Integrate mainline
[p5sagit/p5-mst-13.2.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, *nxt2;
1103
1104                     /* Skip open. */
1105                     nxt = regnext(nxt);
1106                     if (!strchr((char*)PL_simple,OP(nxt))
1107                         && !(PL_regkind[(U8)OP(nxt)] == EXACT
1108                              && STR_LEN(nxt) == 1))
1109                         goto nogo;
1110                     nxt2 = nxt;
1111                     nxt = regnext(nxt);
1112                     if (OP(nxt) != CLOSE)
1113                         goto nogo;
1114                     /* Now we know that nxt2 is the only contents: */
1115                     oscan->flags = ARG(nxt);
1116                     OP(oscan) = CURLYN;
1117                     OP(nxt1) = NOTHING; /* was OPEN. */
1118 #ifdef DEBUGGING
1119                     OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1120                     NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1121                     NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1122                     OP(nxt) = OPTIMIZED;        /* was CLOSE. */
1123                     OP(nxt + 1) = OPTIMIZED; /* was count. */
1124                     NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1125 #endif
1126                 }
1127               nogo:
1128
1129                 /* Try optimization CURLYX => CURLYM. */
1130                 if (  OP(oscan) == CURLYX && data
1131                       && !(data->flags & SF_HAS_PAR)
1132                       && !(data->flags & SF_HAS_EVAL)
1133                       && !deltanext  ) {
1134                     /* XXXX How to optimize if data == 0? */
1135                     /* Optimize to a simpler form.  */
1136                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1137                     regnode *nxt2;
1138
1139                     OP(oscan) = CURLYM;
1140                     while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1141                             && (OP(nxt2) != WHILEM))
1142                         nxt = nxt2;
1143                     OP(nxt2)  = SUCCEED; /* Whas WHILEM */
1144                     /* Need to optimize away parenths. */
1145                     if (data->flags & SF_IN_PAR) {
1146                         /* Set the parenth number.  */
1147                         regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1148
1149                         if (OP(nxt) != CLOSE)
1150                             FAIL("Panic opt close");
1151                         oscan->flags = ARG(nxt);
1152                         OP(nxt1) = OPTIMIZED;   /* was OPEN. */
1153                         OP(nxt) = OPTIMIZED;    /* was CLOSE. */
1154 #ifdef DEBUGGING
1155                         OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1156                         OP(nxt + 1) = OPTIMIZED; /* was count. */
1157                         NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1158                         NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1159 #endif
1160 #if 0
1161                         while ( nxt1 && (OP(nxt1) != WHILEM)) {
1162                             regnode *nnxt = regnext(nxt1);
1163                         
1164                             if (nnxt == nxt) {
1165                                 if (reg_off_by_arg[OP(nxt1)])
1166                                     ARG_SET(nxt1, nxt2 - nxt1);
1167                                 else if (nxt2 - nxt1 < U16_MAX)
1168                                     NEXT_OFF(nxt1) = nxt2 - nxt1;
1169                                 else
1170                                     OP(nxt) = NOTHING;  /* Cannot beautify */
1171                             }
1172                             nxt1 = nnxt;
1173                         }
1174 #endif
1175                         /* Optimize again: */
1176                         study_chunk(pRExC_state, &nxt1, &deltanext, nxt,
1177                                     NULL, 0);
1178                     }
1179                     else
1180                         oscan->flags = 0;
1181                 }
1182                 else if ((OP(oscan) == CURLYX)
1183                          && (flags & SCF_WHILEM_VISITED_POS)
1184                          /* See the comment on a similar expression above.
1185                             However, this time it not a subexpression
1186                             we care about, but the expression itself. */
1187                          && (maxcount == REG_INFTY)
1188                          && data && ++data->whilem_c < 16) {
1189                     /* This stays as CURLYX, we can put the count/of pair. */
1190                     /* Find WHILEM (as in regexec.c) */
1191                     regnode *nxt = oscan + NEXT_OFF(oscan);
1192
1193                     if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1194                         nxt += ARG(nxt);
1195                     PREVOPER(nxt)->flags = data->whilem_c
1196                         | (RExC_whilem_seen << 4); /* On WHILEM */
1197                 }
1198                 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
1199                     pars++;
1200                 if (flags & SCF_DO_SUBSTR) {
1201                     SV *last_str = Nullsv;
1202                     int counted = mincount != 0;
1203
1204                     if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1205                         I32 b = pos_before >= data->last_start_min
1206                             ? pos_before : data->last_start_min;
1207                         STRLEN l;
1208                         char *s = SvPV(data->last_found, l);
1209                         I32 old = b - data->last_start_min;
1210
1211                         if (UTF)
1212                             old = utf8_hop((U8*)s, old) - (U8*)s;
1213                         
1214                         l -= old;
1215                         /* Get the added string: */
1216                         last_str = newSVpvn(s  + old, l);
1217                         if (deltanext == 0 && pos_before == b) {
1218                             /* What was added is a constant string */
1219                             if (mincount > 1) {
1220                                 SvGROW(last_str, (mincount * l) + 1);
1221                                 repeatcpy(SvPVX(last_str) + l,
1222                                           SvPVX(last_str), l, mincount - 1);
1223                                 SvCUR(last_str) *= mincount;
1224                                 /* Add additional parts. */
1225                                 SvCUR_set(data->last_found,
1226                                           SvCUR(data->last_found) - l);
1227                                 sv_catsv(data->last_found, last_str);
1228                                 data->last_end += l * (mincount - 1);
1229                             }
1230                         } else {
1231                             /* start offset must point into the last copy */
1232                             data->last_start_min += minnext * (mincount - 1);
1233                             data->last_start_max += is_inf ? 0 : (maxcount - 1)
1234                                 * (minnext + data->pos_delta);
1235                         }
1236                     }
1237                     /* It is counted once already... */
1238                     data->pos_min += minnext * (mincount - counted);
1239                     data->pos_delta += - counted * deltanext +
1240                         (minnext + deltanext) * maxcount - minnext * mincount;
1241                     if (mincount != maxcount) {
1242                          /* Cannot extend fixed substrings found inside
1243                             the group.  */
1244                         scan_commit(pRExC_state,data);
1245                         if (mincount && last_str) {
1246                             sv_setsv(data->last_found, last_str);
1247                             data->last_end = data->pos_min;
1248                             data->last_start_min =
1249                                 data->pos_min - CHR_SVLEN(last_str);
1250                             data->last_start_max = is_inf
1251                                 ? I32_MAX
1252                                 : data->pos_min + data->pos_delta
1253                                 - CHR_SVLEN(last_str);
1254                         }
1255                         data->longest = &(data->longest_float);
1256                     }
1257                     SvREFCNT_dec(last_str);
1258                 }
1259                 if (data && (fl & SF_HAS_EVAL))
1260                     data->flags |= SF_HAS_EVAL;
1261               optimize_curly_tail:
1262                 if (OP(oscan) != CURLYX) {
1263                     while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1264                            && NEXT_OFF(next))
1265                         NEXT_OFF(oscan) += NEXT_OFF(next);
1266                 }
1267                 continue;
1268             default:                    /* REF and CLUMP only? */
1269                 if (flags & SCF_DO_SUBSTR) {
1270                     scan_commit(pRExC_state,data);      /* Cannot expect anything... */
1271                     data->longest = &(data->longest_float);
1272                 }
1273                 is_inf = is_inf_internal = 1;
1274                 if (flags & SCF_DO_STCLASS_OR)
1275                     cl_anything(pRExC_state, data->start_class);
1276                 flags &= ~SCF_DO_STCLASS;
1277                 break;
1278             }
1279         }
1280         else if (strchr((char*)PL_simple,OP(scan))) {
1281             int value = 0;
1282
1283             if (flags & SCF_DO_SUBSTR) {
1284                 scan_commit(pRExC_state,data);
1285                 data->pos_min++;
1286             }
1287             min++;
1288             if (flags & SCF_DO_STCLASS) {
1289                 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1290
1291                 /* Some of the logic below assumes that switching
1292                    locale on will only add false positives. */
1293                 switch (PL_regkind[(U8)OP(scan)]) {
1294                 case SANY:
1295                 default:
1296                   do_default:
1297                     /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1298                     if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1299                         cl_anything(pRExC_state, data->start_class);
1300                     break;
1301                 case REG_ANY:
1302                     if (OP(scan) == SANY)
1303                         goto do_default;
1304                     if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1305                         value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1306                                  || (data->start_class->flags & ANYOF_CLASS));
1307                         cl_anything(pRExC_state, data->start_class);
1308                     }
1309                     if (flags & SCF_DO_STCLASS_AND || !value)
1310                         ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1311                     break;
1312                 case ANYOF:
1313                     if (flags & SCF_DO_STCLASS_AND)
1314                         cl_and(data->start_class,
1315                                (struct regnode_charclass_class*)scan);
1316                     else
1317                         cl_or(pRExC_state, data->start_class,
1318                               (struct regnode_charclass_class*)scan);
1319                     break;
1320                 case ALNUM:
1321                     if (flags & SCF_DO_STCLASS_AND) {
1322                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1323                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1324                             for (value = 0; value < 256; value++)
1325                                 if (!isALNUM(value))
1326                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1327                         }
1328                     }
1329                     else {
1330                         if (data->start_class->flags & ANYOF_LOCALE)
1331                             ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1332                         else {
1333                             for (value = 0; value < 256; value++)
1334                                 if (isALNUM(value))
1335                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1336                         }
1337                     }
1338                     break;
1339                 case ALNUML:
1340                     if (flags & SCF_DO_STCLASS_AND) {
1341                         if (data->start_class->flags & ANYOF_LOCALE)
1342                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1343                     }
1344                     else {
1345                         ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1346                         data->start_class->flags |= ANYOF_LOCALE;
1347                     }
1348                     break;
1349                 case NALNUM:
1350                     if (flags & SCF_DO_STCLASS_AND) {
1351                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1352                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1353                             for (value = 0; value < 256; value++)
1354                                 if (isALNUM(value))
1355                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1356                         }
1357                     }
1358                     else {
1359                         if (data->start_class->flags & ANYOF_LOCALE)
1360                             ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1361                         else {
1362                             for (value = 0; value < 256; value++)
1363                                 if (!isALNUM(value))
1364                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1365                         }
1366                     }
1367                     break;
1368                 case NALNUML:
1369                     if (flags & SCF_DO_STCLASS_AND) {
1370                         if (data->start_class->flags & ANYOF_LOCALE)
1371                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1372                     }
1373                     else {
1374                         data->start_class->flags |= ANYOF_LOCALE;
1375                         ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1376                     }
1377                     break;
1378                 case SPACE:
1379                     if (flags & SCF_DO_STCLASS_AND) {
1380                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1381                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1382                             for (value = 0; value < 256; value++)
1383                                 if (!isSPACE(value))
1384                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1385                         }
1386                     }
1387                     else {
1388                         if (data->start_class->flags & ANYOF_LOCALE)
1389                             ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1390                         else {
1391                             for (value = 0; value < 256; value++)
1392                                 if (isSPACE(value))
1393                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1394                         }
1395                     }
1396                     break;
1397                 case SPACEL:
1398                     if (flags & SCF_DO_STCLASS_AND) {
1399                         if (data->start_class->flags & ANYOF_LOCALE)
1400                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1401                     }
1402                     else {
1403                         data->start_class->flags |= ANYOF_LOCALE;
1404                         ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1405                     }
1406                     break;
1407                 case NSPACE:
1408                     if (flags & SCF_DO_STCLASS_AND) {
1409                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1410                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1411                             for (value = 0; value < 256; value++)
1412                                 if (isSPACE(value))
1413                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1414                         }
1415                     }
1416                     else {
1417                         if (data->start_class->flags & ANYOF_LOCALE)
1418                             ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1419                         else {
1420                             for (value = 0; value < 256; value++)
1421                                 if (!isSPACE(value))
1422                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1423                         }
1424                     }
1425                     break;
1426                 case NSPACEL:
1427                     if (flags & SCF_DO_STCLASS_AND) {
1428                         if (data->start_class->flags & ANYOF_LOCALE) {
1429                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1430                             for (value = 0; value < 256; value++)
1431                                 if (!isSPACE(value))
1432                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1433                         }
1434                     }
1435                     else {
1436                         data->start_class->flags |= ANYOF_LOCALE;
1437                         ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1438                     }
1439                     break;
1440                 case DIGIT:
1441                     if (flags & SCF_DO_STCLASS_AND) {
1442                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1443                         for (value = 0; value < 256; value++)
1444                             if (!isDIGIT(value))
1445                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1446                     }
1447                     else {
1448                         if (data->start_class->flags & ANYOF_LOCALE)
1449                             ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1450                         else {
1451                             for (value = 0; value < 256; value++)
1452                                 if (isDIGIT(value))
1453                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1454                         }
1455                     }
1456                     break;
1457                 case NDIGIT:
1458                     if (flags & SCF_DO_STCLASS_AND) {
1459                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1460                         for (value = 0; value < 256; value++)
1461                             if (isDIGIT(value))
1462                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1463                     }
1464                     else {
1465                         if (data->start_class->flags & ANYOF_LOCALE)
1466                             ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1467                         else {
1468                             for (value = 0; value < 256; value++)
1469                                 if (!isDIGIT(value))
1470                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1471                         }
1472                     }
1473                     break;
1474                 }
1475                 if (flags & SCF_DO_STCLASS_OR)
1476                     cl_and(data->start_class, &and_with);
1477                 flags &= ~SCF_DO_STCLASS;
1478             }
1479         }
1480         else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1481             data->flags |= (OP(scan) == MEOL
1482                             ? SF_BEFORE_MEOL
1483                             : SF_BEFORE_SEOL);
1484         }
1485         else if (  PL_regkind[(U8)OP(scan)] == BRANCHJ
1486                  /* Lookbehind, or need to calculate parens/evals/stclass: */
1487                    && (scan->flags || data || (flags & SCF_DO_STCLASS))
1488                    && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1489             /* Lookahead/lookbehind */
1490             I32 deltanext, minnext, fake = 0;
1491             regnode *nscan;
1492             struct regnode_charclass_class intrnl;
1493             int f = 0;
1494
1495             data_fake.flags = 0;
1496             if (data) {         
1497                 data_fake.whilem_c = data->whilem_c;
1498                 data_fake.last_closep = data->last_closep;
1499             }
1500             else
1501                 data_fake.last_closep = &fake;
1502             if ( flags & SCF_DO_STCLASS && !scan->flags
1503                  && OP(scan) == IFMATCH ) { /* Lookahead */
1504                 cl_init(pRExC_state, &intrnl);
1505                 data_fake.start_class = &intrnl;
1506                 f |= SCF_DO_STCLASS_AND;
1507             }
1508             if (flags & SCF_WHILEM_VISITED_POS)
1509                 f |= SCF_WHILEM_VISITED_POS;
1510             next = regnext(scan);
1511             nscan = NEXTOPER(NEXTOPER(scan));
1512             minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1513             if (scan->flags) {
1514                 if (deltanext) {
1515                     vFAIL("Variable length lookbehind not implemented");
1516                 }
1517                 else if (minnext > U8_MAX) {
1518                     vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1519                 }
1520                 scan->flags = minnext;
1521             }
1522             if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1523                 pars++;
1524             if (data && (data_fake.flags & SF_HAS_EVAL))
1525                 data->flags |= SF_HAS_EVAL;
1526             if (data)
1527                 data->whilem_c = data_fake.whilem_c;
1528             if (f & SCF_DO_STCLASS_AND) {
1529                 int was = (data->start_class->flags & ANYOF_EOS);
1530
1531                 cl_and(data->start_class, &intrnl);
1532                 if (was)
1533                     data->start_class->flags |= ANYOF_EOS;
1534             }
1535         }
1536         else if (OP(scan) == OPEN) {
1537             pars++;
1538         }
1539         else if (OP(scan) == CLOSE) {
1540             if (ARG(scan) == is_par) {
1541                 next = regnext(scan);
1542
1543                 if ( next && (OP(next) != WHILEM) && next < last)
1544                     is_par = 0;         /* Disable optimization */
1545             }
1546             if (data)
1547                 *(data->last_closep) = ARG(scan);
1548         }
1549         else if (OP(scan) == EVAL) {
1550                 if (data)
1551                     data->flags |= SF_HAS_EVAL;
1552         }
1553         else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1554                 if (flags & SCF_DO_SUBSTR) {
1555                     scan_commit(pRExC_state,data);
1556                     data->longest = &(data->longest_float);
1557                 }
1558                 is_inf = is_inf_internal = 1;
1559                 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1560                     cl_anything(pRExC_state, data->start_class);
1561                 flags &= ~SCF_DO_STCLASS;
1562         }
1563         /* Else: zero-length, ignore. */
1564         scan = regnext(scan);
1565     }
1566
1567   finish:
1568     *scanp = scan;
1569     *deltap = is_inf_internal ? I32_MAX : delta;
1570     if (flags & SCF_DO_SUBSTR && is_inf)
1571         data->pos_delta = I32_MAX - data->pos_min;
1572     if (is_par > U8_MAX)
1573         is_par = 0;
1574     if (is_par && pars==1 && data) {
1575         data->flags |= SF_IN_PAR;
1576         data->flags &= ~SF_HAS_PAR;
1577     }
1578     else if (pars && data) {
1579         data->flags |= SF_HAS_PAR;
1580         data->flags &= ~SF_IN_PAR;
1581     }
1582     if (flags & SCF_DO_STCLASS_OR)
1583         cl_and(data->start_class, &and_with);
1584     return min;
1585 }
1586
1587 STATIC I32
1588 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1589 {
1590     if (RExC_rx->data) {
1591         Renewc(RExC_rx->data,
1592                sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1),
1593                char, struct reg_data);
1594         Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1595         RExC_rx->data->count += n;
1596     }
1597     else {
1598         Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1599              char, struct reg_data);
1600         New(1208, RExC_rx->data->what, n, U8);
1601         RExC_rx->data->count = n;
1602     }
1603     Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1604     return RExC_rx->data->count - n;
1605 }
1606
1607 void
1608 Perl_reginitcolors(pTHX)
1609 {
1610     int i = 0;
1611     char *s = PerlEnv_getenv("PERL_RE_COLORS");
1612         
1613     if (s) {
1614         PL_colors[0] = s = savepv(s);
1615         while (++i < 6) {
1616             s = strchr(s, '\t');
1617             if (s) {
1618                 *s = '\0';
1619                 PL_colors[i] = ++s;
1620             }
1621             else
1622                 PL_colors[i] = s = "";
1623         }
1624     } else {
1625         while (i < 6)
1626             PL_colors[i++] = "";
1627     }
1628     PL_colorset = 1;
1629 }
1630
1631
1632 /*
1633  - pregcomp - compile a regular expression into internal code
1634  *
1635  * We can't allocate space until we know how big the compiled form will be,
1636  * but we can't compile it (and thus know how big it is) until we've got a
1637  * place to put the code.  So we cheat:  we compile it twice, once with code
1638  * generation turned off and size counting turned on, and once "for real".
1639  * This also means that we don't allocate space until we are sure that the
1640  * thing really will compile successfully, and we never have to move the
1641  * code and thus invalidate pointers into it.  (Note that it has to be in
1642  * one piece because free() must be able to free it all.) [NB: not true in perl]
1643  *
1644  * Beware that the optimization-preparation code in here knows about some
1645  * of the structure of the compiled regexp.  [I'll say.]
1646  */
1647 regexp *
1648 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1649 {
1650     register regexp *r;
1651     regnode *scan;
1652     regnode *first;
1653     I32 flags;
1654     I32 minlen = 0;
1655     I32 sawplus = 0;
1656     I32 sawopen = 0;
1657     scan_data_t data;
1658     RExC_state_t RExC_state;
1659     RExC_state_t *pRExC_state = &RExC_state;
1660
1661     if (exp == NULL)
1662         FAIL("NULL regexp argument");
1663
1664     /* XXXX This looks very suspicious... */
1665     if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1666         RExC_utf8 = 1;
1667     else
1668         RExC_utf8 = 0;
1669
1670     RExC_precomp = exp;
1671     DEBUG_r(if (!PL_colorset) reginitcolors());
1672     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1673                       PL_colors[4],PL_colors[5],PL_colors[0],
1674                       (int)(xend - exp), RExC_precomp, PL_colors[1]));
1675     RExC_flags16 = pm->op_pmflags;
1676     RExC_sawback = 0;
1677
1678     RExC_seen = 0;
1679     RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1680     RExC_seen_evals = 0;
1681     RExC_extralen = 0;
1682
1683     /* First pass: determine size, legality. */
1684     RExC_parse = exp;
1685     RExC_start = exp;
1686     RExC_end = xend;
1687     RExC_naughty = 0;
1688     RExC_npar = 1;
1689     RExC_size = 0L;
1690     RExC_emit = &PL_regdummy;
1691     RExC_whilem_seen = 0;
1692 #if 0 /* REGC() is (currently) a NOP at the first pass.
1693        * Clever compilers notice this and complain. --jhi */
1694     REGC((U8)REG_MAGIC, (char*)RExC_emit);
1695 #endif
1696     if (reg(pRExC_state, 0, &flags) == NULL) {
1697         RExC_precomp = Nullch;
1698         return(NULL);
1699     }
1700     DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1701
1702     /* Small enough for pointer-storage convention?
1703        If extralen==0, this means that we will not need long jumps. */
1704     if (RExC_size >= 0x10000L && RExC_extralen)
1705         RExC_size += RExC_extralen;
1706     else
1707         RExC_extralen = 0;
1708     if (RExC_whilem_seen > 15)
1709         RExC_whilem_seen = 15;
1710
1711     /* Allocate space and initialize. */
1712     Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1713          char, regexp);
1714     if (r == NULL)
1715         FAIL("Regexp out of space");
1716
1717 #ifdef DEBUGGING
1718     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1719     Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1720 #endif
1721     r->refcnt = 1;
1722     r->prelen = xend - exp;
1723     r->precomp = savepvn(RExC_precomp, r->prelen);
1724     r->subbeg = NULL;
1725     r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1726     r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1727
1728     r->substrs = 0;                     /* Useful during FAIL. */
1729     r->startp = 0;                      /* Useful during FAIL. */
1730     r->endp = 0;                        /* Useful during FAIL. */
1731
1732     Newz(1304, r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
1733     if (r->offsets) {
1734       r->offsets[0] = RExC_size; 
1735     }
1736     DEBUG_r(PerlIO_printf(Perl_debug_log, 
1737                           "%s %"UVuf" bytes for offset annotations.\n", 
1738                           r->offsets ? "Got" : "Couldn't get", 
1739                           (UV)((2*RExC_size+1) * sizeof(U32))));
1740
1741     RExC_rx = r;
1742
1743     /* Second pass: emit code. */
1744     RExC_parse = exp;
1745     RExC_end = xend;
1746     RExC_naughty = 0;
1747     RExC_npar = 1;
1748     RExC_emit_start = r->program;
1749     RExC_emit = r->program;
1750     /* Store the count of eval-groups for security checks: */
1751     RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1752     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1753     r->data = 0;
1754     if (reg(pRExC_state, 0, &flags) == NULL)
1755         return(NULL);
1756
1757     /* Dig out information for optimizations. */
1758     r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1759     pm->op_pmflags = RExC_flags16;
1760     if (UTF)
1761         r->reganch |= ROPT_UTF8;
1762     r->regstclass = NULL;
1763     if (RExC_naughty >= 10)     /* Probably an expensive pattern. */
1764         r->reganch |= ROPT_NAUGHTY;
1765     scan = r->program + 1;              /* First BRANCH. */
1766
1767     /* XXXX To minimize changes to RE engine we always allocate
1768        3-units-long substrs field. */
1769     Newz(1004, r->substrs, 1, struct reg_substr_data);
1770
1771     StructCopy(&zero_scan_data, &data, scan_data_t);
1772     /* XXXX Should not we check for something else?  Usually it is OPEN1... */
1773     if (OP(scan) != BRANCH) {   /* Only one top-level choice. */
1774         I32 fake;
1775         STRLEN longest_float_length, longest_fixed_length;
1776         struct regnode_charclass_class ch_class;
1777         int stclass_flag;
1778         I32 last_close = 0;
1779
1780         first = scan;
1781         /* Skip introductions and multiplicators >= 1. */
1782         while ((OP(first) == OPEN && (sawopen = 1)) ||
1783                /* An OR of *one* alternative - should not happen now. */
1784             (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1785             (OP(first) == PLUS) ||
1786             (OP(first) == MINMOD) ||
1787                /* An {n,m} with n>0 */
1788             (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1789                 if (OP(first) == PLUS)
1790                     sawplus = 1;
1791                 else
1792                     first += regarglen[(U8)OP(first)];
1793                 first = NEXTOPER(first);
1794         }
1795
1796         /* Starting-point info. */
1797       again:
1798         if (PL_regkind[(U8)OP(first)] == EXACT) {
1799             if (OP(first) == EXACT)
1800                 ;       /* Empty, get anchored substr later. */
1801             else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1802                 r->regstclass = first;
1803         }
1804         else if (strchr((char*)PL_simple,OP(first)))
1805             r->regstclass = first;
1806         else if (PL_regkind[(U8)OP(first)] == BOUND ||
1807                  PL_regkind[(U8)OP(first)] == NBOUND)
1808             r->regstclass = first;
1809         else if (PL_regkind[(U8)OP(first)] == BOL) {
1810             r->reganch |= (OP(first) == MBOL
1811                            ? ROPT_ANCH_MBOL
1812                            : (OP(first) == SBOL
1813                               ? ROPT_ANCH_SBOL
1814                               : ROPT_ANCH_BOL));
1815             first = NEXTOPER(first);
1816             goto again;
1817         }
1818         else if (OP(first) == GPOS) {
1819             r->reganch |= ROPT_ANCH_GPOS;
1820             first = NEXTOPER(first);
1821             goto again;
1822         }
1823         else if ((OP(first) == STAR &&
1824             PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1825             !(r->reganch & ROPT_ANCH) )
1826         {
1827             /* turn .* into ^.* with an implied $*=1 */
1828             int type = OP(NEXTOPER(first));
1829
1830             if (type == REG_ANY)
1831                 type = ROPT_ANCH_MBOL;
1832             else
1833                 type = ROPT_ANCH_SBOL;
1834
1835             r->reganch |= type | ROPT_IMPLICIT;
1836             first = NEXTOPER(first);
1837             goto again;
1838         }
1839         if (sawplus && (!sawopen || !RExC_sawback)
1840             && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1841             /* x+ must match at the 1st pos of run of x's */
1842             r->reganch |= ROPT_SKIP;
1843
1844         /* Scan is after the zeroth branch, first is atomic matcher. */
1845         DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
1846                               (IV)(first - scan + 1)));
1847         /*
1848         * If there's something expensive in the r.e., find the
1849         * longest literal string that must appear and make it the
1850         * regmust.  Resolve ties in favor of later strings, since
1851         * the regstart check works with the beginning of the r.e.
1852         * and avoiding duplication strengthens checking.  Not a
1853         * strong reason, but sufficient in the absence of others.
1854         * [Now we resolve ties in favor of the earlier string if
1855         * it happens that c_offset_min has been invalidated, since the
1856         * earlier string may buy us something the later one won't.]
1857         */
1858         minlen = 0;
1859
1860         data.longest_fixed = newSVpvn("",0);
1861         data.longest_float = newSVpvn("",0);
1862         data.last_found = newSVpvn("",0);
1863         data.longest = &(data.longest_fixed);
1864         first = scan;
1865         if (!r->regstclass) {
1866             cl_init(pRExC_state, &ch_class);
1867             data.start_class = &ch_class;
1868             stclass_flag = SCF_DO_STCLASS_AND;
1869         } else                          /* XXXX Check for BOUND? */
1870             stclass_flag = 0;
1871         data.last_closep = &last_close;
1872
1873         minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1874                              &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1875         if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1876              && data.last_start_min == 0 && data.last_end > 0
1877              && !RExC_seen_zerolen
1878              && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1879             r->reganch |= ROPT_CHECK_ALL;
1880         scan_commit(pRExC_state, &data);
1881         SvREFCNT_dec(data.last_found);
1882
1883         longest_float_length = CHR_SVLEN(data.longest_float);
1884         if (longest_float_length
1885             || (data.flags & SF_FL_BEFORE_EOL
1886                 && (!(data.flags & SF_FL_BEFORE_MEOL)
1887                     || (RExC_flags16 & PMf_MULTILINE)))) {
1888             int t;
1889
1890             if (SvCUR(data.longest_fixed)                       /* ok to leave SvCUR */
1891                 && data.offset_fixed == data.offset_float_min
1892                 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1893                     goto remove_float;          /* As in (a)+. */
1894
1895             r->float_substr = data.longest_float;
1896             r->float_min_offset = data.offset_float_min;
1897             r->float_max_offset = data.offset_float_max;
1898             t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1899                        && (!(data.flags & SF_FL_BEFORE_MEOL)
1900                            || (RExC_flags16 & PMf_MULTILINE)));
1901             fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1902         }
1903         else {
1904           remove_float:
1905             r->float_substr = Nullsv;
1906             SvREFCNT_dec(data.longest_float);
1907             longest_float_length = 0;
1908         }
1909
1910         longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1911         if (longest_fixed_length
1912             || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1913                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1914                     || (RExC_flags16 & PMf_MULTILINE)))) {
1915             int t;
1916
1917             r->anchored_substr = data.longest_fixed;
1918             r->anchored_offset = data.offset_fixed;
1919             t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1920                  && (!(data.flags & SF_FIX_BEFORE_MEOL)
1921                      || (RExC_flags16 & PMf_MULTILINE)));
1922             fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1923         }
1924         else {
1925             r->anchored_substr = Nullsv;
1926             SvREFCNT_dec(data.longest_fixed);
1927             longest_fixed_length = 0;
1928         }
1929         if (r->regstclass
1930             && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1931             r->regstclass = NULL;
1932         if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1933             && !(data.start_class->flags & ANYOF_EOS)
1934             && !cl_is_anything(data.start_class)) {
1935             I32 n = add_data(pRExC_state, 1, "f");
1936
1937             New(1006, RExC_rx->data->data[n], 1,
1938                 struct regnode_charclass_class);
1939             StructCopy(data.start_class,
1940                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1941                        struct regnode_charclass_class);
1942             r->regstclass = (regnode*)RExC_rx->data->data[n];
1943             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1944             PL_regdata = r->data; /* for regprop() */
1945             DEBUG_r({ SV *sv = sv_newmortal();
1946                       regprop(sv, (regnode*)data.start_class);
1947                       PerlIO_printf(Perl_debug_log,
1948                                     "synthetic stclass `%s'.\n",
1949                                     SvPVX(sv));});
1950         }
1951
1952         /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1953         if (longest_fixed_length > longest_float_length) {
1954             r->check_substr = r->anchored_substr;
1955             r->check_offset_min = r->check_offset_max = r->anchored_offset;
1956             if (r->reganch & ROPT_ANCH_SINGLE)
1957                 r->reganch |= ROPT_NOSCAN;
1958         }
1959         else {
1960             r->check_substr = r->float_substr;
1961             r->check_offset_min = data.offset_float_min;
1962             r->check_offset_max = data.offset_float_max;
1963         }
1964         /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1965            This should be changed ASAP!  */
1966         if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1967             r->reganch |= RE_USE_INTUIT;
1968             if (SvTAIL(r->check_substr))
1969                 r->reganch |= RE_INTUIT_TAIL;
1970         }
1971     }
1972     else {
1973         /* Several toplevels. Best we can is to set minlen. */
1974         I32 fake;
1975         struct regnode_charclass_class ch_class;
1976         I32 last_close = 0;
1977         
1978         DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1979         scan = r->program + 1;
1980         cl_init(pRExC_state, &ch_class);
1981         data.start_class = &ch_class;
1982         data.last_closep = &last_close;
1983         minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1984         r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1985         if (!(data.start_class->flags & ANYOF_EOS)
1986             && !cl_is_anything(data.start_class)) {
1987             I32 n = add_data(pRExC_state, 1, "f");
1988
1989             New(1006, RExC_rx->data->data[n], 1,
1990                 struct regnode_charclass_class);
1991             StructCopy(data.start_class,
1992                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1993                        struct regnode_charclass_class);
1994             r->regstclass = (regnode*)RExC_rx->data->data[n];
1995             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1996             DEBUG_r({ SV* sv = sv_newmortal();
1997                       regprop(sv, (regnode*)data.start_class);
1998                       PerlIO_printf(Perl_debug_log,
1999                                     "synthetic stclass `%s'.\n",
2000                                     SvPVX(sv));});
2001         }
2002     }
2003
2004     r->minlen = minlen;
2005     if (RExC_seen & REG_SEEN_GPOS)
2006         r->reganch |= ROPT_GPOS_SEEN;
2007     if (RExC_seen & REG_SEEN_LOOKBEHIND)
2008         r->reganch |= ROPT_LOOKBEHIND_SEEN;
2009     if (RExC_seen & REG_SEEN_EVAL)
2010         r->reganch |= ROPT_EVAL_SEEN;
2011     if (RExC_seen & REG_SEEN_SANY)
2012         r->reganch |= ROPT_SANY_SEEN;
2013     Newz(1002, r->startp, RExC_npar, I32);
2014     Newz(1002, r->endp, RExC_npar, I32);
2015     PL_regdata = r->data; /* for regprop() */
2016     DEBUG_r(regdump(r));
2017     return(r);
2018 }
2019
2020 /*
2021  - reg - regular expression, i.e. main body or parenthesized thing
2022  *
2023  * Caller must absorb opening parenthesis.
2024  *
2025  * Combining parenthesis handling with the base level of regular expression
2026  * is a trifle forced, but the need to tie the tails of the branches to what
2027  * follows makes it hard to avoid.
2028  */
2029 STATIC regnode *
2030 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2031     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2032 {
2033     register regnode *ret;              /* Will be the head of the group. */
2034     register regnode *br;
2035     register regnode *lastbr;
2036     register regnode *ender = 0;
2037     register I32 parno = 0;
2038     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2039
2040     /* for (?g), (?gc), and (?o) warnings; warning
2041        about (?c) will warn about (?g) -- japhy    */
2042
2043     I32 wastedflags = 0x00,
2044         wasted_o    = 0x01,
2045         wasted_g    = 0x02,
2046         wasted_gc   = 0x02 | 0x04,
2047         wasted_c    = 0x04;
2048
2049     char * parse_start = RExC_parse; /* MJD */
2050     char *oregcomp_parse = RExC_parse;
2051     char c;
2052
2053     *flagp = 0;                         /* Tentatively. */
2054
2055
2056     /* Make an OPEN node, if parenthesized. */
2057     if (paren) {
2058         if (*RExC_parse == '?') { /* (?...) */
2059             U16 posflags = 0, negflags = 0;
2060             U16 *flagsp = &posflags;
2061             int logical = 0;
2062             char *seqstart = RExC_parse;
2063
2064             RExC_parse++;
2065             paren = *RExC_parse++;
2066             ret = NULL;                 /* For look-ahead/behind. */
2067             switch (paren) {
2068             case '<':           /* (?<...) */
2069                 RExC_seen |= REG_SEEN_LOOKBEHIND;
2070                 if (*RExC_parse == '!')
2071                     paren = ',';
2072                 if (*RExC_parse != '=' && *RExC_parse != '!')
2073                     goto unknown;
2074                 RExC_parse++;
2075             case '=':           /* (?=...) */
2076             case '!':           /* (?!...) */
2077                 RExC_seen_zerolen++;
2078             case ':':           /* (?:...) */
2079             case '>':           /* (?>...) */
2080                 break;
2081             case '$':           /* (?$...) */
2082             case '@':           /* (?@...) */
2083                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2084                 break;
2085             case '#':           /* (?#...) */
2086                 while (*RExC_parse && *RExC_parse != ')')
2087                     RExC_parse++;
2088                 if (*RExC_parse != ')')
2089                     FAIL("Sequence (?#... not terminated");
2090                 nextchar(pRExC_state);
2091                 *flagp = TRYAGAIN;
2092                 return NULL;
2093             case 'p':           /* (?p...) */
2094                 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
2095                     vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
2096                 /* FALL THROUGH*/
2097             case '?':           /* (??...) */
2098                 logical = 1;
2099                 paren = *RExC_parse++;
2100                 /* FALL THROUGH */
2101             case '{':           /* (?{...}) */
2102             {
2103                 I32 count = 1, n = 0;
2104                 char c;
2105                 char *s = RExC_parse;
2106                 SV *sv;
2107                 OP_4tree *sop, *rop;
2108
2109                 RExC_seen_zerolen++;
2110                 RExC_seen |= REG_SEEN_EVAL;
2111                 while (count && (c = *RExC_parse)) {
2112                     if (c == '\\' && RExC_parse[1])
2113                         RExC_parse++;
2114                     else if (c == '{')
2115                         count++;
2116                     else if (c == '}')
2117                         count--;
2118                     RExC_parse++;
2119                 }
2120                 if (*RExC_parse != ')')
2121                 {
2122                     RExC_parse = s;             
2123                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2124                 }
2125                 if (!SIZE_ONLY) {
2126                     AV *av;
2127                 
2128                     if (RExC_parse - 1 - s)
2129                         sv = newSVpvn(s, RExC_parse - 1 - s);
2130                     else
2131                         sv = newSVpvn("", 0);
2132
2133                     ENTER;
2134                     Perl_save_re_context(aTHX);
2135                     rop = sv_compile_2op(sv, &sop, "re", &av);
2136                     LEAVE;
2137
2138                     n = add_data(pRExC_state, 3, "nop");
2139                     RExC_rx->data->data[n] = (void*)rop;
2140                     RExC_rx->data->data[n+1] = (void*)sop;
2141                     RExC_rx->data->data[n+2] = (void*)av;
2142                     SvREFCNT_dec(sv);
2143                 }
2144                 else {                                          /* First pass */
2145                     if (PL_reginterp_cnt < ++RExC_seen_evals
2146                         && PL_curcop != &PL_compiling)
2147                         /* No compiled RE interpolated, has runtime
2148                            components ===> unsafe.  */
2149                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2150                     if (PL_tainting && PL_tainted)
2151                         FAIL("Eval-group in insecure regular expression");
2152                 }
2153                 
2154                 nextchar(pRExC_state);
2155                 if (logical) {
2156                     ret = reg_node(pRExC_state, LOGICAL);
2157                     if (!SIZE_ONLY)
2158                         ret->flags = 2;
2159                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2160                     /* deal with the length of this later - MJD */
2161                     return ret;
2162                 }
2163                 return reganode(pRExC_state, EVAL, n);
2164             }
2165             case '(':           /* (?(?{...})...) and (?(?=...)...) */
2166             {
2167                 if (RExC_parse[0] == '?') {        /* (?(?...)) */
2168                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2169                         || RExC_parse[1] == '<'
2170                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2171                         I32 flag;
2172                         
2173                         ret = reg_node(pRExC_state, LOGICAL);
2174                         if (!SIZE_ONLY)
2175                             ret->flags = 1;
2176                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2177                         goto insert_if;
2178                     }
2179                 }
2180                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2181                     /* (?(1)...) */
2182                     parno = atoi(RExC_parse++);
2183
2184                     while (isDIGIT(*RExC_parse))
2185                         RExC_parse++;
2186                     ret = reganode(pRExC_state, GROUPP, parno);
2187                     
2188                     if ((c = *nextchar(pRExC_state)) != ')')
2189                         vFAIL("Switch condition not recognized");
2190                   insert_if:
2191                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2192                     br = regbranch(pRExC_state, &flags, 1);
2193                     if (br == NULL)
2194                         br = reganode(pRExC_state, LONGJMP, 0);
2195                     else
2196                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2197                     c = *nextchar(pRExC_state);
2198                     if (flags&HASWIDTH)
2199                         *flagp |= HASWIDTH;
2200                     if (c == '|') {
2201                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2202                         regbranch(pRExC_state, &flags, 1);
2203                         regtail(pRExC_state, ret, lastbr);
2204                         if (flags&HASWIDTH)
2205                             *flagp |= HASWIDTH;
2206                         c = *nextchar(pRExC_state);
2207                     }
2208                     else
2209                         lastbr = NULL;
2210                     if (c != ')')
2211                         vFAIL("Switch (?(condition)... contains too many branches");
2212                     ender = reg_node(pRExC_state, TAIL);
2213                     regtail(pRExC_state, br, ender);
2214                     if (lastbr) {
2215                         regtail(pRExC_state, lastbr, ender);
2216                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2217                     }
2218                     else
2219                         regtail(pRExC_state, ret, ender);
2220                     return ret;
2221                 }
2222                 else {
2223                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2224                 }
2225             }
2226             case 0:
2227                 RExC_parse--; /* for vFAIL to print correctly */
2228                 vFAIL("Sequence (? incomplete");
2229                 break;
2230             default:
2231                 --RExC_parse;
2232               parse_flags:      /* (?i) */
2233                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2234                     /* (?g), (?gc) and (?o) are useless here
2235                        and must be globally applied -- japhy */
2236
2237                     if (*RExC_parse == 'o' || *RExC_parse == 'g') {
2238                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2239                             I32 wflagbit = *RExC_parse == 'o' ? wasted_o : wasted_g;
2240                             if (! (wastedflags & wflagbit) ) {
2241                                 wastedflags |= wflagbit;
2242                                 vWARN5(
2243                                     RExC_parse + 1,
2244                                     "Useless (%s%c) - %suse /%c modifier",
2245                                     flagsp == &negflags ? "?-" : "?",
2246                                     *RExC_parse,
2247                                     flagsp == &negflags ? "don't " : "",
2248                                     *RExC_parse
2249                                 );
2250                             }
2251                         }
2252                     }
2253                     else if (*RExC_parse == 'c') {
2254                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2255                             if (! (wastedflags & wasted_c) ) {
2256                                 wastedflags |= wasted_gc;
2257                                 vWARN3(
2258                                     RExC_parse + 1,
2259                                     "Useless (%sc) - %suse /gc modifier",
2260                                     flagsp == &negflags ? "?-" : "?",
2261                                     flagsp == &negflags ? "don't " : ""
2262                                 );
2263                             }
2264                         }
2265                     }
2266                     else { pmflag(flagsp, *RExC_parse); }
2267
2268                     ++RExC_parse;
2269                 }
2270                 if (*RExC_parse == '-') {
2271                     flagsp = &negflags;
2272                     wastedflags = 0;  /* reset so (?g-c) warns twice */
2273                     ++RExC_parse;
2274                     goto parse_flags;
2275                 }
2276                 RExC_flags16 |= posflags;
2277                 RExC_flags16 &= ~negflags;
2278                 if (*RExC_parse == ':') {
2279                     RExC_parse++;
2280                     paren = ':';
2281                     break;
2282                 }               
2283               unknown:
2284                 if (*RExC_parse != ')') {
2285                     RExC_parse++;
2286                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2287                 }
2288                 nextchar(pRExC_state);
2289                 *flagp = TRYAGAIN;
2290                 return NULL;
2291             }
2292         }
2293         else {                  /* (...) */
2294             parno = RExC_npar;
2295             RExC_npar++;
2296             ret = reganode(pRExC_state, OPEN, parno);
2297             Set_Node_Length(ret, 1); /* MJD */
2298             Set_Node_Offset(ret, RExC_parse); /* MJD */
2299             open = 1;
2300         }
2301     }
2302     else                        /* ! paren */
2303         ret = NULL;
2304
2305     /* Pick up the branches, linking them together. */
2306     parse_start = RExC_parse;   /* MJD */
2307     br = regbranch(pRExC_state, &flags, 1);
2308     /*     branch_len = (paren != 0); */
2309     
2310     if (br == NULL)
2311         return(NULL);
2312     if (*RExC_parse == '|') {
2313         if (!SIZE_ONLY && RExC_extralen) {
2314             reginsert(pRExC_state, BRANCHJ, br);
2315         }
2316         else {                  /* MJD */
2317             reginsert(pRExC_state, BRANCH, br);
2318             Set_Node_Length(br, paren != 0);
2319             Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2320         }
2321         have_branch = 1;
2322         if (SIZE_ONLY)
2323             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2324     }
2325     else if (paren == ':') {
2326         *flagp |= flags&SIMPLE;
2327     }
2328     if (open) {                         /* Starts with OPEN. */
2329         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2330     }
2331     else if (paren != '?')              /* Not Conditional */
2332         ret = br;
2333     if (flags&HASWIDTH)
2334         *flagp |= HASWIDTH;
2335     *flagp |= flags&SPSTART;
2336     lastbr = br;
2337     while (*RExC_parse == '|') {
2338         if (!SIZE_ONLY && RExC_extralen) {
2339             ender = reganode(pRExC_state, LONGJMP,0);
2340             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2341         }
2342         if (SIZE_ONLY)
2343             RExC_extralen += 2;         /* Account for LONGJMP. */
2344         nextchar(pRExC_state);
2345         br = regbranch(pRExC_state, &flags, 0);
2346         
2347         if (br == NULL)
2348             return(NULL);
2349         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2350         lastbr = br;
2351         if (flags&HASWIDTH)
2352             *flagp |= HASWIDTH;
2353         *flagp |= flags&SPSTART;
2354     }
2355
2356     if (have_branch || paren != ':') {
2357         /* Make a closing node, and hook it on the end. */
2358         switch (paren) {
2359         case ':':
2360             ender = reg_node(pRExC_state, TAIL);
2361             break;
2362         case 1:
2363             ender = reganode(pRExC_state, CLOSE, parno);
2364             Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2365             Set_Node_Length(ender,1); /* MJD */
2366             break;
2367         case '<':
2368         case ',':
2369         case '=':
2370         case '!':
2371             *flagp &= ~HASWIDTH;
2372             /* FALL THROUGH */
2373         case '>':
2374             ender = reg_node(pRExC_state, SUCCEED);
2375             break;
2376         case 0:
2377             ender = reg_node(pRExC_state, END);
2378             break;
2379         }
2380         regtail(pRExC_state, lastbr, ender);
2381
2382         if (have_branch) {
2383             /* Hook the tails of the branches to the closing node. */
2384             for (br = ret; br != NULL; br = regnext(br)) {
2385                 regoptail(pRExC_state, br, ender);
2386             }
2387         }
2388     }
2389
2390     {
2391         char *p;
2392         static char parens[] = "=!<,>";
2393
2394         if (paren && (p = strchr(parens, paren))) {
2395             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2396             int flag = (p - parens) > 1;
2397
2398             if (paren == '>')
2399                 node = SUSPEND, flag = 0;
2400             reginsert(pRExC_state, node,ret);
2401             ret->flags = flag;
2402             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2403         }
2404     }
2405
2406     /* Check for proper termination. */
2407     if (paren) {
2408         RExC_flags16 = oregflags;
2409         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2410             RExC_parse = oregcomp_parse;
2411             vFAIL("Unmatched (");
2412         }
2413     }
2414     else if (!paren && RExC_parse < RExC_end) {
2415         if (*RExC_parse == ')') {
2416             RExC_parse++;
2417             vFAIL("Unmatched )");
2418         }
2419         else
2420             FAIL("Junk on end of regexp");      /* "Can't happen". */
2421         /* NOTREACHED */
2422     }
2423
2424     return(ret);
2425 }
2426
2427 /*
2428  - regbranch - one alternative of an | operator
2429  *
2430  * Implements the concatenation operator.
2431  */
2432 STATIC regnode *
2433 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2434 {
2435     register regnode *ret;
2436     register regnode *chain = NULL;
2437     register regnode *latest;
2438     I32 flags = 0, c = 0;
2439
2440     if (first)
2441         ret = NULL;
2442     else {
2443         if (!SIZE_ONLY && RExC_extralen)
2444             ret = reganode(pRExC_state, BRANCHJ,0);
2445         else {
2446             ret = reg_node(pRExC_state, BRANCH);
2447             Set_Node_Length(ret, 1);
2448         }
2449     }
2450         
2451     if (!first && SIZE_ONLY)
2452         RExC_extralen += 1;                     /* BRANCHJ */
2453
2454     *flagp = WORST;                     /* Tentatively. */
2455
2456     RExC_parse--;
2457     nextchar(pRExC_state);
2458     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2459         flags &= ~TRYAGAIN;
2460         latest = regpiece(pRExC_state, &flags);
2461         if (latest == NULL) {
2462             if (flags & TRYAGAIN)
2463                 continue;
2464             return(NULL);
2465         }
2466         else if (ret == NULL)
2467             ret = latest;
2468         *flagp |= flags&HASWIDTH;
2469         if (chain == NULL)      /* First piece. */
2470             *flagp |= flags&SPSTART;
2471         else {
2472             RExC_naughty++;
2473             regtail(pRExC_state, chain, latest);
2474         }
2475         chain = latest;
2476         c++;
2477     }
2478     if (chain == NULL) {        /* Loop ran zero times. */
2479         chain = reg_node(pRExC_state, NOTHING);
2480         if (ret == NULL)
2481             ret = chain;
2482     }
2483     if (c == 1) {
2484         *flagp |= flags&SIMPLE;
2485     }
2486
2487     return(ret);
2488 }
2489
2490 /*
2491  - regpiece - something followed by possible [*+?]
2492  *
2493  * Note that the branching code sequences used for ? and the general cases
2494  * of * and + are somewhat optimized:  they use the same NOTHING node as
2495  * both the endmarker for their branch list and the body of the last branch.
2496  * It might seem that this node could be dispensed with entirely, but the
2497  * endmarker role is not redundant.
2498  */
2499 STATIC regnode *
2500 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2501 {
2502     register regnode *ret;
2503     register char op;
2504     register char *next;
2505     I32 flags;
2506     char *origparse = RExC_parse;
2507     char *maxpos;
2508     I32 min;
2509     I32 max = REG_INFTY;
2510     char *parse_start;
2511
2512     ret = regatom(pRExC_state, &flags);
2513     if (ret == NULL) {
2514         if (flags & TRYAGAIN)
2515             *flagp |= TRYAGAIN;
2516         return(NULL);
2517     }
2518
2519     op = *RExC_parse;
2520
2521     if (op == '{' && regcurly(RExC_parse)) {
2522         parse_start = RExC_parse; /* MJD */
2523         next = RExC_parse + 1;
2524         maxpos = Nullch;
2525         while (isDIGIT(*next) || *next == ',') {
2526             if (*next == ',') {
2527                 if (maxpos)
2528                     break;
2529                 else
2530                     maxpos = next;
2531             }
2532             next++;
2533         }
2534         if (*next == '}') {             /* got one */
2535             if (!maxpos)
2536                 maxpos = next;
2537             RExC_parse++;
2538             min = atoi(RExC_parse);
2539             if (*maxpos == ',')
2540                 maxpos++;
2541             else
2542                 maxpos = RExC_parse;
2543             max = atoi(maxpos);
2544             if (!max && *maxpos != '0')
2545                 max = REG_INFTY;                /* meaning "infinity" */
2546             else if (max >= REG_INFTY)
2547                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2548             RExC_parse = next;
2549             nextchar(pRExC_state);
2550
2551         do_curly:
2552             if ((flags&SIMPLE)) {
2553                 RExC_naughty += 2 + RExC_naughty / 2;
2554                 reginsert(pRExC_state, CURLY, ret);
2555                 Set_Node_Offset(ret, parse_start+1); /* MJD */
2556                 Set_Node_Cur_Length(ret);
2557             }
2558             else {
2559                 regnode *w = reg_node(pRExC_state, WHILEM);
2560
2561                 w->flags = 0;
2562                 regtail(pRExC_state, ret, w);
2563                 if (!SIZE_ONLY && RExC_extralen) {
2564                     reginsert(pRExC_state, LONGJMP,ret);
2565                     reginsert(pRExC_state, NOTHING,ret);
2566                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2567                 }
2568                 reginsert(pRExC_state, CURLYX,ret);
2569                                 /* MJD hk */
2570                 Set_Node_Offset(ret, parse_start+1);
2571                 Set_Node_Length(ret, 
2572                                 op == '{' ? (RExC_parse - parse_start) : 1);
2573                 
2574                 if (!SIZE_ONLY && RExC_extralen)
2575                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2576                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2577                 if (SIZE_ONLY)
2578                     RExC_whilem_seen++, RExC_extralen += 3;
2579                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2580             }
2581             ret->flags = 0;
2582
2583             if (min > 0)
2584                 *flagp = WORST;
2585             if (max > 0)
2586                 *flagp |= HASWIDTH;
2587             if (max && max < min)
2588                 vFAIL("Can't do {n,m} with n > m");
2589             if (!SIZE_ONLY) {
2590                 ARG1_SET(ret, min);
2591                 ARG2_SET(ret, max);
2592             }
2593
2594             goto nest_check;
2595         }
2596     }
2597
2598     if (!ISMULT1(op)) {
2599         *flagp = flags;
2600         return(ret);
2601     }
2602
2603 #if 0                           /* Now runtime fix should be reliable. */
2604
2605     /* if this is reinstated, don't forget to put this back into perldiag:
2606
2607             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2608
2609            (F) The part of the regexp subject to either the * or + quantifier
2610            could match an empty string. The {#} shows in the regular
2611            expression about where the problem was discovered.
2612
2613     */
2614
2615     if (!(flags&HASWIDTH) && op != '?')
2616       vFAIL("Regexp *+ operand could be empty");
2617 #endif
2618
2619     parse_start = RExC_parse;
2620     nextchar(pRExC_state);
2621
2622     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2623
2624     if (op == '*' && (flags&SIMPLE)) {
2625         reginsert(pRExC_state, STAR, ret);
2626         ret->flags = 0;
2627         RExC_naughty += 4;
2628     }
2629     else if (op == '*') {
2630         min = 0;
2631         goto do_curly;
2632     }
2633     else if (op == '+' && (flags&SIMPLE)) {
2634         reginsert(pRExC_state, PLUS, ret);
2635         ret->flags = 0;
2636         RExC_naughty += 3;
2637     }
2638     else if (op == '+') {
2639         min = 1;
2640         goto do_curly;
2641     }
2642     else if (op == '?') {
2643         min = 0; max = 1;
2644         goto do_curly;
2645     }
2646   nest_check:
2647     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2648         vWARN3(RExC_parse,
2649                "%.*s matches null string many times",
2650                RExC_parse - origparse,
2651                origparse);
2652     }
2653
2654     if (*RExC_parse == '?') {
2655         nextchar(pRExC_state);
2656         reginsert(pRExC_state, MINMOD, ret);
2657         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2658     }
2659     if (ISMULT2(RExC_parse)) {
2660         RExC_parse++;
2661         vFAIL("Nested quantifiers");
2662     }
2663
2664     return(ret);
2665 }
2666
2667 /*
2668  - regatom - the lowest level
2669  *
2670  * Optimization:  gobbles an entire sequence of ordinary characters so that
2671  * it can turn them into a single node, which is smaller to store and
2672  * faster to run.  Backslashed characters are exceptions, each becoming a
2673  * separate node; the code is simpler that way and it's not worth fixing.
2674  *
2675  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2676 STATIC regnode *
2677 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2678 {
2679     register regnode *ret = 0;
2680     I32 flags;
2681
2682     *flagp = WORST;             /* Tentatively. */
2683
2684 tryagain:
2685     switch (*RExC_parse) {
2686     case '^':
2687         RExC_seen_zerolen++;
2688         nextchar(pRExC_state);
2689         if (RExC_flags16 & PMf_MULTILINE)
2690             ret = reg_node(pRExC_state, MBOL);
2691         else if (RExC_flags16 & PMf_SINGLELINE)
2692             ret = reg_node(pRExC_state, SBOL);
2693         else
2694             ret = reg_node(pRExC_state, BOL);
2695         Set_Node_Length(ret, 1); /* MJD */
2696         break;
2697     case '$':
2698         nextchar(pRExC_state);
2699         if (*RExC_parse)
2700             RExC_seen_zerolen++;
2701         if (RExC_flags16 & PMf_MULTILINE)
2702             ret = reg_node(pRExC_state, MEOL);
2703         else if (RExC_flags16 & PMf_SINGLELINE)
2704             ret = reg_node(pRExC_state, SEOL);
2705         else
2706             ret = reg_node(pRExC_state, EOL);
2707         Set_Node_Length(ret, 1); /* MJD */
2708         break;
2709     case '.':
2710         nextchar(pRExC_state);
2711         if (RExC_flags16 & PMf_SINGLELINE)
2712             ret = reg_node(pRExC_state, SANY);
2713         else
2714             ret = reg_node(pRExC_state, REG_ANY);
2715         *flagp |= HASWIDTH|SIMPLE;
2716         RExC_naughty++;
2717         Set_Node_Length(ret, 1); /* MJD */
2718         break;
2719     case '[':
2720     {
2721         char *oregcomp_parse = ++RExC_parse;
2722         ret = regclass(pRExC_state);
2723         if (*RExC_parse != ']') {
2724             RExC_parse = oregcomp_parse;
2725             vFAIL("Unmatched [");
2726         }
2727         nextchar(pRExC_state);
2728         *flagp |= HASWIDTH|SIMPLE;
2729         Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2730         break;
2731     }
2732     case '(':
2733         nextchar(pRExC_state);
2734         ret = reg(pRExC_state, 1, &flags);
2735         if (ret == NULL) {
2736                 if (flags & TRYAGAIN) {
2737                     if (RExC_parse == RExC_end) {
2738                          /* Make parent create an empty node if needed. */
2739                         *flagp |= TRYAGAIN;
2740                         return(NULL);
2741                     }
2742                     goto tryagain;
2743                 }
2744                 return(NULL);
2745         }
2746         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2747         break;
2748     case '|':
2749     case ')':
2750         if (flags & TRYAGAIN) {
2751             *flagp |= TRYAGAIN;
2752             return NULL;
2753         }
2754         vFAIL("Internal urp");
2755                                 /* Supposed to be caught earlier. */
2756         break;
2757     case '{':
2758         if (!regcurly(RExC_parse)) {
2759             RExC_parse++;
2760             goto defchar;
2761         }
2762         /* FALL THROUGH */
2763     case '?':
2764     case '+':
2765     case '*':
2766         RExC_parse++;
2767         vFAIL("Quantifier follows nothing");
2768         break;
2769     case '\\':
2770         switch (*++RExC_parse) {
2771         case 'A':
2772             RExC_seen_zerolen++;
2773             ret = reg_node(pRExC_state, SBOL);
2774             *flagp |= SIMPLE;
2775             nextchar(pRExC_state);
2776             Set_Node_Length(ret, 2); /* MJD */
2777             break;
2778         case 'G':
2779             ret = reg_node(pRExC_state, GPOS);
2780             RExC_seen |= REG_SEEN_GPOS;
2781             *flagp |= SIMPLE;
2782             nextchar(pRExC_state);
2783             Set_Node_Length(ret, 2); /* MJD */
2784             break;
2785         case 'Z':
2786             ret = reg_node(pRExC_state, SEOL);
2787             *flagp |= SIMPLE;
2788             nextchar(pRExC_state);
2789             break;
2790         case 'z':
2791             ret = reg_node(pRExC_state, EOS);
2792             *flagp |= SIMPLE;
2793             RExC_seen_zerolen++;                /* Do not optimize RE away */
2794             nextchar(pRExC_state);
2795             Set_Node_Length(ret, 2); /* MJD */
2796             break;
2797         case 'C':
2798             ret = reg_node(pRExC_state, SANY);
2799             RExC_seen |= REG_SEEN_SANY;
2800             *flagp |= HASWIDTH|SIMPLE;
2801             nextchar(pRExC_state);
2802             Set_Node_Length(ret, 2); /* MJD */
2803             break;
2804         case 'X':
2805             ret = reg_node(pRExC_state, CLUMP);
2806             *flagp |= HASWIDTH;
2807             nextchar(pRExC_state);
2808             Set_Node_Length(ret, 2); /* MJD */
2809             break;
2810         case 'w':
2811             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2812             *flagp |= HASWIDTH|SIMPLE;
2813             nextchar(pRExC_state);
2814             Set_Node_Length(ret, 2); /* MJD */
2815             break;
2816         case 'W':
2817             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2818             *flagp |= HASWIDTH|SIMPLE;
2819             nextchar(pRExC_state);
2820             Set_Node_Length(ret, 2); /* MJD */
2821             break;
2822         case 'b':
2823             RExC_seen_zerolen++;
2824             RExC_seen |= REG_SEEN_LOOKBEHIND;
2825             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2826             *flagp |= SIMPLE;
2827             nextchar(pRExC_state);
2828             Set_Node_Length(ret, 2); /* MJD */
2829             break;
2830         case 'B':
2831             RExC_seen_zerolen++;
2832             RExC_seen |= REG_SEEN_LOOKBEHIND;
2833             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2834             *flagp |= SIMPLE;
2835             nextchar(pRExC_state);
2836             Set_Node_Length(ret, 2); /* MJD */
2837             break;
2838         case 's':
2839             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2840             *flagp |= HASWIDTH|SIMPLE;
2841             nextchar(pRExC_state);
2842             Set_Node_Length(ret, 2); /* MJD */
2843             break;
2844         case 'S':
2845             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
2846             *flagp |= HASWIDTH|SIMPLE;
2847             nextchar(pRExC_state);
2848             Set_Node_Length(ret, 2); /* MJD */
2849             break;
2850         case 'd':
2851             ret = reg_node(pRExC_state, DIGIT);
2852             *flagp |= HASWIDTH|SIMPLE;
2853             nextchar(pRExC_state);
2854             Set_Node_Length(ret, 2); /* MJD */
2855             break;
2856         case 'D':
2857             ret = reg_node(pRExC_state, NDIGIT);
2858             *flagp |= HASWIDTH|SIMPLE;
2859             nextchar(pRExC_state);
2860             Set_Node_Length(ret, 2); /* MJD */
2861             break;
2862         case 'p':
2863         case 'P':
2864             {   
2865                 char* oldregxend = RExC_end;
2866                 char* parse_start = RExC_parse;
2867
2868                 if (RExC_parse[1] == '{') {
2869                   /* a lovely hack--pretend we saw [\pX] instead */
2870                     RExC_end = strchr(RExC_parse, '}');
2871                     if (!RExC_end) {
2872                         RExC_parse += 2;
2873                         RExC_end = oldregxend;
2874                         vFAIL("Missing right brace on \\p{}");
2875                     }
2876                     RExC_end++;
2877                 }
2878                 else
2879                     RExC_end = RExC_parse + 2;
2880                 RExC_parse--;
2881
2882                 ret = regclass(pRExC_state);
2883
2884                 RExC_end = oldregxend;
2885                 RExC_parse--;
2886                 Set_Node_Cur_Length(ret); /* MJD */
2887                 nextchar(pRExC_state);
2888                 *flagp |= HASWIDTH|SIMPLE;
2889             }
2890             break;
2891         case 'n':
2892         case 'r':
2893         case 't':
2894         case 'f':
2895         case 'e':
2896         case 'a':
2897         case 'x':
2898         case 'c':
2899         case '0':
2900             goto defchar;
2901         case '1': case '2': case '3': case '4':
2902         case '5': case '6': case '7': case '8': case '9':
2903             {
2904                 I32 num = atoi(RExC_parse);
2905
2906                 if (num > 9 && num >= RExC_npar)
2907                     goto defchar;
2908                 else {
2909                     char * parse_start = RExC_parse - 1; /* MJD */
2910                     while (isDIGIT(*RExC_parse))
2911                         RExC_parse++;
2912
2913                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2914                         vFAIL("Reference to nonexistent group");
2915                     RExC_sawback = 1;
2916                     ret = reganode(pRExC_state, FOLD
2917                                    ? (LOC ? REFFL : REFF)
2918                                    : REF, num);
2919                     *flagp |= HASWIDTH;
2920                     
2921                     /* override incorrect value set in reganode MJD */
2922                     Set_Node_Offset(ret, parse_start+1); 
2923                     Set_Node_Cur_Length(ret); /* MJD */
2924                     RExC_parse--;
2925                     nextchar(pRExC_state);
2926                 }
2927             }
2928             break;
2929         case '\0':
2930             if (RExC_parse >= RExC_end)
2931                 FAIL("Trailing \\");
2932             /* FALL THROUGH */
2933         default:
2934             /* Do not generate `unrecognized' warnings here, we fall
2935                back into the quick-grab loop below */
2936             goto defchar;
2937         }
2938         break;
2939
2940     case '#':
2941         if (RExC_flags16 & PMf_EXTENDED) {
2942             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2943             if (RExC_parse < RExC_end)
2944                 goto tryagain;
2945         }
2946         /* FALL THROUGH */
2947
2948     default: {
2949             register STRLEN len;
2950             register UV ender;
2951             register char *p;
2952             char *oldp, *s;
2953             STRLEN numlen;
2954             char *parse_start = RExC_parse - 1;
2955
2956             RExC_parse++;
2957
2958         defchar:
2959             ret = reg_node(pRExC_state, FOLD
2960                           ? (LOC ? EXACTFL : EXACTF)
2961                           : EXACT);
2962             s = STRING(ret);
2963             for (len = 0, p = RExC_parse - 1;
2964               len < 127 && p < RExC_end;
2965               len++)
2966             {
2967                 oldp = p;
2968
2969                 if (RExC_flags16 & PMf_EXTENDED)
2970                     p = regwhite(p, RExC_end);
2971                 switch (*p) {
2972                 case '^':
2973                 case '$':
2974                 case '.':
2975                 case '[':
2976                 case '(':
2977                 case ')':
2978                 case '|':
2979                     goto loopdone;
2980                 case '\\':
2981                     switch (*++p) {
2982                     case 'A':
2983                     case 'G':
2984                     case 'Z':
2985                     case 'z':
2986                     case 'w':
2987                     case 'W':
2988                     case 'b':
2989                     case 'B':
2990                     case 's':
2991                     case 'S':
2992                     case 'd':
2993                     case 'D':
2994                     case 'p':
2995                     case 'P':
2996                         --p;
2997                         goto loopdone;
2998                     case 'n':
2999                         ender = '\n';
3000                         p++;
3001                         break;
3002                     case 'r':
3003                         ender = '\r';
3004                         p++;
3005                         break;
3006                     case 't':
3007                         ender = '\t';
3008                         p++;
3009                         break;
3010                     case 'f':
3011                         ender = '\f';
3012                         p++;
3013                         break;
3014                     case 'e':
3015                           ender = ASCII_TO_NATIVE('\033');
3016                         p++;
3017                         break;
3018                     case 'a':
3019                           ender = ASCII_TO_NATIVE('\007');
3020                         p++;
3021                         break;
3022                     case 'x':
3023                         if (*++p == '{') {
3024                             char* e = strchr(p, '}');
3025         
3026                             if (!e) {
3027                                 RExC_parse = p + 1;
3028                                 vFAIL("Missing right brace on \\x{}");
3029                             }
3030                             else {
3031                                 numlen = 1;     /* allow underscores */
3032                                 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
3033                                 if (ender > 0xff)
3034                                     RExC_utf8 = 1;
3035                                 /* numlen is generous */
3036                                 if (numlen + len >= 127) {
3037                                     p--;
3038                                     goto loopdone;
3039                                 }
3040                                 p = e + 1;
3041                             }
3042                         }
3043                         else {
3044                             numlen = 0;         /* disallow underscores */
3045                             ender = (UV)scan_hex(p, 2, &numlen);
3046                             p += numlen;
3047                         }
3048                         break;
3049                     case 'c':
3050                         p++;
3051                         ender = UCHARAT(p++);
3052                         ender = toCTRL(ender);
3053                         break;
3054                     case '0': case '1': case '2': case '3':case '4':
3055                     case '5': case '6': case '7': case '8':case '9':
3056                         if (*p == '0' ||
3057                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
3058                             numlen = 0;         /* disallow underscores */
3059                             ender = (UV)scan_oct(p, 3, &numlen);
3060                             p += numlen;
3061                         }
3062                         else {
3063                             --p;
3064                             goto loopdone;
3065                         }
3066                         break;
3067                     case '\0':
3068                         if (p >= RExC_end)
3069                             FAIL("Trailing \\");
3070                         /* FALL THROUGH */
3071                     default:
3072                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3073                             vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
3074                         goto normal_default;
3075                     }
3076                     break;
3077                 default:
3078                   normal_default:
3079                     if (UTF8_IS_START(*p) && UTF) {
3080                         ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3081                                                &numlen, 0);
3082                         p += numlen;
3083                     }
3084                     else
3085                         ender = *p++;
3086                     break;
3087                 }
3088                 if (RExC_flags16 & PMf_EXTENDED)
3089                     p = regwhite(p, RExC_end);
3090                 if (UTF && FOLD) {
3091                     if (LOC)
3092                         ender = toLOWER_LC_uvchr(ender);
3093                     else
3094                         ender = toLOWER_uni(ender);
3095                 }
3096                 if (ISMULT2(p)) { /* Back off on ?+*. */
3097                     if (len)
3098                         p = oldp;
3099                     else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3100                         reguni(pRExC_state, ender, s, &numlen);
3101                         s += numlen;
3102                         len += numlen;
3103                     }
3104                     else {
3105                         len++;
3106                         REGC(ender, s++);
3107                     }
3108                     break;
3109                 }
3110                 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3111                     reguni(pRExC_state, ender, s, &numlen);
3112                     s += numlen;
3113                     len += numlen - 1;
3114                 }
3115                 else
3116                     REGC(ender, s++);
3117             }
3118         loopdone:
3119             RExC_parse = p - 1;
3120             Set_Node_Cur_Length(ret); /* MJD */
3121             nextchar(pRExC_state);
3122             {
3123                 /* len is STRLEN which is unsigned, need to copy to signed */
3124                 IV iv = len;
3125                 if (iv < 0)
3126                     vFAIL("Internal disaster");
3127             }
3128             if (len > 0)
3129                 *flagp |= HASWIDTH;
3130             if (len == 1)
3131                 *flagp |= SIMPLE;
3132             if (!SIZE_ONLY)
3133                 STR_LEN(ret) = len;
3134             if (SIZE_ONLY)
3135                 RExC_size += STR_SZ(len);
3136             else
3137                 RExC_emit += STR_SZ(len);
3138         }
3139         break;
3140     }
3141
3142     return(ret);
3143 }
3144
3145 STATIC char *
3146 S_regwhite(pTHX_ char *p, char *e)
3147 {
3148     while (p < e) {
3149         if (isSPACE(*p))
3150             ++p;
3151         else if (*p == '#') {
3152             do {
3153                 p++;
3154             } while (p < e && *p != '\n');
3155         }
3156         else
3157             break;
3158     }
3159     return p;
3160 }
3161
3162 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3163    Character classes ([:foo:]) can also be negated ([:^foo:]).
3164    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3165    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3166    but trigger warnings because they are currently unimplemented. */
3167 STATIC I32
3168 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3169 {
3170     char *posixcc = 0;
3171     I32 namedclass = OOB_NAMEDCLASS;
3172
3173     if (value == '[' && RExC_parse + 1 < RExC_end &&
3174         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3175         (*RExC_parse == ':' ||
3176          *RExC_parse == '=' ||
3177          *RExC_parse == '.')) {
3178         char  c = *RExC_parse;
3179         char* s = RExC_parse++;
3180         
3181         while (RExC_parse < RExC_end && *RExC_parse != c)
3182             RExC_parse++;
3183         if (RExC_parse == RExC_end)
3184             /* Grandfather lone [:, [=, [. */
3185             RExC_parse = s;
3186         else {
3187             char* t = RExC_parse++; /* skip over the c */
3188
3189             if (*RExC_parse == ']') {
3190                 RExC_parse++; /* skip over the ending ] */
3191                 posixcc = s + 1;
3192                 if (*s == ':') {
3193                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3194                     I32 skip = 5; /* the most common skip */
3195
3196                     switch (*posixcc) {
3197                     case 'a':
3198                         if (strnEQ(posixcc, "alnum", 5))
3199                             namedclass =
3200                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3201                         else if (strnEQ(posixcc, "alpha", 5))
3202                             namedclass =
3203                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3204                         else if (strnEQ(posixcc, "ascii", 5))
3205                             namedclass =
3206                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3207                         break;
3208                     case 'b':
3209                         if (strnEQ(posixcc, "blank", 5))
3210                             namedclass =
3211                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3212                         break;
3213                     case 'c':
3214                         if (strnEQ(posixcc, "cntrl", 5))
3215                             namedclass =
3216                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3217                         break;
3218                     case 'd':
3219                         if (strnEQ(posixcc, "digit", 5))
3220                             namedclass =
3221                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3222                         break;
3223                     case 'g':
3224                         if (strnEQ(posixcc, "graph", 5))
3225                             namedclass =
3226                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3227                         break;
3228                     case 'l':
3229                         if (strnEQ(posixcc, "lower", 5))
3230                             namedclass =
3231                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3232                         break;
3233                     case 'p':
3234                         if (strnEQ(posixcc, "print", 5))
3235                             namedclass =
3236                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3237                         else if (strnEQ(posixcc, "punct", 5))
3238                             namedclass =
3239                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3240                         break;
3241                     case 's':
3242                         if (strnEQ(posixcc, "space", 5))
3243                             namedclass =
3244                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3245                         break;
3246                     case 'u':
3247                         if (strnEQ(posixcc, "upper", 5))
3248                             namedclass =
3249                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3250                         break;
3251                     case 'w': /* this is not POSIX, this is the Perl \w */
3252                         if (strnEQ(posixcc, "word", 4)) {
3253                             namedclass =
3254                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3255                             skip = 4;
3256                         }
3257                         break;
3258                     case 'x':
3259                         if (strnEQ(posixcc, "xdigit", 6)) {
3260                             namedclass =
3261                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3262                             skip = 6;
3263                         }
3264                         break;
3265                     }
3266                     if (namedclass == OOB_NAMEDCLASS ||
3267                         posixcc[skip] != ':' ||
3268                         posixcc[skip+1] != ']')
3269                     {
3270                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3271                                       t - s - 1, s + 1);
3272                     }
3273                 } else if (!SIZE_ONLY) {
3274                     /* [[=foo=]] and [[.foo.]] are still future. */
3275
3276                     /* adjust RExC_parse so the warning shows after
3277                        the class closes */
3278                     while (*RExC_parse && *RExC_parse != ']')
3279                         RExC_parse++;
3280                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3281                 }
3282             } else {
3283                 /* Maternal grandfather:
3284                  * "[:" ending in ":" but not in ":]" */
3285                 RExC_parse = s;
3286             }
3287         }
3288     }
3289
3290     return namedclass;
3291 }
3292
3293 STATIC void
3294 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3295 {
3296     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3297         (*RExC_parse == ':' ||
3298          *RExC_parse == '=' ||
3299          *RExC_parse == '.')) {
3300         char *s = RExC_parse;
3301         char  c = *s++;
3302
3303         while(*s && isALNUM(*s))
3304             s++;
3305         if (*s && c == *s && s[1] == ']') {
3306             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3307
3308             /* [[=foo=]] and [[.foo.]] are still future. */
3309             if (c == '=' || c == '.')
3310             {
3311                 /* adjust RExC_parse so the error shows after
3312                    the class closes */
3313                 while (*RExC_parse && *RExC_parse++ != ']')
3314                     ;
3315                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3316             }
3317         }
3318     }
3319 }
3320
3321 STATIC regnode *
3322 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3323 {
3324     register UV value;
3325     register IV prevvalue = OOB_UNICODE;
3326     register IV range = 0;
3327     register regnode *ret;
3328     STRLEN numlen;
3329     IV namedclass;
3330     char *rangebegin = 0;
3331     bool need_class = 0;
3332     SV *listsv = Nullsv;
3333     register char *e;
3334     UV n;
3335     bool optimize_invert = TRUE;
3336
3337     ret = reganode(pRExC_state, ANYOF, 0);
3338
3339     if (!SIZE_ONLY)
3340         ANYOF_FLAGS(ret) = 0;
3341
3342     if (*RExC_parse == '^') {   /* Complement of range. */
3343         RExC_naughty++;
3344         RExC_parse++;
3345         if (!SIZE_ONLY)
3346             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3347     }
3348
3349     if (SIZE_ONLY)
3350         RExC_size += ANYOF_SKIP;
3351     else {
3352         RExC_emit += ANYOF_SKIP;
3353         if (FOLD)
3354             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3355         if (LOC)
3356             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3357         ANYOF_BITMAP_ZERO(ret);
3358         listsv = newSVpvn("# comment\n", 10);
3359     }
3360
3361     if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3362         checkposixcc(pRExC_state);
3363
3364     if (*RExC_parse == ']' || *RExC_parse == '-')
3365         goto charclassloop;             /* allow 1st char to be ] or - */
3366
3367     while (RExC_parse < RExC_end && *RExC_parse != ']') {
3368
3369     charclassloop:
3370
3371         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3372
3373         if (!range)
3374             rangebegin = RExC_parse;
3375         if (UTF) {
3376             value = utf8n_to_uvchr((U8*)RExC_parse,
3377                                    RExC_end - RExC_parse,
3378                                    &numlen, 0);
3379             RExC_parse += numlen;
3380         }
3381         else
3382             value = UCHARAT(RExC_parse++);
3383         if (value == '[')
3384             namedclass = regpposixcc(pRExC_state, value);
3385         else if (value == '\\') {
3386             if (UTF) {
3387                 value = utf8n_to_uvchr((U8*)RExC_parse,
3388                                    RExC_end - RExC_parse,
3389                                    &numlen, 0);
3390                 RExC_parse += numlen;
3391             }
3392             else
3393                 value = UCHARAT(RExC_parse++);
3394             /* Some compilers cannot handle switching on 64-bit integer
3395              * values, therefore value cannot be an UV.  Yes, this will
3396              * be a problem later if we want switch on Unicode.
3397              * A similar issue a little bit later when switching on
3398              * namedclass. --jhi */
3399             switch ((I32)value) {
3400             case 'w':   namedclass = ANYOF_ALNUM;       break;
3401             case 'W':   namedclass = ANYOF_NALNUM;      break;
3402             case 's':   namedclass = ANYOF_SPACE;       break;
3403             case 'S':   namedclass = ANYOF_NSPACE;      break;
3404             case 'd':   namedclass = ANYOF_DIGIT;       break;
3405             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3406             case 'p':
3407             case 'P':
3408                 if (*RExC_parse == '{') {
3409                     e = strchr(RExC_parse++, '}');
3410                     if (!e)
3411                         vFAIL("Missing right brace on \\p{}");
3412                     n = e - RExC_parse;
3413                 }
3414                 else {
3415                     e = RExC_parse;
3416                     n = 1;
3417                 }
3418                 if (!SIZE_ONLY) {
3419                     if (value == 'p')
3420                         Perl_sv_catpvf(aTHX_ listsv,
3421                                        "+utf8::%.*s\n", (int)n, RExC_parse);
3422                     else
3423                         Perl_sv_catpvf(aTHX_ listsv,
3424                                        "!utf8::%.*s\n", (int)n, RExC_parse);
3425                 }
3426                 RExC_parse = e + 1;
3427                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3428                 continue;
3429             case 'n':   value = '\n';                   break;
3430             case 'r':   value = '\r';                   break;
3431             case 't':   value = '\t';                   break;
3432             case 'f':   value = '\f';                   break;
3433             case 'b':   value = '\b';                   break;
3434             case 'e':   value = ASCII_TO_NATIVE('\033');break;
3435             case 'a':   value = ASCII_TO_NATIVE('\007');break;
3436             case 'x':
3437                 if (*RExC_parse == '{') {
3438                     e = strchr(RExC_parse++, '}');
3439                     if (!e)
3440                         vFAIL("Missing right brace on \\x{}");
3441                     numlen = 1;         /* allow underscores */
3442                     value = (UV)scan_hex(RExC_parse,
3443                                          e - RExC_parse,
3444                                          &numlen);
3445                     RExC_parse = e + 1;
3446                 }
3447                 else {
3448                     numlen = 0;         /* disallow underscores */
3449                     value = (UV)scan_hex(RExC_parse, 2, &numlen);
3450                     RExC_parse += numlen;
3451                 }
3452                 break;
3453             case 'c':
3454                 value = UCHARAT(RExC_parse++);
3455                 value = toCTRL(value);
3456                 break;
3457             case '0': case '1': case '2': case '3': case '4':
3458             case '5': case '6': case '7': case '8': case '9':
3459                 numlen = 0;             /* disallow underscores */
3460                 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3461                 RExC_parse += numlen;
3462                 break;
3463             default:
3464                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3465                     vWARN2(RExC_parse,
3466                            "Unrecognized escape \\%c in character class passed through",
3467                            (int)value);
3468                 break;
3469             }
3470         } /* end of \blah */
3471
3472         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3473
3474             if (!SIZE_ONLY && !need_class)
3475                 ANYOF_CLASS_ZERO(ret);
3476
3477             need_class = 1;
3478
3479             /* a bad range like a-\d, a-[:digit:] ? */
3480             if (range) {
3481                 if (!SIZE_ONLY) {
3482                     if (ckWARN(WARN_REGEXP))
3483                         vWARN4(RExC_parse,
3484                                "False [] range \"%*.*s\"",
3485                                RExC_parse - rangebegin,
3486                                RExC_parse - rangebegin,
3487                                rangebegin);
3488                     if (prevvalue < 256) {
3489                         ANYOF_BITMAP_SET(ret, prevvalue);
3490                         ANYOF_BITMAP_SET(ret, '-');
3491                     }
3492                     else {
3493                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3494                         Perl_sv_catpvf(aTHX_ listsv,
3495                                        "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3496                     }
3497                 }
3498
3499                 range = 0; /* this was not a true range */
3500             }
3501
3502             if (!SIZE_ONLY) {
3503                 if (namedclass > OOB_NAMEDCLASS)
3504                     optimize_invert = FALSE;
3505                 /* Possible truncation here but in some 64-bit environments
3506                  * the compiler gets heartburn about switch on 64-bit values.
3507                  * A similar issue a little earlier when switching on value.
3508                  * --jhi */
3509                 switch ((I32)namedclass) {
3510                 case ANYOF_ALNUM:
3511                     if (LOC)
3512                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3513                     else {
3514                         for (value = 0; value < 256; value++)
3515                             if (isALNUM(value))
3516                                 ANYOF_BITMAP_SET(ret, value);
3517                     }
3518                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3519                     break;
3520                 case ANYOF_NALNUM:
3521                     if (LOC)
3522                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
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_ALNUMC:
3531                     if (LOC)
3532                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3533                     else {
3534                         for (value = 0; value < 256; value++)
3535                             if (isALNUMC(value))
3536                                 ANYOF_BITMAP_SET(ret, value);
3537                     }
3538                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3539                     break;
3540                 case ANYOF_NALNUMC:
3541                     if (LOC)
3542                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
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_ALPHA:
3551                     if (LOC)
3552                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3553                     else {
3554                         for (value = 0; value < 256; value++)
3555                             if (isALPHA(value))
3556                                 ANYOF_BITMAP_SET(ret, value);
3557                     }
3558                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3559                     break;
3560                 case ANYOF_NALPHA:
3561                     if (LOC)
3562                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
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_ASCII:
3571                     if (LOC)
3572                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3573                     else {
3574 #ifndef EBCDIC
3575                         for (value = 0; value < 128; value++)
3576                             ANYOF_BITMAP_SET(ret, value);
3577 #else  /* EBCDIC */
3578                         for (value = 0; value < 256; value++) {
3579                             if (isASCII(value))
3580                                 ANYOF_BITMAP_SET(ret, value);
3581                         }
3582 #endif /* EBCDIC */
3583                     }
3584                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3585                     break;
3586                 case ANYOF_NASCII:
3587                     if (LOC)
3588                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3589                     else {
3590 #ifndef EBCDIC
3591                         for (value = 128; value < 256; value++)
3592                             ANYOF_BITMAP_SET(ret, value);
3593 #else  /* EBCDIC */
3594                         for (value = 0; value < 256; value++) {
3595                             if (!isASCII(value))
3596                                 ANYOF_BITMAP_SET(ret, value);
3597                         }
3598 #endif /* EBCDIC */
3599                     }
3600                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3601                     break;
3602                 case ANYOF_BLANK:
3603                     if (LOC)
3604                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3605                     else {
3606                         for (value = 0; value < 256; value++)
3607                             if (isBLANK(value))
3608                                 ANYOF_BITMAP_SET(ret, value);
3609                     }
3610                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3611                     break;
3612                 case ANYOF_NBLANK:
3613                     if (LOC)
3614                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
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_CNTRL:
3623                     if (LOC)
3624                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3625                     else {
3626                         for (value = 0; value < 256; value++)
3627                             if (isCNTRL(value))
3628                                 ANYOF_BITMAP_SET(ret, value);
3629                     }
3630                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3631                     break;
3632                 case ANYOF_NCNTRL:
3633                     if (LOC)
3634                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
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_DIGIT:
3643                     if (LOC)
3644                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3645                     else {
3646                         /* consecutive digits assumed */
3647                         for (value = '0'; value <= '9'; value++)
3648                             ANYOF_BITMAP_SET(ret, value);
3649                     }
3650                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3651                     break;
3652                 case ANYOF_NDIGIT:
3653                     if (LOC)
3654                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3655                     else {
3656                         /* consecutive digits assumed */
3657                         for (value = 0; value < '0'; value++)
3658                             ANYOF_BITMAP_SET(ret, value);
3659                         for (value = '9' + 1; value < 256; value++)
3660                             ANYOF_BITMAP_SET(ret, value);
3661                     }
3662                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3663                     break;
3664                 case ANYOF_GRAPH:
3665                     if (LOC)
3666                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3667                     else {
3668                         for (value = 0; value < 256; value++)
3669                             if (isGRAPH(value))
3670                                 ANYOF_BITMAP_SET(ret, value);
3671                     }
3672                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3673                     break;
3674                 case ANYOF_NGRAPH:
3675                     if (LOC)
3676                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
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_LOWER:
3685                     if (LOC)
3686                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3687                     else {
3688                         for (value = 0; value < 256; value++)
3689                             if (isLOWER(value))
3690                                 ANYOF_BITMAP_SET(ret, value);
3691                     }
3692                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3693                     break;
3694                 case ANYOF_NLOWER:
3695                     if (LOC)
3696                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
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_PRINT:
3705                     if (LOC)
3706                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3707                     else {
3708                         for (value = 0; value < 256; value++)
3709                             if (isPRINT(value))
3710                                 ANYOF_BITMAP_SET(ret, value);
3711                     }
3712                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3713                     break;
3714                 case ANYOF_NPRINT:
3715                     if (LOC)
3716                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
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_PSXSPC:
3725                     if (LOC)
3726                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3727                     else {
3728                         for (value = 0; value < 256; value++)
3729                             if (isPSXSPC(value))
3730                                 ANYOF_BITMAP_SET(ret, value);
3731                     }
3732                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3733                     break;
3734                 case ANYOF_NPSXSPC:
3735                     if (LOC)
3736                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
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_PUNCT:
3745                     if (LOC)
3746                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3747                     else {
3748                         for (value = 0; value < 256; value++)
3749                             if (isPUNCT(value))
3750                                 ANYOF_BITMAP_SET(ret, value);
3751                     }
3752                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3753                     break;
3754                 case ANYOF_NPUNCT:
3755                     if (LOC)
3756                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
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_SPACE:
3765                     if (LOC)
3766                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3767                     else {
3768                         for (value = 0; value < 256; value++)
3769                             if (isSPACE(value))
3770                                 ANYOF_BITMAP_SET(ret, value);
3771                     }
3772                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3773                     break;
3774                 case ANYOF_NSPACE:
3775                     if (LOC)
3776                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
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_UPPER:
3785                     if (LOC)
3786                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3787                     else {
3788                         for (value = 0; value < 256; value++)
3789                             if (isUPPER(value))
3790                                 ANYOF_BITMAP_SET(ret, value);
3791                     }
3792                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3793                     break;
3794                 case ANYOF_NUPPER:
3795                     if (LOC)
3796                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
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_XDIGIT:
3805                     if (LOC)
3806                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3807                     else {
3808                         for (value = 0; value < 256; value++)
3809                             if (isXDIGIT(value))
3810                                 ANYOF_BITMAP_SET(ret, value);
3811                     }
3812                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3813                     break;
3814                 case ANYOF_NXDIGIT:
3815                     if (LOC)
3816                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
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                 default:
3825                     vFAIL("Invalid [::] class");
3826                     break;
3827                 }
3828                 if (LOC)
3829                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3830                 continue;
3831             }
3832         } /* end of namedclass \blah */
3833
3834         if (range) {
3835             if (prevvalue > value) /* b-a */ {
3836                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3837                               RExC_parse - rangebegin,
3838                               RExC_parse - rangebegin,
3839                               rangebegin);
3840                 range = 0; /* not a valid range */
3841             }
3842         }
3843         else {
3844             prevvalue = value; /* save the beginning of the range */
3845             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3846                 RExC_parse[1] != ']') {
3847                 RExC_parse++;
3848
3849                 /* a bad range like \w-, [:word:]- ? */
3850                 if (namedclass > OOB_NAMEDCLASS) {
3851                     if (ckWARN(WARN_REGEXP))
3852                         vWARN4(RExC_parse,
3853                                "False [] range \"%*.*s\"",
3854                                RExC_parse - rangebegin,
3855                                RExC_parse - rangebegin,
3856                                rangebegin);
3857                     if (!SIZE_ONLY)
3858                         ANYOF_BITMAP_SET(ret, '-');
3859                 } else
3860                     range = 1;  /* yeah, it's a range! */
3861                 continue;       /* but do it the next time */
3862             }
3863         }
3864
3865         /* now is the next time */
3866         if (!SIZE_ONLY) {
3867             IV i;
3868
3869             if (prevvalue < 256) {
3870                 IV ceilvalue = value < 256 ? value : 255;
3871
3872 #ifdef EBCDIC
3873                 if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3874                     (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3875                 {
3876                     if (isLOWER(prevvalue)) {
3877                         for (i = prevvalue; i <= ceilvalue; i++)
3878                             if (isLOWER(i))
3879                                 ANYOF_BITMAP_SET(ret, i);
3880                     } else {
3881                         for (i = prevvalue; i <= ceilvalue; i++)
3882                             if (isUPPER(i))
3883                                 ANYOF_BITMAP_SET(ret, i);
3884                     }
3885                 }
3886                 else
3887 #endif
3888                     for (i = prevvalue; i <= ceilvalue; i++)
3889                         ANYOF_BITMAP_SET(ret, i);
3890           }
3891           if (value > 255) {
3892                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3893                 if (prevvalue < value)
3894                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3895                                    (UV)prevvalue, (UV)value);
3896                 else if (prevvalue == value)
3897                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3898                                    (UV)value);
3899             }
3900         }
3901
3902         range = 0; /* this range (if it was one) is done now */
3903     }
3904
3905     if (need_class) {
3906         ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3907         if (SIZE_ONLY)
3908             RExC_size += ANYOF_CLASS_ADD_SKIP;
3909         else
3910             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3911     }
3912
3913     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3914     if (!SIZE_ONLY &&
3915          /* If the only flag is folding (plus possibly inversion). */
3916         ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
3917        ) {
3918         for (value = 0; value < 256; ++value) {
3919             if (ANYOF_BITMAP_TEST(ret, value)) {
3920                 IV fold = PL_fold[value];
3921
3922                 if (fold != value)
3923                     ANYOF_BITMAP_SET(ret, fold);
3924             }
3925         }
3926         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3927     }
3928
3929     /* optimize inverted simple patterns (e.g. [^a-z]) */
3930     if (!SIZE_ONLY && optimize_invert &&
3931         /* If the only flag is inversion. */
3932         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3933         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3934             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3935         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3936     }
3937
3938     if (!SIZE_ONLY) {
3939         AV *av = newAV();
3940         SV *rv;
3941
3942         av_store(av, 0, listsv);
3943         av_store(av, 1, NULL);
3944         rv = newRV_noinc((SV*)av);
3945         n = add_data(pRExC_state, 1, "s");
3946         RExC_rx->data->data[n] = (void*)rv;
3947         ARG_SET(ret, n);
3948     }
3949
3950     return ret;
3951 }
3952
3953 STATIC char*
3954 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3955 {
3956     char* retval = RExC_parse++;
3957
3958     for (;;) {
3959         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3960                 RExC_parse[2] == '#') {
3961             while (*RExC_parse && *RExC_parse != ')')
3962                 RExC_parse++;
3963             RExC_parse++;
3964             continue;
3965         }
3966         if (RExC_flags16 & PMf_EXTENDED) {
3967             if (isSPACE(*RExC_parse)) {
3968                 RExC_parse++;
3969                 continue;
3970             }
3971             else if (*RExC_parse == '#') {
3972                 while (*RExC_parse && *RExC_parse != '\n')
3973                     RExC_parse++;
3974                 RExC_parse++;
3975                 continue;
3976             }
3977         }
3978         return retval;
3979     }
3980 }
3981
3982 /*
3983 - reg_node - emit a node
3984 */
3985 STATIC regnode *                        /* Location. */
3986 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3987 {
3988     register regnode *ret;
3989     register regnode *ptr;
3990
3991     ret = RExC_emit;
3992     if (SIZE_ONLY) {
3993         SIZE_ALIGN(RExC_size);
3994         RExC_size += 1;
3995         return(ret);
3996     }
3997
3998     NODE_ALIGN_FILL(ret);
3999     ptr = ret;
4000     FILL_ADVANCE_NODE(ptr, op);
4001     if (RExC_offsets) {         /* MJD */
4002       MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n", 
4003               "reg_node", __LINE__, 
4004               reg_name[op],
4005               RExC_emit - RExC_emit_start > RExC_offsets[0] 
4006               ? "Overwriting end of array!\n" : "OK",
4007               RExC_emit - RExC_emit_start,
4008               RExC_parse - RExC_start,
4009               RExC_offsets[0])); 
4010       Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
4011     }
4012             
4013     RExC_emit = ptr;
4014
4015     return(ret);
4016 }
4017
4018 /*
4019 - reganode - emit a node with an argument
4020 */
4021 STATIC regnode *                        /* Location. */
4022 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
4023 {
4024     register regnode *ret;
4025     register regnode *ptr;
4026
4027     ret = RExC_emit;
4028     if (SIZE_ONLY) {
4029         SIZE_ALIGN(RExC_size);
4030         RExC_size += 2;
4031         return(ret);
4032     }
4033
4034     NODE_ALIGN_FILL(ret);
4035     ptr = ret;
4036     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
4037     if (RExC_offsets) {         /* MJD */
4038       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4039               "reganode",
4040               RExC_emit - RExC_emit_start > RExC_offsets[0] ? 
4041               "Overwriting end of array!\n" : "OK",
4042               RExC_emit - RExC_emit_start,
4043               RExC_parse - RExC_start,
4044               RExC_offsets[0])); 
4045       Set_Cur_Node_Offset;
4046     }
4047             
4048     RExC_emit = ptr;
4049
4050     return(ret);
4051 }
4052
4053 /*
4054 - reguni - emit (if appropriate) a Unicode character
4055 */
4056 STATIC void
4057 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4058 {
4059     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4060 }
4061
4062 /*
4063 - reginsert - insert an operator in front of already-emitted operand
4064 *
4065 * Means relocating the operand.
4066 */
4067 STATIC void
4068 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4069 {
4070     register regnode *src;
4071     register regnode *dst;
4072     register regnode *place;
4073     register int offset = regarglen[(U8)op];
4074
4075 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4076
4077     if (SIZE_ONLY) {
4078         RExC_size += NODE_STEP_REGNODE + offset;
4079         return;
4080     }
4081
4082     src = RExC_emit;
4083     RExC_emit += NODE_STEP_REGNODE + offset;
4084     dst = RExC_emit;
4085     while (src > opnd) {
4086         StructCopy(--src, --dst, regnode);
4087         if (RExC_offsets) {     /* MJD 20010112 */
4088           MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n", 
4089                   "reg_insert",
4090                   dst - RExC_emit_start > RExC_offsets[0] 
4091                   ? "Overwriting end of array!\n" : "OK",
4092                   src - RExC_emit_start,
4093                   dst - RExC_emit_start,
4094                   RExC_offsets[0])); 
4095           Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4096           Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4097         }
4098     }
4099     
4100
4101     place = opnd;               /* Op node, where operand used to be. */
4102     if (RExC_offsets) {         /* MJD */
4103       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4104               "reginsert",
4105               place - RExC_emit_start > RExC_offsets[0] 
4106               ? "Overwriting end of array!\n" : "OK",
4107               place - RExC_emit_start,
4108               RExC_parse - RExC_start,
4109               RExC_offsets[0])); 
4110       Set_Node_Offset(place, RExC_parse);
4111     }
4112     src = NEXTOPER(place);
4113     FILL_ADVANCE_NODE(place, op);
4114     Zero(src, offset, regnode);
4115 }
4116
4117 /*
4118 - regtail - set the next-pointer at the end of a node chain of p to val.
4119 */
4120 STATIC void
4121 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4122 {
4123     register regnode *scan;
4124     register regnode *temp;
4125
4126     if (SIZE_ONLY)
4127         return;
4128
4129     /* Find last node. */
4130     scan = p;
4131     for (;;) {
4132         temp = regnext(scan);
4133         if (temp == NULL)
4134             break;
4135         scan = temp;
4136     }
4137
4138     if (reg_off_by_arg[OP(scan)]) {
4139         ARG_SET(scan, val - scan);
4140     }
4141     else {
4142         NEXT_OFF(scan) = val - scan;
4143     }
4144 }
4145
4146 /*
4147 - regoptail - regtail on operand of first argument; nop if operandless
4148 */
4149 STATIC void
4150 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4151 {
4152     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4153     if (p == NULL || SIZE_ONLY)
4154         return;
4155     if (PL_regkind[(U8)OP(p)] == BRANCH) {
4156         regtail(pRExC_state, NEXTOPER(p), val);
4157     }
4158     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4159         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4160     }
4161     else
4162         return;
4163 }
4164
4165 /*
4166  - regcurly - a little FSA that accepts {\d+,?\d*}
4167  */
4168 STATIC I32
4169 S_regcurly(pTHX_ register char *s)
4170 {
4171     if (*s++ != '{')
4172         return FALSE;
4173     if (!isDIGIT(*s))
4174         return FALSE;
4175     while (isDIGIT(*s))
4176         s++;
4177     if (*s == ',')
4178         s++;
4179     while (isDIGIT(*s))
4180         s++;
4181     if (*s != '}')
4182         return FALSE;
4183     return TRUE;
4184 }
4185
4186
4187 #ifdef DEBUGGING
4188
4189 STATIC regnode *
4190 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4191 {
4192     register U8 op = EXACT;     /* Arbitrary non-END op. */
4193     register regnode *next;
4194
4195     while (op != END && (!last || node < last)) {
4196         /* While that wasn't END last time... */
4197
4198         NODE_ALIGN(node);
4199         op = OP(node);
4200         if (op == CLOSE)
4201             l--;        
4202         next = regnext(node);
4203         /* Where, what. */
4204         if (OP(node) == OPTIMIZED)
4205             goto after_print;
4206         regprop(sv, node);
4207         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4208                       (int)(2*l + 1), "", SvPVX(sv));
4209         if (next == NULL)               /* Next ptr. */
4210             PerlIO_printf(Perl_debug_log, "(0)");
4211         else
4212             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4213         (void)PerlIO_putc(Perl_debug_log, '\n');
4214       after_print:
4215         if (PL_regkind[(U8)op] == BRANCHJ) {
4216             register regnode *nnode = (OP(next) == LONGJMP
4217                                        ? regnext(next)
4218                                        : next);
4219             if (last && nnode > last)
4220                 nnode = last;
4221             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4222         }
4223         else if (PL_regkind[(U8)op] == BRANCH) {
4224             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4225         }
4226         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4227             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4228                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4229         }
4230         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4231             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4232                              next, sv, l + 1);
4233         }
4234         else if ( op == PLUS || op == STAR) {
4235             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4236         }
4237         else if (op == ANYOF) {
4238             /* arglen 1 + class block */
4239             node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4240                     ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4241             node = NEXTOPER(node);
4242         }
4243         else if (PL_regkind[(U8)op] == EXACT) {
4244             /* Literal string, where present. */
4245             node += NODE_SZ_STR(node) - 1;
4246             node = NEXTOPER(node);
4247         }
4248         else {
4249             node = NEXTOPER(node);
4250             node += regarglen[(U8)op];
4251         }
4252         if (op == CURLYX || op == OPEN)
4253             l++;
4254         else if (op == WHILEM)
4255             l--;
4256     }
4257     return node;
4258 }
4259
4260 #endif  /* DEBUGGING */
4261
4262 /*
4263  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4264  */
4265 void
4266 Perl_regdump(pTHX_ regexp *r)
4267 {
4268 #ifdef DEBUGGING
4269     SV *sv = sv_newmortal();
4270
4271     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4272
4273     /* Header fields of interest. */
4274     if (r->anchored_substr)
4275         PerlIO_printf(Perl_debug_log,
4276                       "anchored `%s%.*s%s'%s at %"IVdf" ",
4277                       PL_colors[0],
4278                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4279                       SvPVX(r->anchored_substr),
4280                       PL_colors[1],
4281                       SvTAIL(r->anchored_substr) ? "$" : "",
4282                       (IV)r->anchored_offset);
4283     if (r->float_substr)
4284         PerlIO_printf(Perl_debug_log,
4285                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4286                       PL_colors[0],
4287                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4288                       SvPVX(r->float_substr),
4289                       PL_colors[1],
4290                       SvTAIL(r->float_substr) ? "$" : "",
4291                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4292     if (r->check_substr)
4293         PerlIO_printf(Perl_debug_log,
4294                       r->check_substr == r->float_substr
4295                       ? "(checking floating" : "(checking anchored");
4296     if (r->reganch & ROPT_NOSCAN)
4297         PerlIO_printf(Perl_debug_log, " noscan");
4298     if (r->reganch & ROPT_CHECK_ALL)
4299         PerlIO_printf(Perl_debug_log, " isall");
4300     if (r->check_substr)
4301         PerlIO_printf(Perl_debug_log, ") ");
4302
4303     if (r->regstclass) {
4304         regprop(sv, r->regstclass);
4305         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4306     }
4307     if (r->reganch & ROPT_ANCH) {
4308         PerlIO_printf(Perl_debug_log, "anchored");
4309         if (r->reganch & ROPT_ANCH_BOL)
4310             PerlIO_printf(Perl_debug_log, "(BOL)");
4311         if (r->reganch & ROPT_ANCH_MBOL)
4312             PerlIO_printf(Perl_debug_log, "(MBOL)");
4313         if (r->reganch & ROPT_ANCH_SBOL)
4314             PerlIO_printf(Perl_debug_log, "(SBOL)");
4315         if (r->reganch & ROPT_ANCH_GPOS)
4316             PerlIO_printf(Perl_debug_log, "(GPOS)");
4317         PerlIO_putc(Perl_debug_log, ' ');
4318     }
4319     if (r->reganch & ROPT_GPOS_SEEN)
4320         PerlIO_printf(Perl_debug_log, "GPOS ");
4321     if (r->reganch & ROPT_SKIP)
4322         PerlIO_printf(Perl_debug_log, "plus ");
4323     if (r->reganch & ROPT_IMPLICIT)
4324         PerlIO_printf(Perl_debug_log, "implicit ");
4325     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4326     if (r->reganch & ROPT_EVAL_SEEN)
4327         PerlIO_printf(Perl_debug_log, "with eval ");
4328     PerlIO_printf(Perl_debug_log, "\n");
4329     if (r->offsets) {
4330       U32 i;
4331       U32 len = r->offsets[0];
4332       PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
4333       for (i = 1; i <= len; i++)
4334         PerlIO_printf(Perl_debug_log, "%"UVuf"[%"UVuf"] ", 
4335                       (UV)r->offsets[i*2-1], 
4336                       (UV)r->offsets[i*2]);
4337       PerlIO_printf(Perl_debug_log, "\n");
4338     }
4339 #endif  /* DEBUGGING */
4340 }
4341
4342 #ifdef DEBUGGING
4343
4344 STATIC void
4345 S_put_byte(pTHX_ SV *sv, int c)
4346 {
4347     if (isCNTRL(c) || c == 127 || c == 255 || !isPRINT(c))
4348         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4349     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4350         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4351     else
4352         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4353 }
4354
4355 #endif  /* DEBUGGING */
4356
4357 /*
4358 - regprop - printable representation of opcode
4359 */
4360 void
4361 Perl_regprop(pTHX_ SV *sv, regnode *o)
4362 {
4363 #ifdef DEBUGGING
4364     register int k;
4365
4366     sv_setpvn(sv, "", 0);
4367     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4368         /* It would be nice to FAIL() here, but this may be called from
4369            regexec.c, and it would be hard to supply pRExC_state. */
4370         Perl_croak(aTHX_ "Corrupted regexp opcode");
4371     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4372
4373     k = PL_regkind[(U8)OP(o)];
4374
4375     if (k == EXACT)
4376         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4377                        STR_LEN(o), STRING(o), PL_colors[1]);
4378     else if (k == CURLY) {
4379         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4380             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4381         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4382     }
4383     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4384         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4385     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4386         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4387     else if (k == LOGICAL)
4388         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4389     else if (k == ANYOF) {
4390         int i, rangestart = -1;
4391         U8 flags = ANYOF_FLAGS(o);
4392         const char * const anyofs[] = { /* Should be syncronized with
4393                                          * ANYOF_ #xdefines in regcomp.h */
4394             "\\w",
4395             "\\W",
4396             "\\s",
4397             "\\S",
4398             "\\d",
4399             "\\D",
4400             "[:alnum:]",
4401             "[:^alnum:]",
4402             "[:alpha:]",
4403             "[:^alpha:]",
4404             "[:ascii:]",
4405             "[:^ascii:]",
4406             "[:ctrl:]",
4407             "[:^ctrl:]",
4408             "[:graph:]",
4409             "[:^graph:]",
4410             "[:lower:]",
4411             "[:^lower:]",
4412             "[:print:]",
4413             "[:^print:]",
4414             "[:punct:]",
4415             "[:^punct:]",
4416             "[:upper:]",
4417             "[:^upper:]",
4418             "[:xdigit:]",
4419             "[:^xdigit:]",
4420             "[:space:]",
4421             "[:^space:]",
4422             "[:blank:]",
4423             "[:^blank:]"
4424         };
4425
4426         if (flags & ANYOF_LOCALE)
4427             sv_catpv(sv, "{loc}");
4428         if (flags & ANYOF_FOLD)
4429             sv_catpv(sv, "{i}");
4430         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4431         if (flags & ANYOF_INVERT)
4432             sv_catpv(sv, "^");
4433         for (i = 0; i <= 256; i++) {
4434             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4435                 if (rangestart == -1)
4436                     rangestart = i;
4437             } else if (rangestart != -1) {
4438                 if (i <= rangestart + 3)
4439                     for (; rangestart < i; rangestart++)
4440                         put_byte(sv, rangestart);
4441                 else {
4442                     put_byte(sv, rangestart);
4443                     sv_catpv(sv, "-");
4444                     put_byte(sv, i - 1);
4445                 }
4446                 rangestart = -1;
4447             }
4448         }
4449
4450         if (o->flags & ANYOF_CLASS)
4451             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4452                 if (ANYOF_CLASS_TEST(o,i))
4453                     sv_catpv(sv, anyofs[i]);
4454
4455         if (flags & ANYOF_UNICODE)
4456             sv_catpv(sv, "{unicode}");
4457         else if (flags & ANYOF_UNICODE_ALL)
4458             sv_catpv(sv, "{all-unicode}");
4459
4460         {
4461             SV *lv;
4462             SV *sw = regclass_swash(o, FALSE, &lv);
4463         
4464             if (lv) {
4465                 if (sw) {
4466                     UV i;
4467                     U8 s[UTF8_MAXLEN+1];
4468                 
4469                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4470                         U8 *e = uvchr_to_utf8(s, i);
4471                         
4472                         if (i < 256 && swash_fetch(sw, s, TRUE)) {
4473                             if (rangestart == -1)
4474                                 rangestart = i;
4475                         } else if (rangestart != -1) {
4476                             U8 *p;
4477                         
4478                             if (i <= rangestart + 3)
4479                                 for (; rangestart < i; rangestart++) {
4480                                     for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4481                                         put_byte(sv, *p);
4482                                 }
4483                             else {
4484                                 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4485                                     put_byte(sv, *p);
4486                                 sv_catpv(sv, "-");
4487                                     for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4488                                         put_byte(sv, *p);
4489                                 }
4490                                 rangestart = -1;
4491                             }
4492                         }
4493                         
4494                     sv_catpv(sv, "..."); /* et cetera */
4495                 }
4496
4497                 {
4498                     char *s = savepv(SvPVX(lv));
4499                     char *origs = s;
4500                 
4501                     while(*s && *s != '\n') s++;
4502                 
4503                     if (*s == '\n') {
4504                         char *t = ++s;
4505                         
4506                         while (*s) {
4507                             if (*s == '\n')
4508                                 *s = ' ';
4509                             s++;
4510                         }
4511                         if (s[-1] == ' ')
4512                             s[-1] = 0;
4513                         
4514                         sv_catpv(sv, t);
4515                     }
4516                 
4517                     Safefree(origs);
4518                 }
4519             }
4520         }
4521
4522         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4523     }
4524     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4525         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4526 #endif  /* DEBUGGING */
4527 }
4528
4529 SV *
4530 Perl_re_intuit_string(pTHX_ regexp *prog)
4531 {                               /* Assume that RE_INTUIT is set */
4532     DEBUG_r(
4533         {   STRLEN n_a;
4534             char *s = SvPV(prog->check_substr,n_a);
4535
4536             if (!PL_colorset) reginitcolors();
4537             PerlIO_printf(Perl_debug_log,
4538                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4539                       PL_colors[4],PL_colors[5],PL_colors[0],
4540                       s,
4541                       PL_colors[1],
4542                       (strlen(s) > 60 ? "..." : ""));
4543         } );
4544
4545     return prog->check_substr;
4546 }
4547
4548 void
4549 Perl_pregfree(pTHX_ struct regexp *r)
4550 {
4551     DEBUG_r(if (!PL_colorset) reginitcolors());
4552
4553     if (!r || (--r->refcnt > 0))
4554         return;
4555     DEBUG_r(PerlIO_printf(Perl_debug_log,
4556                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4557                       PL_colors[4],PL_colors[5],PL_colors[0],
4558                       r->precomp,
4559                       PL_colors[1],
4560                       (strlen(r->precomp) > 60 ? "..." : "")));
4561
4562     if (r->precomp)
4563         Safefree(r->precomp);
4564     if (r->offsets)             /* 20010421 MJD */
4565         Safefree(r->offsets);
4566     if (RX_MATCH_COPIED(r))
4567         Safefree(r->subbeg);
4568     if (r->substrs) {
4569         if (r->anchored_substr)
4570             SvREFCNT_dec(r->anchored_substr);
4571         if (r->float_substr)
4572             SvREFCNT_dec(r->float_substr);
4573         Safefree(r->substrs);
4574     }
4575     if (r->data) {
4576         int n = r->data->count;
4577         AV* new_comppad = NULL;
4578         AV* old_comppad;
4579         SV** old_curpad;
4580
4581         while (--n >= 0) {
4582             switch (r->data->what[n]) {
4583             case 's':
4584                 SvREFCNT_dec((SV*)r->data->data[n]);
4585                 break;
4586             case 'f':
4587                 Safefree(r->data->data[n]);
4588                 break;
4589             case 'p':
4590                 new_comppad = (AV*)r->data->data[n];
4591                 break;
4592             case 'o':
4593                 if (new_comppad == NULL)
4594                     Perl_croak(aTHX_ "panic: pregfree comppad");
4595                 old_comppad = PL_comppad;
4596                 old_curpad = PL_curpad;
4597                 /* Watch out for global destruction's random ordering. */
4598                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4599                     PL_comppad = new_comppad;
4600                     PL_curpad = AvARRAY(new_comppad);
4601                 }
4602                 else
4603                     PL_curpad = NULL;
4604                 op_free((OP_4tree*)r->data->data[n]);
4605                 PL_comppad = old_comppad;
4606                 PL_curpad = old_curpad;
4607                 SvREFCNT_dec((SV*)new_comppad);
4608                 new_comppad = NULL;
4609                 break;
4610             case 'n':
4611                 break;
4612             default:
4613                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4614             }
4615         }
4616         Safefree(r->data->what);
4617         Safefree(r->data);
4618     }
4619     Safefree(r->startp);
4620     Safefree(r->endp);
4621     Safefree(r);
4622 }
4623
4624 /*
4625  - regnext - dig the "next" pointer out of a node
4626  *
4627  * [Note, when REGALIGN is defined there are two places in regmatch()
4628  * that bypass this code for speed.]
4629  */
4630 regnode *
4631 Perl_regnext(pTHX_ register regnode *p)
4632 {
4633     register I32 offset;
4634
4635     if (p == &PL_regdummy)
4636         return(NULL);
4637
4638     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4639     if (offset == 0)
4640         return(NULL);
4641
4642     return(p+offset);
4643 }
4644
4645 STATIC void     
4646 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4647 {
4648     va_list args;
4649     STRLEN l1 = strlen(pat1);
4650     STRLEN l2 = strlen(pat2);
4651     char buf[512];
4652     SV *msv;
4653     char *message;
4654
4655     if (l1 > 510)
4656         l1 = 510;
4657     if (l1 + l2 > 510)
4658         l2 = 510 - l1;
4659     Copy(pat1, buf, l1 , char);
4660     Copy(pat2, buf + l1, l2 , char);
4661     buf[l1 + l2] = '\n';
4662     buf[l1 + l2 + 1] = '\0';
4663 #ifdef I_STDARG
4664     /* ANSI variant takes additional second argument */
4665     va_start(args, pat2);
4666 #else
4667     va_start(args);
4668 #endif
4669     msv = vmess(buf, &args);
4670     va_end(args);
4671     message = SvPV(msv,l1);
4672     if (l1 > 512)
4673         l1 = 512;
4674     Copy(message, buf, l1 , char);
4675     buf[l1] = '\0';                     /* Overwrite \n */
4676     Perl_croak(aTHX_ "%s", buf);
4677 }
4678
4679 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4680
4681 void
4682 Perl_save_re_context(pTHX)
4683 {
4684 #if 0
4685     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4686     SAVEI32(RExC_npar);         /* () count. */
4687     SAVEI32(RExC_size);         /* Code size. */
4688     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4689     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4690     SAVEI32(RExC_seen);         /* from regcomp.c */
4691     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4692     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4693     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4694     SAVEPPTR(RExC_end);         /* End of input for compile */
4695     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4696 #endif
4697
4698     SAVEI32(PL_reg_flags);              /* from regexec.c */
4699     SAVEPPTR(PL_bostr);
4700     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4701     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4702     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4703     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4704     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4705     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4706     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4707     SAVEGENERICPV(PL_reg_start_tmp);            /* from regexec.c */
4708     PL_reg_start_tmp = 0;
4709     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4710     PL_reg_start_tmpl = 0;
4711     SAVEVPTR(PL_regdata);
4712     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4713     SAVEI32(PL_regnarrate);             /* from regexec.c */
4714     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4715     SAVEINT(PL_regindent);              /* from regexec.c */
4716     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4717     SAVEVPTR(PL_curcop);
4718     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4719     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4720     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4721     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4722     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4723     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4724     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4725     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4726     SAVEI32(PL_regnpar);                /* () count. */
4727     SAVEI32(PL_regsize);                /* from regexec.c */
4728 #ifdef DEBUGGING
4729     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */
4730 #endif
4731 }
4732
4733 #ifdef PERL_OBJECT
4734 #include "XSUB.h"
4735 #undef this
4736 #define this pPerl
4737 #endif
4738
4739 static void
4740 clear_re(pTHXo_ void *r)
4741 {
4742     ReREFCNT_dec((regexp *)r);
4743 }
4744