This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Win32/distclean.bat - Down-case all MANIFEST and File::Find results so
[perl5.git] / regcomp.c
1 /*    regcomp.c
2  */
3
4 /*
5  * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
6  */
7
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9  * confused with the original package (see point 3 below).  Thanks, Henry!
10  */
11
12 /* Additional note: this code is very heavily munged from Henry's version
13  * in places.  In some spots I've traded clarity for efficiency, so don't
14  * blame Henry for some of the lack of readability.
15  */
16
17 /* The names of the functions have been changed from regcomp and
18  * regexec to  pregcomp and pregexec in order to avoid conflicts
19  * with the POSIX routines of the same names.
20 */
21
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 #  ifndef PERL_IN_XSUB_RE
25 #    define PERL_IN_XSUB_RE
26 #  endif
27 /* need access to debugger hooks */
28 #  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
29 #    define DEBUGGING
30 #  endif
31 #endif
32
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 #  define Perl_pregcomp my_regcomp
36 #  define Perl_regdump my_regdump
37 #  define Perl_regprop my_regprop
38 #  define Perl_pregfree my_regfree
39 #  define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 #  define Perl_regnext my_regnext
42 #  define Perl_save_re_context my_save_re_context
43 #  define Perl_reginitcolors my_reginitcolors
44
45 #  define PERL_NO_GET_CONTEXT
46 #endif
47
48 /*SUPPRESS 112*/
49 /*
50  * pregcomp and pregexec -- regsub and regerror are not used in perl
51  *
52  *      Copyright (c) 1986 by University of Toronto.
53  *      Written by Henry Spencer.  Not derived from licensed software.
54  *
55  *      Permission is granted to anyone to use this software for any
56  *      purpose on any computer system, and to redistribute it freely,
57  *      subject to the following restrictions:
58  *
59  *      1. The author is not responsible for the consequences of use of
60  *              this software, no matter how awful, even if they arise
61  *              from defects in it.
62  *
63  *      2. The origin of this software must not be misrepresented, either
64  *              by explicit claim or by omission.
65  *
66  *      3. Altered versions must be plainly marked as such, and must not
67  *              be misrepresented as being the original software.
68  *
69  *
70  ****    Alterations to Henry's code are...
71  ****
72  ****    Copyright (c) 1991-2001, Larry Wall
73  ****
74  ****    You may distribute under the terms of either the GNU General Public
75  ****    License or the Artistic License, as specified in the README file.
76
77  *
78  * Beware that some of this code is subtly aware of the way operator
79  * precedence is structured in regular expressions.  Serious changes in
80  * regular-expression syntax might require a total rethink.
81  */
82 #include "EXTERN.h"
83 #define PERL_IN_REGCOMP_C
84 #include "perl.h"
85
86 #ifdef PERL_IN_XSUB_RE
87 #  if defined(PERL_CAPI) || defined(PERL_OBJECT)
88 #    include "XSUB.h"
89 #  endif
90 #else
91 #  include "INTERN.h"
92 #endif
93
94 #define REG_COMP_C
95 #include "regcomp.h"
96
97 #ifdef op
98 #undef op
99 #endif /* op */
100
101 #ifdef MSDOS
102 # if defined(BUGGY_MSC6)
103  /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
104  # pragma optimize("a",off)
105  /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
106  # pragma optimize("w",on )
107 # endif /* BUGGY_MSC6 */
108 #endif /* MSDOS */
109
110 #ifndef STATIC
111 #define STATIC  static
112 #endif
113
114 typedef struct RExC_state_t {
115     U16         flags16;                /* are we folding, multilining? */
116     char        *precomp;               /* uncompiled string. */
117     regexp      *rx;
118     char        *start;                 /* Start of input for compile */
119     char        *end;                   /* End of input for compile */
120     char        *parse;                 /* Input-scan pointer. */
121     I32         whilem_seen;            /* number of WHILEM in this expr */
122     regnode     *emit_start;            /* Start of emitted-code area */
123     regnode     *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
124     I32         naughty;                /* How bad is this pattern? */
125     I32         sawback;                /* Did we see \1, ...? */
126     U32         seen;
127     I32         size;                   /* Code size. */
128     I32         npar;                   /* () count. */
129     I32         extralen;
130     I32         seen_zerolen;
131     I32         seen_evals;
132     I32         utf8;
133 #if ADD_TO_REGEXEC
134     char        *starttry;              /* -Dr: where regtry was called. */
135 #define RExC_starttry   (pRExC_state->starttry)
136 #endif
137 } RExC_state_t;
138
139 #define RExC_flags16    (pRExC_state->flags16)
140 #define RExC_precomp    (pRExC_state->precomp)
141 #define RExC_rx         (pRExC_state->rx)
142 #define RExC_start      (pRExC_state->start)
143 #define RExC_end        (pRExC_state->end)
144 #define RExC_parse      (pRExC_state->parse)
145 #define RExC_whilem_seen        (pRExC_state->whilem_seen)
146 #define RExC_offsets    (pRExC_state->rx->offsets) /* I am not like the others */
147 #define RExC_emit       (pRExC_state->emit)
148 #define RExC_emit_start (pRExC_state->emit_start)
149 #define RExC_naughty    (pRExC_state->naughty)
150 #define RExC_sawback    (pRExC_state->sawback)
151 #define RExC_seen       (pRExC_state->seen)
152 #define RExC_size       (pRExC_state->size)
153 #define RExC_npar       (pRExC_state->npar)
154 #define RExC_extralen   (pRExC_state->extralen)
155 #define RExC_seen_zerolen       (pRExC_state->seen_zerolen)
156 #define RExC_seen_evals (pRExC_state->seen_evals)
157 #define RExC_utf8       (pRExC_state->utf8)
158
159 #define ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
160 #define ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
161         ((*s) == '{' && regcurly(s)))
162
163 #ifdef SPSTART
164 #undef SPSTART          /* dratted cpp namespace... */
165 #endif
166 /*
167  * Flags to be passed up and down.
168  */
169 #define WORST           0       /* Worst case. */
170 #define HASWIDTH        0x1     /* Known to match non-null strings. */
171 #define SIMPLE          0x2     /* Simple enough to be STAR/PLUS operand. */
172 #define SPSTART         0x4     /* Starts with * or +. */
173 #define TRYAGAIN        0x8     /* Weeded out a declaration. */
174
175 /* Length of a variant. */
176
177 typedef struct scan_data_t {
178     I32 len_min;
179     I32 len_delta;
180     I32 pos_min;
181     I32 pos_delta;
182     SV *last_found;
183     I32 last_end;                       /* min value, <0 unless valid. */
184     I32 last_start_min;
185     I32 last_start_max;
186     SV **longest;                       /* Either &l_fixed, or &l_float. */
187     SV *longest_fixed;
188     I32 offset_fixed;
189     SV *longest_float;
190     I32 offset_float_min;
191     I32 offset_float_max;
192     I32 flags;
193     I32 whilem_c;
194     I32 *last_closep;
195     struct regnode_charclass_class *start_class;
196 } scan_data_t;
197
198 /*
199  * Forward declarations for pregcomp()'s friends.
200  */
201
202 static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
203                                       0, 0, 0, 0, 0, 0};
204
205 #define SF_BEFORE_EOL           (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
206 #define SF_BEFORE_SEOL          0x1
207 #define SF_BEFORE_MEOL          0x2
208 #define SF_FIX_BEFORE_EOL       (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
209 #define SF_FL_BEFORE_EOL        (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
210
211 #ifdef NO_UNARY_PLUS
212 #  define SF_FIX_SHIFT_EOL      (0+2)
213 #  define SF_FL_SHIFT_EOL               (0+4)
214 #else
215 #  define SF_FIX_SHIFT_EOL      (+2)
216 #  define SF_FL_SHIFT_EOL               (+4)
217 #endif
218
219 #define SF_FIX_BEFORE_SEOL      (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
220 #define SF_FIX_BEFORE_MEOL      (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
221
222 #define SF_FL_BEFORE_SEOL       (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
223 #define SF_FL_BEFORE_MEOL       (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
224 #define SF_IS_INF               0x40
225 #define SF_HAS_PAR              0x80
226 #define SF_IN_PAR               0x100
227 #define SF_HAS_EVAL             0x200
228 #define SCF_DO_SUBSTR           0x400
229 #define SCF_DO_STCLASS_AND      0x0800
230 #define SCF_DO_STCLASS_OR       0x1000
231 #define SCF_DO_STCLASS          (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
232 #define SCF_WHILEM_VISITED_POS  0x2000
233
234 #define UTF RExC_utf8
235 #define LOC (RExC_flags16 & PMf_LOCALE)
236 #define FOLD (RExC_flags16 & PMf_FOLD)
237
238 #define OOB_UNICODE             12345678
239 #define OOB_NAMEDCLASS          -1
240
241 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
242 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
243
244
245 /* length of regex to show in messages that don't mark a position within */
246 #define RegexLengthToShowInErrorMessages 127
247
248 /*
249  * If MARKER[12] are adjusted, be sure to adjust the constants at the top
250  * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
251  * op/pragma/warn/regcomp.
252  */
253 #define MARKER1 "<-- HERE"    /* marker as it appears in the description */
254 #define MARKER2 " <-- HERE "  /* marker as it appears within the regex */
255
256 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
257
258 /*
259  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
260  * arg. Show regex, up to a maximum length. If it's too long, chop and add
261  * "...".
262  */
263 #define FAIL(msg)                                                             \
264     STMT_START {                                                             \
265         char *ellipses = "";                                                 \
266         unsigned len = strlen(RExC_precomp);                                \
267                                                                              \
268         if (!SIZE_ONLY)                                                      \
269             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
270                                                                              \
271         if (len > RegexLengthToShowInErrorMessages) {                        \
272             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
273             len = RegexLengthToShowInErrorMessages - 10;                     \
274             ellipses = "...";                                                \
275         }                                                                    \
276         Perl_croak(aTHX_ "%s in regex m/%.*s%s/",                            \
277                    msg, (int)len, RExC_precomp, ellipses);                  \
278     } STMT_END
279
280 /*
281  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
282  * args. Show regex, up to a maximum length. If it's too long, chop and add
283  * "...".
284  */
285 #define FAIL2(pat,msg)                                                        \
286     STMT_START {                                                             \
287         char *ellipses = "";                                                 \
288         unsigned len = strlen(RExC_precomp);                                \
289                                                                              \
290         if (!SIZE_ONLY)                                                      \
291             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
292                                                                              \
293         if (len > RegexLengthToShowInErrorMessages) {                        \
294             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
295             len = RegexLengthToShowInErrorMessages - 10;                     \
296             ellipses = "...";                                                \
297         }                                                                    \
298         S_re_croak2(aTHX_ pat, " in regex m/%.*s%s/",                        \
299                     msg, (int)len, RExC_precomp, ellipses);                \
300     } STMT_END
301
302
303 /*
304  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
305  */
306 #define Simple_vFAIL(m)                                                      \
307     STMT_START {                                                             \
308       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
309                                                                              \
310       Perl_croak(aTHX_ "%s" REPORT_LOCATION,               \
311                  m, (int)offset, RExC_precomp, RExC_precomp + offset);     \
312     } STMT_END
313
314 /*
315  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
316  */
317 #define vFAIL(m)                                                             \
318     STMT_START {                                                             \
319       if (!SIZE_ONLY)                                                        \
320             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
321       Simple_vFAIL(m);                                                       \
322     } STMT_END
323
324 /*
325  * Like Simple_vFAIL(), but accepts two arguments.
326  */
327 #define Simple_vFAIL2(m,a1)                                                  \
328     STMT_START {                                                             \
329       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
330                                                                              \
331       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1,       \
332                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
333     } STMT_END
334
335 /*
336  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
337  */
338 #define vFAIL2(m,a1)                                                         \
339     STMT_START {                                                             \
340       if (!SIZE_ONLY)                                                        \
341             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
342       Simple_vFAIL2(m, a1);                                                  \
343     } STMT_END
344
345
346 /*
347  * Like Simple_vFAIL(), but accepts three arguments.
348  */
349 #define Simple_vFAIL3(m, a1, a2)                                             \
350     STMT_START {                                                             \
351       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
352                                                                              \
353       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2,   \
354                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
355     } STMT_END
356
357 /*
358  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
359  */
360 #define vFAIL3(m,a1,a2)                                                      \
361     STMT_START {                                                             \
362       if (!SIZE_ONLY)                                                        \
363             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
364       Simple_vFAIL3(m, a1, a2);                                              \
365     } STMT_END
366
367 /*
368  * Like Simple_vFAIL(), but accepts four arguments.
369  */
370 #define Simple_vFAIL4(m, a1, a2, a3)                                         \
371     STMT_START {                                                             \
372       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
373                                                                              \
374       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3,\
375                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
376     } STMT_END
377
378 /*
379  * Like Simple_vFAIL(), but accepts five arguments.
380  */
381 #define Simple_vFAIL5(m, a1, a2, a3, a4)                                     \
382     STMT_START {                                                             \
383       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
384       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, a4,\
385                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
386     } STMT_END
387
388
389 #define vWARN(loc,m)                                                         \
390     STMT_START {                                                             \
391         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
392         Perl_warner(aTHX_ WARN_REGEXP, "%s" REPORT_LOCATION,\
393                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
394     } STMT_END                                                               \
395
396 #define vWARNdep(loc,m)                                                         \
397     STMT_START {                                                             \
398         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
399         int warn_cat = ckWARN(WARN_REGEXP) ? WARN_REGEXP : WARN_DEPRECATED;  \
400         Perl_warner(aTHX_ warn_cat, "%s" REPORT_LOCATION,\
401                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
402     } STMT_END                                                               \
403
404
405 #define vWARN2(loc, m, a1)                                                   \
406     STMT_START {                                                             \
407         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
408         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
409                  a1,                                                         \
410                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
411     } STMT_END
412
413 #define vWARN3(loc, m, a1, a2)                                               \
414     STMT_START {                                                             \
415       unsigned offset = strlen(RExC_precomp) - (RExC_end - (loc));        \
416         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,                    \
417                  a1, a2,                                                     \
418                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
419     } STMT_END
420
421 #define vWARN4(loc, m, a1, a2, a3)                                           \
422     STMT_START {                                                             \
423       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));            \
424         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
425                  a1, a2, a3,                                                 \
426                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
427     } STMT_END
428
429 /* used for the parse_flags section for (?c) -- japhy */
430 #define vWARN5(loc, m, a1, a2, a3, a4)                                       \
431   STMT_START {                                                   \
432       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));   \
433         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,      \
434                  a1, a2, a3, a4,                                 \
435                  (int)offset, RExC_precomp, RExC_precomp + offset);  \
436     } STMT_END
437
438
439 /* Allow for side effects in s */
440 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (void)(s);} STMT_END
441
442 /* Macros for recording node offsets.   20001227 mjd@plover.com 
443  * Nodes are numbered 1, 2, 3, 4.  Node #n's position is recorded in
444  * element 2*n-1 of the array.  Element #2n holds the byte length node #n.
445  * Element 0 holds the number n.
446  */
447
448 #define MJD_OFFSET_DEBUG(x)
449 /* #define MJD_OFFSET_DEBUG(x) fprintf x */
450
451
452 #  define Set_Node_Offset_To_R(node,byte)                           \
453    STMT_START {                                        \
454      if (! SIZE_ONLY) {                                  \
455        if((node) < 0) {                   \
456          Perl_croak(aTHX_ "value of node is %d in Offset macro", node); \
457        } else {                                                        \
458          RExC_offsets[2*(node)-1] = (byte);                               \
459        }                                                               \
460      }                                                                 \
461    } STMT_END
462
463 #  define Set_Node_Offset(node,byte) Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
464 #  define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
465
466 #  define Set_Node_Length_To_R(node,len)                            \
467    STMT_START {                                        \
468      if (! SIZE_ONLY) {                                  \
469        MJD_OFFSET_DEBUG((stderr, "** (%d) size of node %d is %d.\n", __LINE__, (node), (len))); \
470        if((node) < 0) {                   \
471          Perl_croak(aTHX_ "value of node is %d in Length macro", node); \
472        } else {                                                        \
473          RExC_offsets[2*(node)] = (len);                               \
474        }                                                               \
475      }                                                                 \
476    } STMT_END
477
478 #  define Set_Node_Length(node,len) Set_Node_Length_To_R((node)-RExC_emit_start, len)
479 #  define Set_Cur_Node_Length(len)  Set_Node_Length(RExC_emit, len)
480 #  define Set_Node_Cur_Length(node)   Set_Node_Length(node, RExC_parse - parse_start)
481
482 /* Get offsets and lengths */
483 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
484 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
485
486 static void clear_re(pTHXo_ void *r);
487
488 /* Mark that we cannot extend a found fixed substring at this point.
489    Updata the longest found anchored substring and the longest found
490    floating substrings if needed. */
491
492 STATIC void
493 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
494 {
495     STRLEN l = CHR_SVLEN(data->last_found);
496     STRLEN old_l = CHR_SVLEN(*data->longest);
497
498     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
499         sv_setsv(*data->longest, data->last_found);
500         if (*data->longest == data->longest_fixed) {
501             data->offset_fixed = l ? data->last_start_min : data->pos_min;
502             if (data->flags & SF_BEFORE_EOL)
503                 data->flags
504                     |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
505             else
506                 data->flags &= ~SF_FIX_BEFORE_EOL;
507         }
508         else {
509             data->offset_float_min = l ? data->last_start_min : data->pos_min;
510             data->offset_float_max = (l
511                                       ? data->last_start_max
512                                       : data->pos_min + data->pos_delta);
513             if (data->flags & SF_BEFORE_EOL)
514                 data->flags
515                     |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
516             else
517                 data->flags &= ~SF_FL_BEFORE_EOL;
518         }
519     }
520     SvCUR_set(data->last_found, 0);
521     data->last_end = -1;
522     data->flags &= ~SF_BEFORE_EOL;
523 }
524
525 /* Can match anything (initialization) */
526 STATIC void
527 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
528 {
529     int value;
530
531     ANYOF_CLASS_ZERO(cl);
532     for (value = 0; value < 256; ++value)
533         ANYOF_BITMAP_SET(cl, value);
534     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
535     if (LOC)
536         cl->flags |= ANYOF_LOCALE;
537 }
538
539 /* Can match anything (initialization) */
540 STATIC int
541 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
542 {
543     int value;
544
545     for (value = 0; value <= ANYOF_MAX; value += 2)
546         if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
547             return 1;
548     if (!(cl->flags & ANYOF_UNICODE_ALL))
549         return 0;
550     for (value = 0; value < 256; ++value)
551         if (!ANYOF_BITMAP_TEST(cl, value))
552             return 0;
553     return 1;
554 }
555
556 /* Can match anything (initialization) */
557 STATIC void
558 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
559 {
560     Zero(cl, 1, struct regnode_charclass_class);
561     cl->type = ANYOF;
562     cl_anything(pRExC_state, cl);
563 }
564
565 STATIC void
566 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
567 {
568     Zero(cl, 1, struct regnode_charclass_class);
569     cl->type = ANYOF;
570     cl_anything(pRExC_state, cl);
571     if (LOC)
572         cl->flags |= ANYOF_LOCALE;
573 }
574
575 /* 'And' a given class with another one.  Can create false positives */
576 /* We assume that cl is not inverted */
577 STATIC void
578 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
579          struct regnode_charclass_class *and_with)
580 {
581     if (!(and_with->flags & ANYOF_CLASS)
582         && !(cl->flags & ANYOF_CLASS)
583         && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
584         && !(and_with->flags & ANYOF_FOLD)
585         && !(cl->flags & ANYOF_FOLD)) {
586         int i;
587
588         if (and_with->flags & ANYOF_INVERT)
589             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
590                 cl->bitmap[i] &= ~and_with->bitmap[i];
591         else
592             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
593                 cl->bitmap[i] &= and_with->bitmap[i];
594     } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
595     if (!(and_with->flags & ANYOF_EOS))
596         cl->flags &= ~ANYOF_EOS;
597
598     if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
599         cl->flags &= ~ANYOF_UNICODE_ALL;
600         cl->flags |= ANYOF_UNICODE;
601         ARG_SET(cl, ARG(and_with));
602     }
603     if (!(and_with->flags & ANYOF_UNICODE_ALL))
604         cl->flags &= ~ANYOF_UNICODE_ALL;
605     if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
606         cl->flags &= ~ANYOF_UNICODE;
607 }
608
609 /* 'OR' a given class with another one.  Can create false positives */
610 /* We assume that cl is not inverted */
611 STATIC void
612 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
613 {
614     if (or_with->flags & ANYOF_INVERT) {
615         /* We do not use
616          * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
617          *   <= (B1 | !B2) | (CL1 | !CL2)
618          * which is wasteful if CL2 is small, but we ignore CL2:
619          *   (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
620          * XXXX Can we handle case-fold?  Unclear:
621          *   (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
622          *   (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
623          */
624         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
625              && !(or_with->flags & ANYOF_FOLD)
626              && !(cl->flags & ANYOF_FOLD) ) {
627             int i;
628
629             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
630                 cl->bitmap[i] |= ~or_with->bitmap[i];
631         } /* XXXX: logic is complicated otherwise */
632         else {
633             cl_anything(pRExC_state, cl);
634         }
635     } else {
636         /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
637         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
638              && (!(or_with->flags & ANYOF_FOLD)
639                  || (cl->flags & ANYOF_FOLD)) ) {
640             int i;
641
642             /* OR char bitmap and class bitmap separately */
643             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
644                 cl->bitmap[i] |= or_with->bitmap[i];
645             if (or_with->flags & ANYOF_CLASS) {
646                 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
647                     cl->classflags[i] |= or_with->classflags[i];
648                 cl->flags |= ANYOF_CLASS;
649             }
650         }
651         else { /* XXXX: logic is complicated, leave it along for a moment. */
652             cl_anything(pRExC_state, cl);
653         }
654     }
655     if (or_with->flags & ANYOF_EOS)
656         cl->flags |= ANYOF_EOS;
657
658     if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
659         ARG(cl) != ARG(or_with)) {
660         cl->flags |= ANYOF_UNICODE_ALL;
661         cl->flags &= ~ANYOF_UNICODE;
662     }
663     if (or_with->flags & ANYOF_UNICODE_ALL) {
664         cl->flags |= ANYOF_UNICODE_ALL;
665         cl->flags &= ~ANYOF_UNICODE;
666     }
667 }
668
669 /* REx optimizer.  Converts nodes into quickier variants "in place".
670    Finds fixed substrings.  */
671
672 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
673    to the position after last scanned or to NULL. */
674
675 STATIC I32
676 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
677                         /* scanp: Start here (read-write). */
678                         /* deltap: Write maxlen-minlen here. */
679                         /* last: Stop before this one. */
680 {
681     I32 min = 0, pars = 0, code;
682     regnode *scan = *scanp, *next;
683     I32 delta = 0;
684     int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
685     int is_inf_internal = 0;            /* The studied chunk is infinite */
686     I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
687     scan_data_t data_fake;
688     struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
689
690     while (scan && OP(scan) != END && scan < last) {
691         /* Peephole optimizer: */
692
693         if (PL_regkind[(U8)OP(scan)] == EXACT) {
694             /* Merge several consecutive EXACTish nodes into one. */
695             regnode *n = regnext(scan);
696             U32 stringok = 1;
697 #ifdef DEBUGGING
698             regnode *stop = scan;
699 #endif
700
701             next = scan + NODE_SZ_STR(scan);
702             /* Skip NOTHING, merge EXACT*. */
703             while (n &&
704                    ( PL_regkind[(U8)OP(n)] == NOTHING ||
705                      (stringok && (OP(n) == OP(scan))))
706                    && NEXT_OFF(n)
707                    && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
708                 if (OP(n) == TAIL || n > next)
709                     stringok = 0;
710                 if (PL_regkind[(U8)OP(n)] == NOTHING) {
711                     NEXT_OFF(scan) += NEXT_OFF(n);
712                     next = n + NODE_STEP_REGNODE;
713 #ifdef DEBUGGING
714                     if (stringok)
715                         stop = n;
716 #endif
717                     n = regnext(n);
718                 }
719                 else if (stringok) {
720                     int oldl = STR_LEN(scan);
721                     regnode *nnext = regnext(n);
722
723                     if (oldl + STR_LEN(n) > U8_MAX)
724                         break;
725                     NEXT_OFF(scan) += NEXT_OFF(n);
726                     STR_LEN(scan) += STR_LEN(n);
727                     next = n + NODE_SZ_STR(n);
728                     /* Now we can overwrite *n : */
729                     Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
730 #ifdef DEBUGGING
731                     stop = next - 1;
732 #endif
733                     n = nnext;
734                 }
735             }
736 #ifdef DEBUGGING
737             /* Allow dumping */
738             n = scan + NODE_SZ_STR(scan);
739             while (n <= stop) {
740                 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
741                     OP(n) = OPTIMIZED;
742                     NEXT_OFF(n) = 0;
743                 }
744                 n++;
745             }
746 #endif
747         }
748         /* Follow the next-chain of the current node and optimize
749            away all the NOTHINGs from it.  */
750         if (OP(scan) != CURLYX) {
751             int max = (reg_off_by_arg[OP(scan)]
752                        ? I32_MAX
753                        /* I32 may be smaller than U16 on CRAYs! */
754                        : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
755             int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
756             int noff;
757             regnode *n = scan;
758         
759             /* Skip NOTHING and LONGJMP. */
760             while ((n = regnext(n))
761                    && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
762                        || ((OP(n) == LONGJMP) && (noff = ARG(n))))
763                    && off + noff < max)
764                 off += noff;
765             if (reg_off_by_arg[OP(scan)])
766                 ARG(scan) = off;
767             else
768                 NEXT_OFF(scan) = off;
769         }
770         /* The principal pseudo-switch.  Cannot be a switch, since we
771            look into several different things.  */
772         if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
773                    || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
774             next = regnext(scan);
775             code = OP(scan);
776         
777             if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
778                 I32 max1 = 0, min1 = I32_MAX, num = 0;
779                 struct regnode_charclass_class accum;
780                 
781                 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
782                     scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
783                 if (flags & SCF_DO_STCLASS)
784                     cl_init_zero(pRExC_state, &accum);
785                 while (OP(scan) == code) {
786                     I32 deltanext, minnext, f = 0, fake;
787                     struct regnode_charclass_class this_class;
788
789                     num++;
790                     data_fake.flags = 0;
791                     if (data) {         
792                         data_fake.whilem_c = data->whilem_c;
793                         data_fake.last_closep = data->last_closep;
794                     }
795                     else
796                         data_fake.last_closep = &fake;
797                     next = regnext(scan);
798                     scan = NEXTOPER(scan);
799                     if (code != BRANCH)
800                         scan = NEXTOPER(scan);
801                     if (flags & SCF_DO_STCLASS) {
802                         cl_init(pRExC_state, &this_class);
803                         data_fake.start_class = &this_class;
804                         f = SCF_DO_STCLASS_AND;
805                     }           
806                     if (flags & SCF_WHILEM_VISITED_POS)
807                         f |= SCF_WHILEM_VISITED_POS;
808                     /* we suppose the run is continuous, last=next...*/
809                     minnext = study_chunk(pRExC_state, &scan, &deltanext,
810                                           next, &data_fake, f);
811                     if (min1 > minnext)
812                         min1 = minnext;
813                     if (max1 < minnext + deltanext)
814                         max1 = minnext + deltanext;
815                     if (deltanext == I32_MAX)
816                         is_inf = is_inf_internal = 1;
817                     scan = next;
818                     if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
819                         pars++;
820                     if (data && (data_fake.flags & SF_HAS_EVAL))
821                         data->flags |= SF_HAS_EVAL;
822                     if (data)
823                         data->whilem_c = data_fake.whilem_c;
824                     if (flags & SCF_DO_STCLASS)
825                         cl_or(pRExC_state, &accum, &this_class);
826                     if (code == SUSPEND)
827                         break;
828                 }
829                 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
830                     min1 = 0;
831                 if (flags & SCF_DO_SUBSTR) {
832                     data->pos_min += min1;
833                     data->pos_delta += max1 - min1;
834                     if (max1 != min1 || is_inf)
835                         data->longest = &(data->longest_float);
836                 }
837                 min += min1;
838                 delta += max1 - min1;
839                 if (flags & SCF_DO_STCLASS_OR) {
840                     cl_or(pRExC_state, data->start_class, &accum);
841                     if (min1) {
842                         cl_and(data->start_class, &and_with);
843                         flags &= ~SCF_DO_STCLASS;
844                     }
845                 }
846                 else if (flags & SCF_DO_STCLASS_AND) {
847                     if (min1) {
848                         cl_and(data->start_class, &accum);
849                         flags &= ~SCF_DO_STCLASS;
850                     }
851                     else {
852                         /* Switch to OR mode: cache the old value of
853                          * data->start_class */
854                         StructCopy(data->start_class, &and_with,
855                                    struct regnode_charclass_class);
856                         flags &= ~SCF_DO_STCLASS_AND;
857                         StructCopy(&accum, data->start_class,
858                                    struct regnode_charclass_class);
859                         flags |= SCF_DO_STCLASS_OR;
860                         data->start_class->flags |= ANYOF_EOS;
861                     }
862                 }
863             }
864             else if (code == BRANCHJ)   /* single branch is optimized. */
865                 scan = NEXTOPER(NEXTOPER(scan));
866             else                        /* single branch is optimized. */
867                 scan = NEXTOPER(scan);
868             continue;
869         }
870         else if (OP(scan) == EXACT) {
871             I32 l = STR_LEN(scan);
872             UV uc = *((U8*)STRING(scan));
873             if (UTF) {
874                 U8 *s = (U8*)STRING(scan);
875                 l = utf8_length(s, s + l);
876                 uc = utf8_to_uvchr(s, NULL);
877             }
878             min += l;
879             if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
880                 /* The code below prefers earlier match for fixed
881                    offset, later match for variable offset.  */
882                 if (data->last_end == -1) { /* Update the start info. */
883                     data->last_start_min = data->pos_min;
884                     data->last_start_max = is_inf
885                         ? I32_MAX : data->pos_min + data->pos_delta;
886                 }
887                 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
888                 data->last_end = data->pos_min + l;
889                 data->pos_min += l; /* As in the first entry. */
890                 data->flags &= ~SF_BEFORE_EOL;
891             }
892             if (flags & SCF_DO_STCLASS_AND) {
893                 /* Check whether it is compatible with what we know already! */
894                 int compat = 1;
895
896                 if (uc >= 0x100 ||
897                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
898                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
899                     && (!(data->start_class->flags & ANYOF_FOLD)
900                         || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
901                     )
902                     compat = 0;
903                 ANYOF_CLASS_ZERO(data->start_class);
904                 ANYOF_BITMAP_ZERO(data->start_class);
905                 if (compat)
906                     ANYOF_BITMAP_SET(data->start_class, uc);
907                 data->start_class->flags &= ~ANYOF_EOS;
908                 if (uc < 0x100)
909                   data->start_class->flags &= ~ANYOF_UNICODE_ALL;
910             }
911             else if (flags & SCF_DO_STCLASS_OR) {
912                 /* false positive possible if the class is case-folded */
913                 if (uc < 0x100)
914                     ANYOF_BITMAP_SET(data->start_class, uc);
915                 else
916                     data->start_class->flags |= ANYOF_UNICODE_ALL;
917                 data->start_class->flags &= ~ANYOF_EOS;
918                 cl_and(data->start_class, &and_with);
919             }
920             flags &= ~SCF_DO_STCLASS;
921         }
922         else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
923             I32 l = STR_LEN(scan);
924             UV uc = *((U8*)STRING(scan));
925
926             /* Search for fixed substrings supports EXACT only. */
927             if (flags & SCF_DO_SUBSTR)
928                 scan_commit(pRExC_state, data);
929             if (UTF) {
930                 U8 *s = (U8 *)STRING(scan);
931                 l = utf8_length(s, s + l);
932                 uc = utf8_to_uvchr(s, NULL);
933             }
934             min += l;
935             if (data && (flags & SCF_DO_SUBSTR))
936                 data->pos_min += l;
937             if (flags & SCF_DO_STCLASS_AND) {
938                 /* Check whether it is compatible with what we know already! */
939                 int compat = 1;
940
941                 if (uc >= 0x100 ||
942                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
943                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
944                      && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
945                     compat = 0;
946                 ANYOF_CLASS_ZERO(data->start_class);
947                 ANYOF_BITMAP_ZERO(data->start_class);
948                 if (compat) {
949                     ANYOF_BITMAP_SET(data->start_class, uc);
950                     data->start_class->flags &= ~ANYOF_EOS;
951                     data->start_class->flags |= ANYOF_FOLD;
952                     if (OP(scan) == EXACTFL)
953                         data->start_class->flags |= ANYOF_LOCALE;
954                 }
955             }
956             else if (flags & SCF_DO_STCLASS_OR) {
957                 if (data->start_class->flags & ANYOF_FOLD) {
958                     /* false positive possible if the class is case-folded.
959                        Assume that the locale settings are the same... */
960                     if (uc < 0x100)
961                         ANYOF_BITMAP_SET(data->start_class, uc);
962                     data->start_class->flags &= ~ANYOF_EOS;
963                 }
964                 cl_and(data->start_class, &and_with);
965             }
966             flags &= ~SCF_DO_STCLASS;
967         }
968         else if (strchr((char*)PL_varies,OP(scan))) {
969             I32 mincount, maxcount, minnext, deltanext, fl = 0;
970             I32 f = flags, pos_before = 0;
971             regnode *oscan = scan;
972             struct regnode_charclass_class this_class;
973             struct regnode_charclass_class *oclass = NULL;
974
975             switch (PL_regkind[(U8)OP(scan)]) {
976             case WHILEM:                /* End of (?:...)* . */
977                 scan = NEXTOPER(scan);
978                 goto finish;
979             case PLUS:
980                 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
981                     next = NEXTOPER(scan);
982                     if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
983                         mincount = 1;
984                         maxcount = REG_INFTY;
985                         next = regnext(scan);
986                         scan = NEXTOPER(scan);
987                         goto do_curly;
988                     }
989                 }
990                 if (flags & SCF_DO_SUBSTR)
991                     data->pos_min++;
992                 min++;
993                 /* Fall through. */
994             case STAR:
995                 if (flags & SCF_DO_STCLASS) {
996                     mincount = 0;
997                     maxcount = REG_INFTY;
998                     next = regnext(scan);
999                     scan = NEXTOPER(scan);
1000                     goto do_curly;
1001                 }
1002                 is_inf = is_inf_internal = 1;
1003                 scan = regnext(scan);
1004                 if (flags & SCF_DO_SUBSTR) {
1005                     scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
1006                     data->longest = &(data->longest_float);
1007                 }
1008                 goto optimize_curly_tail;
1009             case CURLY:
1010                 mincount = ARG1(scan);
1011                 maxcount = ARG2(scan);
1012                 next = regnext(scan);
1013                 if (OP(scan) == CURLYX) {
1014                     I32 lp = (data ? *(data->last_closep) : 0);
1015
1016                     scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
1017                 }
1018                 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
1019               do_curly:
1020                 if (flags & SCF_DO_SUBSTR) {
1021                     if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
1022                     pos_before = data->pos_min;
1023                 }
1024                 if (data) {
1025                     fl = data->flags;
1026                     data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
1027                     if (is_inf)
1028                         data->flags |= SF_IS_INF;
1029                 }
1030                 if (flags & SCF_DO_STCLASS) {
1031                     cl_init(pRExC_state, &this_class);
1032                     oclass = data->start_class;
1033                     data->start_class = &this_class;
1034                     f |= SCF_DO_STCLASS_AND;
1035                     f &= ~SCF_DO_STCLASS_OR;
1036                 }
1037                 /* These are the cases when once a subexpression
1038                    fails at a particular position, it cannot succeed
1039                    even after backtracking at the enclosing scope.
1040                 
1041                    XXXX what if minimal match and we are at the
1042                         initial run of {n,m}? */
1043                 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
1044                     f &= ~SCF_WHILEM_VISITED_POS;
1045
1046                 /* This will finish on WHILEM, setting scan, or on NULL: */
1047                 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data,
1048                                       mincount == 0
1049                                         ? (f & ~SCF_DO_SUBSTR) : f);
1050
1051                 if (flags & SCF_DO_STCLASS)
1052                     data->start_class = oclass;
1053                 if (mincount == 0 || minnext == 0) {
1054                     if (flags & SCF_DO_STCLASS_OR) {
1055                         cl_or(pRExC_state, data->start_class, &this_class);
1056                     }
1057                     else if (flags & SCF_DO_STCLASS_AND) {
1058                         /* Switch to OR mode: cache the old value of
1059                          * data->start_class */
1060                         StructCopy(data->start_class, &and_with,
1061                                    struct regnode_charclass_class);
1062                         flags &= ~SCF_DO_STCLASS_AND;
1063                         StructCopy(&this_class, data->start_class,
1064                                    struct regnode_charclass_class);
1065                         flags |= SCF_DO_STCLASS_OR;
1066                         data->start_class->flags |= ANYOF_EOS;
1067                     }
1068                 } else {                /* Non-zero len */
1069                     if (flags & SCF_DO_STCLASS_OR) {
1070                         cl_or(pRExC_state, data->start_class, &this_class);
1071                         cl_and(data->start_class, &and_with);
1072                     }
1073                     else if (flags & SCF_DO_STCLASS_AND)
1074                         cl_and(data->start_class, &this_class);
1075                     flags &= ~SCF_DO_STCLASS;
1076                 }
1077                 if (!scan)              /* It was not CURLYX, but CURLY. */
1078                     scan = next;
1079                 if (ckWARN(WARN_REGEXP)
1080                     && (minnext == 0) && (deltanext == 0)
1081                     && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1082                     && maxcount <= REG_INFTY/3) /* Complement check for big count */
1083                 {
1084                     vWARN(RExC_parse,
1085                           "Quantifier unexpected on zero-length expression");
1086                 }
1087
1088                 min += minnext * mincount;
1089                 is_inf_internal |= ((maxcount == REG_INFTY
1090                                      && (minnext + deltanext) > 0)
1091                                     || deltanext == I32_MAX);
1092                 is_inf |= is_inf_internal;
1093                 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1094
1095                 /* Try powerful optimization CURLYX => CURLYN. */
1096                 if (  OP(oscan) == CURLYX && data
1097                       && data->flags & SF_IN_PAR
1098                       && !(data->flags & SF_HAS_EVAL)
1099                       && !deltanext && minnext == 1 ) {
1100                     /* Try to optimize to CURLYN.  */
1101                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1102                     regnode *nxt1 = nxt;
1103 #ifdef DEBUGGING
1104                     regnode *nxt2;
1105 #endif
1106
1107                     /* Skip open. */
1108                     nxt = regnext(nxt);
1109                     if (!strchr((char*)PL_simple,OP(nxt))
1110                         && !(PL_regkind[(U8)OP(nxt)] == EXACT
1111                              && STR_LEN(nxt) == 1))
1112                         goto nogo;
1113 #ifdef DEBUGGING
1114                     nxt2 = nxt;
1115 #endif
1116                     nxt = regnext(nxt);
1117                     if (OP(nxt) != CLOSE)
1118                         goto nogo;
1119                     /* Now we know that nxt2 is the only contents: */
1120                     oscan->flags = ARG(nxt);
1121                     OP(oscan) = CURLYN;
1122                     OP(nxt1) = NOTHING; /* was OPEN. */
1123 #ifdef DEBUGGING
1124                     OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1125                     NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1126                     NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1127                     OP(nxt) = OPTIMIZED;        /* was CLOSE. */
1128                     OP(nxt + 1) = OPTIMIZED; /* was count. */
1129                     NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1130 #endif
1131                 }
1132               nogo:
1133
1134                 /* Try optimization CURLYX => CURLYM. */
1135                 if (  OP(oscan) == CURLYX && data
1136                       && !(data->flags & SF_HAS_PAR)
1137                       && !(data->flags & SF_HAS_EVAL)
1138                       && !deltanext  ) {
1139                     /* XXXX How to optimize if data == 0? */
1140                     /* Optimize to a simpler form.  */
1141                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1142                     regnode *nxt2;
1143
1144                     OP(oscan) = CURLYM;
1145                     while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1146                             && (OP(nxt2) != WHILEM))
1147                         nxt = nxt2;
1148                     OP(nxt2)  = SUCCEED; /* Whas WHILEM */
1149                     /* Need to optimize away parenths. */
1150                     if (data->flags & SF_IN_PAR) {
1151                         /* Set the parenth number.  */
1152                         regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1153
1154                         if (OP(nxt) != CLOSE)
1155                             FAIL("Panic opt close");
1156                         oscan->flags = ARG(nxt);
1157                         OP(nxt1) = OPTIMIZED;   /* was OPEN. */
1158                         OP(nxt) = OPTIMIZED;    /* was CLOSE. */
1159 #ifdef DEBUGGING
1160                         OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1161                         OP(nxt + 1) = OPTIMIZED; /* was count. */
1162                         NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1163                         NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1164 #endif
1165 #if 0
1166                         while ( nxt1 && (OP(nxt1) != WHILEM)) {
1167                             regnode *nnxt = regnext(nxt1);
1168                         
1169                             if (nnxt == nxt) {
1170                                 if (reg_off_by_arg[OP(nxt1)])
1171                                     ARG_SET(nxt1, nxt2 - nxt1);
1172                                 else if (nxt2 - nxt1 < U16_MAX)
1173                                     NEXT_OFF(nxt1) = nxt2 - nxt1;
1174                                 else
1175                                     OP(nxt) = NOTHING;  /* Cannot beautify */
1176                             }
1177                             nxt1 = nnxt;
1178                         }
1179 #endif
1180                         /* Optimize again: */
1181                         study_chunk(pRExC_state, &nxt1, &deltanext, nxt,
1182                                     NULL, 0);
1183                     }
1184                     else
1185                         oscan->flags = 0;
1186                 }
1187                 else if ((OP(oscan) == CURLYX)
1188                          && (flags & SCF_WHILEM_VISITED_POS)
1189                          /* See the comment on a similar expression above.
1190                             However, this time it not a subexpression
1191                             we care about, but the expression itself. */
1192                          && (maxcount == REG_INFTY)
1193                          && data && ++data->whilem_c < 16) {
1194                     /* This stays as CURLYX, we can put the count/of pair. */
1195                     /* Find WHILEM (as in regexec.c) */
1196                     regnode *nxt = oscan + NEXT_OFF(oscan);
1197
1198                     if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1199                         nxt += ARG(nxt);
1200                     PREVOPER(nxt)->flags = data->whilem_c
1201                         | (RExC_whilem_seen << 4); /* On WHILEM */
1202                 }
1203                 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
1204                     pars++;
1205                 if (flags & SCF_DO_SUBSTR) {
1206                     SV *last_str = Nullsv;
1207                     int counted = mincount != 0;
1208
1209                     if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1210                         I32 b = pos_before >= data->last_start_min
1211                             ? pos_before : data->last_start_min;
1212                         STRLEN l;
1213                         char *s = SvPV(data->last_found, l);
1214                         I32 old = b - data->last_start_min;
1215
1216                         if (UTF)
1217                             old = utf8_hop((U8*)s, old) - (U8*)s;
1218                         
1219                         l -= old;
1220                         /* Get the added string: */
1221                         last_str = newSVpvn(s  + old, l);
1222                         if (deltanext == 0 && pos_before == b) {
1223                             /* What was added is a constant string */
1224                             if (mincount > 1) {
1225                                 SvGROW(last_str, (mincount * l) + 1);
1226                                 repeatcpy(SvPVX(last_str) + l,
1227                                           SvPVX(last_str), l, mincount - 1);
1228                                 SvCUR(last_str) *= mincount;
1229                                 /* Add additional parts. */
1230                                 SvCUR_set(data->last_found,
1231                                           SvCUR(data->last_found) - l);
1232                                 sv_catsv(data->last_found, last_str);
1233                                 data->last_end += l * (mincount - 1);
1234                             }
1235                         } else {
1236                             /* start offset must point into the last copy */
1237                             data->last_start_min += minnext * (mincount - 1);
1238                             data->last_start_max += is_inf ? 0 : (maxcount - 1)
1239                                 * (minnext + data->pos_delta);
1240                         }
1241                     }
1242                     /* It is counted once already... */
1243                     data->pos_min += minnext * (mincount - counted);
1244                     data->pos_delta += - counted * deltanext +
1245                         (minnext + deltanext) * maxcount - minnext * mincount;
1246                     if (mincount != maxcount) {
1247                          /* Cannot extend fixed substrings found inside
1248                             the group.  */
1249                         scan_commit(pRExC_state,data);
1250                         if (mincount && last_str) {
1251                             sv_setsv(data->last_found, last_str);
1252                             data->last_end = data->pos_min;
1253                             data->last_start_min =
1254                                 data->pos_min - CHR_SVLEN(last_str);
1255                             data->last_start_max = is_inf
1256                                 ? I32_MAX
1257                                 : data->pos_min + data->pos_delta
1258                                 - CHR_SVLEN(last_str);
1259                         }
1260                         data->longest = &(data->longest_float);
1261                     }
1262                     SvREFCNT_dec(last_str);
1263                 }
1264                 if (data && (fl & SF_HAS_EVAL))
1265                     data->flags |= SF_HAS_EVAL;
1266               optimize_curly_tail:
1267                 if (OP(oscan) != CURLYX) {
1268                     while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1269                            && NEXT_OFF(next))
1270                         NEXT_OFF(oscan) += NEXT_OFF(next);
1271                 }
1272                 continue;
1273             default:                    /* REF and CLUMP only? */
1274                 if (flags & SCF_DO_SUBSTR) {
1275                     scan_commit(pRExC_state,data);      /* Cannot expect anything... */
1276                     data->longest = &(data->longest_float);
1277                 }
1278                 is_inf = is_inf_internal = 1;
1279                 if (flags & SCF_DO_STCLASS_OR)
1280                     cl_anything(pRExC_state, data->start_class);
1281                 flags &= ~SCF_DO_STCLASS;
1282                 break;
1283             }
1284         }
1285         else if (strchr((char*)PL_simple,OP(scan))) {
1286             int value = 0;
1287
1288             if (flags & SCF_DO_SUBSTR) {
1289                 scan_commit(pRExC_state,data);
1290                 data->pos_min++;
1291             }
1292             min++;
1293             if (flags & SCF_DO_STCLASS) {
1294                 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1295
1296                 /* Some of the logic below assumes that switching
1297                    locale on will only add false positives. */
1298                 switch (PL_regkind[(U8)OP(scan)]) {
1299                 case SANY:
1300                 default:
1301                   do_default:
1302                     /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1303                     if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1304                         cl_anything(pRExC_state, data->start_class);
1305                     break;
1306                 case REG_ANY:
1307                     if (OP(scan) == SANY)
1308                         goto do_default;
1309                     if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1310                         value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1311                                  || (data->start_class->flags & ANYOF_CLASS));
1312                         cl_anything(pRExC_state, data->start_class);
1313                     }
1314                     if (flags & SCF_DO_STCLASS_AND || !value)
1315                         ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1316                     break;
1317                 case ANYOF:
1318                     if (flags & SCF_DO_STCLASS_AND)
1319                         cl_and(data->start_class,
1320                                (struct regnode_charclass_class*)scan);
1321                     else
1322                         cl_or(pRExC_state, data->start_class,
1323                               (struct regnode_charclass_class*)scan);
1324                     break;
1325                 case ALNUM:
1326                     if (flags & SCF_DO_STCLASS_AND) {
1327                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1328                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1329                             for (value = 0; value < 256; value++)
1330                                 if (!isALNUM(value))
1331                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1332                         }
1333                     }
1334                     else {
1335                         if (data->start_class->flags & ANYOF_LOCALE)
1336                             ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1337                         else {
1338                             for (value = 0; value < 256; value++)
1339                                 if (isALNUM(value))
1340                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1341                         }
1342                     }
1343                     break;
1344                 case ALNUML:
1345                     if (flags & SCF_DO_STCLASS_AND) {
1346                         if (data->start_class->flags & ANYOF_LOCALE)
1347                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1348                     }
1349                     else {
1350                         ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1351                         data->start_class->flags |= ANYOF_LOCALE;
1352                     }
1353                     break;
1354                 case NALNUM:
1355                     if (flags & SCF_DO_STCLASS_AND) {
1356                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1357                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1358                             for (value = 0; value < 256; value++)
1359                                 if (isALNUM(value))
1360                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1361                         }
1362                     }
1363                     else {
1364                         if (data->start_class->flags & ANYOF_LOCALE)
1365                             ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1366                         else {
1367                             for (value = 0; value < 256; value++)
1368                                 if (!isALNUM(value))
1369                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1370                         }
1371                     }
1372                     break;
1373                 case NALNUML:
1374                     if (flags & SCF_DO_STCLASS_AND) {
1375                         if (data->start_class->flags & ANYOF_LOCALE)
1376                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1377                     }
1378                     else {
1379                         data->start_class->flags |= ANYOF_LOCALE;
1380                         ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1381                     }
1382                     break;
1383                 case SPACE:
1384                     if (flags & SCF_DO_STCLASS_AND) {
1385                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1386                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1387                             for (value = 0; value < 256; value++)
1388                                 if (!isSPACE(value))
1389                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1390                         }
1391                     }
1392                     else {
1393                         if (data->start_class->flags & ANYOF_LOCALE)
1394                             ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1395                         else {
1396                             for (value = 0; value < 256; value++)
1397                                 if (isSPACE(value))
1398                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1399                         }
1400                     }
1401                     break;
1402                 case SPACEL:
1403                     if (flags & SCF_DO_STCLASS_AND) {
1404                         if (data->start_class->flags & ANYOF_LOCALE)
1405                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1406                     }
1407                     else {
1408                         data->start_class->flags |= ANYOF_LOCALE;
1409                         ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1410                     }
1411                     break;
1412                 case NSPACE:
1413                     if (flags & SCF_DO_STCLASS_AND) {
1414                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1415                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1416                             for (value = 0; value < 256; value++)
1417                                 if (isSPACE(value))
1418                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1419                         }
1420                     }
1421                     else {
1422                         if (data->start_class->flags & ANYOF_LOCALE)
1423                             ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1424                         else {
1425                             for (value = 0; value < 256; value++)
1426                                 if (!isSPACE(value))
1427                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1428                         }
1429                     }
1430                     break;
1431                 case NSPACEL:
1432                     if (flags & SCF_DO_STCLASS_AND) {
1433                         if (data->start_class->flags & ANYOF_LOCALE) {
1434                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1435                             for (value = 0; value < 256; value++)
1436                                 if (!isSPACE(value))
1437                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1438                         }
1439                     }
1440                     else {
1441                         data->start_class->flags |= ANYOF_LOCALE;
1442                         ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1443                     }
1444                     break;
1445                 case DIGIT:
1446                     if (flags & SCF_DO_STCLASS_AND) {
1447                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1448                         for (value = 0; value < 256; value++)
1449                             if (!isDIGIT(value))
1450                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1451                     }
1452                     else {
1453                         if (data->start_class->flags & ANYOF_LOCALE)
1454                             ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1455                         else {
1456                             for (value = 0; value < 256; value++)
1457                                 if (isDIGIT(value))
1458                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1459                         }
1460                     }
1461                     break;
1462                 case NDIGIT:
1463                     if (flags & SCF_DO_STCLASS_AND) {
1464                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1465                         for (value = 0; value < 256; value++)
1466                             if (isDIGIT(value))
1467                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1468                     }
1469                     else {
1470                         if (data->start_class->flags & ANYOF_LOCALE)
1471                             ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1472                         else {
1473                             for (value = 0; value < 256; value++)
1474                                 if (!isDIGIT(value))
1475                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1476                         }
1477                     }
1478                     break;
1479                 }
1480                 if (flags & SCF_DO_STCLASS_OR)
1481                     cl_and(data->start_class, &and_with);
1482                 flags &= ~SCF_DO_STCLASS;
1483             }
1484         }
1485         else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1486             data->flags |= (OP(scan) == MEOL
1487                             ? SF_BEFORE_MEOL
1488                             : SF_BEFORE_SEOL);
1489         }
1490         else if (  PL_regkind[(U8)OP(scan)] == BRANCHJ
1491                  /* Lookbehind, or need to calculate parens/evals/stclass: */
1492                    && (scan->flags || data || (flags & SCF_DO_STCLASS))
1493                    && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1494             /* Lookahead/lookbehind */
1495             I32 deltanext, minnext, fake = 0;
1496             regnode *nscan;
1497             struct regnode_charclass_class intrnl;
1498             int f = 0;
1499
1500             data_fake.flags = 0;
1501             if (data) {         
1502                 data_fake.whilem_c = data->whilem_c;
1503                 data_fake.last_closep = data->last_closep;
1504             }
1505             else
1506                 data_fake.last_closep = &fake;
1507             if ( flags & SCF_DO_STCLASS && !scan->flags
1508                  && OP(scan) == IFMATCH ) { /* Lookahead */
1509                 cl_init(pRExC_state, &intrnl);
1510                 data_fake.start_class = &intrnl;
1511                 f |= SCF_DO_STCLASS_AND;
1512             }
1513             if (flags & SCF_WHILEM_VISITED_POS)
1514                 f |= SCF_WHILEM_VISITED_POS;
1515             next = regnext(scan);
1516             nscan = NEXTOPER(NEXTOPER(scan));
1517             minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1518             if (scan->flags) {
1519                 if (deltanext) {
1520                     vFAIL("Variable length lookbehind not implemented");
1521                 }
1522                 else if (minnext > U8_MAX) {
1523                     vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1524                 }
1525                 scan->flags = minnext;
1526             }
1527             if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1528                 pars++;
1529             if (data && (data_fake.flags & SF_HAS_EVAL))
1530                 data->flags |= SF_HAS_EVAL;
1531             if (data)
1532                 data->whilem_c = data_fake.whilem_c;
1533             if (f & SCF_DO_STCLASS_AND) {
1534                 int was = (data->start_class->flags & ANYOF_EOS);
1535
1536                 cl_and(data->start_class, &intrnl);
1537                 if (was)
1538                     data->start_class->flags |= ANYOF_EOS;
1539             }
1540         }
1541         else if (OP(scan) == OPEN) {
1542             pars++;
1543         }
1544         else if (OP(scan) == CLOSE) {
1545             if (ARG(scan) == is_par) {
1546                 next = regnext(scan);
1547
1548                 if ( next && (OP(next) != WHILEM) && next < last)
1549                     is_par = 0;         /* Disable optimization */
1550             }
1551             if (data)
1552                 *(data->last_closep) = ARG(scan);
1553         }
1554         else if (OP(scan) == EVAL) {
1555                 if (data)
1556                     data->flags |= SF_HAS_EVAL;
1557         }
1558         else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1559                 if (flags & SCF_DO_SUBSTR) {
1560                     scan_commit(pRExC_state,data);
1561                     data->longest = &(data->longest_float);
1562                 }
1563                 is_inf = is_inf_internal = 1;
1564                 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1565                     cl_anything(pRExC_state, data->start_class);
1566                 flags &= ~SCF_DO_STCLASS;
1567         }
1568         /* Else: zero-length, ignore. */
1569         scan = regnext(scan);
1570     }
1571
1572   finish:
1573     *scanp = scan;
1574     *deltap = is_inf_internal ? I32_MAX : delta;
1575     if (flags & SCF_DO_SUBSTR && is_inf)
1576         data->pos_delta = I32_MAX - data->pos_min;
1577     if (is_par > U8_MAX)
1578         is_par = 0;
1579     if (is_par && pars==1 && data) {
1580         data->flags |= SF_IN_PAR;
1581         data->flags &= ~SF_HAS_PAR;
1582     }
1583     else if (pars && data) {
1584         data->flags |= SF_HAS_PAR;
1585         data->flags &= ~SF_IN_PAR;
1586     }
1587     if (flags & SCF_DO_STCLASS_OR)
1588         cl_and(data->start_class, &and_with);
1589     return min;
1590 }
1591
1592 STATIC I32
1593 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1594 {
1595     if (RExC_rx->data) {
1596         Renewc(RExC_rx->data,
1597                sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1),
1598                char, struct reg_data);
1599         Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1600         RExC_rx->data->count += n;
1601     }
1602     else {
1603         Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1604              char, struct reg_data);
1605         New(1208, RExC_rx->data->what, n, U8);
1606         RExC_rx->data->count = n;
1607     }
1608     Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1609     return RExC_rx->data->count - n;
1610 }
1611
1612 void
1613 Perl_reginitcolors(pTHX)
1614 {
1615     int i = 0;
1616     char *s = PerlEnv_getenv("PERL_RE_COLORS");
1617         
1618     if (s) {
1619         PL_colors[0] = s = savepv(s);
1620         while (++i < 6) {
1621             s = strchr(s, '\t');
1622             if (s) {
1623                 *s = '\0';
1624                 PL_colors[i] = ++s;
1625             }
1626             else
1627                 PL_colors[i] = s = "";
1628         }
1629     } else {
1630         while (i < 6)
1631             PL_colors[i++] = "";
1632     }
1633     PL_colorset = 1;
1634 }
1635
1636
1637 /*
1638  - pregcomp - compile a regular expression into internal code
1639  *
1640  * We can't allocate space until we know how big the compiled form will be,
1641  * but we can't compile it (and thus know how big it is) until we've got a
1642  * place to put the code.  So we cheat:  we compile it twice, once with code
1643  * generation turned off and size counting turned on, and once "for real".
1644  * This also means that we don't allocate space until we are sure that the
1645  * thing really will compile successfully, and we never have to move the
1646  * code and thus invalidate pointers into it.  (Note that it has to be in
1647  * one piece because free() must be able to free it all.) [NB: not true in perl]
1648  *
1649  * Beware that the optimization-preparation code in here knows about some
1650  * of the structure of the compiled regexp.  [I'll say.]
1651  */
1652 regexp *
1653 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1654 {
1655     register regexp *r;
1656     regnode *scan;
1657     regnode *first;
1658     I32 flags;
1659     I32 minlen = 0;
1660     I32 sawplus = 0;
1661     I32 sawopen = 0;
1662     scan_data_t data;
1663     RExC_state_t RExC_state;
1664     RExC_state_t *pRExC_state = &RExC_state;
1665
1666     if (exp == NULL)
1667         FAIL("NULL regexp argument");
1668
1669     /* XXXX This looks very suspicious... */
1670     if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1671         RExC_utf8 = 1;
1672     else
1673         RExC_utf8 = 0;
1674
1675     RExC_precomp = exp;
1676     DEBUG_r(if (!PL_colorset) reginitcolors());
1677     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1678                       PL_colors[4],PL_colors[5],PL_colors[0],
1679                       (int)(xend - exp), RExC_precomp, PL_colors[1]));
1680     RExC_flags16 = pm->op_pmflags;
1681     RExC_sawback = 0;
1682
1683     RExC_seen = 0;
1684     RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1685     RExC_seen_evals = 0;
1686     RExC_extralen = 0;
1687
1688     /* First pass: determine size, legality. */
1689     RExC_parse = exp;
1690     RExC_start = exp;
1691     RExC_end = xend;
1692     RExC_naughty = 0;
1693     RExC_npar = 1;
1694     RExC_size = 0L;
1695     RExC_emit = &PL_regdummy;
1696     RExC_whilem_seen = 0;
1697 #if 0 /* REGC() is (currently) a NOP at the first pass.
1698        * Clever compilers notice this and complain. --jhi */
1699     REGC((U8)REG_MAGIC, (char*)RExC_emit);
1700 #endif
1701     if (reg(pRExC_state, 0, &flags) == NULL) {
1702         RExC_precomp = Nullch;
1703         return(NULL);
1704     }
1705     DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1706
1707     /* Small enough for pointer-storage convention?
1708        If extralen==0, this means that we will not need long jumps. */
1709     if (RExC_size >= 0x10000L && RExC_extralen)
1710         RExC_size += RExC_extralen;
1711     else
1712         RExC_extralen = 0;
1713     if (RExC_whilem_seen > 15)
1714         RExC_whilem_seen = 15;
1715
1716     /* Allocate space and initialize. */
1717     Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1718          char, regexp);
1719     if (r == NULL)
1720         FAIL("Regexp out of space");
1721
1722 #ifdef DEBUGGING
1723     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1724     Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1725 #endif
1726     r->refcnt = 1;
1727     r->prelen = xend - exp;
1728     r->precomp = savepvn(RExC_precomp, r->prelen);
1729     r->subbeg = NULL;
1730     r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1731     r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1732
1733     r->substrs = 0;                     /* Useful during FAIL. */
1734     r->startp = 0;                      /* Useful during FAIL. */
1735     r->endp = 0;                        /* Useful during FAIL. */
1736
1737     Newz(1304, r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
1738     if (r->offsets) {
1739       r->offsets[0] = RExC_size; 
1740     }
1741     DEBUG_r(PerlIO_printf(Perl_debug_log, 
1742                           "%s %"UVuf" bytes for offset annotations.\n", 
1743                           r->offsets ? "Got" : "Couldn't get", 
1744                           (UV)((2*RExC_size+1) * sizeof(U32))));
1745
1746     RExC_rx = r;
1747
1748     /* Second pass: emit code. */
1749     RExC_parse = exp;
1750     RExC_end = xend;
1751     RExC_naughty = 0;
1752     RExC_npar = 1;
1753     RExC_emit_start = r->program;
1754     RExC_emit = r->program;
1755     /* Store the count of eval-groups for security checks: */
1756     RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1757     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1758     r->data = 0;
1759     if (reg(pRExC_state, 0, &flags) == NULL)
1760         return(NULL);
1761
1762     /* Dig out information for optimizations. */
1763     r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1764     pm->op_pmflags = RExC_flags16;
1765     if (UTF)
1766         r->reganch |= ROPT_UTF8;
1767     r->regstclass = NULL;
1768     if (RExC_naughty >= 10)     /* Probably an expensive pattern. */
1769         r->reganch |= ROPT_NAUGHTY;
1770     scan = r->program + 1;              /* First BRANCH. */
1771
1772     /* XXXX To minimize changes to RE engine we always allocate
1773        3-units-long substrs field. */
1774     Newz(1004, r->substrs, 1, struct reg_substr_data);
1775
1776     StructCopy(&zero_scan_data, &data, scan_data_t);
1777     /* XXXX Should not we check for something else?  Usually it is OPEN1... */
1778     if (OP(scan) != BRANCH) {   /* Only one top-level choice. */
1779         I32 fake;
1780         STRLEN longest_float_length, longest_fixed_length;
1781         struct regnode_charclass_class ch_class;
1782         int stclass_flag;
1783         I32 last_close = 0;
1784
1785         first = scan;
1786         /* Skip introductions and multiplicators >= 1. */
1787         while ((OP(first) == OPEN && (sawopen = 1)) ||
1788                /* An OR of *one* alternative - should not happen now. */
1789             (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1790             (OP(first) == PLUS) ||
1791             (OP(first) == MINMOD) ||
1792                /* An {n,m} with n>0 */
1793             (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1794                 if (OP(first) == PLUS)
1795                     sawplus = 1;
1796                 else
1797                     first += regarglen[(U8)OP(first)];
1798                 first = NEXTOPER(first);
1799         }
1800
1801         /* Starting-point info. */
1802       again:
1803         if (PL_regkind[(U8)OP(first)] == EXACT) {
1804             if (OP(first) == EXACT)
1805                 ;       /* Empty, get anchored substr later. */
1806             else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1807                 r->regstclass = first;
1808         }
1809         else if (strchr((char*)PL_simple,OP(first)))
1810             r->regstclass = first;
1811         else if (PL_regkind[(U8)OP(first)] == BOUND ||
1812                  PL_regkind[(U8)OP(first)] == NBOUND)
1813             r->regstclass = first;
1814         else if (PL_regkind[(U8)OP(first)] == BOL) {
1815             r->reganch |= (OP(first) == MBOL
1816                            ? ROPT_ANCH_MBOL
1817                            : (OP(first) == SBOL
1818                               ? ROPT_ANCH_SBOL
1819                               : ROPT_ANCH_BOL));
1820             first = NEXTOPER(first);
1821             goto again;
1822         }
1823         else if (OP(first) == GPOS) {
1824             r->reganch |= ROPT_ANCH_GPOS;
1825             first = NEXTOPER(first);
1826             goto again;
1827         }
1828         else if ((OP(first) == STAR &&
1829             PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1830             !(r->reganch & ROPT_ANCH) )
1831         {
1832             /* turn .* into ^.* with an implied $*=1 */
1833             int type = OP(NEXTOPER(first));
1834
1835             if (type == REG_ANY)
1836                 type = ROPT_ANCH_MBOL;
1837             else
1838                 type = ROPT_ANCH_SBOL;
1839
1840             r->reganch |= type | ROPT_IMPLICIT;
1841             first = NEXTOPER(first);
1842             goto again;
1843         }
1844         if (sawplus && (!sawopen || !RExC_sawback)
1845             && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1846             /* x+ must match at the 1st pos of run of x's */
1847             r->reganch |= ROPT_SKIP;
1848
1849         /* Scan is after the zeroth branch, first is atomic matcher. */
1850         DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
1851                               (IV)(first - scan + 1)));
1852         /*
1853         * If there's something expensive in the r.e., find the
1854         * longest literal string that must appear and make it the
1855         * regmust.  Resolve ties in favor of later strings, since
1856         * the regstart check works with the beginning of the r.e.
1857         * and avoiding duplication strengthens checking.  Not a
1858         * strong reason, but sufficient in the absence of others.
1859         * [Now we resolve ties in favor of the earlier string if
1860         * it happens that c_offset_min has been invalidated, since the
1861         * earlier string may buy us something the later one won't.]
1862         */
1863         minlen = 0;
1864
1865         data.longest_fixed = newSVpvn("",0);
1866         data.longest_float = newSVpvn("",0);
1867         data.last_found = newSVpvn("",0);
1868         data.longest = &(data.longest_fixed);
1869         first = scan;
1870         if (!r->regstclass) {
1871             cl_init(pRExC_state, &ch_class);
1872             data.start_class = &ch_class;
1873             stclass_flag = SCF_DO_STCLASS_AND;
1874         } else                          /* XXXX Check for BOUND? */
1875             stclass_flag = 0;
1876         data.last_closep = &last_close;
1877
1878         minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1879                              &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1880         if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1881              && data.last_start_min == 0 && data.last_end > 0
1882              && !RExC_seen_zerolen
1883              && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1884             r->reganch |= ROPT_CHECK_ALL;
1885         scan_commit(pRExC_state, &data);
1886         SvREFCNT_dec(data.last_found);
1887
1888         longest_float_length = CHR_SVLEN(data.longest_float);
1889         if (longest_float_length
1890             || (data.flags & SF_FL_BEFORE_EOL
1891                 && (!(data.flags & SF_FL_BEFORE_MEOL)
1892                     || (RExC_flags16 & PMf_MULTILINE)))) {
1893             int t;
1894
1895             if (SvCUR(data.longest_fixed)                       /* ok to leave SvCUR */
1896                 && data.offset_fixed == data.offset_float_min
1897                 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1898                     goto remove_float;          /* As in (a)+. */
1899
1900             r->float_substr = data.longest_float;
1901             r->float_min_offset = data.offset_float_min;
1902             r->float_max_offset = data.offset_float_max;
1903             t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1904                        && (!(data.flags & SF_FL_BEFORE_MEOL)
1905                            || (RExC_flags16 & PMf_MULTILINE)));
1906             fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1907         }
1908         else {
1909           remove_float:
1910             r->float_substr = Nullsv;
1911             SvREFCNT_dec(data.longest_float);
1912             longest_float_length = 0;
1913         }
1914
1915         longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1916         if (longest_fixed_length
1917             || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1918                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1919                     || (RExC_flags16 & PMf_MULTILINE)))) {
1920             int t;
1921
1922             r->anchored_substr = data.longest_fixed;
1923             r->anchored_offset = data.offset_fixed;
1924             t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1925                  && (!(data.flags & SF_FIX_BEFORE_MEOL)
1926                      || (RExC_flags16 & PMf_MULTILINE)));
1927             fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1928         }
1929         else {
1930             r->anchored_substr = Nullsv;
1931             SvREFCNT_dec(data.longest_fixed);
1932             longest_fixed_length = 0;
1933         }
1934         if (r->regstclass
1935             && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1936             r->regstclass = NULL;
1937         if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1938             && !(data.start_class->flags & ANYOF_EOS)
1939             && !cl_is_anything(data.start_class)) {
1940             I32 n = add_data(pRExC_state, 1, "f");
1941
1942             New(1006, RExC_rx->data->data[n], 1,
1943                 struct regnode_charclass_class);
1944             StructCopy(data.start_class,
1945                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1946                        struct regnode_charclass_class);
1947             r->regstclass = (regnode*)RExC_rx->data->data[n];
1948             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1949             PL_regdata = r->data; /* for regprop() */
1950             DEBUG_r({ SV *sv = sv_newmortal();
1951                       regprop(sv, (regnode*)data.start_class);
1952                       PerlIO_printf(Perl_debug_log,
1953                                     "synthetic stclass `%s'.\n",
1954                                     SvPVX(sv));});
1955         }
1956
1957         /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1958         if (longest_fixed_length > longest_float_length) {
1959             r->check_substr = r->anchored_substr;
1960             r->check_offset_min = r->check_offset_max = r->anchored_offset;
1961             if (r->reganch & ROPT_ANCH_SINGLE)
1962                 r->reganch |= ROPT_NOSCAN;
1963         }
1964         else {
1965             r->check_substr = r->float_substr;
1966             r->check_offset_min = data.offset_float_min;
1967             r->check_offset_max = data.offset_float_max;
1968         }
1969         /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1970            This should be changed ASAP!  */
1971         if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1972             r->reganch |= RE_USE_INTUIT;
1973             if (SvTAIL(r->check_substr))
1974                 r->reganch |= RE_INTUIT_TAIL;
1975         }
1976     }
1977     else {
1978         /* Several toplevels. Best we can is to set minlen. */
1979         I32 fake;
1980         struct regnode_charclass_class ch_class;
1981         I32 last_close = 0;
1982         
1983         DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1984         scan = r->program + 1;
1985         cl_init(pRExC_state, &ch_class);
1986         data.start_class = &ch_class;
1987         data.last_closep = &last_close;
1988         minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1989         r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1990         if (!(data.start_class->flags & ANYOF_EOS)
1991             && !cl_is_anything(data.start_class)) {
1992             I32 n = add_data(pRExC_state, 1, "f");
1993
1994             New(1006, RExC_rx->data->data[n], 1,
1995                 struct regnode_charclass_class);
1996             StructCopy(data.start_class,
1997                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1998                        struct regnode_charclass_class);
1999             r->regstclass = (regnode*)RExC_rx->data->data[n];
2000             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
2001             DEBUG_r({ SV* sv = sv_newmortal();
2002                       regprop(sv, (regnode*)data.start_class);
2003                       PerlIO_printf(Perl_debug_log,
2004                                     "synthetic stclass `%s'.\n",
2005                                     SvPVX(sv));});
2006         }
2007     }
2008
2009     r->minlen = minlen;
2010     if (RExC_seen & REG_SEEN_GPOS)
2011         r->reganch |= ROPT_GPOS_SEEN;
2012     if (RExC_seen & REG_SEEN_LOOKBEHIND)
2013         r->reganch |= ROPT_LOOKBEHIND_SEEN;
2014     if (RExC_seen & REG_SEEN_EVAL)
2015         r->reganch |= ROPT_EVAL_SEEN;
2016     if (RExC_seen & REG_SEEN_SANY)
2017         r->reganch |= ROPT_SANY_SEEN;
2018     Newz(1002, r->startp, RExC_npar, I32);
2019     Newz(1002, r->endp, RExC_npar, I32);
2020     PL_regdata = r->data; /* for regprop() */
2021     DEBUG_r(regdump(r));
2022     return(r);
2023 }
2024
2025 /*
2026  - reg - regular expression, i.e. main body or parenthesized thing
2027  *
2028  * Caller must absorb opening parenthesis.
2029  *
2030  * Combining parenthesis handling with the base level of regular expression
2031  * is a trifle forced, but the need to tie the tails of the branches to what
2032  * follows makes it hard to avoid.
2033  */
2034 STATIC regnode *
2035 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2036     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2037 {
2038     register regnode *ret;              /* Will be the head of the group. */
2039     register regnode *br;
2040     register regnode *lastbr;
2041     register regnode *ender = 0;
2042     register I32 parno = 0;
2043     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2044
2045     /* for (?g), (?gc), and (?o) warnings; warning
2046        about (?c) will warn about (?g) -- japhy    */
2047
2048     I32 wastedflags = 0x00,
2049         wasted_o    = 0x01,
2050         wasted_g    = 0x02,
2051         wasted_gc   = 0x02 | 0x04,
2052         wasted_c    = 0x04;
2053
2054     char * parse_start = RExC_parse; /* MJD */
2055     char *oregcomp_parse = RExC_parse;
2056     char c;
2057
2058     *flagp = 0;                         /* Tentatively. */
2059
2060
2061     /* Make an OPEN node, if parenthesized. */
2062     if (paren) {
2063         if (*RExC_parse == '?') { /* (?...) */
2064             U16 posflags = 0, negflags = 0;
2065             U16 *flagsp = &posflags;
2066             int logical = 0;
2067             char *seqstart = RExC_parse;
2068
2069             RExC_parse++;
2070             paren = *RExC_parse++;
2071             ret = NULL;                 /* For look-ahead/behind. */
2072             switch (paren) {
2073             case '<':           /* (?<...) */
2074                 RExC_seen |= REG_SEEN_LOOKBEHIND;
2075                 if (*RExC_parse == '!')
2076                     paren = ',';
2077                 if (*RExC_parse != '=' && *RExC_parse != '!')
2078                     goto unknown;
2079                 RExC_parse++;
2080             case '=':           /* (?=...) */
2081             case '!':           /* (?!...) */
2082                 RExC_seen_zerolen++;
2083             case ':':           /* (?:...) */
2084             case '>':           /* (?>...) */
2085                 break;
2086             case '$':           /* (?$...) */
2087             case '@':           /* (?@...) */
2088                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2089                 break;
2090             case '#':           /* (?#...) */
2091                 while (*RExC_parse && *RExC_parse != ')')
2092                     RExC_parse++;
2093                 if (*RExC_parse != ')')
2094                     FAIL("Sequence (?#... not terminated");
2095                 nextchar(pRExC_state);
2096                 *flagp = TRYAGAIN;
2097                 return NULL;
2098             case 'p':           /* (?p...) */
2099                 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
2100                     vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
2101                 /* FALL THROUGH*/
2102             case '?':           /* (??...) */
2103                 logical = 1;
2104                 paren = *RExC_parse++;
2105                 /* FALL THROUGH */
2106             case '{':           /* (?{...}) */
2107             {
2108                 I32 count = 1, n = 0;
2109                 char c;
2110                 char *s = RExC_parse;
2111                 SV *sv;
2112                 OP_4tree *sop, *rop;
2113
2114                 RExC_seen_zerolen++;
2115                 RExC_seen |= REG_SEEN_EVAL;
2116                 while (count && (c = *RExC_parse)) {
2117                     if (c == '\\' && RExC_parse[1])
2118                         RExC_parse++;
2119                     else if (c == '{')
2120                         count++;
2121                     else if (c == '}')
2122                         count--;
2123                     RExC_parse++;
2124                 }
2125                 if (*RExC_parse != ')')
2126                 {
2127                     RExC_parse = s;             
2128                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2129                 }
2130                 if (!SIZE_ONLY) {
2131                     AV *av;
2132                 
2133                     if (RExC_parse - 1 - s)
2134                         sv = newSVpvn(s, RExC_parse - 1 - s);
2135                     else
2136                         sv = newSVpvn("", 0);
2137
2138                     ENTER;
2139                     Perl_save_re_context(aTHX);
2140                     rop = sv_compile_2op(sv, &sop, "re", &av);
2141                     LEAVE;
2142
2143                     n = add_data(pRExC_state, 3, "nop");
2144                     RExC_rx->data->data[n] = (void*)rop;
2145                     RExC_rx->data->data[n+1] = (void*)sop;
2146                     RExC_rx->data->data[n+2] = (void*)av;
2147                     SvREFCNT_dec(sv);
2148                 }
2149                 else {                                          /* First pass */
2150                     if (PL_reginterp_cnt < ++RExC_seen_evals
2151                         && PL_curcop != &PL_compiling)
2152                         /* No compiled RE interpolated, has runtime
2153                            components ===> unsafe.  */
2154                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2155                     if (PL_tainting && PL_tainted)
2156                         FAIL("Eval-group in insecure regular expression");
2157                 }
2158                 
2159                 nextchar(pRExC_state);
2160                 if (logical) {
2161                     ret = reg_node(pRExC_state, LOGICAL);
2162                     if (!SIZE_ONLY)
2163                         ret->flags = 2;
2164                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2165                     /* deal with the length of this later - MJD */
2166                     return ret;
2167                 }
2168                 return reganode(pRExC_state, EVAL, n);
2169             }
2170             case '(':           /* (?(?{...})...) and (?(?=...)...) */
2171             {
2172                 if (RExC_parse[0] == '?') {        /* (?(?...)) */
2173                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2174                         || RExC_parse[1] == '<'
2175                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2176                         I32 flag;
2177                         
2178                         ret = reg_node(pRExC_state, LOGICAL);
2179                         if (!SIZE_ONLY)
2180                             ret->flags = 1;
2181                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2182                         goto insert_if;
2183                     }
2184                 }
2185                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2186                     /* (?(1)...) */
2187                     parno = atoi(RExC_parse++);
2188
2189                     while (isDIGIT(*RExC_parse))
2190                         RExC_parse++;
2191                     ret = reganode(pRExC_state, GROUPP, parno);
2192                     
2193                     if ((c = *nextchar(pRExC_state)) != ')')
2194                         vFAIL("Switch condition not recognized");
2195                   insert_if:
2196                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2197                     br = regbranch(pRExC_state, &flags, 1);
2198                     if (br == NULL)
2199                         br = reganode(pRExC_state, LONGJMP, 0);
2200                     else
2201                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2202                     c = *nextchar(pRExC_state);
2203                     if (flags&HASWIDTH)
2204                         *flagp |= HASWIDTH;
2205                     if (c == '|') {
2206                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2207                         regbranch(pRExC_state, &flags, 1);
2208                         regtail(pRExC_state, ret, lastbr);
2209                         if (flags&HASWIDTH)
2210                             *flagp |= HASWIDTH;
2211                         c = *nextchar(pRExC_state);
2212                     }
2213                     else
2214                         lastbr = NULL;
2215                     if (c != ')')
2216                         vFAIL("Switch (?(condition)... contains too many branches");
2217                     ender = reg_node(pRExC_state, TAIL);
2218                     regtail(pRExC_state, br, ender);
2219                     if (lastbr) {
2220                         regtail(pRExC_state, lastbr, ender);
2221                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2222                     }
2223                     else
2224                         regtail(pRExC_state, ret, ender);
2225                     return ret;
2226                 }
2227                 else {
2228                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2229                 }
2230             }
2231             case 0:
2232                 RExC_parse--; /* for vFAIL to print correctly */
2233                 vFAIL("Sequence (? incomplete");
2234                 break;
2235             default:
2236                 --RExC_parse;
2237               parse_flags:      /* (?i) */
2238                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2239                     /* (?g), (?gc) and (?o) are useless here
2240                        and must be globally applied -- japhy */
2241
2242                     if (*RExC_parse == 'o' || *RExC_parse == 'g') {
2243                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2244                             I32 wflagbit = *RExC_parse == 'o' ? wasted_o : wasted_g;
2245                             if (! (wastedflags & wflagbit) ) {
2246                                 wastedflags |= wflagbit;
2247                                 vWARN5(
2248                                     RExC_parse + 1,
2249                                     "Useless (%s%c) - %suse /%c modifier",
2250                                     flagsp == &negflags ? "?-" : "?",
2251                                     *RExC_parse,
2252                                     flagsp == &negflags ? "don't " : "",
2253                                     *RExC_parse
2254                                 );
2255                             }
2256                         }
2257                     }
2258                     else if (*RExC_parse == 'c') {
2259                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2260                             if (! (wastedflags & wasted_c) ) {
2261                                 wastedflags |= wasted_gc;
2262                                 vWARN3(
2263                                     RExC_parse + 1,
2264                                     "Useless (%sc) - %suse /gc modifier",
2265                                     flagsp == &negflags ? "?-" : "?",
2266                                     flagsp == &negflags ? "don't " : ""
2267                                 );
2268                             }
2269                         }
2270                     }
2271                     else { pmflag(flagsp, *RExC_parse); }
2272
2273                     ++RExC_parse;
2274                 }
2275                 if (*RExC_parse == '-') {
2276                     flagsp = &negflags;
2277                     wastedflags = 0;  /* reset so (?g-c) warns twice */
2278                     ++RExC_parse;
2279                     goto parse_flags;
2280                 }
2281                 RExC_flags16 |= posflags;
2282                 RExC_flags16 &= ~negflags;
2283                 if (*RExC_parse == ':') {
2284                     RExC_parse++;
2285                     paren = ':';
2286                     break;
2287                 }               
2288               unknown:
2289                 if (*RExC_parse != ')') {
2290                     RExC_parse++;
2291                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2292                 }
2293                 nextchar(pRExC_state);
2294                 *flagp = TRYAGAIN;
2295                 return NULL;
2296             }
2297         }
2298         else {                  /* (...) */
2299             parno = RExC_npar;
2300             RExC_npar++;
2301             ret = reganode(pRExC_state, OPEN, parno);
2302             Set_Node_Length(ret, 1); /* MJD */
2303             Set_Node_Offset(ret, RExC_parse); /* MJD */
2304             open = 1;
2305         }
2306     }
2307     else                        /* ! paren */
2308         ret = NULL;
2309
2310     /* Pick up the branches, linking them together. */
2311     parse_start = RExC_parse;   /* MJD */
2312     br = regbranch(pRExC_state, &flags, 1);
2313     /*     branch_len = (paren != 0); */
2314     
2315     if (br == NULL)
2316         return(NULL);
2317     if (*RExC_parse == '|') {
2318         if (!SIZE_ONLY && RExC_extralen) {
2319             reginsert(pRExC_state, BRANCHJ, br);
2320         }
2321         else {                  /* MJD */
2322             reginsert(pRExC_state, BRANCH, br);
2323             Set_Node_Length(br, paren != 0);
2324             Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2325         }
2326         have_branch = 1;
2327         if (SIZE_ONLY)
2328             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2329     }
2330     else if (paren == ':') {
2331         *flagp |= flags&SIMPLE;
2332     }
2333     if (open) {                         /* Starts with OPEN. */
2334         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2335     }
2336     else if (paren != '?')              /* Not Conditional */
2337         ret = br;
2338     if (flags&HASWIDTH)
2339         *flagp |= HASWIDTH;
2340     *flagp |= flags&SPSTART;
2341     lastbr = br;
2342     while (*RExC_parse == '|') {
2343         if (!SIZE_ONLY && RExC_extralen) {
2344             ender = reganode(pRExC_state, LONGJMP,0);
2345             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2346         }
2347         if (SIZE_ONLY)
2348             RExC_extralen += 2;         /* Account for LONGJMP. */
2349         nextchar(pRExC_state);
2350         br = regbranch(pRExC_state, &flags, 0);
2351         
2352         if (br == NULL)
2353             return(NULL);
2354         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2355         lastbr = br;
2356         if (flags&HASWIDTH)
2357             *flagp |= HASWIDTH;
2358         *flagp |= flags&SPSTART;
2359     }
2360
2361     if (have_branch || paren != ':') {
2362         /* Make a closing node, and hook it on the end. */
2363         switch (paren) {
2364         case ':':
2365             ender = reg_node(pRExC_state, TAIL);
2366             break;
2367         case 1:
2368             ender = reganode(pRExC_state, CLOSE, parno);
2369             Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2370             Set_Node_Length(ender,1); /* MJD */
2371             break;
2372         case '<':
2373         case ',':
2374         case '=':
2375         case '!':
2376             *flagp &= ~HASWIDTH;
2377             /* FALL THROUGH */
2378         case '>':
2379             ender = reg_node(pRExC_state, SUCCEED);
2380             break;
2381         case 0:
2382             ender = reg_node(pRExC_state, END);
2383             break;
2384         }
2385         regtail(pRExC_state, lastbr, ender);
2386
2387         if (have_branch) {
2388             /* Hook the tails of the branches to the closing node. */
2389             for (br = ret; br != NULL; br = regnext(br)) {
2390                 regoptail(pRExC_state, br, ender);
2391             }
2392         }
2393     }
2394
2395     {
2396         char *p;
2397         static char parens[] = "=!<,>";
2398
2399         if (paren && (p = strchr(parens, paren))) {
2400             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2401             int flag = (p - parens) > 1;
2402
2403             if (paren == '>')
2404                 node = SUSPEND, flag = 0;
2405             reginsert(pRExC_state, node,ret);
2406             ret->flags = flag;
2407             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2408         }
2409     }
2410
2411     /* Check for proper termination. */
2412     if (paren) {
2413         RExC_flags16 = oregflags;
2414         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2415             RExC_parse = oregcomp_parse;
2416             vFAIL("Unmatched (");
2417         }
2418     }
2419     else if (!paren && RExC_parse < RExC_end) {
2420         if (*RExC_parse == ')') {
2421             RExC_parse++;
2422             vFAIL("Unmatched )");
2423         }
2424         else
2425             FAIL("Junk on end of regexp");      /* "Can't happen". */
2426         /* NOTREACHED */
2427     }
2428
2429     return(ret);
2430 }
2431
2432 /*
2433  - regbranch - one alternative of an | operator
2434  *
2435  * Implements the concatenation operator.
2436  */
2437 STATIC regnode *
2438 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2439 {
2440     register regnode *ret;
2441     register regnode *chain = NULL;
2442     register regnode *latest;
2443     I32 flags = 0, c = 0;
2444
2445     if (first)
2446         ret = NULL;
2447     else {
2448         if (!SIZE_ONLY && RExC_extralen)
2449             ret = reganode(pRExC_state, BRANCHJ,0);
2450         else {
2451             ret = reg_node(pRExC_state, BRANCH);
2452             Set_Node_Length(ret, 1);
2453         }
2454     }
2455         
2456     if (!first && SIZE_ONLY)
2457         RExC_extralen += 1;                     /* BRANCHJ */
2458
2459     *flagp = WORST;                     /* Tentatively. */
2460
2461     RExC_parse--;
2462     nextchar(pRExC_state);
2463     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2464         flags &= ~TRYAGAIN;
2465         latest = regpiece(pRExC_state, &flags);
2466         if (latest == NULL) {
2467             if (flags & TRYAGAIN)
2468                 continue;
2469             return(NULL);
2470         }
2471         else if (ret == NULL)
2472             ret = latest;
2473         *flagp |= flags&HASWIDTH;
2474         if (chain == NULL)      /* First piece. */
2475             *flagp |= flags&SPSTART;
2476         else {
2477             RExC_naughty++;
2478             regtail(pRExC_state, chain, latest);
2479         }
2480         chain = latest;
2481         c++;
2482     }
2483     if (chain == NULL) {        /* Loop ran zero times. */
2484         chain = reg_node(pRExC_state, NOTHING);
2485         if (ret == NULL)
2486             ret = chain;
2487     }
2488     if (c == 1) {
2489         *flagp |= flags&SIMPLE;
2490     }
2491
2492     return(ret);
2493 }
2494
2495 /*
2496  - regpiece - something followed by possible [*+?]
2497  *
2498  * Note that the branching code sequences used for ? and the general cases
2499  * of * and + are somewhat optimized:  they use the same NOTHING node as
2500  * both the endmarker for their branch list and the body of the last branch.
2501  * It might seem that this node could be dispensed with entirely, but the
2502  * endmarker role is not redundant.
2503  */
2504 STATIC regnode *
2505 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2506 {
2507     register regnode *ret;
2508     register char op;
2509     register char *next;
2510     I32 flags;
2511     char *origparse = RExC_parse;
2512     char *maxpos;
2513     I32 min;
2514     I32 max = REG_INFTY;
2515     char *parse_start;
2516
2517     ret = regatom(pRExC_state, &flags);
2518     if (ret == NULL) {
2519         if (flags & TRYAGAIN)
2520             *flagp |= TRYAGAIN;
2521         return(NULL);
2522     }
2523
2524     op = *RExC_parse;
2525
2526     if (op == '{' && regcurly(RExC_parse)) {
2527         parse_start = RExC_parse; /* MJD */
2528         next = RExC_parse + 1;
2529         maxpos = Nullch;
2530         while (isDIGIT(*next) || *next == ',') {
2531             if (*next == ',') {
2532                 if (maxpos)
2533                     break;
2534                 else
2535                     maxpos = next;
2536             }
2537             next++;
2538         }
2539         if (*next == '}') {             /* got one */
2540             if (!maxpos)
2541                 maxpos = next;
2542             RExC_parse++;
2543             min = atoi(RExC_parse);
2544             if (*maxpos == ',')
2545                 maxpos++;
2546             else
2547                 maxpos = RExC_parse;
2548             max = atoi(maxpos);
2549             if (!max && *maxpos != '0')
2550                 max = REG_INFTY;                /* meaning "infinity" */
2551             else if (max >= REG_INFTY)
2552                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2553             RExC_parse = next;
2554             nextchar(pRExC_state);
2555
2556         do_curly:
2557             if ((flags&SIMPLE)) {
2558                 RExC_naughty += 2 + RExC_naughty / 2;
2559                 reginsert(pRExC_state, CURLY, ret);
2560                 Set_Node_Offset(ret, parse_start+1); /* MJD */
2561                 Set_Node_Cur_Length(ret);
2562             }
2563             else {
2564                 regnode *w = reg_node(pRExC_state, WHILEM);
2565
2566                 w->flags = 0;
2567                 regtail(pRExC_state, ret, w);
2568                 if (!SIZE_ONLY && RExC_extralen) {
2569                     reginsert(pRExC_state, LONGJMP,ret);
2570                     reginsert(pRExC_state, NOTHING,ret);
2571                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2572                 }
2573                 reginsert(pRExC_state, CURLYX,ret);
2574                                 /* MJD hk */
2575                 Set_Node_Offset(ret, parse_start+1);
2576                 Set_Node_Length(ret, 
2577                                 op == '{' ? (RExC_parse - parse_start) : 1);
2578                 
2579                 if (!SIZE_ONLY && RExC_extralen)
2580                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2581                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2582                 if (SIZE_ONLY)
2583                     RExC_whilem_seen++, RExC_extralen += 3;
2584                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2585             }
2586             ret->flags = 0;
2587
2588             if (min > 0)
2589                 *flagp = WORST;
2590             if (max > 0)
2591                 *flagp |= HASWIDTH;
2592             if (max && max < min)
2593                 vFAIL("Can't do {n,m} with n > m");
2594             if (!SIZE_ONLY) {
2595                 ARG1_SET(ret, min);
2596                 ARG2_SET(ret, max);
2597             }
2598
2599             goto nest_check;
2600         }
2601     }
2602
2603     if (!ISMULT1(op)) {
2604         *flagp = flags;
2605         return(ret);
2606     }
2607
2608 #if 0                           /* Now runtime fix should be reliable. */
2609
2610     /* if this is reinstated, don't forget to put this back into perldiag:
2611
2612             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2613
2614            (F) The part of the regexp subject to either the * or + quantifier
2615            could match an empty string. The {#} shows in the regular
2616            expression about where the problem was discovered.
2617
2618     */
2619
2620     if (!(flags&HASWIDTH) && op != '?')
2621       vFAIL("Regexp *+ operand could be empty");
2622 #endif
2623
2624     parse_start = RExC_parse;
2625     nextchar(pRExC_state);
2626
2627     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2628
2629     if (op == '*' && (flags&SIMPLE)) {
2630         reginsert(pRExC_state, STAR, ret);
2631         ret->flags = 0;
2632         RExC_naughty += 4;
2633     }
2634     else if (op == '*') {
2635         min = 0;
2636         goto do_curly;
2637     }
2638     else if (op == '+' && (flags&SIMPLE)) {
2639         reginsert(pRExC_state, PLUS, ret);
2640         ret->flags = 0;
2641         RExC_naughty += 3;
2642     }
2643     else if (op == '+') {
2644         min = 1;
2645         goto do_curly;
2646     }
2647     else if (op == '?') {
2648         min = 0; max = 1;
2649         goto do_curly;
2650     }
2651   nest_check:
2652     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2653         vWARN3(RExC_parse,
2654                "%.*s matches null string many times",
2655                RExC_parse - origparse,
2656                origparse);
2657     }
2658
2659     if (*RExC_parse == '?') {
2660         nextchar(pRExC_state);
2661         reginsert(pRExC_state, MINMOD, ret);
2662         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2663     }
2664     if (ISMULT2(RExC_parse)) {
2665         RExC_parse++;
2666         vFAIL("Nested quantifiers");
2667     }
2668
2669     return(ret);
2670 }
2671
2672 /*
2673  - regatom - the lowest level
2674  *
2675  * Optimization:  gobbles an entire sequence of ordinary characters so that
2676  * it can turn them into a single node, which is smaller to store and
2677  * faster to run.  Backslashed characters are exceptions, each becoming a
2678  * separate node; the code is simpler that way and it's not worth fixing.
2679  *
2680  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2681 STATIC regnode *
2682 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2683 {
2684     register regnode *ret = 0;
2685     I32 flags;
2686     char *parse_start = 0;
2687
2688     *flagp = WORST;             /* Tentatively. */
2689
2690 tryagain:
2691     switch (*RExC_parse) {
2692     case '^':
2693         RExC_seen_zerolen++;
2694         nextchar(pRExC_state);
2695         if (RExC_flags16 & PMf_MULTILINE)
2696             ret = reg_node(pRExC_state, MBOL);
2697         else if (RExC_flags16 & PMf_SINGLELINE)
2698             ret = reg_node(pRExC_state, SBOL);
2699         else
2700             ret = reg_node(pRExC_state, BOL);
2701         Set_Node_Length(ret, 1); /* MJD */
2702         break;
2703     case '$':
2704         nextchar(pRExC_state);
2705         if (*RExC_parse)
2706             RExC_seen_zerolen++;
2707         if (RExC_flags16 & PMf_MULTILINE)
2708             ret = reg_node(pRExC_state, MEOL);
2709         else if (RExC_flags16 & PMf_SINGLELINE)
2710             ret = reg_node(pRExC_state, SEOL);
2711         else
2712             ret = reg_node(pRExC_state, EOL);
2713         Set_Node_Length(ret, 1); /* MJD */
2714         break;
2715     case '.':
2716         nextchar(pRExC_state);
2717         if (RExC_flags16 & PMf_SINGLELINE)
2718             ret = reg_node(pRExC_state, SANY);
2719         else
2720             ret = reg_node(pRExC_state, REG_ANY);
2721         *flagp |= HASWIDTH|SIMPLE;
2722         RExC_naughty++;
2723         Set_Node_Length(ret, 1); /* MJD */
2724         break;
2725     case '[':
2726     {
2727         char *oregcomp_parse = ++RExC_parse;
2728         ret = regclass(pRExC_state);
2729         if (*RExC_parse != ']') {
2730             RExC_parse = oregcomp_parse;
2731             vFAIL("Unmatched [");
2732         }
2733         nextchar(pRExC_state);
2734         *flagp |= HASWIDTH|SIMPLE;
2735         Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2736         break;
2737     }
2738     case '(':
2739         nextchar(pRExC_state);
2740         ret = reg(pRExC_state, 1, &flags);
2741         if (ret == NULL) {
2742                 if (flags & TRYAGAIN) {
2743                     if (RExC_parse == RExC_end) {
2744                          /* Make parent create an empty node if needed. */
2745                         *flagp |= TRYAGAIN;
2746                         return(NULL);
2747                     }
2748                     goto tryagain;
2749                 }
2750                 return(NULL);
2751         }
2752         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2753         break;
2754     case '|':
2755     case ')':
2756         if (flags & TRYAGAIN) {
2757             *flagp |= TRYAGAIN;
2758             return NULL;
2759         }
2760         vFAIL("Internal urp");
2761                                 /* Supposed to be caught earlier. */
2762         break;
2763     case '{':
2764         if (!regcurly(RExC_parse)) {
2765             RExC_parse++;
2766             goto defchar;
2767         }
2768         /* FALL THROUGH */
2769     case '?':
2770     case '+':
2771     case '*':
2772         RExC_parse++;
2773         vFAIL("Quantifier follows nothing");
2774         break;
2775     case '\\':
2776         switch (*++RExC_parse) {
2777         case 'A':
2778             RExC_seen_zerolen++;
2779             ret = reg_node(pRExC_state, SBOL);
2780             *flagp |= SIMPLE;
2781             nextchar(pRExC_state);
2782             Set_Node_Length(ret, 2); /* MJD */
2783             break;
2784         case 'G':
2785             ret = reg_node(pRExC_state, GPOS);
2786             RExC_seen |= REG_SEEN_GPOS;
2787             *flagp |= SIMPLE;
2788             nextchar(pRExC_state);
2789             Set_Node_Length(ret, 2); /* MJD */
2790             break;
2791         case 'Z':
2792             ret = reg_node(pRExC_state, SEOL);
2793             *flagp |= SIMPLE;
2794             nextchar(pRExC_state);
2795             break;
2796         case 'z':
2797             ret = reg_node(pRExC_state, EOS);
2798             *flagp |= SIMPLE;
2799             RExC_seen_zerolen++;                /* Do not optimize RE away */
2800             nextchar(pRExC_state);
2801             Set_Node_Length(ret, 2); /* MJD */
2802             break;
2803         case 'C':
2804             ret = reg_node(pRExC_state, SANY);
2805             RExC_seen |= REG_SEEN_SANY;
2806             *flagp |= HASWIDTH|SIMPLE;
2807             nextchar(pRExC_state);
2808             Set_Node_Length(ret, 2); /* MJD */
2809             break;
2810         case 'X':
2811             ret = reg_node(pRExC_state, CLUMP);
2812             *flagp |= HASWIDTH;
2813             nextchar(pRExC_state);
2814             Set_Node_Length(ret, 2); /* MJD */
2815             break;
2816         case 'w':
2817             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2818             *flagp |= HASWIDTH|SIMPLE;
2819             nextchar(pRExC_state);
2820             Set_Node_Length(ret, 2); /* MJD */
2821             break;
2822         case 'W':
2823             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2824             *flagp |= HASWIDTH|SIMPLE;
2825             nextchar(pRExC_state);
2826             Set_Node_Length(ret, 2); /* MJD */
2827             break;
2828         case 'b':
2829             RExC_seen_zerolen++;
2830             RExC_seen |= REG_SEEN_LOOKBEHIND;
2831             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2832             *flagp |= SIMPLE;
2833             nextchar(pRExC_state);
2834             Set_Node_Length(ret, 2); /* MJD */
2835             break;
2836         case 'B':
2837             RExC_seen_zerolen++;
2838             RExC_seen |= REG_SEEN_LOOKBEHIND;
2839             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2840             *flagp |= SIMPLE;
2841             nextchar(pRExC_state);
2842             Set_Node_Length(ret, 2); /* MJD */
2843             break;
2844         case 's':
2845             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2846             *flagp |= HASWIDTH|SIMPLE;
2847             nextchar(pRExC_state);
2848             Set_Node_Length(ret, 2); /* MJD */
2849             break;
2850         case 'S':
2851             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
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, DIGIT);
2858             *flagp |= HASWIDTH|SIMPLE;
2859             nextchar(pRExC_state);
2860             Set_Node_Length(ret, 2); /* MJD */
2861             break;
2862         case 'D':
2863             ret = reg_node(pRExC_state, NDIGIT);
2864             *flagp |= HASWIDTH|SIMPLE;
2865             nextchar(pRExC_state);
2866             Set_Node_Length(ret, 2); /* MJD */
2867             break;
2868         case 'p':
2869         case 'P':
2870             {   
2871                 char* oldregxend = RExC_end;
2872                 char* parse_start = RExC_parse;
2873
2874                 if (RExC_parse[1] == '{') {
2875                   /* a lovely hack--pretend we saw [\pX] instead */
2876                     RExC_end = strchr(RExC_parse, '}');
2877                     if (!RExC_end) {
2878                         RExC_parse += 2;
2879                         RExC_end = oldregxend;
2880                         vFAIL("Missing right brace on \\p{}");
2881                     }
2882                     RExC_end++;
2883                 }
2884                 else
2885                     RExC_end = RExC_parse + 2;
2886                 RExC_parse--;
2887
2888                 ret = regclass(pRExC_state);
2889
2890                 RExC_end = oldregxend;
2891                 RExC_parse--;
2892                 Set_Node_Cur_Length(ret); /* MJD */
2893                 nextchar(pRExC_state);
2894                 *flagp |= HASWIDTH|SIMPLE;
2895             }
2896             break;
2897         case 'n':
2898         case 'r':
2899         case 't':
2900         case 'f':
2901         case 'e':
2902         case 'a':
2903         case 'x':
2904         case 'c':
2905         case '0':
2906             goto defchar;
2907         case '1': case '2': case '3': case '4':
2908         case '5': case '6': case '7': case '8': case '9':
2909             {
2910                 I32 num = atoi(RExC_parse);
2911
2912                 if (num > 9 && num >= RExC_npar)
2913                     goto defchar;
2914                 else {
2915                     char * parse_start = RExC_parse - 1; /* MJD */
2916                     while (isDIGIT(*RExC_parse))
2917                         RExC_parse++;
2918
2919                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2920                         vFAIL("Reference to nonexistent group");
2921                     RExC_sawback = 1;
2922                     ret = reganode(pRExC_state, FOLD
2923                                    ? (LOC ? REFFL : REFF)
2924                                    : REF, num);
2925                     *flagp |= HASWIDTH;
2926                     
2927                     /* override incorrect value set in reganode MJD */
2928                     Set_Node_Offset(ret, parse_start+1); 
2929                     Set_Node_Cur_Length(ret); /* MJD */
2930                     RExC_parse--;
2931                     nextchar(pRExC_state);
2932                 }
2933             }
2934             break;
2935         case '\0':
2936             if (RExC_parse >= RExC_end)
2937                 FAIL("Trailing \\");
2938             /* FALL THROUGH */
2939         default:
2940             /* Do not generate `unrecognized' warnings here, we fall
2941                back into the quick-grab loop below */
2942             goto defchar;
2943         }
2944         break;
2945
2946     case '#':
2947         if (RExC_flags16 & PMf_EXTENDED) {
2948             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2949             if (RExC_parse < RExC_end)
2950                 goto tryagain;
2951         }
2952         /* FALL THROUGH */
2953
2954     default: {
2955             register STRLEN len;
2956             register UV ender;
2957             register char *p;
2958             char *oldp, *s;
2959             STRLEN numlen;
2960
2961             parse_start = RExC_parse - 1;
2962
2963             RExC_parse++;
2964
2965         defchar:
2966             ret = reg_node(pRExC_state, FOLD
2967                           ? (LOC ? EXACTFL : EXACTF)
2968                           : EXACT);
2969             s = STRING(ret);
2970             for (len = 0, p = RExC_parse - 1;
2971               len < 127 && p < RExC_end;
2972               len++)
2973             {
2974                 oldp = p;
2975
2976                 if (RExC_flags16 & PMf_EXTENDED)
2977                     p = regwhite(p, RExC_end);
2978                 switch (*p) {
2979                 case '^':
2980                 case '$':
2981                 case '.':
2982                 case '[':
2983                 case '(':
2984                 case ')':
2985                 case '|':
2986                     goto loopdone;
2987                 case '\\':
2988                     switch (*++p) {
2989                     case 'A':
2990                     case 'G':
2991                     case 'Z':
2992                     case 'z':
2993                     case 'w':
2994                     case 'W':
2995                     case 'b':
2996                     case 'B':
2997                     case 's':
2998                     case 'S':
2999                     case 'd':
3000                     case 'D':
3001                     case 'p':
3002                     case 'P':
3003                         --p;
3004                         goto loopdone;
3005                     case 'n':
3006                         ender = '\n';
3007                         p++;
3008                         break;
3009                     case 'r':
3010                         ender = '\r';
3011                         p++;
3012                         break;
3013                     case 't':
3014                         ender = '\t';
3015                         p++;
3016                         break;
3017                     case 'f':
3018                         ender = '\f';
3019                         p++;
3020                         break;
3021                     case 'e':
3022                           ender = ASCII_TO_NATIVE('\033');
3023                         p++;
3024                         break;
3025                     case 'a':
3026                           ender = ASCII_TO_NATIVE('\007');
3027                         p++;
3028                         break;
3029                     case 'x':
3030                         if (*++p == '{') {
3031                             char* e = strchr(p, '}');
3032         
3033                             if (!e) {
3034                                 RExC_parse = p + 1;
3035                                 vFAIL("Missing right brace on \\x{}");
3036                             }
3037                             else {
3038                                 numlen = 1;     /* allow underscores */
3039                                 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
3040                                 if (ender > 0xff)
3041                                     RExC_utf8 = 1;
3042                                 /* numlen is generous */
3043                                 if (numlen + len >= 127) {
3044                                     p--;
3045                                     goto loopdone;
3046                                 }
3047                                 p = e + 1;
3048                             }
3049                         }
3050                         else {
3051                             numlen = 0;         /* disallow underscores */
3052                             ender = (UV)scan_hex(p, 2, &numlen);
3053                             p += numlen;
3054                         }
3055                         break;
3056                     case 'c':
3057                         p++;
3058                         ender = UCHARAT(p++);
3059                         ender = toCTRL(ender);
3060                         break;
3061                     case '0': case '1': case '2': case '3':case '4':
3062                     case '5': case '6': case '7': case '8':case '9':
3063                         if (*p == '0' ||
3064                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
3065                             numlen = 0;         /* disallow underscores */
3066                             ender = (UV)scan_oct(p, 3, &numlen);
3067                             p += numlen;
3068                         }
3069                         else {
3070                             --p;
3071                             goto loopdone;
3072                         }
3073                         break;
3074                     case '\0':
3075                         if (p >= RExC_end)
3076                             FAIL("Trailing \\");
3077                         /* FALL THROUGH */
3078                     default:
3079                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3080                             vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
3081                         goto normal_default;
3082                     }
3083                     break;
3084                 default:
3085                   normal_default:
3086                     if (UTF8_IS_START(*p) && UTF) {
3087                         ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3088                                                &numlen, 0);
3089                         p += numlen;
3090                     }
3091                     else
3092                         ender = *p++;
3093                     break;
3094                 }
3095                 if (RExC_flags16 & PMf_EXTENDED)
3096                     p = regwhite(p, RExC_end);
3097                 if (UTF && FOLD) {
3098                     if (LOC)
3099                         ender = toLOWER_LC_uvchr(ender);
3100                     else
3101                         ender = toLOWER_uni(ender);
3102                 }
3103                 if (ISMULT2(p)) { /* Back off on ?+*. */
3104                     if (len)
3105                         p = oldp;
3106                     else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3107                         reguni(pRExC_state, ender, s, &numlen);
3108                         s += numlen;
3109                         len += numlen;
3110                     }
3111                     else {
3112                         len++;
3113                         REGC(ender, s++);
3114                     }
3115                     break;
3116                 }
3117                 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3118                     reguni(pRExC_state, ender, s, &numlen);
3119                     s += numlen;
3120                     len += numlen - 1;
3121                 }
3122                 else
3123                     REGC(ender, s++);
3124             }
3125         loopdone:
3126             RExC_parse = p - 1;
3127             Set_Node_Cur_Length(ret); /* MJD */
3128             nextchar(pRExC_state);
3129             {
3130                 /* len is STRLEN which is unsigned, need to copy to signed */
3131                 IV iv = len;
3132                 if (iv < 0)
3133                     vFAIL("Internal disaster");
3134             }
3135             if (len > 0)
3136                 *flagp |= HASWIDTH;
3137             if (len == 1)
3138                 *flagp |= SIMPLE;
3139             if (!SIZE_ONLY)
3140                 STR_LEN(ret) = len;
3141             if (SIZE_ONLY)
3142                 RExC_size += STR_SZ(len);
3143             else
3144                 RExC_emit += STR_SZ(len);
3145         }
3146         break;
3147     }
3148
3149     return(ret);
3150 }
3151
3152 STATIC char *
3153 S_regwhite(pTHX_ char *p, char *e)
3154 {
3155     while (p < e) {
3156         if (isSPACE(*p))
3157             ++p;
3158         else if (*p == '#') {
3159             do {
3160                 p++;
3161             } while (p < e && *p != '\n');
3162         }
3163         else
3164             break;
3165     }
3166     return p;
3167 }
3168
3169 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3170    Character classes ([:foo:]) can also be negated ([:^foo:]).
3171    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3172    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3173    but trigger warnings because they are currently unimplemented. */
3174 STATIC I32
3175 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3176 {
3177     char *posixcc = 0;
3178     I32 namedclass = OOB_NAMEDCLASS;
3179
3180     if (value == '[' && RExC_parse + 1 < RExC_end &&
3181         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3182         (*RExC_parse == ':' ||
3183          *RExC_parse == '=' ||
3184          *RExC_parse == '.')) {
3185         char  c = *RExC_parse;
3186         char* s = RExC_parse++;
3187         
3188         while (RExC_parse < RExC_end && *RExC_parse != c)
3189             RExC_parse++;
3190         if (RExC_parse == RExC_end)
3191             /* Grandfather lone [:, [=, [. */
3192             RExC_parse = s;
3193         else {
3194             char* t = RExC_parse++; /* skip over the c */
3195
3196             if (*RExC_parse == ']') {
3197                 RExC_parse++; /* skip over the ending ] */
3198                 posixcc = s + 1;
3199                 if (*s == ':') {
3200                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3201                     I32 skip = 5; /* the most common skip */
3202
3203                     switch (*posixcc) {
3204                     case 'a':
3205                         if (strnEQ(posixcc, "alnum", 5))
3206                             namedclass =
3207                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3208                         else if (strnEQ(posixcc, "alpha", 5))
3209                             namedclass =
3210                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3211                         else if (strnEQ(posixcc, "ascii", 5))
3212                             namedclass =
3213                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3214                         break;
3215                     case 'b':
3216                         if (strnEQ(posixcc, "blank", 5))
3217                             namedclass =
3218                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3219                         break;
3220                     case 'c':
3221                         if (strnEQ(posixcc, "cntrl", 5))
3222                             namedclass =
3223                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3224                         break;
3225                     case 'd':
3226                         if (strnEQ(posixcc, "digit", 5))
3227                             namedclass =
3228                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3229                         break;
3230                     case 'g':
3231                         if (strnEQ(posixcc, "graph", 5))
3232                             namedclass =
3233                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3234                         break;
3235                     case 'l':
3236                         if (strnEQ(posixcc, "lower", 5))
3237                             namedclass =
3238                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3239                         break;
3240                     case 'p':
3241                         if (strnEQ(posixcc, "print", 5))
3242                             namedclass =
3243                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3244                         else if (strnEQ(posixcc, "punct", 5))
3245                             namedclass =
3246                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3247                         break;
3248                     case 's':
3249                         if (strnEQ(posixcc, "space", 5))
3250                             namedclass =
3251                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3252                         break;
3253                     case 'u':
3254                         if (strnEQ(posixcc, "upper", 5))
3255                             namedclass =
3256                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3257                         break;
3258                     case 'w': /* this is not POSIX, this is the Perl \w */
3259                         if (strnEQ(posixcc, "word", 4)) {
3260                             namedclass =
3261                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3262                             skip = 4;
3263                         }
3264                         break;
3265                     case 'x':
3266                         if (strnEQ(posixcc, "xdigit", 6)) {
3267                             namedclass =
3268                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3269                             skip = 6;
3270                         }
3271                         break;
3272                     }
3273                     if (namedclass == OOB_NAMEDCLASS ||
3274                         posixcc[skip] != ':' ||
3275                         posixcc[skip+1] != ']')
3276                     {
3277                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3278                                       t - s - 1, s + 1);
3279                     }
3280                 } else if (!SIZE_ONLY) {
3281                     /* [[=foo=]] and [[.foo.]] are still future. */
3282
3283                     /* adjust RExC_parse so the warning shows after
3284                        the class closes */
3285                     while (*RExC_parse && *RExC_parse != ']')
3286                         RExC_parse++;
3287                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3288                 }
3289             } else {
3290                 /* Maternal grandfather:
3291                  * "[:" ending in ":" but not in ":]" */
3292                 RExC_parse = s;
3293             }
3294         }
3295     }
3296
3297     return namedclass;
3298 }
3299
3300 STATIC void
3301 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3302 {
3303     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3304         (*RExC_parse == ':' ||
3305          *RExC_parse == '=' ||
3306          *RExC_parse == '.')) {
3307         char *s = RExC_parse;
3308         char  c = *s++;
3309
3310         while(*s && isALNUM(*s))
3311             s++;
3312         if (*s && c == *s && s[1] == ']') {
3313             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3314
3315             /* [[=foo=]] and [[.foo.]] are still future. */
3316             if (c == '=' || c == '.')
3317             {
3318                 /* adjust RExC_parse so the error shows after
3319                    the class closes */
3320                 while (*RExC_parse && *RExC_parse++ != ']')
3321                     ;
3322                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3323             }
3324         }
3325     }
3326 }
3327
3328 STATIC regnode *
3329 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3330 {
3331     register UV value;
3332     register IV prevvalue = OOB_UNICODE;
3333     register IV range = 0;
3334     register regnode *ret;
3335     STRLEN numlen;
3336     IV namedclass;
3337     char *rangebegin = 0;
3338     bool need_class = 0;
3339     SV *listsv = Nullsv;
3340     register char *e;
3341     UV n;
3342     bool optimize_invert = TRUE;
3343
3344     ret = reganode(pRExC_state, ANYOF, 0);
3345
3346     if (!SIZE_ONLY)
3347         ANYOF_FLAGS(ret) = 0;
3348
3349     if (*RExC_parse == '^') {   /* Complement of range. */
3350         RExC_naughty++;
3351         RExC_parse++;
3352         if (!SIZE_ONLY)
3353             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3354     }
3355
3356     if (SIZE_ONLY)
3357         RExC_size += ANYOF_SKIP;
3358     else {
3359         RExC_emit += ANYOF_SKIP;
3360         if (FOLD)
3361             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3362         if (LOC)
3363             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3364         ANYOF_BITMAP_ZERO(ret);
3365         listsv = newSVpvn("# comment\n", 10);
3366     }
3367
3368     if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3369         checkposixcc(pRExC_state);
3370
3371     if (*RExC_parse == ']' || *RExC_parse == '-')
3372         goto charclassloop;             /* allow 1st char to be ] or - */
3373
3374     while (RExC_parse < RExC_end && *RExC_parse != ']') {
3375
3376     charclassloop:
3377
3378         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3379
3380         if (!range)
3381             rangebegin = RExC_parse;
3382         if (UTF) {
3383             value = utf8n_to_uvchr((U8*)RExC_parse,
3384                                    RExC_end - RExC_parse,
3385                                    &numlen, 0);
3386             RExC_parse += numlen;
3387         }
3388         else
3389             value = UCHARAT(RExC_parse++);
3390         if (value == '[')
3391             namedclass = regpposixcc(pRExC_state, value);
3392         else if (value == '\\') {
3393             if (UTF) {
3394                 value = utf8n_to_uvchr((U8*)RExC_parse,
3395                                    RExC_end - RExC_parse,
3396                                    &numlen, 0);
3397                 RExC_parse += numlen;
3398             }
3399             else
3400                 value = UCHARAT(RExC_parse++);
3401             /* Some compilers cannot handle switching on 64-bit integer
3402              * values, therefore value cannot be an UV.  Yes, this will
3403              * be a problem later if we want switch on Unicode.
3404              * A similar issue a little bit later when switching on
3405              * namedclass. --jhi */
3406             switch ((I32)value) {
3407             case 'w':   namedclass = ANYOF_ALNUM;       break;
3408             case 'W':   namedclass = ANYOF_NALNUM;      break;
3409             case 's':   namedclass = ANYOF_SPACE;       break;
3410             case 'S':   namedclass = ANYOF_NSPACE;      break;
3411             case 'd':   namedclass = ANYOF_DIGIT;       break;
3412             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3413             case 'p':
3414             case 'P':
3415                 if (*RExC_parse == '{') {
3416                     e = strchr(RExC_parse++, '}');
3417                     if (!e)
3418                         vFAIL("Missing right brace on \\p{}");
3419                     n = e - RExC_parse;
3420                 }
3421                 else {
3422                     e = RExC_parse;
3423                     n = 1;
3424                 }
3425                 if (!SIZE_ONLY) {
3426                     if (value == 'p')
3427                         Perl_sv_catpvf(aTHX_ listsv,
3428                                        "+utf8::%.*s\n", (int)n, RExC_parse);
3429                     else
3430                         Perl_sv_catpvf(aTHX_ listsv,
3431                                        "!utf8::%.*s\n", (int)n, RExC_parse);
3432                 }
3433                 RExC_parse = e + 1;
3434                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3435                 continue;
3436             case 'n':   value = '\n';                   break;
3437             case 'r':   value = '\r';                   break;
3438             case 't':   value = '\t';                   break;
3439             case 'f':   value = '\f';                   break;
3440             case 'b':   value = '\b';                   break;
3441             case 'e':   value = ASCII_TO_NATIVE('\033');break;
3442             case 'a':   value = ASCII_TO_NATIVE('\007');break;
3443             case 'x':
3444                 if (*RExC_parse == '{') {
3445                     e = strchr(RExC_parse++, '}');
3446                     if (!e)
3447                         vFAIL("Missing right brace on \\x{}");
3448                     numlen = 1;         /* allow underscores */
3449                     value = (UV)scan_hex(RExC_parse,
3450                                          e - RExC_parse,
3451                                          &numlen);
3452                     RExC_parse = e + 1;
3453                 }
3454                 else {
3455                     numlen = 0;         /* disallow underscores */
3456                     value = (UV)scan_hex(RExC_parse, 2, &numlen);
3457                     RExC_parse += numlen;
3458                 }
3459                 break;
3460             case 'c':
3461                 value = UCHARAT(RExC_parse++);
3462                 value = toCTRL(value);
3463                 break;
3464             case '0': case '1': case '2': case '3': case '4':
3465             case '5': case '6': case '7': case '8': case '9':
3466                 numlen = 0;             /* disallow underscores */
3467                 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3468                 RExC_parse += numlen;
3469                 break;
3470             default:
3471                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3472                     vWARN2(RExC_parse,
3473                            "Unrecognized escape \\%c in character class passed through",
3474                            (int)value);
3475                 break;
3476             }
3477         } /* end of \blah */
3478
3479         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3480
3481             if (!SIZE_ONLY && !need_class)
3482                 ANYOF_CLASS_ZERO(ret);
3483
3484             need_class = 1;
3485
3486             /* a bad range like a-\d, a-[:digit:] ? */
3487             if (range) {
3488                 if (!SIZE_ONLY) {
3489                     if (ckWARN(WARN_REGEXP))
3490                         vWARN4(RExC_parse,
3491                                "False [] range \"%*.*s\"",
3492                                RExC_parse - rangebegin,
3493                                RExC_parse - rangebegin,
3494                                rangebegin);
3495                     if (prevvalue < 256) {
3496                         ANYOF_BITMAP_SET(ret, prevvalue);
3497                         ANYOF_BITMAP_SET(ret, '-');
3498                     }
3499                     else {
3500                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3501                         Perl_sv_catpvf(aTHX_ listsv,
3502                                        "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3503                     }
3504                 }
3505
3506                 range = 0; /* this was not a true range */
3507             }
3508
3509             if (!SIZE_ONLY) {
3510                 if (namedclass > OOB_NAMEDCLASS)
3511                     optimize_invert = FALSE;
3512                 /* Possible truncation here but in some 64-bit environments
3513                  * the compiler gets heartburn about switch on 64-bit values.
3514                  * A similar issue a little earlier when switching on value.
3515                  * --jhi */
3516                 switch ((I32)namedclass) {
3517                 case ANYOF_ALNUM:
3518                     if (LOC)
3519                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3520                     else {
3521                         for (value = 0; value < 256; value++)
3522                             if (isALNUM(value))
3523                                 ANYOF_BITMAP_SET(ret, value);
3524                     }
3525                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3526                     break;
3527                 case ANYOF_NALNUM:
3528                     if (LOC)
3529                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3530                     else {
3531                         for (value = 0; value < 256; value++)
3532                             if (!isALNUM(value))
3533                                 ANYOF_BITMAP_SET(ret, value);
3534                     }
3535                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3536                     break;
3537                 case ANYOF_ALNUMC:
3538                     if (LOC)
3539                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3540                     else {
3541                         for (value = 0; value < 256; value++)
3542                             if (isALNUMC(value))
3543                                 ANYOF_BITMAP_SET(ret, value);
3544                     }
3545                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3546                     break;
3547                 case ANYOF_NALNUMC:
3548                     if (LOC)
3549                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3550                     else {
3551                         for (value = 0; value < 256; value++)
3552                             if (!isALNUMC(value))
3553                                 ANYOF_BITMAP_SET(ret, value);
3554                     }
3555                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3556                     break;
3557                 case ANYOF_ALPHA:
3558                     if (LOC)
3559                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3560                     else {
3561                         for (value = 0; value < 256; value++)
3562                             if (isALPHA(value))
3563                                 ANYOF_BITMAP_SET(ret, value);
3564                     }
3565                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3566                     break;
3567                 case ANYOF_NALPHA:
3568                     if (LOC)
3569                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3570                     else {
3571                         for (value = 0; value < 256; value++)
3572                             if (!isALPHA(value))
3573                                 ANYOF_BITMAP_SET(ret, value);
3574                     }
3575                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3576                     break;
3577                 case ANYOF_ASCII:
3578                     if (LOC)
3579                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3580                     else {
3581 #ifndef EBCDIC
3582                         for (value = 0; value < 128; value++)
3583                             ANYOF_BITMAP_SET(ret, value);
3584 #else  /* EBCDIC */
3585                         for (value = 0; value < 256; value++) {
3586                             if (isASCII(value))
3587                                 ANYOF_BITMAP_SET(ret, value);
3588                         }
3589 #endif /* EBCDIC */
3590                     }
3591                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3592                     break;
3593                 case ANYOF_NASCII:
3594                     if (LOC)
3595                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3596                     else {
3597 #ifndef EBCDIC
3598                         for (value = 128; value < 256; value++)
3599                             ANYOF_BITMAP_SET(ret, value);
3600 #else  /* EBCDIC */
3601                         for (value = 0; value < 256; value++) {
3602                             if (!isASCII(value))
3603                                 ANYOF_BITMAP_SET(ret, value);
3604                         }
3605 #endif /* EBCDIC */
3606                     }
3607                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3608                     break;
3609                 case ANYOF_BLANK:
3610                     if (LOC)
3611                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3612                     else {
3613                         for (value = 0; value < 256; value++)
3614                             if (isBLANK(value))
3615                                 ANYOF_BITMAP_SET(ret, value);
3616                     }
3617                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3618                     break;
3619                 case ANYOF_NBLANK:
3620                     if (LOC)
3621                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3622                     else {
3623                         for (value = 0; value < 256; value++)
3624                             if (!isBLANK(value))
3625                                 ANYOF_BITMAP_SET(ret, value);
3626                     }
3627                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3628                     break;
3629                 case ANYOF_CNTRL:
3630                     if (LOC)
3631                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3632                     else {
3633                         for (value = 0; value < 256; value++)
3634                             if (isCNTRL(value))
3635                                 ANYOF_BITMAP_SET(ret, value);
3636                     }
3637                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3638                     break;
3639                 case ANYOF_NCNTRL:
3640                     if (LOC)
3641                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3642                     else {
3643                         for (value = 0; value < 256; value++)
3644                             if (!isCNTRL(value))
3645                                 ANYOF_BITMAP_SET(ret, value);
3646                     }
3647                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3648                     break;
3649                 case ANYOF_DIGIT:
3650                     if (LOC)
3651                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3652                     else {
3653                         /* consecutive digits assumed */
3654                         for (value = '0'; value <= '9'; value++)
3655                             ANYOF_BITMAP_SET(ret, value);
3656                     }
3657                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3658                     break;
3659                 case ANYOF_NDIGIT:
3660                     if (LOC)
3661                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3662                     else {
3663                         /* consecutive digits assumed */
3664                         for (value = 0; value < '0'; value++)
3665                             ANYOF_BITMAP_SET(ret, value);
3666                         for (value = '9' + 1; value < 256; value++)
3667                             ANYOF_BITMAP_SET(ret, value);
3668                     }
3669                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3670                     break;
3671                 case ANYOF_GRAPH:
3672                     if (LOC)
3673                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3674                     else {
3675                         for (value = 0; value < 256; value++)
3676                             if (isGRAPH(value))
3677                                 ANYOF_BITMAP_SET(ret, value);
3678                     }
3679                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3680                     break;
3681                 case ANYOF_NGRAPH:
3682                     if (LOC)
3683                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3684                     else {
3685                         for (value = 0; value < 256; value++)
3686                             if (!isGRAPH(value))
3687                                 ANYOF_BITMAP_SET(ret, value);
3688                     }
3689                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3690                     break;
3691                 case ANYOF_LOWER:
3692                     if (LOC)
3693                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3694                     else {
3695                         for (value = 0; value < 256; value++)
3696                             if (isLOWER(value))
3697                                 ANYOF_BITMAP_SET(ret, value);
3698                     }
3699                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3700                     break;
3701                 case ANYOF_NLOWER:
3702                     if (LOC)
3703                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3704                     else {
3705                         for (value = 0; value < 256; value++)
3706                             if (!isLOWER(value))
3707                                 ANYOF_BITMAP_SET(ret, value);
3708                     }
3709                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3710                     break;
3711                 case ANYOF_PRINT:
3712                     if (LOC)
3713                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3714                     else {
3715                         for (value = 0; value < 256; value++)
3716                             if (isPRINT(value))
3717                                 ANYOF_BITMAP_SET(ret, value);
3718                     }
3719                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3720                     break;
3721                 case ANYOF_NPRINT:
3722                     if (LOC)
3723                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3724                     else {
3725                         for (value = 0; value < 256; value++)
3726                             if (!isPRINT(value))
3727                                 ANYOF_BITMAP_SET(ret, value);
3728                     }
3729                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3730                     break;
3731                 case ANYOF_PSXSPC:
3732                     if (LOC)
3733                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3734                     else {
3735                         for (value = 0; value < 256; value++)
3736                             if (isPSXSPC(value))
3737                                 ANYOF_BITMAP_SET(ret, value);
3738                     }
3739                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3740                     break;
3741                 case ANYOF_NPSXSPC:
3742                     if (LOC)
3743                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3744                     else {
3745                         for (value = 0; value < 256; value++)
3746                             if (!isPSXSPC(value))
3747                                 ANYOF_BITMAP_SET(ret, value);
3748                     }
3749                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3750                     break;
3751                 case ANYOF_PUNCT:
3752                     if (LOC)
3753                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3754                     else {
3755                         for (value = 0; value < 256; value++)
3756                             if (isPUNCT(value))
3757                                 ANYOF_BITMAP_SET(ret, value);
3758                     }
3759                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3760                     break;
3761                 case ANYOF_NPUNCT:
3762                     if (LOC)
3763                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3764                     else {
3765                         for (value = 0; value < 256; value++)
3766                             if (!isPUNCT(value))
3767                                 ANYOF_BITMAP_SET(ret, value);
3768                     }
3769                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3770                     break;
3771                 case ANYOF_SPACE:
3772                     if (LOC)
3773                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3774                     else {
3775                         for (value = 0; value < 256; value++)
3776                             if (isSPACE(value))
3777                                 ANYOF_BITMAP_SET(ret, value);
3778                     }
3779                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3780                     break;
3781                 case ANYOF_NSPACE:
3782                     if (LOC)
3783                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3784                     else {
3785                         for (value = 0; value < 256; value++)
3786                             if (!isSPACE(value))
3787                                 ANYOF_BITMAP_SET(ret, value);
3788                     }
3789                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3790                     break;
3791                 case ANYOF_UPPER:
3792                     if (LOC)
3793                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3794                     else {
3795                         for (value = 0; value < 256; value++)
3796                             if (isUPPER(value))
3797                                 ANYOF_BITMAP_SET(ret, value);
3798                     }
3799                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3800                     break;
3801                 case ANYOF_NUPPER:
3802                     if (LOC)
3803                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3804                     else {
3805                         for (value = 0; value < 256; value++)
3806                             if (!isUPPER(value))
3807                                 ANYOF_BITMAP_SET(ret, value);
3808                     }
3809                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3810                     break;
3811                 case ANYOF_XDIGIT:
3812                     if (LOC)
3813                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3814                     else {
3815                         for (value = 0; value < 256; value++)
3816                             if (isXDIGIT(value))
3817                                 ANYOF_BITMAP_SET(ret, value);
3818                     }
3819                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3820                     break;
3821                 case ANYOF_NXDIGIT:
3822                     if (LOC)
3823                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3824                     else {
3825                         for (value = 0; value < 256; value++)
3826                             if (!isXDIGIT(value))
3827                                 ANYOF_BITMAP_SET(ret, value);
3828                     }
3829                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3830                     break;
3831                 default:
3832                     vFAIL("Invalid [::] class");
3833                     break;
3834                 }
3835                 if (LOC)
3836                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3837                 continue;
3838             }
3839         } /* end of namedclass \blah */
3840
3841         if (range) {
3842             if (prevvalue > value) /* b-a */ {
3843                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3844                               RExC_parse - rangebegin,
3845                               RExC_parse - rangebegin,
3846                               rangebegin);
3847                 range = 0; /* not a valid range */
3848             }
3849         }
3850         else {
3851             prevvalue = value; /* save the beginning of the range */
3852             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3853                 RExC_parse[1] != ']') {
3854                 RExC_parse++;
3855
3856                 /* a bad range like \w-, [:word:]- ? */
3857                 if (namedclass > OOB_NAMEDCLASS) {
3858                     if (ckWARN(WARN_REGEXP))
3859                         vWARN4(RExC_parse,
3860                                "False [] range \"%*.*s\"",
3861                                RExC_parse - rangebegin,
3862                                RExC_parse - rangebegin,
3863                                rangebegin);
3864                     if (!SIZE_ONLY)
3865                         ANYOF_BITMAP_SET(ret, '-');
3866                 } else
3867                     range = 1;  /* yeah, it's a range! */
3868                 continue;       /* but do it the next time */
3869             }
3870         }
3871
3872         /* now is the next time */
3873         if (!SIZE_ONLY) {
3874             IV i;
3875
3876             if (prevvalue < 256) {
3877                 IV ceilvalue = value < 256 ? value : 255;
3878
3879 #ifdef EBCDIC
3880                 if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3881                     (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3882                 {
3883                     if (isLOWER(prevvalue)) {
3884                         for (i = prevvalue; i <= ceilvalue; i++)
3885                             if (isLOWER(i))
3886                                 ANYOF_BITMAP_SET(ret, i);
3887                     } else {
3888                         for (i = prevvalue; i <= ceilvalue; i++)
3889                             if (isUPPER(i))
3890                                 ANYOF_BITMAP_SET(ret, i);
3891                     }
3892                 }
3893                 else
3894 #endif
3895                     for (i = prevvalue; i <= ceilvalue; i++)
3896                         ANYOF_BITMAP_SET(ret, i);
3897           }
3898           if (value > 255) {
3899                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3900                 if (prevvalue < value)
3901                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3902                                    (UV)prevvalue, (UV)value);
3903                 else if (prevvalue == value)
3904                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3905                                    (UV)value);
3906             }
3907         }
3908
3909         range = 0; /* this range (if it was one) is done now */
3910     }
3911
3912     if (need_class) {
3913         ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3914         if (SIZE_ONLY)
3915             RExC_size += ANYOF_CLASS_ADD_SKIP;
3916         else
3917             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3918     }
3919
3920     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3921     if (!SIZE_ONLY &&
3922          /* If the only flag is folding (plus possibly inversion). */
3923         ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
3924        ) {
3925         for (value = 0; value < 256; ++value) {
3926             if (ANYOF_BITMAP_TEST(ret, value)) {
3927                 IV fold = PL_fold[value];
3928
3929                 if (fold != value)
3930                     ANYOF_BITMAP_SET(ret, fold);
3931             }
3932         }
3933         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3934     }
3935
3936     /* optimize inverted simple patterns (e.g. [^a-z]) */
3937     if (!SIZE_ONLY && optimize_invert &&
3938         /* If the only flag is inversion. */
3939         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3940         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3941             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3942         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3943     }
3944
3945     if (!SIZE_ONLY) {
3946         AV *av = newAV();
3947         SV *rv;
3948
3949         av_store(av, 0, listsv);
3950         av_store(av, 1, NULL);
3951         rv = newRV_noinc((SV*)av);
3952         n = add_data(pRExC_state, 1, "s");
3953         RExC_rx->data->data[n] = (void*)rv;
3954         ARG_SET(ret, n);
3955     }
3956
3957     return ret;
3958 }
3959
3960 STATIC char*
3961 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3962 {
3963     char* retval = RExC_parse++;
3964
3965     for (;;) {
3966         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3967                 RExC_parse[2] == '#') {
3968             while (*RExC_parse && *RExC_parse != ')')
3969                 RExC_parse++;
3970             RExC_parse++;
3971             continue;
3972         }
3973         if (RExC_flags16 & PMf_EXTENDED) {
3974             if (isSPACE(*RExC_parse)) {
3975                 RExC_parse++;
3976                 continue;
3977             }
3978             else if (*RExC_parse == '#') {
3979                 while (*RExC_parse && *RExC_parse != '\n')
3980                     RExC_parse++;
3981                 RExC_parse++;
3982                 continue;
3983             }
3984         }
3985         return retval;
3986     }
3987 }
3988
3989 /*
3990 - reg_node - emit a node
3991 */
3992 STATIC regnode *                        /* Location. */
3993 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3994 {
3995     register regnode *ret;
3996     register regnode *ptr;
3997
3998     ret = RExC_emit;
3999     if (SIZE_ONLY) {
4000         SIZE_ALIGN(RExC_size);
4001         RExC_size += 1;
4002         return(ret);
4003     }
4004
4005     NODE_ALIGN_FILL(ret);
4006     ptr = ret;
4007     FILL_ADVANCE_NODE(ptr, op);
4008     if (RExC_offsets) {         /* MJD */
4009       MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n", 
4010               "reg_node", __LINE__, 
4011               reg_name[op],
4012               RExC_emit - RExC_emit_start > RExC_offsets[0] 
4013               ? "Overwriting end of array!\n" : "OK",
4014               RExC_emit - RExC_emit_start,
4015               RExC_parse - RExC_start,
4016               RExC_offsets[0])); 
4017       Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
4018     }
4019             
4020     RExC_emit = ptr;
4021
4022     return(ret);
4023 }
4024
4025 /*
4026 - reganode - emit a node with an argument
4027 */
4028 STATIC regnode *                        /* Location. */
4029 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
4030 {
4031     register regnode *ret;
4032     register regnode *ptr;
4033
4034     ret = RExC_emit;
4035     if (SIZE_ONLY) {
4036         SIZE_ALIGN(RExC_size);
4037         RExC_size += 2;
4038         return(ret);
4039     }
4040
4041     NODE_ALIGN_FILL(ret);
4042     ptr = ret;
4043     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
4044     if (RExC_offsets) {         /* MJD */
4045       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4046               "reganode",
4047               RExC_emit - RExC_emit_start > RExC_offsets[0] ? 
4048               "Overwriting end of array!\n" : "OK",
4049               RExC_emit - RExC_emit_start,
4050               RExC_parse - RExC_start,
4051               RExC_offsets[0])); 
4052       Set_Cur_Node_Offset;
4053     }
4054             
4055     RExC_emit = ptr;
4056
4057     return(ret);
4058 }
4059
4060 /*
4061 - reguni - emit (if appropriate) a Unicode character
4062 */
4063 STATIC void
4064 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4065 {
4066     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4067 }
4068
4069 /*
4070 - reginsert - insert an operator in front of already-emitted operand
4071 *
4072 * Means relocating the operand.
4073 */
4074 STATIC void
4075 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4076 {
4077     register regnode *src;
4078     register regnode *dst;
4079     register regnode *place;
4080     register int offset = regarglen[(U8)op];
4081
4082 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4083
4084     if (SIZE_ONLY) {
4085         RExC_size += NODE_STEP_REGNODE + offset;
4086         return;
4087     }
4088
4089     src = RExC_emit;
4090     RExC_emit += NODE_STEP_REGNODE + offset;
4091     dst = RExC_emit;
4092     while (src > opnd) {
4093         StructCopy(--src, --dst, regnode);
4094         if (RExC_offsets) {     /* MJD 20010112 */
4095           MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n", 
4096                   "reg_insert",
4097                   dst - RExC_emit_start > RExC_offsets[0] 
4098                   ? "Overwriting end of array!\n" : "OK",
4099                   src - RExC_emit_start,
4100                   dst - RExC_emit_start,
4101                   RExC_offsets[0])); 
4102           Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4103           Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4104         }
4105     }
4106     
4107
4108     place = opnd;               /* Op node, where operand used to be. */
4109     if (RExC_offsets) {         /* MJD */
4110       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4111               "reginsert",
4112               place - RExC_emit_start > RExC_offsets[0] 
4113               ? "Overwriting end of array!\n" : "OK",
4114               place - RExC_emit_start,
4115               RExC_parse - RExC_start,
4116               RExC_offsets[0])); 
4117       Set_Node_Offset(place, RExC_parse);
4118     }
4119     src = NEXTOPER(place);
4120     FILL_ADVANCE_NODE(place, op);
4121     Zero(src, offset, regnode);
4122 }
4123
4124 /*
4125 - regtail - set the next-pointer at the end of a node chain of p to val.
4126 */
4127 STATIC void
4128 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4129 {
4130     register regnode *scan;
4131     register regnode *temp;
4132
4133     if (SIZE_ONLY)
4134         return;
4135
4136     /* Find last node. */
4137     scan = p;
4138     for (;;) {
4139         temp = regnext(scan);
4140         if (temp == NULL)
4141             break;
4142         scan = temp;
4143     }
4144
4145     if (reg_off_by_arg[OP(scan)]) {
4146         ARG_SET(scan, val - scan);
4147     }
4148     else {
4149         NEXT_OFF(scan) = val - scan;
4150     }
4151 }
4152
4153 /*
4154 - regoptail - regtail on operand of first argument; nop if operandless
4155 */
4156 STATIC void
4157 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4158 {
4159     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4160     if (p == NULL || SIZE_ONLY)
4161         return;
4162     if (PL_regkind[(U8)OP(p)] == BRANCH) {
4163         regtail(pRExC_state, NEXTOPER(p), val);
4164     }
4165     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4166         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4167     }
4168     else
4169         return;
4170 }
4171
4172 /*
4173  - regcurly - a little FSA that accepts {\d+,?\d*}
4174  */
4175 STATIC I32
4176 S_regcurly(pTHX_ register char *s)
4177 {
4178     if (*s++ != '{')
4179         return FALSE;
4180     if (!isDIGIT(*s))
4181         return FALSE;
4182     while (isDIGIT(*s))
4183         s++;
4184     if (*s == ',')
4185         s++;
4186     while (isDIGIT(*s))
4187         s++;
4188     if (*s != '}')
4189         return FALSE;
4190     return TRUE;
4191 }
4192
4193
4194 #ifdef DEBUGGING
4195
4196 STATIC regnode *
4197 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4198 {
4199     register U8 op = EXACT;     /* Arbitrary non-END op. */
4200     register regnode *next;
4201
4202     while (op != END && (!last || node < last)) {
4203         /* While that wasn't END last time... */
4204
4205         NODE_ALIGN(node);
4206         op = OP(node);
4207         if (op == CLOSE)
4208             l--;        
4209         next = regnext(node);
4210         /* Where, what. */
4211         if (OP(node) == OPTIMIZED)
4212             goto after_print;
4213         regprop(sv, node);
4214         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4215                       (int)(2*l + 1), "", SvPVX(sv));
4216         if (next == NULL)               /* Next ptr. */
4217             PerlIO_printf(Perl_debug_log, "(0)");
4218         else
4219             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4220         (void)PerlIO_putc(Perl_debug_log, '\n');
4221       after_print:
4222         if (PL_regkind[(U8)op] == BRANCHJ) {
4223             register regnode *nnode = (OP(next) == LONGJMP
4224                                        ? regnext(next)
4225                                        : next);
4226             if (last && nnode > last)
4227                 nnode = last;
4228             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4229         }
4230         else if (PL_regkind[(U8)op] == BRANCH) {
4231             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4232         }
4233         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4234             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4235                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4236         }
4237         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4238             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4239                              next, sv, l + 1);
4240         }
4241         else if ( op == PLUS || op == STAR) {
4242             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4243         }
4244         else if (op == ANYOF) {
4245             /* arglen 1 + class block */
4246             node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4247                     ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4248             node = NEXTOPER(node);
4249         }
4250         else if (PL_regkind[(U8)op] == EXACT) {
4251             /* Literal string, where present. */
4252             node += NODE_SZ_STR(node) - 1;
4253             node = NEXTOPER(node);
4254         }
4255         else {
4256             node = NEXTOPER(node);
4257             node += regarglen[(U8)op];
4258         }
4259         if (op == CURLYX || op == OPEN)
4260             l++;
4261         else if (op == WHILEM)
4262             l--;
4263     }
4264     return node;
4265 }
4266
4267 #endif  /* DEBUGGING */
4268
4269 /*
4270  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4271  */
4272 void
4273 Perl_regdump(pTHX_ regexp *r)
4274 {
4275 #ifdef DEBUGGING
4276     SV *sv = sv_newmortal();
4277
4278     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4279
4280     /* Header fields of interest. */
4281     if (r->anchored_substr)
4282         PerlIO_printf(Perl_debug_log,
4283                       "anchored `%s%.*s%s'%s at %"IVdf" ",
4284                       PL_colors[0],
4285                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4286                       SvPVX(r->anchored_substr),
4287                       PL_colors[1],
4288                       SvTAIL(r->anchored_substr) ? "$" : "",
4289                       (IV)r->anchored_offset);
4290     if (r->float_substr)
4291         PerlIO_printf(Perl_debug_log,
4292                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4293                       PL_colors[0],
4294                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4295                       SvPVX(r->float_substr),
4296                       PL_colors[1],
4297                       SvTAIL(r->float_substr) ? "$" : "",
4298                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4299     if (r->check_substr)
4300         PerlIO_printf(Perl_debug_log,
4301                       r->check_substr == r->float_substr
4302                       ? "(checking floating" : "(checking anchored");
4303     if (r->reganch & ROPT_NOSCAN)
4304         PerlIO_printf(Perl_debug_log, " noscan");
4305     if (r->reganch & ROPT_CHECK_ALL)
4306         PerlIO_printf(Perl_debug_log, " isall");
4307     if (r->check_substr)
4308         PerlIO_printf(Perl_debug_log, ") ");
4309
4310     if (r->regstclass) {
4311         regprop(sv, r->regstclass);
4312         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4313     }
4314     if (r->reganch & ROPT_ANCH) {
4315         PerlIO_printf(Perl_debug_log, "anchored");
4316         if (r->reganch & ROPT_ANCH_BOL)
4317             PerlIO_printf(Perl_debug_log, "(BOL)");
4318         if (r->reganch & ROPT_ANCH_MBOL)
4319             PerlIO_printf(Perl_debug_log, "(MBOL)");
4320         if (r->reganch & ROPT_ANCH_SBOL)
4321             PerlIO_printf(Perl_debug_log, "(SBOL)");
4322         if (r->reganch & ROPT_ANCH_GPOS)
4323             PerlIO_printf(Perl_debug_log, "(GPOS)");
4324         PerlIO_putc(Perl_debug_log, ' ');
4325     }
4326     if (r->reganch & ROPT_GPOS_SEEN)
4327         PerlIO_printf(Perl_debug_log, "GPOS ");
4328     if (r->reganch & ROPT_SKIP)
4329         PerlIO_printf(Perl_debug_log, "plus ");
4330     if (r->reganch & ROPT_IMPLICIT)
4331         PerlIO_printf(Perl_debug_log, "implicit ");
4332     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4333     if (r->reganch & ROPT_EVAL_SEEN)
4334         PerlIO_printf(Perl_debug_log, "with eval ");
4335     PerlIO_printf(Perl_debug_log, "\n");
4336     if (r->offsets) {
4337       U32 i;
4338       U32 len = r->offsets[0];
4339       PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
4340       for (i = 1; i <= len; i++)
4341         PerlIO_printf(Perl_debug_log, "%"UVuf"[%"UVuf"] ", 
4342                       (UV)r->offsets[i*2-1], 
4343                       (UV)r->offsets[i*2]);
4344       PerlIO_printf(Perl_debug_log, "\n");
4345     }
4346 #endif  /* DEBUGGING */
4347 }
4348
4349 #ifdef DEBUGGING
4350
4351 STATIC void
4352 S_put_byte(pTHX_ SV *sv, int c)
4353 {
4354     if (isCNTRL(c) || c == 127 || c == 255 || !isPRINT(c))
4355         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4356     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4357         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4358     else
4359         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4360 }
4361
4362 #endif  /* DEBUGGING */
4363
4364 /*
4365 - regprop - printable representation of opcode
4366 */
4367 void
4368 Perl_regprop(pTHX_ SV *sv, regnode *o)
4369 {
4370 #ifdef DEBUGGING
4371     register int k;
4372
4373     sv_setpvn(sv, "", 0);
4374     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4375         /* It would be nice to FAIL() here, but this may be called from
4376            regexec.c, and it would be hard to supply pRExC_state. */
4377         Perl_croak(aTHX_ "Corrupted regexp opcode");
4378     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4379
4380     k = PL_regkind[(U8)OP(o)];
4381
4382     if (k == EXACT)
4383         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4384                        STR_LEN(o), STRING(o), PL_colors[1]);
4385     else if (k == CURLY) {
4386         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4387             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4388         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4389     }
4390     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4391         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4392     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4393         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4394     else if (k == LOGICAL)
4395         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4396     else if (k == ANYOF) {
4397         int i, rangestart = -1;
4398         U8 flags = ANYOF_FLAGS(o);
4399         const char * const anyofs[] = { /* Should be syncronized with
4400                                          * ANYOF_ #xdefines in regcomp.h */
4401             "\\w",
4402             "\\W",
4403             "\\s",
4404             "\\S",
4405             "\\d",
4406             "\\D",
4407             "[:alnum:]",
4408             "[:^alnum:]",
4409             "[:alpha:]",
4410             "[:^alpha:]",
4411             "[:ascii:]",
4412             "[:^ascii:]",
4413             "[:ctrl:]",
4414             "[:^ctrl:]",
4415             "[:graph:]",
4416             "[:^graph:]",
4417             "[:lower:]",
4418             "[:^lower:]",
4419             "[:print:]",
4420             "[:^print:]",
4421             "[:punct:]",
4422             "[:^punct:]",
4423             "[:upper:]",
4424             "[:^upper:]",
4425             "[:xdigit:]",
4426             "[:^xdigit:]",
4427             "[:space:]",
4428             "[:^space:]",
4429             "[:blank:]",
4430             "[:^blank:]"
4431         };
4432
4433         if (flags & ANYOF_LOCALE)
4434             sv_catpv(sv, "{loc}");
4435         if (flags & ANYOF_FOLD)
4436             sv_catpv(sv, "{i}");
4437         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4438         if (flags & ANYOF_INVERT)
4439             sv_catpv(sv, "^");
4440         for (i = 0; i <= 256; i++) {
4441             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4442                 if (rangestart == -1)
4443                     rangestart = i;
4444             } else if (rangestart != -1) {
4445                 if (i <= rangestart + 3)
4446                     for (; rangestart < i; rangestart++)
4447                         put_byte(sv, rangestart);
4448                 else {
4449                     put_byte(sv, rangestart);
4450                     sv_catpv(sv, "-");
4451                     put_byte(sv, i - 1);
4452                 }
4453                 rangestart = -1;
4454             }
4455         }
4456
4457         if (o->flags & ANYOF_CLASS)
4458             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4459                 if (ANYOF_CLASS_TEST(o,i))
4460                     sv_catpv(sv, anyofs[i]);
4461
4462         if (flags & ANYOF_UNICODE)
4463             sv_catpv(sv, "{unicode}");
4464         else if (flags & ANYOF_UNICODE_ALL)
4465             sv_catpv(sv, "{all-unicode}");
4466
4467         {
4468             SV *lv;
4469             SV *sw = regclass_swash(o, FALSE, &lv);
4470         
4471             if (lv) {
4472                 if (sw) {
4473                     UV i;
4474                     U8 s[UTF8_MAXLEN+1];
4475                 
4476                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4477                         U8 *e = uvchr_to_utf8(s, i);
4478                         
4479                         if (i < 256 && swash_fetch(sw, s, TRUE)) {
4480                             if (rangestart == -1)
4481                                 rangestart = i;
4482                         } else if (rangestart != -1) {
4483                             U8 *p;
4484                         
4485                             if (i <= rangestart + 3)
4486                                 for (; rangestart < i; rangestart++) {
4487                                     for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4488                                         put_byte(sv, *p);
4489                                 }
4490                             else {
4491                                 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4492                                     put_byte(sv, *p);
4493                                 sv_catpv(sv, "-");
4494                                     for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4495                                         put_byte(sv, *p);
4496                                 }
4497                                 rangestart = -1;
4498                             }
4499                         }
4500                         
4501                     sv_catpv(sv, "..."); /* et cetera */
4502                 }
4503
4504                 {
4505                     char *s = savepv(SvPVX(lv));
4506                     char *origs = s;
4507                 
4508                     while(*s && *s != '\n') s++;
4509                 
4510                     if (*s == '\n') {
4511                         char *t = ++s;
4512                         
4513                         while (*s) {
4514                             if (*s == '\n')
4515                                 *s = ' ';
4516                             s++;
4517                         }
4518                         if (s[-1] == ' ')
4519                             s[-1] = 0;
4520                         
4521                         sv_catpv(sv, t);
4522                     }
4523                 
4524                     Safefree(origs);
4525                 }
4526             }
4527         }
4528
4529         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4530     }
4531     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4532         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4533 #endif  /* DEBUGGING */
4534 }
4535
4536 SV *
4537 Perl_re_intuit_string(pTHX_ regexp *prog)
4538 {                               /* Assume that RE_INTUIT is set */
4539     DEBUG_r(
4540         {   STRLEN n_a;
4541             char *s = SvPV(prog->check_substr,n_a);
4542
4543             if (!PL_colorset) reginitcolors();
4544             PerlIO_printf(Perl_debug_log,
4545                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4546                       PL_colors[4],PL_colors[5],PL_colors[0],
4547                       s,
4548                       PL_colors[1],
4549                       (strlen(s) > 60 ? "..." : ""));
4550         } );
4551
4552     return prog->check_substr;
4553 }
4554
4555 void
4556 Perl_pregfree(pTHX_ struct regexp *r)
4557 {
4558     DEBUG_r(if (!PL_colorset) reginitcolors());
4559
4560     if (!r || (--r->refcnt > 0))
4561         return;
4562     DEBUG_r(PerlIO_printf(Perl_debug_log,
4563                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4564                       PL_colors[4],PL_colors[5],PL_colors[0],
4565                       r->precomp,
4566                       PL_colors[1],
4567                       (strlen(r->precomp) > 60 ? "..." : "")));
4568
4569     if (r->precomp)
4570         Safefree(r->precomp);
4571     if (r->offsets)             /* 20010421 MJD */
4572         Safefree(r->offsets);
4573     if (RX_MATCH_COPIED(r))
4574         Safefree(r->subbeg);
4575     if (r->substrs) {
4576         if (r->anchored_substr)
4577             SvREFCNT_dec(r->anchored_substr);
4578         if (r->float_substr)
4579             SvREFCNT_dec(r->float_substr);
4580         Safefree(r->substrs);
4581     }
4582     if (r->data) {
4583         int n = r->data->count;
4584         AV* new_comppad = NULL;
4585         AV* old_comppad;
4586         SV** old_curpad;
4587
4588         while (--n >= 0) {
4589           /* If you add a ->what type here, update the comment in regcomp.h */
4590             switch (r->data->what[n]) {
4591             case 's':
4592                 SvREFCNT_dec((SV*)r->data->data[n]);
4593                 break;
4594             case 'f':
4595                 Safefree(r->data->data[n]);
4596                 break;
4597             case 'p':
4598                 new_comppad = (AV*)r->data->data[n];
4599                 break;
4600             case 'o':
4601                 if (new_comppad == NULL)
4602                     Perl_croak(aTHX_ "panic: pregfree comppad");
4603                 old_comppad = PL_comppad;
4604                 old_curpad = PL_curpad;
4605                 /* Watch out for global destruction's random ordering. */
4606                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4607                     PL_comppad = new_comppad;
4608                     PL_curpad = AvARRAY(new_comppad);
4609                 }
4610                 else
4611                     PL_curpad = NULL;
4612                 op_free((OP_4tree*)r->data->data[n]);
4613                 PL_comppad = old_comppad;
4614                 PL_curpad = old_curpad;
4615                 SvREFCNT_dec((SV*)new_comppad);
4616                 new_comppad = NULL;
4617                 break;
4618             case 'n':
4619                 break;
4620             default:
4621                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4622             }
4623         }
4624         Safefree(r->data->what);
4625         Safefree(r->data);
4626     }
4627     Safefree(r->startp);
4628     Safefree(r->endp);
4629     Safefree(r);
4630 }
4631
4632 /*
4633  - regnext - dig the "next" pointer out of a node
4634  *
4635  * [Note, when REGALIGN is defined there are two places in regmatch()
4636  * that bypass this code for speed.]
4637  */
4638 regnode *
4639 Perl_regnext(pTHX_ register regnode *p)
4640 {
4641     register I32 offset;
4642
4643     if (p == &PL_regdummy)
4644         return(NULL);
4645
4646     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4647     if (offset == 0)
4648         return(NULL);
4649
4650     return(p+offset);
4651 }
4652
4653 STATIC void     
4654 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4655 {
4656     va_list args;
4657     STRLEN l1 = strlen(pat1);
4658     STRLEN l2 = strlen(pat2);
4659     char buf[512];
4660     SV *msv;
4661     char *message;
4662
4663     if (l1 > 510)
4664         l1 = 510;
4665     if (l1 + l2 > 510)
4666         l2 = 510 - l1;
4667     Copy(pat1, buf, l1 , char);
4668     Copy(pat2, buf + l1, l2 , char);
4669     buf[l1 + l2] = '\n';
4670     buf[l1 + l2 + 1] = '\0';
4671 #ifdef I_STDARG
4672     /* ANSI variant takes additional second argument */
4673     va_start(args, pat2);
4674 #else
4675     va_start(args);
4676 #endif
4677     msv = vmess(buf, &args);
4678     va_end(args);
4679     message = SvPV(msv,l1);
4680     if (l1 > 512)
4681         l1 = 512;
4682     Copy(message, buf, l1 , char);
4683     buf[l1] = '\0';                     /* Overwrite \n */
4684     Perl_croak(aTHX_ "%s", buf);
4685 }
4686
4687 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4688
4689 void
4690 Perl_save_re_context(pTHX)
4691 {
4692 #if 0
4693     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4694     SAVEI32(RExC_npar);         /* () count. */
4695     SAVEI32(RExC_size);         /* Code size. */
4696     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4697     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4698     SAVEI32(RExC_seen);         /* from regcomp.c */
4699     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4700     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4701     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4702     SAVEPPTR(RExC_end);         /* End of input for compile */
4703     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4704 #endif
4705
4706     SAVEI32(PL_reg_flags);              /* from regexec.c */
4707     SAVEPPTR(PL_bostr);
4708     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4709     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4710     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4711     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4712     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4713     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4714     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4715     SAVEGENERICPV(PL_reg_start_tmp);            /* from regexec.c */
4716     PL_reg_start_tmp = 0;
4717     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4718     PL_reg_start_tmpl = 0;
4719     SAVEVPTR(PL_regdata);
4720     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4721     SAVEI32(PL_regnarrate);             /* from regexec.c */
4722     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4723     SAVEINT(PL_regindent);              /* from regexec.c */
4724     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4725     SAVEVPTR(PL_curcop);
4726     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4727     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4728     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4729     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4730     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4731     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4732     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4733     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4734     SAVEI32(PL_regnpar);                /* () count. */
4735     SAVEI32(PL_regsize);                /* from regexec.c */
4736 #ifdef DEBUGGING
4737     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */
4738 #endif
4739 }
4740
4741 #ifdef PERL_OBJECT
4742 #include "XSUB.h"
4743 #undef this
4744 #define this pPerl
4745 #endif
4746
4747 static void
4748 clear_re(pTHXo_ void *r)
4749 {
4750     ReREFCNT_dec((regexp *)r);
4751 }
4752