, 1); /* MJD */
2717         break;
2718     case '.':
2719         nextchar(pRExC_state);
2720         if (RExC_flags16 & PMf_SINGLELINE)
2721             ret = reg_node(pRExC_state, SANY);
2722         else
2723             ret = reg_node(pRExC_state, REG_ANY);
2724         *flagp |= HASWIDTH|SIMPLE;
2725         RExC_naughty++;
2726         Set_Node_Length(ret, 1); /* MJD */
2727         break;
2728     case '[':
2729     {
2730         char *oregcomp_parse = ++RExC_parse;
2731         ret = regclass(pRExC_state);
2732         if (*RExC_parse != ']') {
2733             RExC_parse = oregcomp_parse;
2734             vFAIL("Unmatched [");
2735         }
2736         nextchar(pRExC_state);
2737         *flagp |= HASWIDTH|SIMPLE;
2738         Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2739         break;
2740     }
2741     case '(':
2742         nextchar(pRExC_state);
2743         ret = reg(pRExC_state, 1, &flags);
2744         if (ret == NULL) {
2745                 if (flags & TRYAGAIN) {
2746                     if (RExC_parse == RExC_end) {
2747                          /* Make parent create an empty node if needed. */
2748                         *flagp |= TRYAGAIN;
2749                         return(NULL);
2750                     }
2751                     goto tryagain;
2752                 }
2753                 return(NULL);
2754         }
2755         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2756         break;
2757     case '|':
2758     case ')':
2759         if (flags & TRYAGAIN) {
2760             *flagp |= TRYAGAIN;
2761             return NULL;
2762         }
2763         vFAIL("Internal urp");
2764                                 /* Supposed to be caught earlier. */
2765         break;
2766     case '{':
2767         if (!regcurly(RExC_parse)) {
2768             RExC_parse++;
2769             goto defchar;
2770         }
2771         /* FALL THROUGH */
2772     case '?':
2773     case '+':
2774     case '*':
2775         RExC_parse++;
2776         vFAIL("Quantifier follows nothing");
2777         break;
2778     case '\\':
2779         switch (*++RExC_parse) {
2780         case 'A':
2781             RExC_seen_zerolen++;
2782             ret = reg_node(pRExC_state, SBOL);
2783             *flagp |= SIMPLE;
2784             nextchar(pRExC_state);
2785             Set_Node_Length(ret, 2); /* MJD */
2786             break;
2787         case 'G':
2788             ret = reg_node(pRExC_state, GPOS);
2789             RExC_seen |= REG_SEEN_GPOS;
2790             *flagp |= SIMPLE;
2791             nextchar(pRExC_state);
2792             Set_Node_Length(ret, 2); /* MJD */
2793             break;
2794         case 'Z':
2795             ret = reg_node(pRExC_state, SEOL);
2796             *flagp |= SIMPLE;
2797             nextchar(pRExC_state);
2798             break;
2799         case 'z':
2800             ret = reg_node(pRExC_state, EOS);
2801             *flagp |= SIMPLE;
2802             RExC_seen_zerolen++;                /* Do not optimize RE away */
2803             nextchar(pRExC_state);
2804             Set_Node_Length(ret, 2); /* MJD */
2805             break;
2806         case 'C':
2807             ret = reg_node(pRExC_state, CANY);
2808             RExC_seen |= REG_SEEN_CANY;
2809             *flagp |= HASWIDTH|SIMPLE;
2810             nextchar(pRExC_state);
2811             Set_Node_Length(ret, 2); /* MJD */
2812             break;
2813         case 'X':
2814             ret = reg_node(pRExC_state, CLUMP);
2815             *flagp |= HASWIDTH;
2816             nextchar(pRExC_state);
2817             Set_Node_Length(ret, 2); /* MJD */
2818             break;
2819         case 'w':
2820             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2821             *flagp |= HASWIDTH|SIMPLE;
2822             nextchar(pRExC_state);
2823             Set_Node_Length(ret, 2); /* MJD */
2824             break;
2825         case 'W':
2826             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2827             *flagp |= HASWIDTH|SIMPLE;
2828             nextchar(pRExC_state);
2829             Set_Node_Length(ret, 2); /* MJD */
2830             break;
2831         case 'b':
2832             RExC_seen_zerolen++;
2833             RExC_seen |= REG_SEEN_LOOKBEHIND;
2834             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2835             *flagp |= SIMPLE;
2836             nextchar(pRExC_state);
2837             Set_Node_Length(ret, 2); /* MJD */
2838             break;
2839         case 'B':
2840             RExC_seen_zerolen++;
2841             RExC_seen |= REG_SEEN_LOOKBEHIND;
2842             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2843             *flagp |= SIMPLE;
2844             nextchar(pRExC_state);
2845             Set_Node_Length(ret, 2); /* MJD */
2846             break;
2847         case 's':
2848             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2849             *flagp |= HASWIDTH|SIMPLE;
2850             nextchar(pRExC_state);
2851             Set_Node_Length(ret, 2); /* MJD */
2852             break;
2853         case 'S':
2854             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
2855             *flagp |= HASWIDTH|SIMPLE;
2856             nextchar(pRExC_state);
2857             Set_Node_Length(ret, 2); /* MJD */
2858             break;
2859         case 'd':
2860             ret = reg_node(pRExC_state, DIGIT);
2861             *flagp |= HASWIDTH|SIMPLE;
2862             nextchar(pRExC_state);
2863             Set_Node_Length(ret, 2); /* MJD */
2864             break;
2865         case 'D':
2866             ret = reg_node(pRExC_state, NDIGIT);
2867             *flagp |= HASWIDTH|SIMPLE;
2868             nextchar(pRExC_state);
2869             Set_Node_Length(ret, 2); /* MJD */
2870             break;
2871         case 'p':
2872         case 'P':
2873             {   
2874                 char* oldregxend = RExC_end;
2875                 char* parse_start = RExC_parse;
2876
2877                 if (RExC_parse[1] == '{') {
2878                   /* a lovely hack--pretend we saw [\pX] instead */
2879                     RExC_end = strchr(RExC_parse, '}');
2880                     if (!RExC_end) {
2881                         RExC_parse += 2;
2882                         RExC_end = oldregxend;
2883                         vFAIL("Missing right brace on \\p{}");
2884                     }
2885                     RExC_end++;
2886                 }
2887                 else
2888                     RExC_end = RExC_parse + 2;
2889                 RExC_parse--;
2890
2891                 ret = regclass(pRExC_state);
2892
2893                 RExC_end = oldregxend;
2894                 RExC_parse--;
2895                 Set_Node_Cur_Length(ret); /* MJD */
2896                 nextchar(pRExC_state);
2897                 *flagp |= HASWIDTH|SIMPLE;
2898             }
2899             break;
2900         case 'n':
2901         case 'r':
2902         case 't':
2903         case 'f':
2904         case 'e':
2905         case 'a':
2906         case 'x':
2907         case 'c':
2908         case '0':
2909             goto defchar;
2910         case '1': case '2': case '3': case '4':
2911         case '5': case '6': case '7': case '8': case '9':
2912             {
2913                 I32 num = atoi(RExC_parse);
2914
2915                 if (num > 9 && num >= RExC_npar)
2916                     goto defchar;
2917                 else {
2918                     char * parse_start = RExC_parse - 1; /* MJD */
2919                     while (isDIGIT(*RExC_parse))
2920                         RExC_parse++;
2921
2922                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2923                         vFAIL("Reference to nonexistent group");
2924                     RExC_sawback = 1;
2925                     ret = reganode(pRExC_state, FOLD
2926                                    ? (LOC ? REFFL : REFF)
2927                                    : REF, num);
2928                     *flagp |= HASWIDTH;
2929                     
2930                     /* override incorrect value set in reganode MJD */
2931                     Set_Node_Offset(ret, parse_start+1); 
2932                     Set_Node_Cur_Length(ret); /* MJD */
2933                     RExC_parse--;
2934                     nextchar(pRExC_state);
2935                 }
2936             }
2937             break;
2938         case '\0':
2939             if (RExC_parse >= RExC_end)
2940                 FAIL("Trailing \\");
2941             /* FALL THROUGH */
2942         default:
2943             /* Do not generate `unrecognized' warnings here, we fall
2944                back into the quick-grab loop below */
2945             goto defchar;
2946         }
2947         break;
2948
2949     case '#':
2950         if (RExC_flags16 & PMf_EXTENDED) {
2951             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2952             if (RExC_parse < RExC_end)
2953                 goto tryagain;
2954         }
2955         /* FALL THROUGH */
2956
2957     default: {
2958             register STRLEN len;
2959             register UV ender;
2960             register char *p;
2961             char *oldp, *s;
2962             STRLEN numlen;
2963
2964             parse_start = RExC_parse - 1;
2965
2966             RExC_parse++;
2967
2968         defchar:
2969             ret = reg_node(pRExC_state, FOLD
2970                           ? (LOC ? EXACTFL : EXACTF)
2971                           : EXACT);
2972             s = STRING(ret);
2973             for (len = 0, p = RExC_parse - 1;
2974               len < 127 && p < RExC_end;
2975               len++)
2976             {
2977                 oldp = p;
2978
2979                 if (RExC_flags16 & PMf_EXTENDED)
2980                     p = regwhite(p, RExC_end);
2981                 switch (*p) {
2982                 case '^':
2983                 case '$':
2984                 case '.':
2985                 case '[':
2986                 case '(':
2987                 case ')':
2988                 case '|':
2989                     goto loopdone;
2990                 case '\\':
2991                     switch (*++p) {
2992                     case 'A':
2993                     case 'G':
2994                     case 'Z':
2995                     case 'z':
2996                     case 'w':
2997                     case 'W':
2998                     case 'b':
2999                     case 'B':
3000                     case 's':
3001                     case 'S':
3002                     case 'd':
3003                     case 'D':
3004                     case 'p':
3005                     case 'P':
3006                         --p;
3007                         goto loopdone;
3008                     case 'n':
3009                         ender = '\n';
3010                         p++;
3011                         break;
3012                     case 'r':
3013                         ender = '\r';
3014                         p++;
3015                         break;
3016                     case 't':
3017                         ender = '\t';
3018                         p++;
3019                         break;
3020                     case 'f':
3021                         ender = '\f';
3022                         p++;
3023                         break;
3024                     case 'e':
3025                           ender = ASCII_TO_NATIVE('\033');
3026                         p++;
3027                         break;
3028                     case 'a':
3029                           ender = ASCII_TO_NATIVE('\007');
3030                         p++;
3031                         break;
3032                     case 'x':
3033                         if (*++p == '{') {
3034                             char* e = strchr(p, '}');
3035         
3036                             if (!e) {
3037                                 RExC_parse = p + 1;
3038                                 vFAIL("Missing right brace on \\x{}");
3039                             }
3040                             else {
3041                                 numlen = 1;     /* allow underscores */
3042                                 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
3043                                 if (ender > 0xff)
3044                                     RExC_utf8 = 1;
3045                                 /* numlen is generous */
3046                                 if (numlen + len >= 127) {
3047                                     p--;
3048                                     goto loopdone;
3049                                 }
3050                                 p = e + 1;
3051                             }
3052                         }
3053                         else {
3054                             numlen = 0;         /* disallow underscores */
3055                             ender = (UV)scan_hex(p, 2, &numlen);
3056                             p += numlen;
3057                         }
3058                         break;
3059                     case 'c':
3060                         p++;
3061                         ender = UCHARAT(p++);
3062                         ender = toCTRL(ender);
3063                         break;
3064                     case '0': case '1': case '2': case '3':case '4':
3065                     case '5': case '6': case '7': case '8':case '9':
3066                         if (*p == '0' ||
3067                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
3068                             numlen = 0;         /* disallow underscores */
3069                             ender = (UV)scan_oct(p, 3, &numlen);
3070                             p += numlen;
3071                         }
3072                         else {
3073                             --p;
3074                             goto loopdone;
3075                         }
3076                         break;
3077                     case '\0':
3078                         if (p >= RExC_end)
3079                             FAIL("Trailing \\");
3080                         /* FALL THROUGH */
3081                     default:
3082                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3083                             vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
3084                         goto normal_default;
3085                     }
3086                     break;
3087                 default:
3088                   normal_default:
3089                     if (UTF8_IS_START(*p) && UTF) {
3090                         ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3091                                                &numlen, 0);
3092                         p += numlen;
3093                     }
3094                     else
3095                         ender = *p++;
3096                     break;
3097                 }
3098                 if (RExC_flags16 & PMf_EXTENDED)
3099                     p = regwhite(p, RExC_end);
3100                 if (UTF && FOLD) {
3101                     if (LOC)
3102                         ender = toLOWER_LC_uvchr(ender);
3103                     else
3104                         ender = toLOWER_uni(ender);
3105                 }
3106                 if (ISMULT2(p)) { /* Back off on ?+*. */
3107                     if (len)
3108                         p = oldp;
3109                     else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3110                         reguni(pRExC_state, ender, s, &numlen);
3111                         s += numlen;
3112                         len += numlen;
3113                     }
3114                     else {
3115                         len++;
3116                         REGC(ender, s++);
3117                     }
3118                     break;
3119                 }
3120                 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3121                     reguni(pRExC_state, ender, s, &numlen);
3122                     s += numlen;
3123                     len += numlen - 1;
3124                 }
3125                 else
3126                     REGC(ender, s++);
3127             }
3128         loopdone:
3129             RExC_parse = p - 1;
3130             Set_Node_Cur_Length(ret); /* MJD */
3131             nextchar(pRExC_state);
3132             {
3133                 /* len is STRLEN which is unsigned, need to copy to signed */
3134                 IV iv = len;
3135                 if (iv < 0)
3136                     vFAIL("Internal disaster");
3137             }
3138             if (len > 0)
3139                 *flagp |= HASWIDTH;
3140             if (len == 1)
3141                 *flagp |= SIMPLE;
3142             if (!SIZE_ONLY)
3143                 STR_LEN(ret) = len;
3144             if (SIZE_ONLY)
3145                 RExC_size += STR_SZ(len);
3146             else
3147                 RExC_emit += STR_SZ(len);
3148         }
3149         break;
3150     }
3151
3152     return(ret);
3153 }
3154
3155 STATIC char *
3156 S_regwhite(pTHX_ char *p, char *e)
3157 {
3158     while (p < e) {
3159         if (isSPACE(*p))
3160             ++p;
3161         else if (*p == '#') {
3162             do {
3163                 p++;
3164             } while (p < e && *p != '\n');
3165         }
3166         else
3167             break;
3168     }
3169     return p;
3170 }
3171
3172 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3173    Character classes ([:foo:]) can also be negated ([:^foo:]).
3174    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3175    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3176    but trigger warnings because they are currently unimplemented. */
3177 STATIC I32
3178 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3179 {
3180     char *posixcc = 0;
3181     I32 namedclass = OOB_NAMEDCLASS;
3182
3183     if (value == '[' && RExC_parse + 1 < RExC_end &&
3184         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3185         (*RExC_parse == ':' ||
3186          *RExC_parse == '=' ||
3187          *RExC_parse == '.')) {
3188         char  c = *RExC_parse;
3189         char* s = RExC_parse++;
3190         
3191         while (RExC_parse < RExC_end && *RExC_parse != c)
3192             RExC_parse++;
3193         if (RExC_parse == RExC_end)
3194             /* Grandfather lone [:, [=, [. */
3195             RExC_parse = s;
3196         else {
3197             char* t = RExC_parse++; /* skip over the c */
3198
3199             if (*RExC_parse == ']') {
3200                 RExC_parse++; /* skip over the ending ] */
3201                 posixcc = s + 1;
3202                 if (*s == ':') {
3203                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3204                     I32 skip = 5; /* the most common skip */
3205
3206                     switch (*posixcc) {
3207                     case 'a':
3208                         if (strnEQ(posixcc, "alnum", 5))
3209                             namedclass =
3210                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3211                         else if (strnEQ(posixcc, "alpha", 5))
3212                             namedclass =
3213                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3214                         else if (strnEQ(posixcc, "ascii", 5))
3215                             namedclass =
3216                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3217                         break;
3218                     case 'b':
3219                         if (strnEQ(posixcc, "blank", 5))
3220                             namedclass =
3221                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3222                         break;
3223                     case 'c':
3224                         if (strnEQ(posixcc, "cntrl", 5))
3225                             namedclass =
3226                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3227                         break;
3228                     case 'd':
3229                         if (strnEQ(posixcc, "digit", 5))
3230                             namedclass =
3231                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3232                         break;
3233                     case 'g':
3234                         if (strnEQ(posixcc, "graph", 5))
3235                             namedclass =
3236                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3237                         break;
3238                     case 'l':
3239                         if (strnEQ(posixcc, "lower", 5))
3240                             namedclass =
3241                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3242                         break;
3243                     case 'p':
3244                         if (strnEQ(posixcc, "print", 5))
3245                             namedclass =
3246                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3247                         else if (strnEQ(posixcc, "punct", 5))
3248                             namedclass =
3249                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3250                         break;
3251                     case 's':
3252                         if (strnEQ(posixcc, "space", 5))
3253                             namedclass =
3254                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3255                         break;
3256                     case 'u':
3257                         if (strnEQ(posixcc, "upper", 5))
3258                             namedclass =
3259                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3260                         break;
3261                     case 'w': /* this is not POSIX, this is the Perl \w */
3262                         if (strnEQ(posixcc, "word", 4)) {
3263                             namedclass =
3264                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3265                             skip = 4;
3266                         }
3267                         break;
3268                     case 'x':
3269                         if (strnEQ(posixcc, "xdigit", 6)) {
3270                             namedclass =
3271                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3272                             skip = 6;
3273                         }
3274                         break;
3275                     }
3276                     if (namedclass == OOB_NAMEDCLASS ||
3277                         posixcc[skip] != ':' ||
3278                         posixcc[skip+1] != ']')
3279                     {
3280                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3281                                       t - s - 1, s + 1);
3282                     }
3283                 } else if (!SIZE_ONLY) {
3284                     /* [[=foo=]] and [[.foo.]] are still future. */
3285
3286                     /* adjust RExC_parse so the warning shows after
3287                        the class closes */
3288                     while (*RExC_parse && *RExC_parse != ']')
3289                         RExC_parse++;
3290                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3291                 }
3292             } else {
3293                 /* Maternal grandfather:
3294                  * "[:" ending in ":" but not in ":]" */
3295                 RExC_parse = s;
3296             }
3297         }
3298     }
3299
3300     return namedclass;
3301 }
3302
3303 STATIC void
3304 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3305 {
3306     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3307         (*RExC_parse == ':' ||
3308          *RExC_parse == '=' ||
3309          *RExC_parse == '.')) {
3310         char *s = RExC_parse;
3311         char  c = *s++;
3312
3313         while(*s && isALNUM(*s))
3314             s++;
3315         if (*s && c == *s && s[1] == ']') {
3316             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3317
3318             /* [[=foo=]] and [[.foo.]] are still future. */
3319             if (c == '=' || c == '.')
3320             {
3321                 /* adjust RExC_parse so the error shows after
3322                    the class closes */
3323                 while (*RExC_parse && *RExC_parse++ != ']')
3324                     ;
3325                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3326             }
3327         }
3328     }
3329 }
3330
3331 STATIC regnode *
3332 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3333 {
3334     register UV value;
3335     register IV prevvalue = OOB_UNICODE;
3336     register IV range = 0;
3337     register regnode *ret;
3338     STRLEN numlen;
3339     IV namedclass;
3340     char *rangebegin = 0;
3341     bool need_class = 0;
3342     SV *listsv = Nullsv;
3343     register char *e;
3344     UV n;
3345     bool optimize_invert = TRUE;
3346
3347     ret = reganode(pRExC_state, ANYOF, 0);
3348
3349     if (!SIZE_ONLY)
3350         ANYOF_FLAGS(ret) = 0;
3351
3352     if (*RExC_parse == '^') {   /* Complement of range. */
3353         RExC_naughty++;
3354         RExC_parse++;
3355         if (!SIZE_ONLY)
3356             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3357     }
3358
3359     if (SIZE_ONLY)
3360         RExC_size += ANYOF_SKIP;
3361     else {
3362         RExC_emit += ANYOF_SKIP;
3363         if (FOLD)
3364             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3365         if (LOC)
3366             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3367         ANYOF_BITMAP_ZERO(ret);
3368         listsv = newSVpvn("# comment\n", 10);
3369     }
3370
3371     if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3372         checkposixcc(pRExC_state);
3373
3374     if (*RExC_parse == ']' || *RExC_parse == '-')
3375         goto charclassloop;             /* allow 1st char to be ] or - */
3376
3377     while (RExC_parse < RExC_end && *RExC_parse != ']') {
3378
3379     charclassloop:
3380
3381         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3382
3383         if (!range)
3384             rangebegin = RExC_parse;
3385         if (UTF) {
3386             value = utf8n_to_uvchr((U8*)RExC_parse,
3387                                    RExC_end - RExC_parse,
3388                                    &numlen, 0);
3389             RExC_parse += numlen;
3390         }
3391         else
3392             value = UCHARAT(RExC_parse++);
3393         if (value == '[')
3394             namedclass = regpposixcc(pRExC_state, value);
3395         else if (value == '\\') {
3396             if (UTF) {
3397                 value = utf8n_to_uvchr((U8*)RExC_parse,
3398                                    RExC_end - RExC_parse,
3399                                    &numlen, 0);
3400                 RExC_parse += numlen;
3401             }
3402             else
3403                 value = UCHARAT(RExC_parse++);
3404             /* Some compilers cannot handle switching on 64-bit integer
3405              * values, therefore value cannot be an UV.  Yes, this will
3406              * be a problem later if we want switch on Unicode.
3407              * A similar issue a little bit later when switching on
3408              * namedclass. --jhi */
3409             switch ((I32)value) {
3410             case 'w':   namedclass = ANYOF_ALNUM;       break;
3411             case 'W':   namedclass = ANYOF_NALNUM;      break;
3412             case 's':   namedclass = ANYOF_SPACE;       break;
3413             case 'S':   namedclass = ANYOF_NSPACE;      break;
3414             case 'd':   namedclass = ANYOF_DIGIT;       break;
3415             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3416             case 'p':
3417             case 'P':
3418                 if (*RExC_parse == '{') {
3419                     e = strchr(RExC_parse++, '}');
3420                     if (!e)
3421                         vFAIL("Missing right brace on \\p{}");
3422                     n = e - RExC_parse;
3423                 }
3424                 else {
3425                     e = RExC_parse;
3426                     n = 1;
3427                 }
3428                 if (!SIZE_ONLY) {
3429                     if (value == 'p')
3430                         Perl_sv_catpvf(aTHX_ listsv,
3431                                        "+utf8::%.*s\n", (int)n, RExC_parse);
3432                     else
3433                         Perl_sv_catpvf(aTHX_ listsv,
3434                                        "!utf8::%.*s\n", (int)n, RExC_parse);
3435                 }
3436                 RExC_parse = e + 1;
3437                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3438                 continue;
3439             case 'n':   value = '\n';                   break;
3440             case 'r':   value = '\r';                   break;
3441             case 't':   value = '\t';                   break;
3442             case 'f':   value = '\f';                   break;
3443             case 'b':   value = '\b';                   break;
3444             case 'e':   value = ASCII_TO_NATIVE('\033');break;
3445             case 'a':   value = ASCII_TO_NATIVE('\007');break;
3446             case 'x':
3447                 if (*RExC_parse == '{') {
3448                     e = strchr(RExC_parse++, '}');
3449                     if (!e)
3450                         vFAIL("Missing right brace on \\x{}");
3451                     numlen = 1;         /* allow underscores */
3452                     value = (UV)scan_hex(RExC_parse,
3453                                          e - RExC_parse,
3454                                          &numlen);
3455                     RExC_parse = e + 1;
3456                 }
3457                 else {
3458                     numlen = 0;         /* disallow underscores */
3459                     value = (UV)scan_hex(RExC_parse, 2, &numlen);
3460                     RExC_parse += numlen;
3461                 }
3462                 break;
3463             case 'c':
3464                 value = UCHARAT(RExC_parse++);
3465                 value = toCTRL(value);
3466                 break;
3467             case '0': case '1': case '2': case '3': case '4':
3468             case '5': case '6': case '7': case '8': case '9':
3469                 numlen = 0;             /* disallow underscores */
3470                 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3471                 RExC_parse += numlen;
3472                 break;
3473             default:
3474                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3475                     vWARN2(RExC_parse,
3476                            "Unrecognized escape \\%c in character class passed through",
3477                            (int)value);
3478                 break;
3479             }
3480         } /* end of \blah */
3481
3482         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3483
3484             if (!SIZE_ONLY && !need_class)
3485                 ANYOF_CLASS_ZERO(ret);
3486
3487             need_class = 1;
3488
3489             /* a bad range like a-\d, a-[:digit:] ? */
3490             if (range) {
3491                 if (!SIZE_ONLY) {
3492                     if (ckWARN(WARN_REGEXP))
3493                         vWARN4(RExC_parse,
3494                                "False [] range \"%*.*s\"",
3495                                RExC_parse - rangebegin,
3496                                RExC_parse - rangebegin,
3497                                rangebegin);
3498                     if (prevvalue < 256) {
3499                         ANYOF_BITMAP_SET(ret, prevvalue);
3500                         ANYOF_BITMAP_SET(ret, '-');
3501                     }
3502                     else {
3503                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3504                         Perl_sv_catpvf(aTHX_ listsv,
3505                                        "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3506                     }
3507                 }
3508
3509                 range = 0; /* this was not a true range */
3510             }
3511
3512             if (!SIZE_ONLY) {
3513                 if (namedclass > OOB_NAMEDCLASS)
3514                     optimize_invert = FALSE;
3515                 /* Possible truncation here but in some 64-bit environments
3516                  * the compiler gets heartburn about switch on 64-bit values.
3517                  * A similar issue a little earlier when switching on value.
3518                  * --jhi */
3519                 switch ((I32)namedclass) {
3520                 case ANYOF_ALNUM:
3521                     if (LOC)
3522                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3523                     else {
3524                         for (value = 0; value < 256; value++)
3525                             if (isALNUM(value))
3526                                 ANYOF_BITMAP_SET(ret, value);
3527                     }
3528                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3529                     break;
3530                 case ANYOF_NALNUM:
3531                     if (LOC)
3532                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3533                     else {
3534                         for (value = 0; value < 256; value++)
3535                             if (!isALNUM(value))
3536                                 ANYOF_BITMAP_SET(ret, value);
3537                     }
3538                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3539                     break;
3540                 case ANYOF_ALNUMC:
3541                     if (LOC)
3542                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3543                     else {
3544                         for (value = 0; value < 256; value++)
3545                             if (isALNUMC(value))
3546                                 ANYOF_BITMAP_SET(ret, value);
3547                     }
3548                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3549                     break;
3550                 case ANYOF_NALNUMC:
3551                     if (LOC)
3552                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3553                     else {
3554                         for (value = 0; value < 256; value++)
3555                             if (!isALNUMC(value))
3556                                 ANYOF_BITMAP_SET(ret, value);
3557                     }
3558                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3559                     break;
3560                 case ANYOF_ALPHA:
3561                     if (LOC)
3562                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3563                     else {
3564                         for (value = 0; value < 256; value++)
3565                             if (isALPHA(value))
3566                                 ANYOF_BITMAP_SET(ret, value);
3567                     }
3568                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3569                     break;
3570                 case ANYOF_NALPHA:
3571                     if (LOC)
3572                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3573                     else {
3574                         for (value = 0; value < 256; value++)
3575                             if (!isALPHA(value))
3576                                 ANYOF_BITMAP_SET(ret, value);
3577                     }
3578                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3579                     break;
3580                 case ANYOF_ASCII:
3581                     if (LOC)
3582                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3583                     else {
3584 #ifndef EBCDIC
3585                         for (value = 0; value < 128; value++)
3586                             ANYOF_BITMAP_SET(ret, value);
3587 #else  /* EBCDIC */
3588                         for (value = 0; value < 256; value++) {
3589                             if (isASCII(value))
3590                                 ANYOF_BITMAP_SET(ret, value);
3591                         }
3592 #endif /* EBCDIC */
3593                     }
3594                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3595                     break;
3596                 case ANYOF_NASCII:
3597                     if (LOC)
3598                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3599                     else {
3600 #ifndef EBCDIC
3601                         for (value = 128; value < 256; value++)
3602                             ANYOF_BITMAP_SET(ret, value);
3603 #else  /* EBCDIC */
3604                         for (value = 0; value < 256; value++) {
3605                             if (!isASCII(value))
3606                                 ANYOF_BITMAP_SET(ret, value);
3607                         }
3608 #endif /* EBCDIC */
3609                     }
3610                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3611                     break;
3612                 case ANYOF_BLANK:
3613                     if (LOC)
3614                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3615                     else {
3616                         for (value = 0; value < 256; value++)
3617                             if (isBLANK(value))
3618                                 ANYOF_BITMAP_SET(ret, value);
3619                     }
3620                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3621                     break;
3622                 case ANYOF_NBLANK:
3623                     if (LOC)
3624                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3625                     else {
3626                         for (value = 0; value < 256; value++)
3627                             if (!isBLANK(value))
3628                                 ANYOF_BITMAP_SET(ret, value);
3629                     }
3630                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3631                     break;
3632                 case ANYOF_CNTRL:
3633                     if (LOC)
3634                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3635                     else {
3636                         for (value = 0; value < 256; value++)
3637                             if (isCNTRL(value))
3638                                 ANYOF_BITMAP_SET(ret, value);
3639                     }
3640                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3641                     break;
3642                 case ANYOF_NCNTRL:
3643                     if (LOC)
3644                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3645                     else {
3646                         for (value = 0; value < 256; value++)
3647                             if (!isCNTRL(value))
3648                                 ANYOF_BITMAP_SET(ret, value);
3649                     }
3650                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3651                     break;
3652                 case ANYOF_DIGIT:
3653                     if (LOC)
3654                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3655                     else {
3656                         /* consecutive digits assumed */
3657                         for (value = '0'; value <= '9'; value++)
3658                             ANYOF_BITMAP_SET(ret, value);
3659                     }
3660                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3661                     break;
3662                 case ANYOF_NDIGIT:
3663                     if (LOC)
3664                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3665                     else {
3666                         /* consecutive digits assumed */
3667                         for (value = 0; value < '0'; value++)
3668                             ANYOF_BITMAP_SET(ret, value);
3669                         for (value = '9' + 1; value < 256; value++)
3670                             ANYOF_BITMAP_SET(ret, value);
3671                     }
3672                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3673                     break;
3674                 case ANYOF_GRAPH:
3675                     if (LOC)
3676                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3677                     else {
3678                         for (value = 0; value < 256; value++)
3679                             if (isGRAPH(value))
3680                                 ANYOF_BITMAP_SET(ret, value);
3681                     }
3682                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3683                     break;
3684                 case ANYOF_NGRAPH:
3685                     if (LOC)
3686                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3687                     else {
3688                         for (value = 0; value < 256; value++)
3689                             if (!isGRAPH(value))
3690                                 ANYOF_BITMAP_SET(ret, value);
3691                     }
3692                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3693                     break;
3694                 case ANYOF_LOWER:
3695                     if (LOC)
3696                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3697                     else {
3698                         for (value = 0; value < 256; value++)
3699                             if (isLOWER(value))
3700                                 ANYOF_BITMAP_SET(ret, value);
3701                     }
3702                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3703                     break;
3704                 case ANYOF_NLOWER:
3705                     if (LOC)
3706                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3707                     else {
3708                         for (value = 0; value < 256; value++)
3709                             if (!isLOWER(value))
3710                                 ANYOF_BITMAP_SET(ret, value);
3711                     }
3712                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3713                     break;
3714                 case ANYOF_PRINT:
3715                     if (LOC)
3716                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3717                     else {
3718                         for (value = 0; value < 256; value++)
3719                             if (isPRINT(value))
3720                                 ANYOF_BITMAP_SET(ret, value);
3721                     }
3722                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3723                     break;
3724                 case ANYOF_NPRINT:
3725                     if (LOC)
3726                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3727                     else {
3728                         for (value = 0; value < 256; value++)
3729                             if (!isPRINT(value))
3730                                 ANYOF_BITMAP_SET(ret, value);
3731                     }
3732                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3733                     break;
3734                 case ANYOF_PSXSPC:
3735                     if (LOC)
3736                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3737                     else {
3738                         for (value = 0; value < 256; value++)
3739                             if (isPSXSPC(value))
3740                                 ANYOF_BITMAP_SET(ret, value);
3741                     }
3742                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3743                     break;
3744                 case ANYOF_NPSXSPC:
3745                     if (LOC)
3746                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3747                     else {
3748                         for (value = 0; value < 256; value++)
3749                             if (!isPSXSPC(value))
3750                                 ANYOF_BITMAP_SET(ret, value);
3751                     }
3752                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3753                     break;
3754                 case ANYOF_PUNCT:
3755                     if (LOC)
3756                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3757                     else {
3758                         for (value = 0; value < 256; value++)
3759                             if (isPUNCT(value))
3760                                 ANYOF_BITMAP_SET(ret, value);
3761                     }
3762                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3763                     break;
3764                 case ANYOF_NPUNCT:
3765                     if (LOC)
3766                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3767                     else {
3768                         for (value = 0; value < 256; value++)
3769                             if (!isPUNCT(value))
3770                                 ANYOF_BITMAP_SET(ret, value);
3771                     }
3772                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3773                     break;
3774                 case ANYOF_SPACE:
3775                     if (LOC)
3776                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3777                     else {
3778                         for (value = 0; value < 256; value++)
3779                             if (isSPACE(value))
3780                                 ANYOF_BITMAP_SET(ret, value);
3781                     }
3782                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3783                     break;
3784                 case ANYOF_NSPACE:
3785                     if (LOC)
3786                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3787                     else {
3788                         for (value = 0; value < 256; value++)
3789                             if (!isSPACE(value))
3790                                 ANYOF_BITMAP_SET(ret, value);
3791                     }
3792                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3793                     break;
3794                 case ANYOF_UPPER:
3795                     if (LOC)
3796                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3797                     else {
3798                         for (value = 0; value < 256; value++)
3799                             if (isUPPER(value))
3800                                 ANYOF_BITMAP_SET(ret, value);
3801                     }
3802                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3803                     break;
3804                 case ANYOF_NUPPER:
3805                     if (LOC)
3806                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3807                     else {
3808                         for (value = 0; value < 256; value++)
3809                             if (!isUPPER(value))
3810                                 ANYOF_BITMAP_SET(ret, value);
3811                     }
3812                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3813                     break;
3814                 case ANYOF_XDIGIT:
3815                     if (LOC)
3816                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3817                     else {
3818                         for (value = 0; value < 256; value++)
3819                             if (isXDIGIT(value))
3820                                 ANYOF_BITMAP_SET(ret, value);
3821                     }
3822                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3823                     break;
3824                 case ANYOF_NXDIGIT:
3825                     if (LOC)
3826                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3827                     else {
3828                         for (value = 0; value < 256; value++)
3829                             if (!isXDIGIT(value))
3830                                 ANYOF_BITMAP_SET(ret, value);
3831                     }
3832                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3833                     break;
3834                 default:
3835                     vFAIL("Invalid [::] class");
3836                     break;
3837                 }
3838                 if (LOC)
3839                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3840                 continue;
3841             }
3842         } /* end of namedclass \blah */
3843
3844         if (range) {
3845             if (prevvalue > value) /* b-a */ {
3846                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3847                               RExC_parse - rangebegin,
3848                               RExC_parse - rangebegin,
3849                               rangebegin);
3850                 range = 0; /* not a valid range */
3851             }
3852         }
3853         else {
3854             prevvalue = value; /* save the beginning of the range */
3855             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3856                 RExC_parse[1] != ']') {
3857                 RExC_parse++;
3858
3859                 /* a bad range like \w-, [:word:]- ? */
3860                 if (namedclass > OOB_NAMEDCLASS) {
3861                     if (ckWARN(WARN_REGEXP))
3862                         vWARN4(RExC_parse,
3863                                "False [] range \"%*.*s\"",
3864                                RExC_parse - rangebegin,
3865                                RExC_parse - rangebegin,
3866                                rangebegin);
3867                     if (!SIZE_ONLY)
3868                         ANYOF_BITMAP_SET(ret, '-');
3869                 } else
3870                     range = 1;  /* yeah, it's a range! */
3871                 continue;       /* but do it the next time */
3872             }
3873         }
3874
3875         /* now is the next time */
3876         if (!SIZE_ONLY) {
3877             IV i;
3878
3879             if (prevvalue < 256) {
3880                 IV ceilvalue = value < 256 ? value : 255;
3881
3882 #ifdef EBCDIC
3883                 if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3884                     (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3885                 {
3886                     if (isLOWER(prevvalue)) {
3887                         for (i = prevvalue; i <= ceilvalue; i++)
3888                             if (isLOWER(i))
3889                                 ANYOF_BITMAP_SET(ret, i);
3890                     } else {
3891                         for (i = prevvalue; i <= ceilvalue; i++)
3892                             if (isUPPER(i))
3893                                 ANYOF_BITMAP_SET(ret, i);
3894                     }
3895                 }
3896                 else
3897 #endif
3898                     for (i = prevvalue; i <= ceilvalue; i++)
3899                         ANYOF_BITMAP_SET(ret, i);
3900           }
3901           if (value > 255) {
3902                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3903                 if (prevvalue < value)
3904                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3905                                    (UV)prevvalue, (UV)value);
3906                 else if (prevvalue == value)
3907                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3908                                    (UV)value);
3909             }
3910         }
3911
3912         range = 0; /* this range (if it was one) is done now */
3913     }
3914
3915     if (need_class) {
3916         ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3917         if (SIZE_ONLY)
3918             RExC_size += ANYOF_CLASS_ADD_SKIP;
3919         else
3920             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3921     }
3922
3923     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3924     if (!SIZE_ONLY &&
3925          /* If the only flag is folding (plus possibly inversion). */
3926         ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
3927        ) {
3928         for (value = 0; value < 256; ++value) {
3929             if (ANYOF_BITMAP_TEST(ret, value)) {
3930                 IV fold = PL_fold[value];
3931
3932                 if (fold != value)
3933                     ANYOF_BITMAP_SET(ret, fold);
3934             }
3935         }
3936         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3937     }
3938
3939     /* optimize inverted simple patterns (e.g. [^a-z]) */
3940     if (!SIZE_ONLY && optimize_invert &&
3941         /* If the only flag is inversion. */
3942         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3943         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3944             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3945         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3946     }
3947
3948     if (!SIZE_ONLY) {
3949         AV *av = newAV();
3950         SV *rv;
3951
3952         av_store(av, 0, listsv);
3953         av_store(av, 1, NULL);
3954         rv = newRV_noinc((SV*)av);
3955         n = add_data(pRExC_state, 1, "s");
3956         RExC_rx->data->data[n] = (void*)rv;
3957         ARG_SET(ret, n);
3958     }
3959
3960     return ret;
3961 }
3962
3963 STATIC char*
3964 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3965 {
3966     char* retval = RExC_parse++;
3967
3968     for (;;) {
3969         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3970                 RExC_parse[2] == '#') {
3971             while (*RExC_parse && *RExC_parse != ')')
3972                 RExC_parse++;
3973             RExC_parse++;
3974             continue;
3975         }
3976         if (RExC_flags16 & PMf_EXTENDED) {
3977             if (isSPACE(*RExC_parse)) {
3978                 RExC_parse++;
3979                 continue;
3980             }
3981             else if (*RExC_parse == '#') {
3982                 while (*RExC_parse && *RExC_parse != '\n')
3983                     RExC_parse++;
3984                 RExC_parse++;
3985                 continue;
3986             }
3987         }
3988         return retval;
3989     }
3990 }
3991
3992 /*
3993 - reg_node - emit a node
3994 */
3995 STATIC regnode *                        /* Location. */
3996 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3997 {
3998     register regnode *ret;
3999     register regnode *ptr;
4000
4001     ret = RExC_emit;
4002     if (SIZE_ONLY) {
4003         SIZE_ALIGN(RExC_size);
4004         RExC_size += 1;
4005         return(ret);
4006     }
4007
4008     NODE_ALIGN_FILL(ret);
4009     ptr = ret;
4010     FILL_ADVANCE_NODE(ptr, op);
4011     if (RExC_offsets) {         /* MJD */
4012       MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n", 
4013               "reg_node", __LINE__, 
4014               reg_name[op],
4015               RExC_emit - RExC_emit_start > RExC_offsets[0] 
4016               ? "Overwriting end of array!\n" : "OK",
4017               RExC_emit - RExC_emit_start,
4018               RExC_parse - RExC_start,
4019               RExC_offsets[0])); 
4020       Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
4021     }
4022             
4023     RExC_emit = ptr;
4024
4025     return(ret);
4026 }
4027
4028 /*
4029 - reganode - emit a node with an argument
4030 */
4031 STATIC regnode *                        /* Location. */
4032 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
4033 {
4034     register regnode *ret;
4035     register regnode *ptr;
4036
4037     ret = RExC_emit;
4038     if (SIZE_ONLY) {
4039         SIZE_ALIGN(RExC_size);
4040         RExC_size += 2;
4041         return(ret);
4042     }
4043
4044     NODE_ALIGN_FILL(ret);
4045     ptr = ret;
4046     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
4047     if (RExC_offsets) {         /* MJD */
4048       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4049               "reganode",
4050               RExC_emit - RExC_emit_start > RExC_offsets[0] ? 
4051               "Overwriting end of array!\n" : "OK",
4052               RExC_emit - RExC_emit_start,
4053               RExC_parse - RExC_start,
4054               RExC_offsets[0])); 
4055       Set_Cur_Node_Offset;
4056     }
4057             
4058     RExC_emit = ptr;
4059
4060     return(ret);
4061 }
4062
4063 /*
4064 - reguni - emit (if appropriate) a Unicode character
4065 */
4066 STATIC void
4067 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4068 {
4069     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4070 }
4071
4072 /*
4073 - reginsert - insert an operator in front of already-emitted operand
4074 *
4075 * Means relocating the operand.
4076 */
4077 STATIC void
4078 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4079 {
4080     register regnode *src;
4081     register regnode *dst;
4082     register regnode *place;
4083     register int offset = regarglen[(U8)op];
4084
4085 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4086
4087     if (SIZE_ONLY) {
4088         RExC_size += NODE_STEP_REGNODE + offset;
4089         return;
4090     }
4091
4092     src = RExC_emit;
4093     RExC_emit += NODE_STEP_REGNODE + offset;
4094     dst = RExC_emit;
4095     while (src > opnd) {
4096         StructCopy(--src, --dst, regnode);
4097         if (RExC_offsets) {     /* MJD 20010112 */
4098           MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n", 
4099                   "reg_insert",
4100                   dst - RExC_emit_start > RExC_offsets[0] 
4101                   ? "Overwriting end of array!\n" : "OK",
4102                   src - RExC_emit_start,
4103                   dst - RExC_emit_start,
4104                   RExC_offsets[0])); 
4105           Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4106           Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4107         }
4108     }
4109     
4110
4111     place = opnd;               /* Op node, where operand used to be. */
4112     if (RExC_offsets) {         /* MJD */
4113       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4114               "reginsert",
4115               place - RExC_emit_start > RExC_offsets[0] 
4116               ? "Overwriting end of array!\n" : "OK",
4117               place - RExC_emit_start,
4118               RExC_parse - RExC_start,
4119               RExC_offsets[0])); 
4120       Set_Node_Offset(place, RExC_parse);
4121     }
4122     src = NEXTOPER(place);
4123     FILL_ADVANCE_NODE(place, op);
4124     Zero(src, offset, regnode);
4125 }
4126
4127 /*
4128 - regtail - set the next-pointer at the end of a node chain of p to val.
4129 */
4130 STATIC void
4131 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4132 {
4133     register regnode *scan;
4134     register regnode *temp;
4135
4136     if (SIZE_ONLY)
4137         return;
4138
4139     /* Find last node. */
4140     scan = p;
4141     for (;;) {
4142         temp = regnext(scan);
4143         if (temp == NULL)
4144             break;
4145         scan = temp;
4146     }
4147
4148     if (reg_off_by_arg[OP(scan)]) {
4149         ARG_SET(scan, val - scan);
4150     }
4151     else {
4152         NEXT_OFF(scan) = val - scan;
4153     }
4154 }
4155
4156 /*
4157 - regoptail - regtail on operand of first argument; nop if operandless
4158 */
4159 STATIC void
4160 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4161 {
4162     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4163     if (p == NULL || SIZE_ONLY)
4164         return;
4165     if (PL_regkind[(U8)OP(p)] == BRANCH) {
4166         regtail(pRExC_state, NEXTOPER(p), val);
4167     }
4168     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4169         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4170     }
4171     else
4172         return;
4173 }
4174
4175 /*
4176  - regcurly - a little FSA that accepts {\d+,?\d*}
4177  */
4178 STATIC I32
4179 S_regcurly(pTHX_ register char *s)
4180 {
4181     if (*s++ != '{')
4182         return FALSE;
4183     if (!isDIGIT(*s))
4184         return FALSE;
4185     while (isDIGIT(*s))
4186         s++;
4187     if (*s == ',')
4188         s++;
4189     while (isDIGIT(*s))
4190         s++;
4191     if (*s != '}')
4192         return FALSE;
4193     return TRUE;
4194 }
4195
4196
4197 #ifdef DEBUGGING
4198
4199 STATIC regnode *
4200 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4201 {
4202     register U8 op = EXACT;     /* Arbitrary non-END op. */
4203     register regnode *next;
4204
4205     while (op != END && (!last || node < last)) {
4206         /* While that wasn't END last time... */
4207
4208         NODE_ALIGN(node);
4209         op = OP(node);
4210         if (op == CLOSE)
4211             l--;        
4212         next = regnext(node);
4213         /* Where, what. */
4214         if (OP(node) == OPTIMIZED)
4215             goto after_print;
4216         regprop(sv, node);
4217         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4218                       (int)(2*l + 1), "", SvPVX(sv));
4219         if (next == NULL)               /* Next ptr. */
4220             PerlIO_printf(Perl_debug_log, "(0)");
4221         else
4222             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4223         (void)PerlIO_putc(Perl_debug_log, '\n');
4224       after_print:
4225         if (PL_regkind[(U8)op] == BRANCHJ) {
4226             register regnode *nnode = (OP(next) == LONGJMP
4227                                        ? regnext(next)
4228                                        : next);
4229             if (last && nnode > last)
4230                 nnode = last;
4231             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4232         }
4233         else if (PL_regkind[(U8)op] == BRANCH) {
4234             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4235         }
4236         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4237             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4238                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4239         }
4240         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4241             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4242                              next, sv, l + 1);
4243         }
4244         else if ( op == PLUS || op == STAR) {
4245             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4246         }
4247         else if (op == ANYOF) {
4248             /* arglen 1 + class block */
4249             node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4250                     ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4251             node = NEXTOPER(node);
4252         }
4253         else if (PL_regkind[(U8)op] == EXACT) {
4254             /* Literal string, where present. */
4255             node += NODE_SZ_STR(node) - 1;
4256             node = NEXTOPER(node);
4257         }
4258         else {
4259             node = NEXTOPER(node);
4260             node += regarglen[(U8)op];
4261         }
4262         if (op == CURLYX || op == OPEN)
4263             l++;
4264         else if (op == WHILEM)
4265             l--;
4266     }
4267     return node;
4268 }
4269
4270 #endif  /* DEBUGGING */
4271
4272 /*
4273  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4274  */
4275 void
4276 Perl_regdump(pTHX_ regexp *r)
4277 {
4278 #ifdef DEBUGGING
4279     SV *sv = sv_newmortal();
4280
4281     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4282
4283     /* Header fields of interest. */
4284     if (r->anchored_substr)
4285         PerlIO_printf(Perl_debug_log,
4286                       "anchored `%s%.*s%s'%s at %"IVdf" ",
4287                       PL_colors[0],
4288                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4289                       SvPVX(r->anchored_substr),
4290                       PL_colors[1],
4291                       SvTAIL(r->anchored_substr) ? "$" : "",
4292                       (IV)r->anchored_offset);
4293     if (r->float_substr)
4294         PerlIO_printf(Perl_debug_log,
4295                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4296                       PL_colors[0],
4297                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4298                       SvPVX(r->float_substr),
4299                       PL_colors[1],
4300                       SvTAIL(r->float_substr) ? "$" : "",
4301                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4302     if (r->check_substr)
4303         PerlIO_printf(Perl_debug_log,
4304                       r->check_substr == r->float_substr
4305                       ? "(checking floating" : "(checking anchored");
4306     if (r->reganch & ROPT_NOSCAN)
4307         PerlIO_printf(Perl_debug_log, " noscan");
4308     if (r->reganch & ROPT_CHECK_ALL)
4309         PerlIO_printf(Perl_debug_log, " isall");
4310     if (r->check_substr)
4311         PerlIO_printf(Perl_debug_log, ") ");
4312
4313     if (r->regstclass) {
4314         regprop(sv, r->regstclass);
4315         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4316     }
4317     if (r->reganch & ROPT_ANCH) {
4318         PerlIO_printf(Perl_debug_log, "anchored");
4319         if (r->reganch & ROPT_ANCH_BOL)
4320             PerlIO_printf(Perl_debug_log, "(BOL)");
4321         if (r->reganch & ROPT_ANCH_MBOL)
4322             PerlIO_printf(Perl_debug_log, "(MBOL)");
4323         if (r->reganch & ROPT_ANCH_SBOL)
4324             PerlIO_printf(Perl_debug_log, "(SBOL)");
4325         if (r->reganch & ROPT_ANCH_GPOS)
4326             PerlIO_printf(Perl_debug_log, "(GPOS)");
4327         PerlIO_putc(Perl_debug_log, ' ');
4328     }
4329     if (r->reganch & ROPT_GPOS_SEEN)
4330         PerlIO_printf(Perl_debug_log, "GPOS ");
4331     if (r->reganch & ROPT_SKIP)
4332         PerlIO_printf(Perl_debug_log, "plus ");
4333     if (r->reganch & ROPT_IMPLICIT)
4334         PerlIO_printf(Perl_debug_log, "implicit ");
4335     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4336     if (r->reganch & ROPT_EVAL_SEEN)
4337         PerlIO_printf(Perl_debug_log, "with eval ");
4338     PerlIO_printf(Perl_debug_log, "\n");
4339     if (r->offsets) {
4340       U32 i;
4341       U32 len = r->offsets[0];
4342       PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
4343       for (i = 1; i <= len; i++)
4344         PerlIO_printf(Perl_debug_log, "%"UVuf"[%"UVuf"] ", 
4345                       (UV)r->offsets[i*2-1], 
4346                       (UV)r->offsets[i*2]);
4347       PerlIO_printf(Perl_debug_log, "\n");
4348     }
4349 #endif  /* DEBUGGING */
4350 }
4351
4352 #ifdef DEBUGGING
4353
4354 STATIC void
4355 S_put_byte(pTHX_ SV *sv, int c)
4356 {
4357     if (isCNTRL(c) || c == 255 || !isPRINT(c))
4358         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4359     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4360         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4361     else
4362         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4363 }
4364
4365 #endif  /* DEBUGGING */
4366
4367 /*
4368 - regprop - printable representation of opcode
4369 */
4370 void
4371 Perl_regprop(pTHX_ SV *sv, regnode *o)
4372 {
4373 #ifdef DEBUGGING
4374     register int k;
4375
4376     sv_setpvn(sv, "", 0);
4377     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4378         /* It would be nice to FAIL() here, but this may be called from
4379            regexec.c, and it would be hard to supply pRExC_state. */
4380         Perl_croak(aTHX_ "Corrupted regexp opcode");
4381     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4382
4383     k = PL_regkind[(U8)OP(o)];
4384
4385     if (k == EXACT)
4386         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4387                        STR_LEN(o), STRING(o), PL_colors[1]);
4388     else if (k == CURLY) {
4389         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4390             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4391         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4392     }
4393     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4394         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4395     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4396         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4397     else if (k == LOGICAL)
4398         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4399     else if (k == ANYOF) {
4400         int i, rangestart = -1;
4401         U8 flags = ANYOF_FLAGS(o);
4402         const char * const anyofs[] = { /* Should be syncronized with
4403                                          * ANYOF_ #xdefines in regcomp.h */
4404             "\\w",
4405             "\\W",
4406             "\\s",
4407             "\\S",
4408             "\\d",
4409             "\\D",
4410             "[:alnum:]",
4411             "[:^alnum:]",
4412             "[:alpha:]",
4413             "[:^alpha:]",
4414             "[:ascii:]",
4415             "[:^ascii:]",
4416             "[:ctrl:]",
4417             "[:^ctrl:]",
4418             "[:graph:]",
4419             "[:^graph:]",
4420             "[:lower:]",
4421             "[:^lower:]",
4422             "[:print:]",
4423             "[:^print:]",
4424             "[:punct:]",
4425             "[:^punct:]",
4426             "[:upper:]",
4427             "[:^upper:]",
4428             "[:xdigit:]",
4429             "[:^xdigit:]",
4430             "[:space:]",
4431             "[:^space:]",
4432             "[:blank:]",
4433             "[:^blank:]"
4434         };
4435
4436         if (flags & ANYOF_LOCALE)
4437             sv_catpv(sv, "{loc}");
4438         if (flags & ANYOF_FOLD)
4439             sv_catpv(sv, "{i}");
4440         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4441         if (flags & ANYOF_INVERT)
4442             sv_catpv(sv, "^");
4443         for (i = 0; i <= 256; i++) {
4444             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4445                 if (rangestart == -1)
4446                     rangestart = i;
4447             } else if (rangestart != -1) {
4448                 if (i <= rangestart + 3)
4449                     for (; rangestart < i; rangestart++)
4450                         put_byte(sv, rangestart);
4451                 else {
4452                     put_byte(sv, rangestart);
4453                     sv_catpv(sv, "-");
4454                     put_byte(sv, i - 1);
4455                 }
4456                 rangestart = -1;
4457             }
4458         }
4459
4460         if (o->flags & ANYOF_CLASS)
4461             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4462                 if (ANYOF_CLASS_TEST(o,i))
4463                     sv_catpv(sv, anyofs[i]);
4464
4465         if (flags & ANYOF_UNICODE)
4466             sv_catpv(sv, "{unicode}");
4467         else if (flags & ANYOF_UNICODE_ALL)
4468             sv_catpv(sv, "{all-unicode}");
4469
4470         {
4471             SV *lv;
4472             SV *sw = regclass_swash(o, FALSE, &lv);
4473         
4474             if (lv) {
4475                 if (sw) {
4476                     UV i;
4477                     U8 s[UTF8_MAXLEN+1];
4478                 
4479                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4480                         U8 *e = uvchr_to_utf8(s, i);
4481                         
4482                         if (i < 256 && swash_fetch(sw, s, TRUE)) {
4483                             if (rangestart == -1)
4484                                 rangestart = i;
4485                         } else if (rangestart != -1) {
4486                             U8 *p;
4487                         
4488                             if (i <= rangestart + 3)
4489                                 for (; rangestart < i; rangestart++) {
4490                                     for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4491                                         put_byte(sv, *p);
4492                                 }
4493                             else {
4494                                 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4495                                     put_byte(sv, *p);
4496                                 sv_catpv(sv, "-");
4497                                     for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4498                                         put_byte(sv, *p);
4499                                 }
4500                                 rangestart = -1;
4501                             }
4502                         }
4503                         
4504                     sv_catpv(sv, "..."); /* et cetera */
4505                 }
4506
4507                 {
4508                     char *s = savepv(SvPVX(lv));
4509                     char *origs = s;
4510                 
4511                     while(*s && *s != '\n') s++;
4512                 
4513                     if (*s == '\n') {
4514                         char *t = ++s;
4515                         
4516                         while (*s) {
4517                             if (*s == '\n')
4518                                 *s = ' ';
4519                             s++;
4520                         }
4521                         if (s[-1] == ' ')
4522                             s[-1] = 0;
4523                         
4524                         sv_catpv(sv, t);
4525                     }
4526                 
4527                     Safefree(origs);
4528                 }
4529             }
4530         }
4531
4532         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4533     }
4534     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4535         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4536 #endif  /* DEBUGGING */
4537 }
4538
4539 SV *
4540 Perl_re_intuit_string(pTHX_ regexp *prog)
4541 {                               /* Assume that RE_INTUIT is set */
4542     DEBUG_r(
4543         {   STRLEN n_a;
4544             char *s = SvPV(prog->check_substr,n_a);
4545
4546             if (!PL_colorset) reginitcolors();
4547             PerlIO_printf(Perl_debug_log,
4548                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4549                       PL_colors[4],PL_colors[5],PL_colors[0],
4550                       s,
4551                       PL_colors[1],
4552                       (strlen(s) > 60 ? "..." : ""));
4553         } );
4554
4555     return prog->check_substr;
4556 }
4557
4558 void
4559 Perl_pregfree(pTHX_ struct regexp *r)
4560 {
4561     DEBUG_r(if (!PL_colorset) reginitcolors());
4562
4563     if (!r || (--r->refcnt > 0))
4564         return;
4565     DEBUG_r(PerlIO_printf(Perl_debug_log,
4566                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4567                       PL_colors[4],PL_colors[5],PL_colors[0],
4568                       r->precomp,
4569                       PL_colors[1],
4570                       (strlen(r->precomp) > 60 ? "..." : "")));
4571
4572     if (r->precomp)
4573         Safefree(r->precomp);
4574     if (r->offsets)             /* 20010421 MJD */
4575         Safefree(r->offsets);
4576     if (RX_MATCH_COPIED(r))
4577         Safefree(r->subbeg);
4578     if (r->substrs) {
4579         if (r->anchored_substr)
4580             SvREFCNT_dec(r->anchored_substr);
4581         if (r->float_substr)
4582             SvREFCNT_dec(r->float_substr);
4583         Safefree(r->substrs);
4584     }
4585     if (r->data) {
4586         int n = r->data->count;
4587         AV* new_comppad = NULL;
4588         AV* old_comppad;
4589         SV** old_curpad;
4590
4591         while (--n >= 0) {
4592           /* If you add a ->what type here, update the comment in regcomp.h */
4593             switch (r->data->what[n]) {
4594             case 's':
4595                 SvREFCNT_dec((SV*)r->data->data[n]);
4596                 break;
4597             case 'f':
4598                 Safefree(r->data->data[n]);
4599                 break;
4600             case 'p':
4601                 new_comppad = (AV*)r->data->data[n];
4602                 break;
4603             case 'o':
4604                 if (new_comppad == NULL)
4605                     Perl_croak(aTHX_ "panic: pregfree comppad");
4606                 old_comppad = PL_comppad;
4607                 old_curpad = PL_curpad;
4608                 /* Watch out for global destruction's random ordering. */
4609                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4610                     PL_comppad = new_comppad;
4611                     PL_curpad = AvARRAY(new_comppad);
4612                 }
4613                 else
4614                     PL_curpad = NULL;
4615
4616                 if (!OpREFCNT_dec((OP_4tree*)r->data->data[n])) {
4617                     op_free((OP_4tree*)r->data->data[n]);
4618                 }
4619
4620                 PL_comppad = old_comppad;
4621                 PL_curpad = old_curpad;
4622                 SvREFCNT_dec((SV*)new_comppad);
4623                 new_comppad = NULL;
4624                 break;
4625             case 'n':
4626                 break;
4627             default:
4628                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4629             }
4630         }
4631         Safefree(r->data->what);
4632         Safefree(r->data);
4633     }
4634     Safefree(r->startp);
4635     Safefree(r->endp);
4636     Safefree(r);
4637 }
4638
4639 /*
4640  - regnext - dig the "next" pointer out of a node
4641  *
4642  * [Note, when REGALIGN is defined there are two places in regmatch()
4643  * that bypass this code for speed.]
4644  */
4645 regnode *
4646 Perl_regnext(pTHX_ register regnode *p)
4647 {
4648     register I32 offset;
4649
4650     if (p == &PL_regdummy)
4651         return(NULL);
4652
4653     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4654     if (offset == 0)
4655         return(NULL);
4656
4657     return(p+offset);
4658 }
4659
4660 STATIC void     
4661 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4662 {
4663     va_list args;
4664     STRLEN l1 = strlen(pat1);
4665     STRLEN l2 = strlen(pat2);
4666     char buf[512];
4667     SV *msv;
4668     char *message;
4669
4670     if (l1 > 510)
4671         l1 = 510;
4672     if (l1 + l2 > 510)
4673         l2 = 510 - l1;
4674     Copy(pat1, buf, l1 , char);
4675     Copy(pat2, buf + l1, l2 , char);
4676     buf[l1 + l2] = '\n';
4677     buf[l1 + l2 + 1] = '\0';
4678 #ifdef I_STDARG
4679     /* ANSI variant takes additional second argument */
4680     va_start(args, pat2);
4681 #else
4682     va_start(args);
4683 #endif
4684     msv = vmess(buf, &args);
4685     va_end(args);
4686     message = SvPV(msv,l1);
4687     if (l1 > 512)
4688         l1 = 512;
4689     Copy(message, buf, l1 , char);
4690     buf[l1] = '\0';                     /* Overwrite \n */
4691     Perl_croak(aTHX_ "%s", buf);
4692 }
4693
4694 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4695
4696 void
4697 Perl_save_re_context(pTHX)
4698 {
4699 #if 0
4700     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4701     SAVEI32(RExC_npar);         /* () count. */
4702     SAVEI32(RExC_size);         /* Code size. */
4703     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4704     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4705     SAVEI32(RExC_seen);         /* from regcomp.c */
4706     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4707     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4708     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4709     SAVEPPTR(RExC_end);         /* End of input for compile */
4710     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4711 #endif
4712
4713     SAVEI32(PL_reg_flags);              /* from regexec.c */
4714     SAVEPPTR(PL_bostr);
4715     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4716     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4717     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4718     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4719     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4720     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4721     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4722     SAVEGENERICPV(PL_reg_start_tmp);            /* from regexec.c */
4723     PL_reg_start_tmp = 0;
4724     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4725     PL_reg_start_tmpl = 0;
4726     SAVEVPTR(PL_regdata);
4727     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4728     SAVEI32(PL_regnarrate);             /* from regexec.c */
4729     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4730     SAVEINT(PL_regindent);              /* from regexec.c */
4731     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4732     SAVEVPTR(PL_curcop);
4733     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4734     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4735     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4736     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4737     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4738     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4739     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4740     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4741     SAVEI32(PL_regnpar);                /* () count. */
4742     SAVEI32(PL_regsize);                /* from regexec.c */
4743 #ifdef DEBUGGING
4744     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */
4745 #endif
4746 }
4747
4748 #ifdef PERL_OBJECT
4749 #include "XSUB.h"
4750 #undef this
4751 #define this pPerl
4752 #endif
4753
4754 static void
4755 clear_re(pTHXo_ void *r)
4756 {
4757     ReREFCNT_dec((regexp *)r);
4758 }
4759