Tweak for MULTIPLICITY/USE_PERLIO
[p5sagit/p5-mst-13.2.git] / regcomp.c
1 /*    regcomp.c
2  */
3
4 /*
5  * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
6  */
7
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9  * confused with the original package (see point 3 below).  Thanks, Henry!
10  */
11
12 /* Additional note: this code is very heavily munged from Henry's version
13  * in places.  In some spots I've traded clarity for efficiency, so don't
14  * blame Henry for some of the lack of readability.
15  */
16
17 /* The names of the functions have been changed from regcomp and
18  * regexec to  pregcomp and pregexec in order to avoid conflicts
19  * with the POSIX routines of the same names.
20 */
21
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 #  ifndef PERL_IN_XSUB_RE
25 #    define PERL_IN_XSUB_RE
26 #  endif
27 /* need access to debugger hooks */
28 #  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
29 #    define DEBUGGING
30 #  endif
31 #endif
32
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 #  define Perl_pregcomp my_regcomp
36 #  define Perl_regdump my_regdump
37 #  define Perl_regprop my_regprop
38 #  define Perl_pregfree my_regfree
39 #  define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 #  define Perl_regnext my_regnext
42 #  define Perl_save_re_context my_save_re_context
43 #  define Perl_reginitcolors my_reginitcolors 
44
45 #  define PERL_NO_GET_CONTEXT
46 #endif 
47
48 /*SUPPRESS 112*/
49 /*
50  * pregcomp and pregexec -- regsub and regerror are not used in perl
51  *
52  *      Copyright (c) 1986 by University of Toronto.
53  *      Written by Henry Spencer.  Not derived from licensed software.
54  *
55  *      Permission is granted to anyone to use this software for any
56  *      purpose on any computer system, and to redistribute it freely,
57  *      subject to the following restrictions:
58  *
59  *      1. The author is not responsible for the consequences of use of
60  *              this software, no matter how awful, even if they arise
61  *              from defects in it.
62  *
63  *      2. The origin of this software must not be misrepresented, either
64  *              by explicit claim or by omission.
65  *
66  *      3. Altered versions must be plainly marked as such, and must not
67  *              be misrepresented as being the original software.
68  *
69  *
70  ****    Alterations to Henry's code are...
71  ****
72  ****    Copyright (c) 1991-2000, 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        *end;                   /* End of input for compile */
119     char        *parse;                 /* Input-scan pointer. */
120     I32         whilem_seen;            /* number of WHILEM in this expr */
121     regnode     *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
122     I32         naughty;                /* How bad is this pattern? */
123     I32         sawback;                /* Did we see \1, ...? */
124     U32         seen;
125     I32         size;                   /* Code size. */
126     I32         npar;                   /* () count. */
127     I32         extralen;
128     I32         seen_zerolen;
129     I32         seen_evals;
130     I32         utf8;
131 #if ADD_TO_REGEXEC
132     char        *starttry;              /* -Dr: where regtry was called. */
133 #define RExC_starttry   (pRExC_state->starttry)
134 #endif
135 } RExC_state_t;
136
137 #define RExC_flags16    (pRExC_state->flags16)
138 #define RExC_precomp    (pRExC_state->precomp)
139 #define RExC_rx         (pRExC_state->rx)
140 #define RExC_end        (pRExC_state->end)
141 #define RExC_parse      (pRExC_state->parse)
142 #define RExC_whilem_seen        (pRExC_state->whilem_seen)
143 #define RExC_emit       (pRExC_state->emit)
144 #define RExC_naughty    (pRExC_state->naughty)
145 #define RExC_sawback    (pRExC_state->sawback)
146 #define RExC_seen       (pRExC_state->seen)
147 #define RExC_size       (pRExC_state->size)
148 #define RExC_npar       (pRExC_state->npar)
149 #define RExC_extralen   (pRExC_state->extralen)
150 #define RExC_seen_zerolen       (pRExC_state->seen_zerolen)
151 #define RExC_seen_evals (pRExC_state->seen_evals)
152 #define RExC_utf8       (pRExC_state->utf8)
153
154 #define ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
155 #define ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
156         ((*s) == '{' && regcurly(s)))
157 #ifdef atarist
158 #define PERL_META       "^$.[()|?+*\\"
159 #else
160 #define META    "^$.[()|?+*\\"
161 #endif
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 " before " MARKER1 " mark in regex 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
397 #define vWARN2(loc, m, a1)                                                   \
398     STMT_START {                                                             \
399         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
400         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
401                  a1,                                                         \
402                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
403     } STMT_END
404
405 #define vWARN3(loc, m, a1, a2)                                               \
406     STMT_START {                                                             \
407       unsigned offset = strlen(RExC_precomp) - (RExC_end - (loc));        \
408         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,                    \
409                  a1, a2,                                                     \
410                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
411     } STMT_END
412
413 #define vWARN4(loc, m, a1, a2, a3)                                           \
414     STMT_START {                                                             \
415       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));            \
416         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
417                  a1, a2, a3,                                                 \
418                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
419     } STMT_END
420
421
422 /* Allow for side effects in s */
423 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (s);} STMT_END
424
425 static void clear_re(pTHXo_ void *r);
426
427 /* Mark that we cannot extend a found fixed substring at this point.
428    Updata the longest found anchored substring and the longest found
429    floating substrings if needed. */
430
431 STATIC void
432 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
433 {
434     STRLEN l = CHR_SVLEN(data->last_found);
435     STRLEN old_l = CHR_SVLEN(*data->longest);
436     
437     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
438         sv_setsv(*data->longest, data->last_found);
439         if (*data->longest == data->longest_fixed) {
440             data->offset_fixed = l ? data->last_start_min : data->pos_min;
441             if (data->flags & SF_BEFORE_EOL)
442                 data->flags 
443                     |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
444             else
445                 data->flags &= ~SF_FIX_BEFORE_EOL;
446         }
447         else {
448             data->offset_float_min = l ? data->last_start_min : data->pos_min;
449             data->offset_float_max = (l 
450                                       ? data->last_start_max 
451                                       : data->pos_min + data->pos_delta);
452             if (data->flags & SF_BEFORE_EOL)
453                 data->flags 
454                     |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
455             else
456                 data->flags &= ~SF_FL_BEFORE_EOL;
457         }
458     }
459     SvCUR_set(data->last_found, 0);
460     data->last_end = -1;
461     data->flags &= ~SF_BEFORE_EOL;
462 }
463
464 /* Can match anything (initialization) */
465 STATIC void
466 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
467 {
468     int value;
469
470     ANYOF_CLASS_ZERO(cl);
471     for (value = 0; value < 256; ++value)
472         ANYOF_BITMAP_SET(cl, value);
473     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
474     if (LOC)
475         cl->flags |= ANYOF_LOCALE;
476 }
477
478 /* Can match anything (initialization) */
479 STATIC int
480 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
481 {
482     int value;
483
484     for (value = 0; value <= ANYOF_MAX; value += 2)
485         if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
486             return 1;
487     if (!(cl->flags & ANYOF_UNICODE_ALL))
488         return 0;
489     for (value = 0; value < 256; ++value)
490         if (!ANYOF_BITMAP_TEST(cl, value))
491             return 0;
492     return 1;
493 }
494
495 /* Can match anything (initialization) */
496 STATIC void
497 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
498 {
499     Zero(cl, 1, struct regnode_charclass_class);
500     cl->type = ANYOF;
501     cl_anything(pRExC_state, cl);
502 }
503
504 STATIC void
505 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
506 {
507     Zero(cl, 1, struct regnode_charclass_class);
508     cl->type = ANYOF;
509     cl_anything(pRExC_state, cl);
510     if (LOC)
511         cl->flags |= ANYOF_LOCALE;
512 }
513
514 /* 'And' a given class with another one.  Can create false positives */
515 /* We assume that cl is not inverted */
516 STATIC void
517 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
518          struct regnode_charclass_class *and_with)
519 {
520     if (!(and_with->flags & ANYOF_CLASS)
521         && !(cl->flags & ANYOF_CLASS)
522         && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
523         && !(and_with->flags & ANYOF_FOLD)
524         && !(cl->flags & ANYOF_FOLD)) {
525         int i;
526
527         if (and_with->flags & ANYOF_INVERT)
528             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
529                 cl->bitmap[i] &= ~and_with->bitmap[i];
530         else
531             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
532                 cl->bitmap[i] &= and_with->bitmap[i];
533     } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
534     if (!(and_with->flags & ANYOF_EOS))
535         cl->flags &= ~ANYOF_EOS;
536
537     if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
538         cl->flags &= ~ANYOF_UNICODE_ALL;
539         cl->flags |= ANYOF_UNICODE;
540         ARG_SET(cl, ARG(and_with));
541     }
542     if (!(and_with->flags & ANYOF_UNICODE_ALL))
543         cl->flags &= ~ANYOF_UNICODE_ALL;
544     if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
545         cl->flags &= ~ANYOF_UNICODE;
546 }
547
548 /* 'OR' a given class with another one.  Can create false positives */
549 /* We assume that cl is not inverted */
550 STATIC void
551 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
552 {
553     if (or_with->flags & ANYOF_INVERT) {
554         /* We do not use
555          * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
556          *   <= (B1 | !B2) | (CL1 | !CL2)
557          * which is wasteful if CL2 is small, but we ignore CL2:
558          *   (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
559          * XXXX Can we handle case-fold?  Unclear:
560          *   (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
561          *   (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
562          */
563         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
564              && !(or_with->flags & ANYOF_FOLD)
565              && !(cl->flags & ANYOF_FOLD) ) {
566             int i;
567
568             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
569                 cl->bitmap[i] |= ~or_with->bitmap[i];
570         } /* XXXX: logic is complicated otherwise */
571         else {
572             cl_anything(pRExC_state, cl);
573         }
574     } else {
575         /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
576         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
577              && (!(or_with->flags & ANYOF_FOLD) 
578                  || (cl->flags & ANYOF_FOLD)) ) {
579             int i;
580
581             /* OR char bitmap and class bitmap separately */
582             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
583                 cl->bitmap[i] |= or_with->bitmap[i];
584             if (or_with->flags & ANYOF_CLASS) {
585                 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
586                     cl->classflags[i] |= or_with->classflags[i];
587                 cl->flags |= ANYOF_CLASS;
588             }
589         }
590         else { /* XXXX: logic is complicated, leave it along for a moment. */
591             cl_anything(pRExC_state, cl);
592         }
593     }
594     if (or_with->flags & ANYOF_EOS)
595         cl->flags |= ANYOF_EOS;
596
597     if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
598         ARG(cl) != ARG(or_with)) {
599         cl->flags |= ANYOF_UNICODE_ALL;
600         cl->flags &= ~ANYOF_UNICODE;
601     }
602     if (or_with->flags & ANYOF_UNICODE_ALL) {
603         cl->flags |= ANYOF_UNICODE_ALL;
604         cl->flags &= ~ANYOF_UNICODE;
605     }
606 }
607
608 /* REx optimizer.  Converts nodes into quickier variants "in place".
609    Finds fixed substrings.  */
610
611 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
612    to the position after last scanned or to NULL. */
613
614 STATIC I32
615 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
616                         /* scanp: Start here (read-write). */
617                         /* deltap: Write maxlen-minlen here. */
618                         /* last: Stop before this one. */
619 {
620     I32 min = 0, pars = 0, code;
621     regnode *scan = *scanp, *next;
622     I32 delta = 0;
623     int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
624     int is_inf_internal = 0;            /* The studied chunk is infinite */
625     I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
626     scan_data_t data_fake;
627     struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
628     
629     while (scan && OP(scan) != END && scan < last) {
630         /* Peephole optimizer: */
631
632         if (PL_regkind[(U8)OP(scan)] == EXACT) {
633             /* Merge several consecutive EXACTish nodes into one. */
634             regnode *n = regnext(scan);
635             U32 stringok = 1;
636 #ifdef DEBUGGING
637             regnode *stop = scan;
638 #endif 
639
640             next = scan + NODE_SZ_STR(scan);
641             /* Skip NOTHING, merge EXACT*. */
642             while (n &&
643                    ( PL_regkind[(U8)OP(n)] == NOTHING || 
644                      (stringok && (OP(n) == OP(scan))))
645                    && NEXT_OFF(n)
646                    && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
647                 if (OP(n) == TAIL || n > next)
648                     stringok = 0;
649                 if (PL_regkind[(U8)OP(n)] == NOTHING) {
650                     NEXT_OFF(scan) += NEXT_OFF(n);
651                     next = n + NODE_STEP_REGNODE;
652 #ifdef DEBUGGING
653                     if (stringok)
654                         stop = n;
655 #endif 
656                     n = regnext(n);
657                 }
658                 else {
659                     int oldl = STR_LEN(scan);
660                     regnode *nnext = regnext(n);
661                     
662                     if (oldl + STR_LEN(n) > U8_MAX) 
663                         break;
664                     NEXT_OFF(scan) += NEXT_OFF(n);
665                     STR_LEN(scan) += STR_LEN(n);
666                     next = n + NODE_SZ_STR(n);
667                     /* Now we can overwrite *n : */
668                     Move(STRING(n), STRING(scan) + oldl,
669                          STR_LEN(n), char);
670 #ifdef DEBUGGING
671                     if (stringok)
672                         stop = next - 1;
673 #endif 
674                     n = nnext;
675                 }
676             }
677 #ifdef DEBUGGING
678             /* Allow dumping */
679             n = scan + NODE_SZ_STR(scan);
680             while (n <= stop) {
681                 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
682                     OP(n) = OPTIMIZED;
683                     NEXT_OFF(n) = 0;
684                 }
685                 n++;
686             }
687 #endif
688         }
689         /* Follow the next-chain of the current node and optimize
690            away all the NOTHINGs from it.  */
691         if (OP(scan) != CURLYX) {
692             int max = (reg_off_by_arg[OP(scan)]
693                        ? I32_MAX
694                        /* I32 may be smaller than U16 on CRAYs! */
695                        : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
696             int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
697             int noff;
698             regnode *n = scan;
699             
700             /* Skip NOTHING and LONGJMP. */
701             while ((n = regnext(n))
702                    && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
703                        || ((OP(n) == LONGJMP) && (noff = ARG(n))))
704                    && off + noff < max)
705                 off += noff;
706             if (reg_off_by_arg[OP(scan)])
707                 ARG(scan) = off;
708             else 
709                 NEXT_OFF(scan) = off;
710         }
711         /* The principal pseudo-switch.  Cannot be a switch, since we
712            look into several different things.  */
713         if (OP(scan) == BRANCH || OP(scan) == BRANCHJ 
714                    || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
715             next = regnext(scan);
716             code = OP(scan);
717             
718             if (OP(next) == code || code == IFTHEN || code == SUSPEND) { 
719                 I32 max1 = 0, min1 = I32_MAX, num = 0;
720                 struct regnode_charclass_class accum;
721                 
722                 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
723                     scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
724                 if (flags & SCF_DO_STCLASS)
725                     cl_init_zero(pRExC_state, &accum);
726                 while (OP(scan) == code) {
727                     I32 deltanext, minnext, f = 0, fake;
728                     struct regnode_charclass_class this_class;
729
730                     num++;
731                     data_fake.flags = 0;
732                     if (data) {             
733                         data_fake.whilem_c = data->whilem_c;
734                         data_fake.last_closep = data->last_closep;
735                     }
736                     else
737                         data_fake.last_closep = &fake;
738                     next = regnext(scan);
739                     scan = NEXTOPER(scan);
740                     if (code != BRANCH)
741                         scan = NEXTOPER(scan);
742                     if (flags & SCF_DO_STCLASS) {
743                         cl_init(pRExC_state, &this_class);
744                         data_fake.start_class = &this_class;
745                         f = SCF_DO_STCLASS_AND;
746                     }               
747                     if (flags & SCF_WHILEM_VISITED_POS)
748                         f |= SCF_WHILEM_VISITED_POS;
749                     /* we suppose the run is continuous, last=next...*/
750                     minnext = study_chunk(pRExC_state, &scan, &deltanext,
751                                           next, &data_fake, f);
752                     if (min1 > minnext) 
753                         min1 = minnext;
754                     if (max1 < minnext + deltanext)
755                         max1 = minnext + deltanext;
756                     if (deltanext == I32_MAX)
757                         is_inf = is_inf_internal = 1;
758                     scan = next;
759                     if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
760                         pars++;
761                     if (data && (data_fake.flags & SF_HAS_EVAL))
762                         data->flags |= SF_HAS_EVAL;
763                     if (data)
764                         data->whilem_c = data_fake.whilem_c;
765                     if (flags & SCF_DO_STCLASS)
766                         cl_or(pRExC_state, &accum, &this_class);
767                     if (code == SUSPEND) 
768                         break;
769                 }
770                 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
771                     min1 = 0;
772                 if (flags & SCF_DO_SUBSTR) {
773                     data->pos_min += min1;
774                     data->pos_delta += max1 - min1;
775                     if (max1 != min1 || is_inf)
776                         data->longest = &(data->longest_float);
777                 }
778                 min += min1;
779                 delta += max1 - min1;
780                 if (flags & SCF_DO_STCLASS_OR) {
781                     cl_or(pRExC_state, data->start_class, &accum);
782                     if (min1) {
783                         cl_and(data->start_class, &and_with);
784                         flags &= ~SCF_DO_STCLASS;
785                     }
786                 }
787                 else if (flags & SCF_DO_STCLASS_AND) {
788                     if (min1) {
789                         cl_and(data->start_class, &accum);
790                         flags &= ~SCF_DO_STCLASS;
791                     }
792                     else {
793                         /* Switch to OR mode: cache the old value of 
794                          * data->start_class */
795                         StructCopy(data->start_class, &and_with,
796                                    struct regnode_charclass_class);
797                         flags &= ~SCF_DO_STCLASS_AND;
798                         StructCopy(&accum, data->start_class,
799                                    struct regnode_charclass_class);
800                         flags |= SCF_DO_STCLASS_OR;
801                         data->start_class->flags |= ANYOF_EOS;
802                     }
803                 }
804             }
805             else if (code == BRANCHJ)   /* single branch is optimized. */
806                 scan = NEXTOPER(NEXTOPER(scan));
807             else                        /* single branch is optimized. */
808                 scan = NEXTOPER(scan);
809             continue;
810         }
811         else if (OP(scan) == EXACT) {
812             I32 l = STR_LEN(scan);
813             UV uc = *((U8*)STRING(scan));
814             if (UTF) {
815                 U8 *s = (U8*)STRING(scan);
816                 l = utf8_length(s, s + l);
817                 uc = utf8_to_uv_simple(s, NULL);
818             }
819             min += l;
820             if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
821                 /* The code below prefers earlier match for fixed
822                    offset, later match for variable offset.  */
823                 if (data->last_end == -1) { /* Update the start info. */
824                     data->last_start_min = data->pos_min;
825                     data->last_start_max = is_inf
826                         ? I32_MAX : data->pos_min + data->pos_delta; 
827                 }
828                 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
829                 data->last_end = data->pos_min + l;
830                 data->pos_min += l; /* As in the first entry. */
831                 data->flags &= ~SF_BEFORE_EOL;
832             }
833             if (flags & SCF_DO_STCLASS_AND) {
834                 /* Check whether it is compatible with what we know already! */
835                 int compat = 1;
836
837                 if (uc >= 0x100 ||
838                     !(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) 
839                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
840                     && (!(data->start_class->flags & ANYOF_FOLD)
841                         || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
842                     compat = 0;
843                 ANYOF_CLASS_ZERO(data->start_class);
844                 ANYOF_BITMAP_ZERO(data->start_class);
845                 if (compat)
846                     ANYOF_BITMAP_SET(data->start_class, uc);
847                 data->start_class->flags &= ~ANYOF_EOS;
848             }
849             else if (flags & SCF_DO_STCLASS_OR) {
850                 /* false positive possible if the class is case-folded */
851                 if (uc < 0x100)
852                     ANYOF_BITMAP_SET(data->start_class, uc);    
853                 data->start_class->flags &= ~ANYOF_EOS;
854                 cl_and(data->start_class, &and_with);
855             }
856             flags &= ~SCF_DO_STCLASS;
857         }
858         else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
859             I32 l = STR_LEN(scan);
860             UV uc = *((U8*)STRING(scan));
861
862             /* Search for fixed substrings supports EXACT only. */
863             if (flags & SCF_DO_SUBSTR) 
864                 scan_commit(pRExC_state, data);
865             if (UTF) {
866                 U8 *s = (U8 *)STRING(scan);
867                 l = utf8_length(s, s + l);
868                 uc = utf8_to_uv_simple(s, NULL);
869             }
870             min += l;
871             if (data && (flags & SCF_DO_SUBSTR))
872                 data->pos_min += l;
873             if (flags & SCF_DO_STCLASS_AND) {
874                 /* Check whether it is compatible with what we know already! */
875                 int compat = 1;
876
877                 if (uc >= 0x100 ||
878                     !(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE)) 
879                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
880                     && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc]))
881                     compat = 0;
882                 ANYOF_CLASS_ZERO(data->start_class);
883                 ANYOF_BITMAP_ZERO(data->start_class);
884                 if (compat) {
885                     ANYOF_BITMAP_SET(data->start_class, uc);
886                     data->start_class->flags &= ~ANYOF_EOS;
887                     data->start_class->flags |= ANYOF_FOLD;
888                     if (OP(scan) == EXACTFL)
889                         data->start_class->flags |= ANYOF_LOCALE;
890                 }
891             }
892             else if (flags & SCF_DO_STCLASS_OR) {
893                 if (data->start_class->flags & ANYOF_FOLD) {
894                     /* false positive possible if the class is case-folded.
895                        Assume that the locale settings are the same... */
896                     if (uc < 0x100)
897                         ANYOF_BITMAP_SET(data->start_class, uc);
898                     data->start_class->flags &= ~ANYOF_EOS;
899                 }
900                 cl_and(data->start_class, &and_with);
901             }
902             flags &= ~SCF_DO_STCLASS;
903         }
904         else if (strchr((char*)PL_varies,OP(scan))) {
905             I32 mincount, maxcount, minnext, deltanext, fl;
906             I32 f = flags, pos_before = 0;
907             regnode *oscan = scan;
908             struct regnode_charclass_class this_class;
909             struct regnode_charclass_class *oclass = NULL;
910
911             switch (PL_regkind[(U8)OP(scan)]) {
912             case WHILEM:                /* End of (?:...)* . */
913                 scan = NEXTOPER(scan);
914                 goto finish;
915             case PLUS:
916                 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
917                     next = NEXTOPER(scan);
918                     if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
919                         mincount = 1; 
920                         maxcount = REG_INFTY; 
921                         next = regnext(scan);
922                         scan = NEXTOPER(scan);
923                         goto do_curly;
924                     }
925                 }
926                 if (flags & SCF_DO_SUBSTR)
927                     data->pos_min++;
928                 min++;
929                 /* Fall through. */
930             case STAR:
931                 if (flags & SCF_DO_STCLASS) {
932                     mincount = 0;
933                     maxcount = REG_INFTY; 
934                     next = regnext(scan);
935                     scan = NEXTOPER(scan);
936                     goto do_curly;
937                 }
938                 is_inf = is_inf_internal = 1; 
939                 scan = regnext(scan);
940                 if (flags & SCF_DO_SUBSTR) {
941                     scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
942                     data->longest = &(data->longest_float);
943                 }
944                 goto optimize_curly_tail;
945             case CURLY:
946                 mincount = ARG1(scan); 
947                 maxcount = ARG2(scan);
948                 next = regnext(scan);
949                 if (OP(scan) == CURLYX) {
950                     I32 lp = (data ? *(data->last_closep) : 0);
951
952                     scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
953                 }
954                 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
955               do_curly:
956                 if (flags & SCF_DO_SUBSTR) {
957                     if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
958                     pos_before = data->pos_min;
959                 }
960                 if (data) {
961                     fl = data->flags;
962                     data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
963                     if (is_inf)
964                         data->flags |= SF_IS_INF;
965                 }
966                 if (flags & SCF_DO_STCLASS) {
967                     cl_init(pRExC_state, &this_class);
968                     oclass = data->start_class;
969                     data->start_class = &this_class;
970                     f |= SCF_DO_STCLASS_AND;
971                     f &= ~SCF_DO_STCLASS_OR;
972                 }
973                 /* These are the cases when once a subexpression
974                    fails at a particular position, it cannot succeed
975                    even after backtracking at the enclosing scope.
976                    
977                    XXXX what if minimal match and we are at the
978                         initial run of {n,m}? */
979                 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
980                     f &= ~SCF_WHILEM_VISITED_POS;
981
982                 /* This will finish on WHILEM, setting scan, or on NULL: */
983                 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data, 
984                                       mincount == 0 
985                                         ? (f & ~SCF_DO_SUBSTR) : f);
986
987                 if (flags & SCF_DO_STCLASS)
988                     data->start_class = oclass;
989                 if (mincount == 0 || minnext == 0) {
990                     if (flags & SCF_DO_STCLASS_OR) {
991                         cl_or(pRExC_state, data->start_class, &this_class);
992                     }
993                     else if (flags & SCF_DO_STCLASS_AND) {
994                         /* Switch to OR mode: cache the old value of 
995                          * data->start_class */
996                         StructCopy(data->start_class, &and_with,
997                                    struct regnode_charclass_class);
998                         flags &= ~SCF_DO_STCLASS_AND;
999                         StructCopy(&this_class, data->start_class,
1000                                    struct regnode_charclass_class);
1001                         flags |= SCF_DO_STCLASS_OR;
1002                         data->start_class->flags |= ANYOF_EOS;
1003                     }
1004                 } else {                /* Non-zero len */
1005                     if (flags & SCF_DO_STCLASS_OR) {
1006                         cl_or(pRExC_state, data->start_class, &this_class);
1007                         cl_and(data->start_class, &and_with);
1008                     }
1009                     else if (flags & SCF_DO_STCLASS_AND)
1010                         cl_and(data->start_class, &this_class);
1011                     flags &= ~SCF_DO_STCLASS;
1012                 }
1013                 if (!scan)              /* It was not CURLYX, but CURLY. */
1014                     scan = next;
1015                 if (ckWARN(WARN_REGEXP) && (minnext + deltanext == 0) 
1016                     && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1017                     && maxcount <= REG_INFTY/3) /* Complement check for big count */
1018                 {
1019                     vWARN(RExC_parse,
1020                           "Quantifier unexpected on zero-length expression");
1021                 }
1022
1023                 min += minnext * mincount;
1024                 is_inf_internal |= ((maxcount == REG_INFTY 
1025                                      && (minnext + deltanext) > 0)
1026                                     || deltanext == I32_MAX);
1027                 is_inf |= is_inf_internal;
1028                 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1029
1030                 /* Try powerful optimization CURLYX => CURLYN. */
1031                 if (  OP(oscan) == CURLYX && data 
1032                       && data->flags & SF_IN_PAR
1033                       && !(data->flags & SF_HAS_EVAL)
1034                       && !deltanext && minnext == 1 ) {
1035                     /* Try to optimize to CURLYN.  */
1036                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1037                     regnode *nxt1 = nxt, *nxt2;
1038
1039                     /* Skip open. */
1040                     nxt = regnext(nxt);
1041                     if (!strchr((char*)PL_simple,OP(nxt))
1042                         && !(PL_regkind[(U8)OP(nxt)] == EXACT
1043                              && STR_LEN(nxt) == 1)) 
1044                         goto nogo;
1045                     nxt2 = nxt;
1046                     nxt = regnext(nxt);
1047                     if (OP(nxt) != CLOSE) 
1048                         goto nogo;
1049                     /* Now we know that nxt2 is the only contents: */
1050                     oscan->flags = ARG(nxt);
1051                     OP(oscan) = CURLYN;
1052                     OP(nxt1) = NOTHING; /* was OPEN. */
1053 #ifdef DEBUGGING
1054                     OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1055                     NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1056                     NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1057                     OP(nxt) = OPTIMIZED;        /* was CLOSE. */
1058                     OP(nxt + 1) = OPTIMIZED; /* was count. */
1059                     NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1060 #endif 
1061                 }
1062               nogo:
1063
1064                 /* Try optimization CURLYX => CURLYM. */
1065                 if (  OP(oscan) == CURLYX && data 
1066                       && !(data->flags & SF_HAS_PAR)
1067                       && !(data->flags & SF_HAS_EVAL)
1068                       && !deltanext  ) {
1069                     /* XXXX How to optimize if data == 0? */
1070                     /* Optimize to a simpler form.  */
1071                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1072                     regnode *nxt2;
1073
1074                     OP(oscan) = CURLYM;
1075                     while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1076                             && (OP(nxt2) != WHILEM)) 
1077                         nxt = nxt2;
1078                     OP(nxt2)  = SUCCEED; /* Whas WHILEM */
1079                     /* Need to optimize away parenths. */
1080                     if (data->flags & SF_IN_PAR) {
1081                         /* Set the parenth number.  */
1082                         regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1083
1084                         if (OP(nxt) != CLOSE) 
1085                             FAIL("Panic opt close");
1086                         oscan->flags = ARG(nxt);
1087                         OP(nxt1) = OPTIMIZED;   /* was OPEN. */
1088                         OP(nxt) = OPTIMIZED;    /* was CLOSE. */
1089 #ifdef DEBUGGING
1090                         OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1091                         OP(nxt + 1) = OPTIMIZED; /* was count. */
1092                         NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1093                         NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1094 #endif 
1095 #if 0
1096                         while ( nxt1 && (OP(nxt1) != WHILEM)) {
1097                             regnode *nnxt = regnext(nxt1);
1098                             
1099                             if (nnxt == nxt) {
1100                                 if (reg_off_by_arg[OP(nxt1)])
1101                                     ARG_SET(nxt1, nxt2 - nxt1);
1102                                 else if (nxt2 - nxt1 < U16_MAX)
1103                                     NEXT_OFF(nxt1) = nxt2 - nxt1;
1104                                 else
1105                                     OP(nxt) = NOTHING;  /* Cannot beautify */
1106                             }
1107                             nxt1 = nnxt;
1108                         }
1109 #endif
1110                         /* Optimize again: */
1111                         study_chunk(pRExC_state, &nxt1, &deltanext, nxt, 
1112                                     NULL, 0);
1113                     }
1114                     else
1115                         oscan->flags = 0;
1116                 }
1117                 else if ((OP(oscan) == CURLYX)
1118                          && (flags & SCF_WHILEM_VISITED_POS)
1119                          /* See the comment on a similar expression above.
1120                             However, this time it not a subexpression
1121                             we care about, but the expression itself. */
1122                          && (maxcount == REG_INFTY)
1123                          && data && ++data->whilem_c < 16) {
1124                     /* This stays as CURLYX, we can put the count/of pair. */
1125                     /* Find WHILEM (as in regexec.c) */
1126                     regnode *nxt = oscan + NEXT_OFF(oscan);
1127
1128                     if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1129                         nxt += ARG(nxt);
1130                     PREVOPER(nxt)->flags = data->whilem_c
1131                         | (RExC_whilem_seen << 4); /* On WHILEM */
1132                 }
1133                 if (data && fl & (SF_HAS_PAR|SF_IN_PAR)) 
1134                     pars++;
1135                 if (flags & SCF_DO_SUBSTR) {
1136                     SV *last_str = Nullsv;
1137                     int counted = mincount != 0;
1138
1139                     if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1140                         I32 b = pos_before >= data->last_start_min 
1141                             ? pos_before : data->last_start_min;
1142                         STRLEN l;
1143                         char *s = SvPV(data->last_found, l);
1144                         I32 old = b - data->last_start_min;
1145
1146                         if (UTF)
1147                             old = utf8_hop((U8*)s, old) - (U8*)s;
1148                         
1149                         l -= old;
1150                         /* Get the added string: */
1151                         last_str = newSVpvn(s  + old, l);
1152                         if (deltanext == 0 && pos_before == b) {
1153                             /* What was added is a constant string */
1154                             if (mincount > 1) {
1155                                 SvGROW(last_str, (mincount * l) + 1);
1156                                 repeatcpy(SvPVX(last_str) + l, 
1157                                           SvPVX(last_str), l, mincount - 1);
1158                                 SvCUR(last_str) *= mincount;
1159                                 /* Add additional parts. */
1160                                 SvCUR_set(data->last_found, 
1161                                           SvCUR(data->last_found) - l);
1162                                 sv_catsv(data->last_found, last_str);
1163                                 data->last_end += l * (mincount - 1);
1164                             }
1165                         } else {
1166                             /* start offset must point into the last copy */
1167                             data->last_start_min += minnext * (mincount - 1);
1168                             data->last_start_max += is_inf ? 0 : (maxcount - 1)
1169                                 * (minnext + data->pos_delta);
1170                         }
1171                     }
1172                     /* It is counted once already... */
1173                     data->pos_min += minnext * (mincount - counted);
1174                     data->pos_delta += - counted * deltanext +
1175                         (minnext + deltanext) * maxcount - minnext * mincount;
1176                     if (mincount != maxcount) {
1177                          /* Cannot extend fixed substrings found inside
1178                             the group.  */
1179                         scan_commit(pRExC_state,data);
1180                         if (mincount && last_str) {
1181                             sv_setsv(data->last_found, last_str);
1182                             data->last_end = data->pos_min;
1183                             data->last_start_min = 
1184                                 data->pos_min - CHR_SVLEN(last_str);
1185                             data->last_start_max = is_inf 
1186                                 ? I32_MAX 
1187                                 : data->pos_min + data->pos_delta
1188                                 - CHR_SVLEN(last_str);
1189                         }
1190                         data->longest = &(data->longest_float);
1191                     }
1192                     SvREFCNT_dec(last_str);
1193                 }
1194                 if (data && (fl & SF_HAS_EVAL))
1195                     data->flags |= SF_HAS_EVAL;
1196               optimize_curly_tail:
1197                 if (OP(oscan) != CURLYX) {
1198                     while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1199                            && NEXT_OFF(next))
1200                         NEXT_OFF(oscan) += NEXT_OFF(next);
1201                 }
1202                 continue;
1203             default:                    /* REF and CLUMP only? */
1204                 if (flags & SCF_DO_SUBSTR) {
1205                     scan_commit(pRExC_state,data);      /* Cannot expect anything... */
1206                     data->longest = &(data->longest_float);
1207                 }
1208                 is_inf = is_inf_internal = 1;
1209                 if (flags & SCF_DO_STCLASS_OR)
1210                     cl_anything(pRExC_state, data->start_class);
1211                 flags &= ~SCF_DO_STCLASS;
1212                 break;
1213             }
1214         }
1215         else if (strchr((char*)PL_simple,OP(scan))) {
1216             int value;
1217
1218             if (flags & SCF_DO_SUBSTR) {
1219                 scan_commit(pRExC_state,data);
1220                 data->pos_min++;
1221             }
1222             min++;
1223             if (flags & SCF_DO_STCLASS) {
1224                 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1225
1226                 /* Some of the logic below assumes that switching
1227                    locale on will only add false positives. */
1228                 switch (PL_regkind[(U8)OP(scan)]) {
1229                 case SANY:
1230                 default:
1231                   do_default:
1232                     /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1233                     if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1234                         cl_anything(pRExC_state, data->start_class);
1235                     break;
1236                 case REG_ANY:
1237                     if (OP(scan) == SANY)
1238                         goto do_default;
1239                     if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1240                         value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1241                                  || (data->start_class->flags & ANYOF_CLASS));
1242                         cl_anything(pRExC_state, data->start_class);
1243                     }
1244                     if (flags & SCF_DO_STCLASS_AND || !value)
1245                         ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1246                     break;
1247                 case ANYOF:
1248                     if (flags & SCF_DO_STCLASS_AND)
1249                         cl_and(data->start_class,
1250                                (struct regnode_charclass_class*)scan);
1251                     else
1252                         cl_or(pRExC_state, data->start_class,
1253                               (struct regnode_charclass_class*)scan);
1254                     break;
1255                 case ALNUM:
1256                     if (flags & SCF_DO_STCLASS_AND) {
1257                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1258                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1259                             for (value = 0; value < 256; value++)
1260                                 if (!isALNUM(value))
1261                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1262                         }
1263                     }
1264                     else {
1265                         if (data->start_class->flags & ANYOF_LOCALE)
1266                             ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1267                         else {
1268                             for (value = 0; value < 256; value++)
1269                                 if (isALNUM(value))
1270                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1271                         }
1272                     }
1273                     break;
1274                 case ALNUML:
1275                     if (flags & SCF_DO_STCLASS_AND) {
1276                         if (data->start_class->flags & ANYOF_LOCALE)
1277                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1278                     }
1279                     else {
1280                         ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1281                         data->start_class->flags |= ANYOF_LOCALE;
1282                     }
1283                     break;
1284                 case NALNUM:
1285                     if (flags & SCF_DO_STCLASS_AND) {
1286                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1287                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1288                             for (value = 0; value < 256; value++)
1289                                 if (isALNUM(value))
1290                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1291                         }
1292                     }
1293                     else {
1294                         if (data->start_class->flags & ANYOF_LOCALE)
1295                             ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1296                         else {
1297                             for (value = 0; value < 256; value++)
1298                                 if (!isALNUM(value))
1299                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1300                         }
1301                     }
1302                     break;
1303                 case NALNUML:
1304                     if (flags & SCF_DO_STCLASS_AND) {
1305                         if (data->start_class->flags & ANYOF_LOCALE)
1306                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1307                     }
1308                     else {
1309                         data->start_class->flags |= ANYOF_LOCALE;
1310                         ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1311                     }
1312                     break;
1313                 case SPACE:
1314                     if (flags & SCF_DO_STCLASS_AND) {
1315                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1316                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1317                             for (value = 0; value < 256; value++)
1318                                 if (!isSPACE(value))
1319                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1320                         }
1321                     }
1322                     else {
1323                         if (data->start_class->flags & ANYOF_LOCALE)
1324                             ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1325                         else {
1326                             for (value = 0; value < 256; value++)
1327                                 if (isSPACE(value))
1328                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1329                         }
1330                     }
1331                     break;
1332                 case SPACEL:
1333                     if (flags & SCF_DO_STCLASS_AND) {
1334                         if (data->start_class->flags & ANYOF_LOCALE)
1335                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1336                     }
1337                     else {
1338                         data->start_class->flags |= ANYOF_LOCALE;
1339                         ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1340                     }
1341                     break;
1342                 case NSPACE:
1343                     if (flags & SCF_DO_STCLASS_AND) {
1344                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1345                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1346                             for (value = 0; value < 256; value++)
1347                                 if (isSPACE(value))
1348                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1349                         }
1350                     }
1351                     else {
1352                         if (data->start_class->flags & ANYOF_LOCALE)
1353                             ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1354                         else {
1355                             for (value = 0; value < 256; value++)
1356                                 if (!isSPACE(value))
1357                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1358                         }
1359                     }
1360                     break;
1361                 case NSPACEL:
1362                     if (flags & SCF_DO_STCLASS_AND) {
1363                         if (data->start_class->flags & ANYOF_LOCALE) {
1364                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1365                             for (value = 0; value < 256; value++)
1366                                 if (!isSPACE(value))
1367                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1368                         }
1369                     }
1370                     else {
1371                         data->start_class->flags |= ANYOF_LOCALE;
1372                         ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1373                     }
1374                     break;
1375                 case DIGIT:
1376                     if (flags & SCF_DO_STCLASS_AND) {
1377                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1378                         for (value = 0; value < 256; value++)
1379                             if (!isDIGIT(value))
1380                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1381                     }
1382                     else {
1383                         if (data->start_class->flags & ANYOF_LOCALE)
1384                             ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1385                         else {
1386                             for (value = 0; value < 256; value++)
1387                                 if (isDIGIT(value))
1388                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1389                         }
1390                     }
1391                     break;
1392                 case NDIGIT:
1393                     if (flags & SCF_DO_STCLASS_AND) {
1394                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1395                         for (value = 0; value < 256; value++)
1396                             if (isDIGIT(value))
1397                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1398                     }
1399                     else {
1400                         if (data->start_class->flags & ANYOF_LOCALE)
1401                             ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1402                         else {
1403                             for (value = 0; value < 256; value++)
1404                                 if (!isDIGIT(value))
1405                                     ANYOF_BITMAP_SET(data->start_class, value);                     
1406                         }
1407                     }
1408                     break;
1409                 }
1410                 if (flags & SCF_DO_STCLASS_OR)
1411                     cl_and(data->start_class, &and_with);
1412                 flags &= ~SCF_DO_STCLASS;
1413             }
1414         }
1415         else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1416             data->flags |= (OP(scan) == MEOL
1417                             ? SF_BEFORE_MEOL
1418                             : SF_BEFORE_SEOL);
1419         }
1420         else if (  PL_regkind[(U8)OP(scan)] == BRANCHJ
1421                  /* Lookbehind, or need to calculate parens/evals/stclass: */
1422                    && (scan->flags || data || (flags & SCF_DO_STCLASS))
1423                    && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1424             /* Lookahead/lookbehind */
1425             I32 deltanext, minnext, fake = 0;
1426             regnode *nscan;
1427             struct regnode_charclass_class intrnl;
1428             int f = 0;
1429
1430             data_fake.flags = 0;
1431             if (data) {             
1432                 data_fake.whilem_c = data->whilem_c;
1433                 data_fake.last_closep = data->last_closep;
1434             }
1435             else
1436                 data_fake.last_closep = &fake;
1437             if ( flags & SCF_DO_STCLASS && !scan->flags
1438                  && OP(scan) == IFMATCH ) { /* Lookahead */
1439                 cl_init(pRExC_state, &intrnl);
1440                 data_fake.start_class = &intrnl;
1441                 f |= SCF_DO_STCLASS_AND;
1442             }
1443             if (flags & SCF_WHILEM_VISITED_POS)
1444                 f |= SCF_WHILEM_VISITED_POS;
1445             next = regnext(scan);
1446             nscan = NEXTOPER(NEXTOPER(scan));
1447             minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1448             if (scan->flags) {
1449                 if (deltanext) {
1450                     vFAIL("Variable length lookbehind not implemented");
1451                 }
1452                 else if (minnext > U8_MAX) {
1453                     vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1454                 }
1455                 scan->flags = minnext;
1456             }
1457             if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1458                 pars++;
1459             if (data && (data_fake.flags & SF_HAS_EVAL))
1460                 data->flags |= SF_HAS_EVAL;
1461             if (data)
1462                 data->whilem_c = data_fake.whilem_c;
1463             if (f & SCF_DO_STCLASS_AND) {
1464                 int was = (data->start_class->flags & ANYOF_EOS);
1465
1466                 cl_and(data->start_class, &intrnl);
1467                 if (was)
1468                     data->start_class->flags |= ANYOF_EOS;
1469             }
1470         }
1471         else if (OP(scan) == OPEN) {
1472             pars++;
1473         }
1474         else if (OP(scan) == CLOSE) {
1475             if (ARG(scan) == is_par) {
1476                 next = regnext(scan);
1477
1478                 if ( next && (OP(next) != WHILEM) && next < last)
1479                     is_par = 0;         /* Disable optimization */
1480             }
1481             if (data)
1482                 *(data->last_closep) = ARG(scan);
1483         }
1484         else if (OP(scan) == EVAL) {
1485                 if (data)
1486                     data->flags |= SF_HAS_EVAL;
1487         }
1488         else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1489                 if (flags & SCF_DO_SUBSTR) {
1490                     scan_commit(pRExC_state,data);
1491                     data->longest = &(data->longest_float);
1492                 }
1493                 is_inf = is_inf_internal = 1;
1494                 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1495                     cl_anything(pRExC_state, data->start_class);
1496                 flags &= ~SCF_DO_STCLASS;
1497         }
1498         /* Else: zero-length, ignore. */
1499         scan = regnext(scan);
1500     }
1501
1502   finish:
1503     *scanp = scan;
1504     *deltap = is_inf_internal ? I32_MAX : delta;
1505     if (flags & SCF_DO_SUBSTR && is_inf) 
1506         data->pos_delta = I32_MAX - data->pos_min;
1507     if (is_par > U8_MAX)
1508         is_par = 0;
1509     if (is_par && pars==1 && data) {
1510         data->flags |= SF_IN_PAR;
1511         data->flags &= ~SF_HAS_PAR;
1512     }
1513     else if (pars && data) {
1514         data->flags |= SF_HAS_PAR;
1515         data->flags &= ~SF_IN_PAR;
1516     }
1517     if (flags & SCF_DO_STCLASS_OR)
1518         cl_and(data->start_class, &and_with);
1519     return min;
1520 }
1521
1522 STATIC I32
1523 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1524 {
1525     if (RExC_rx->data) {
1526         Renewc(RExC_rx->data, 
1527                sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1), 
1528                char, struct reg_data);
1529         Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1530         RExC_rx->data->count += n;
1531     }
1532     else {
1533         Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1534              char, struct reg_data);
1535         New(1208, RExC_rx->data->what, n, U8);
1536         RExC_rx->data->count = n;
1537     }
1538     Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1539     return RExC_rx->data->count - n;
1540 }
1541
1542 void
1543 Perl_reginitcolors(pTHX)
1544 {
1545     int i = 0;
1546     char *s = PerlEnv_getenv("PERL_RE_COLORS");
1547             
1548     if (s) {
1549         PL_colors[0] = s = savepv(s);
1550         while (++i < 6) {
1551             s = strchr(s, '\t');
1552             if (s) {
1553                 *s = '\0';
1554                 PL_colors[i] = ++s;
1555             }
1556             else
1557                 PL_colors[i] = s = "";
1558         }
1559     } else {
1560         while (i < 6) 
1561             PL_colors[i++] = "";
1562     }
1563     PL_colorset = 1;
1564 }
1565
1566
1567 /*
1568  - pregcomp - compile a regular expression into internal code
1569  *
1570  * We can't allocate space until we know how big the compiled form will be,
1571  * but we can't compile it (and thus know how big it is) until we've got a
1572  * place to put the code.  So we cheat:  we compile it twice, once with code
1573  * generation turned off and size counting turned on, and once "for real".
1574  * This also means that we don't allocate space until we are sure that the
1575  * thing really will compile successfully, and we never have to move the
1576  * code and thus invalidate pointers into it.  (Note that it has to be in
1577  * one piece because free() must be able to free it all.) [NB: not true in perl]
1578  *
1579  * Beware that the optimization-preparation code in here knows about some
1580  * of the structure of the compiled regexp.  [I'll say.]
1581  */
1582 regexp *
1583 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1584 {
1585     register regexp *r;
1586     regnode *scan;
1587     regnode *first;
1588     I32 flags;
1589     I32 minlen = 0;
1590     I32 sawplus = 0;
1591     I32 sawopen = 0;
1592     scan_data_t data;
1593     RExC_state_t RExC_state;
1594     RExC_state_t *pRExC_state = &RExC_state;
1595
1596     if (exp == NULL)
1597         FAIL("NULL regexp argument");
1598
1599     /* XXXX This looks very suspicious... */
1600     if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1601         RExC_utf8 = 1;
1602     else
1603         RExC_utf8 = 0;
1604
1605     RExC_precomp = savepvn(exp, xend - exp);
1606     DEBUG_r(if (!PL_colorset) reginitcolors());
1607     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1608                       PL_colors[4],PL_colors[5],PL_colors[0],
1609                       (int)(xend - exp), RExC_precomp, PL_colors[1]));
1610     RExC_flags16 = pm->op_pmflags;
1611     RExC_sawback = 0;
1612
1613     RExC_seen = 0;
1614     RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1615     RExC_seen_evals = 0;
1616     RExC_extralen = 0;
1617
1618     /* First pass: determine size, legality. */
1619     RExC_parse = exp;
1620     RExC_end = xend;
1621     RExC_naughty = 0;
1622     RExC_npar = 1;
1623     RExC_size = 0L;
1624     RExC_emit = &PL_regdummy;
1625     RExC_whilem_seen = 0;
1626 #if 0 /* REGC() is (currently) a NOP at the first pass.
1627        * Clever compilers notice this and complain. --jhi */
1628     REGC((U8)REG_MAGIC, (char*)RExC_emit);
1629 #endif
1630     if (reg(pRExC_state, 0, &flags) == NULL) {
1631         Safefree(RExC_precomp);
1632         RExC_precomp = Nullch;
1633         return(NULL);
1634     }
1635     DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1636
1637     /* Small enough for pointer-storage convention?
1638        If extralen==0, this means that we will not need long jumps. */
1639     if (RExC_size >= 0x10000L && RExC_extralen)
1640         RExC_size += RExC_extralen;
1641     else
1642         RExC_extralen = 0;
1643     if (RExC_whilem_seen > 15)
1644         RExC_whilem_seen = 15;
1645
1646     /* Allocate space and initialize. */
1647     Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1648          char, regexp);
1649     if (r == NULL)
1650         FAIL("Regexp out of space");
1651
1652 #ifdef DEBUGGING
1653     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1654     Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1655 #endif
1656     r->refcnt = 1;
1657     r->prelen = xend - exp;
1658     r->precomp = RExC_precomp;
1659     r->subbeg = NULL;
1660     r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1661     r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1662
1663     r->substrs = 0;                     /* Useful during FAIL. */
1664     r->startp = 0;                      /* Useful during FAIL. */
1665     r->endp = 0;                        /* Useful during FAIL. */
1666
1667     RExC_rx = r;
1668
1669     /* Second pass: emit code. */
1670     RExC_parse = exp;
1671     RExC_end = xend;
1672     RExC_naughty = 0;
1673     RExC_npar = 1;
1674     RExC_emit = r->program;
1675     /* Store the count of eval-groups for security checks: */
1676     RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1677     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1678     r->data = 0;
1679     if (reg(pRExC_state, 0, &flags) == NULL)
1680         return(NULL);
1681
1682     /* Dig out information for optimizations. */
1683     r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1684     pm->op_pmflags = RExC_flags16;
1685     if (UTF)
1686         r->reganch |= ROPT_UTF8;
1687     r->regstclass = NULL;
1688     if (RExC_naughty >= 10)     /* Probably an expensive pattern. */
1689         r->reganch |= ROPT_NAUGHTY;
1690     scan = r->program + 1;              /* First BRANCH. */
1691
1692     /* XXXX To minimize changes to RE engine we always allocate
1693        3-units-long substrs field. */
1694     Newz(1004, r->substrs, 1, struct reg_substr_data);
1695
1696     StructCopy(&zero_scan_data, &data, scan_data_t);
1697     /* XXXX Should not we check for something else?  Usually it is OPEN1... */
1698     if (OP(scan) != BRANCH) {   /* Only one top-level choice. */
1699         I32 fake;
1700         STRLEN longest_float_length, longest_fixed_length;
1701         struct regnode_charclass_class ch_class;
1702         int stclass_flag;
1703         I32 last_close = 0;
1704
1705         first = scan;
1706         /* Skip introductions and multiplicators >= 1. */
1707         while ((OP(first) == OPEN && (sawopen = 1)) ||
1708                /* An OR of *one* alternative - should not happen now. */
1709             (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1710             (OP(first) == PLUS) ||
1711             (OP(first) == MINMOD) ||
1712                /* An {n,m} with n>0 */
1713             (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1714                 if (OP(first) == PLUS)
1715                     sawplus = 1;
1716                 else
1717                     first += regarglen[(U8)OP(first)];
1718                 first = NEXTOPER(first);
1719         }
1720
1721         /* Starting-point info. */
1722       again:
1723         if (PL_regkind[(U8)OP(first)] == EXACT) {
1724             if (OP(first) == EXACT)
1725                 ;       /* Empty, get anchored substr later. */
1726             else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1727                 r->regstclass = first;
1728         }
1729         else if (strchr((char*)PL_simple,OP(first)))
1730             r->regstclass = first;
1731         else if (PL_regkind[(U8)OP(first)] == BOUND ||
1732                  PL_regkind[(U8)OP(first)] == NBOUND)
1733             r->regstclass = first;
1734         else if (PL_regkind[(U8)OP(first)] == BOL) {
1735             r->reganch |= (OP(first) == MBOL
1736                            ? ROPT_ANCH_MBOL
1737                            : (OP(first) == SBOL
1738                               ? ROPT_ANCH_SBOL
1739                               : ROPT_ANCH_BOL));
1740             first = NEXTOPER(first);
1741             goto again;
1742         }
1743         else if (OP(first) == GPOS) {
1744             r->reganch |= ROPT_ANCH_GPOS;
1745             first = NEXTOPER(first);
1746             goto again;
1747         }
1748         else if ((OP(first) == STAR &&
1749             PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1750             !(r->reganch & ROPT_ANCH) )
1751         {
1752             /* turn .* into ^.* with an implied $*=1 */
1753             int type = OP(NEXTOPER(first));
1754
1755             if (type == REG_ANY)
1756                 type = ROPT_ANCH_MBOL;
1757             else
1758                 type = ROPT_ANCH_SBOL;
1759
1760             r->reganch |= type | ROPT_IMPLICIT;
1761             first = NEXTOPER(first);
1762             goto again;
1763         }
1764         if (sawplus && (!sawopen || !RExC_sawback) 
1765             && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1766             /* x+ must match at the 1st pos of run of x's */
1767             r->reganch |= ROPT_SKIP;
1768
1769         /* Scan is after the zeroth branch, first is atomic matcher. */
1770         DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n", 
1771                               (IV)(first - scan + 1)));
1772         /*
1773         * If there's something expensive in the r.e., find the
1774         * longest literal string that must appear and make it the
1775         * regmust.  Resolve ties in favor of later strings, since
1776         * the regstart check works with the beginning of the r.e.
1777         * and avoiding duplication strengthens checking.  Not a
1778         * strong reason, but sufficient in the absence of others.
1779         * [Now we resolve ties in favor of the earlier string if
1780         * it happens that c_offset_min has been invalidated, since the
1781         * earlier string may buy us something the later one won't.]
1782         */
1783         minlen = 0;
1784
1785         data.longest_fixed = newSVpvn("",0);
1786         data.longest_float = newSVpvn("",0);
1787         data.last_found = newSVpvn("",0);
1788         data.longest = &(data.longest_fixed);
1789         first = scan;
1790         if (!r->regstclass) {
1791             cl_init(pRExC_state, &ch_class);
1792             data.start_class = &ch_class;
1793             stclass_flag = SCF_DO_STCLASS_AND;
1794         } else                          /* XXXX Check for BOUND? */
1795             stclass_flag = 0;
1796         data.last_closep = &last_close;
1797
1798         minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1799                              &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1800         if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1801              && data.last_start_min == 0 && data.last_end > 0 
1802              && !RExC_seen_zerolen
1803              && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1804             r->reganch |= ROPT_CHECK_ALL;
1805         scan_commit(pRExC_state, &data);
1806         SvREFCNT_dec(data.last_found);
1807
1808         longest_float_length = CHR_SVLEN(data.longest_float);
1809         if (longest_float_length
1810             || (data.flags & SF_FL_BEFORE_EOL
1811                 && (!(data.flags & SF_FL_BEFORE_MEOL)
1812                     || (RExC_flags16 & PMf_MULTILINE)))) {
1813             int t;
1814
1815             if (SvCUR(data.longest_fixed)                       /* ok to leave SvCUR */
1816                 && data.offset_fixed == data.offset_float_min
1817                 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1818                     goto remove_float;          /* As in (a)+. */
1819
1820             r->float_substr = data.longest_float;
1821             r->float_min_offset = data.offset_float_min;
1822             r->float_max_offset = data.offset_float_max;
1823             t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1824                        && (!(data.flags & SF_FL_BEFORE_MEOL)
1825                            || (RExC_flags16 & PMf_MULTILINE)));
1826             fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1827         }
1828         else {
1829           remove_float:
1830             r->float_substr = Nullsv;
1831             SvREFCNT_dec(data.longest_float);
1832             longest_float_length = 0;
1833         }
1834
1835         longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1836         if (longest_fixed_length
1837             || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1838                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1839                     || (RExC_flags16 & PMf_MULTILINE)))) {
1840             int t;
1841
1842             r->anchored_substr = data.longest_fixed;
1843             r->anchored_offset = data.offset_fixed;
1844             t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1845                  && (!(data.flags & SF_FIX_BEFORE_MEOL)
1846                      || (RExC_flags16 & PMf_MULTILINE)));
1847             fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1848         }
1849         else {
1850             r->anchored_substr = Nullsv;
1851             SvREFCNT_dec(data.longest_fixed);
1852             longest_fixed_length = 0;
1853         }
1854         if (r->regstclass 
1855             && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1856             r->regstclass = NULL;
1857         if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1858             && !(data.start_class->flags & ANYOF_EOS)
1859             && !cl_is_anything(data.start_class)) {
1860             SV *sv;
1861             I32 n = add_data(pRExC_state, 1, "f");
1862
1863             New(1006, RExC_rx->data->data[n], 1, 
1864                 struct regnode_charclass_class);
1865             StructCopy(data.start_class,
1866                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1867                        struct regnode_charclass_class);
1868             r->regstclass = (regnode*)RExC_rx->data->data[n];
1869             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1870             PL_regdata = r->data; /* for regprop() */
1871             DEBUG_r((sv = sv_newmortal(),
1872                      regprop(sv, (regnode*)data.start_class),
1873                      PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n",
1874                                    SvPVX(sv))));
1875         }
1876
1877         /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1878         if (longest_fixed_length > longest_float_length) {
1879             r->check_substr = r->anchored_substr;
1880             r->check_offset_min = r->check_offset_max = r->anchored_offset;
1881             if (r->reganch & ROPT_ANCH_SINGLE)
1882                 r->reganch |= ROPT_NOSCAN;
1883         }
1884         else {
1885             r->check_substr = r->float_substr;
1886             r->check_offset_min = data.offset_float_min;
1887             r->check_offset_max = data.offset_float_max;
1888         }
1889         /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1890            This should be changed ASAP!  */
1891         if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1892             r->reganch |= RE_USE_INTUIT;
1893             if (SvTAIL(r->check_substr))
1894                 r->reganch |= RE_INTUIT_TAIL;
1895         }
1896     }
1897     else {
1898         /* Several toplevels. Best we can is to set minlen. */
1899         I32 fake;
1900         struct regnode_charclass_class ch_class;
1901         I32 last_close = 0;
1902         
1903         DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1904         scan = r->program + 1;
1905         cl_init(pRExC_state, &ch_class);
1906         data.start_class = &ch_class;
1907         data.last_closep = &last_close;
1908         minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1909         r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1910         if (!(data.start_class->flags & ANYOF_EOS)
1911             && !cl_is_anything(data.start_class)) {
1912             SV *sv;
1913             I32 n = add_data(pRExC_state, 1, "f");
1914
1915             New(1006, RExC_rx->data->data[n], 1, 
1916                 struct regnode_charclass_class);
1917             StructCopy(data.start_class,
1918                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1919                        struct regnode_charclass_class);
1920             r->regstclass = (regnode*)RExC_rx->data->data[n];
1921             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1922             DEBUG_r((sv = sv_newmortal(),
1923                      regprop(sv, (regnode*)data.start_class),
1924                      PerlIO_printf(Perl_debug_log, "synthetic stclass `%s'.\n",
1925                                    SvPVX(sv))));
1926         }
1927     }
1928
1929     r->minlen = minlen;
1930     if (RExC_seen & REG_SEEN_GPOS) 
1931         r->reganch |= ROPT_GPOS_SEEN;
1932     if (RExC_seen & REG_SEEN_LOOKBEHIND)
1933         r->reganch |= ROPT_LOOKBEHIND_SEEN;
1934     if (RExC_seen & REG_SEEN_EVAL)
1935         r->reganch |= ROPT_EVAL_SEEN;
1936     Newz(1002, r->startp, RExC_npar, I32);
1937     Newz(1002, r->endp, RExC_npar, I32);
1938     PL_regdata = r->data; /* for regprop() */
1939     DEBUG_r(regdump(r));
1940     return(r);
1941 }
1942
1943 /*
1944  - reg - regular expression, i.e. main body or parenthesized thing
1945  *
1946  * Caller must absorb opening parenthesis.
1947  *
1948  * Combining parenthesis handling with the base level of regular expression
1949  * is a trifle forced, but the need to tie the tails of the branches to what
1950  * follows makes it hard to avoid.
1951  */
1952 STATIC regnode *
1953 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
1954     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
1955 {
1956     register regnode *ret;              /* Will be the head of the group. */
1957     register regnode *br;
1958     register regnode *lastbr;
1959     register regnode *ender = 0;
1960     register I32 parno = 0;
1961     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
1962     char *oregcomp_parse = RExC_parse;
1963     char c;
1964
1965     *flagp = 0;                         /* Tentatively. */
1966
1967     /* Make an OPEN node, if parenthesized. */
1968     if (paren) {
1969         if (*RExC_parse == '?') {
1970             U16 posflags = 0, negflags = 0;
1971             U16 *flagsp = &posflags;
1972             int logical = 0;
1973             char *seqstart = RExC_parse;
1974
1975             RExC_parse++;
1976             paren = *RExC_parse++;
1977             ret = NULL;                 /* For look-ahead/behind. */
1978             switch (paren) {
1979             case '<':
1980                 RExC_seen |= REG_SEEN_LOOKBEHIND;
1981                 if (*RExC_parse == '!') 
1982                     paren = ',';
1983                 if (*RExC_parse != '=' && *RExC_parse != '!') 
1984                     goto unknown;
1985                 RExC_parse++;
1986             case '=':
1987             case '!':
1988                 RExC_seen_zerolen++;
1989             case ':':
1990             case '>':
1991                 break;
1992             case '$':
1993             case '@':
1994                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
1995                 break;
1996             case '#':
1997                 while (*RExC_parse && *RExC_parse != ')')
1998                     RExC_parse++;
1999                 if (*RExC_parse != ')')
2000                     FAIL("Sequence (?#... not terminated");
2001                 nextchar(pRExC_state);
2002                 *flagp = TRYAGAIN;
2003                 return NULL;
2004             case 'p':
2005                 if (SIZE_ONLY)
2006                     vWARN(RExC_parse, "(?p{}) is deprecated - use (??{})");
2007                 /* FALL THROUGH*/
2008             case '?':
2009                 logical = 1;
2010                 paren = *RExC_parse++;
2011                 /* FALL THROUGH */
2012             case '{':
2013             {
2014                 I32 count = 1, n = 0;
2015                 char c;
2016                 char *s = RExC_parse;
2017                 SV *sv;
2018                 OP_4tree *sop, *rop;
2019
2020                 RExC_seen_zerolen++;
2021                 RExC_seen |= REG_SEEN_EVAL;
2022                 while (count && (c = *RExC_parse)) {
2023                     if (c == '\\' && RExC_parse[1])
2024                         RExC_parse++;
2025                     else if (c == '{') 
2026                         count++;
2027                     else if (c == '}') 
2028                         count--;
2029                     RExC_parse++;
2030                 }
2031                 if (*RExC_parse != ')')
2032                 {
2033                     RExC_parse = s;                 
2034                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2035                 }
2036                 if (!SIZE_ONLY) {
2037                     AV *av;
2038                     
2039                     if (RExC_parse - 1 - s) 
2040                         sv = newSVpvn(s, RExC_parse - 1 - s);
2041                     else
2042                         sv = newSVpvn("", 0);
2043
2044                     ENTER;
2045                     Perl_save_re_context(aTHX);
2046                     rop = sv_compile_2op(sv, &sop, "re", &av);
2047                     LEAVE;
2048
2049                     n = add_data(pRExC_state, 3, "nop");
2050                     RExC_rx->data->data[n] = (void*)rop;
2051                     RExC_rx->data->data[n+1] = (void*)sop;
2052                     RExC_rx->data->data[n+2] = (void*)av;
2053                     SvREFCNT_dec(sv);
2054                 }
2055                 else {                                          /* First pass */
2056                     if (PL_reginterp_cnt < ++RExC_seen_evals
2057                         && PL_curcop != &PL_compiling)
2058                         /* No compiled RE interpolated, has runtime
2059                            components ===> unsafe.  */
2060                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2061                     if (PL_tainted)
2062                         FAIL("Eval-group in insecure regular expression");
2063                 }
2064                 
2065                 nextchar(pRExC_state);
2066                 if (logical) {
2067                     ret = reg_node(pRExC_state, LOGICAL);
2068                     if (!SIZE_ONLY)
2069                         ret->flags = 2;
2070                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2071                     return ret;
2072                 }
2073                 return reganode(pRExC_state, EVAL, n);
2074             }
2075             case '(':
2076             {
2077                 if (RExC_parse[0] == '?') {
2078                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!' 
2079                         || RExC_parse[1] == '<' 
2080                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2081                         I32 flag;
2082                         
2083                         ret = reg_node(pRExC_state, LOGICAL);
2084                         if (!SIZE_ONLY)
2085                             ret->flags = 1;
2086                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2087                         goto insert_if;
2088                     } 
2089                 }
2090                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2091                     parno = atoi(RExC_parse++);
2092
2093                     while (isDIGIT(*RExC_parse))
2094                         RExC_parse++;
2095                     ret = reganode(pRExC_state, GROUPP, parno);
2096                     if ((c = *nextchar(pRExC_state)) != ')')
2097                         vFAIL("Switch condition not recognized");
2098                   insert_if:
2099                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2100                     br = regbranch(pRExC_state, &flags, 1);
2101                     if (br == NULL)
2102                         br = reganode(pRExC_state, LONGJMP, 0);
2103                     else
2104                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2105                     c = *nextchar(pRExC_state);
2106                     if (flags&HASWIDTH)
2107                         *flagp |= HASWIDTH;
2108                     if (c == '|') {
2109                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2110                         regbranch(pRExC_state, &flags, 1);
2111                         regtail(pRExC_state, ret, lastbr);
2112                         if (flags&HASWIDTH)
2113                             *flagp |= HASWIDTH;
2114                         c = *nextchar(pRExC_state);
2115                     }
2116                     else
2117                         lastbr = NULL;
2118                     if (c != ')')
2119                         vFAIL("Switch (?(condition)... contains too many branches");
2120                     ender = reg_node(pRExC_state, TAIL);
2121                     regtail(pRExC_state, br, ender);
2122                     if (lastbr) {
2123                         regtail(pRExC_state, lastbr, ender);
2124                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2125                     }
2126                     else
2127                         regtail(pRExC_state, ret, ender);
2128                     return ret;
2129                 }
2130                 else {
2131                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2132                 }
2133             }
2134             case 0:
2135                 RExC_parse--; /* for vFAIL to print correctly */
2136                 vFAIL("Sequence (? incomplete");
2137                 break;
2138             default:
2139                 --RExC_parse;
2140               parse_flags:
2141                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2142                     if (*RExC_parse != 'o')
2143                         pmflag(flagsp, *RExC_parse);
2144                     ++RExC_parse;
2145                 }
2146                 if (*RExC_parse == '-') {
2147                     flagsp = &negflags;
2148                     ++RExC_parse;
2149                     goto parse_flags;
2150                 }
2151                 RExC_flags16 |= posflags;
2152                 RExC_flags16 &= ~negflags;
2153                 if (*RExC_parse == ':') {
2154                     RExC_parse++;
2155                     paren = ':';
2156                     break;
2157                 }               
2158               unknown:
2159                 if (*RExC_parse != ')') {
2160                     RExC_parse++;
2161                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2162                 }
2163                 nextchar(pRExC_state);
2164                 *flagp = TRYAGAIN;
2165                 return NULL;
2166             }
2167         }
2168         else {
2169             parno = RExC_npar;
2170             RExC_npar++;
2171             ret = reganode(pRExC_state, OPEN, parno);
2172             open = 1;
2173         }
2174     }
2175     else
2176         ret = NULL;
2177
2178     /* Pick up the branches, linking them together. */
2179     br = regbranch(pRExC_state, &flags, 1);
2180     if (br == NULL)
2181         return(NULL);
2182     if (*RExC_parse == '|') {
2183         if (!SIZE_ONLY && RExC_extralen) {
2184             reginsert(pRExC_state, BRANCHJ, br);
2185         }
2186         else
2187             reginsert(pRExC_state, BRANCH, br);
2188         have_branch = 1;
2189         if (SIZE_ONLY)
2190             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2191     }
2192     else if (paren == ':') {
2193         *flagp |= flags&SIMPLE;
2194     }
2195     if (open) {                         /* Starts with OPEN. */
2196         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2197     }
2198     else if (paren != '?')              /* Not Conditional */
2199         ret = br;
2200     if (flags&HASWIDTH)
2201         *flagp |= HASWIDTH;
2202     *flagp |= flags&SPSTART;
2203     lastbr = br;
2204     while (*RExC_parse == '|') {
2205         if (!SIZE_ONLY && RExC_extralen) {
2206             ender = reganode(pRExC_state, LONGJMP,0);
2207             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2208         }
2209         if (SIZE_ONLY)
2210             RExC_extralen += 2;         /* Account for LONGJMP. */
2211         nextchar(pRExC_state);
2212         br = regbranch(pRExC_state, &flags, 0);
2213         if (br == NULL)
2214             return(NULL);
2215         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2216         lastbr = br;
2217         if (flags&HASWIDTH)
2218             *flagp |= HASWIDTH;
2219         *flagp |= flags&SPSTART;
2220     }
2221
2222     if (have_branch || paren != ':') {
2223         /* Make a closing node, and hook it on the end. */
2224         switch (paren) {
2225         case ':':
2226             ender = reg_node(pRExC_state, TAIL);
2227             break;
2228         case 1:
2229             ender = reganode(pRExC_state, CLOSE, parno);
2230             break;
2231         case '<':
2232         case ',':
2233         case '=':
2234         case '!':
2235             *flagp &= ~HASWIDTH;
2236             /* FALL THROUGH */
2237         case '>':
2238             ender = reg_node(pRExC_state, SUCCEED);
2239             break;
2240         case 0:
2241             ender = reg_node(pRExC_state, END);
2242             break;
2243         }
2244         regtail(pRExC_state, lastbr, ender);
2245
2246         if (have_branch) {
2247             /* Hook the tails of the branches to the closing node. */
2248             for (br = ret; br != NULL; br = regnext(br)) {
2249                 regoptail(pRExC_state, br, ender);
2250             }
2251         }
2252     }
2253
2254     {
2255         char *p;
2256         static char parens[] = "=!<,>";
2257
2258         if (paren && (p = strchr(parens, paren))) {
2259             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2260             int flag = (p - parens) > 1;
2261
2262             if (paren == '>')
2263                 node = SUSPEND, flag = 0;
2264             reginsert(pRExC_state, node,ret);
2265             ret->flags = flag;
2266             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2267         }
2268     }
2269
2270     /* Check for proper termination. */
2271     if (paren) {
2272         RExC_flags16 = oregflags;
2273         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2274             RExC_parse = oregcomp_parse;
2275             vFAIL("Unmatched (");
2276         }
2277     }
2278     else if (!paren && RExC_parse < RExC_end) {
2279         if (*RExC_parse == ')') {
2280             RExC_parse++;
2281             vFAIL("Unmatched )");
2282         }
2283         else
2284             FAIL("Junk on end of regexp");      /* "Can't happen". */
2285         /* NOTREACHED */
2286     }
2287
2288     return(ret);
2289 }
2290
2291 /*
2292  - regbranch - one alternative of an | operator
2293  *
2294  * Implements the concatenation operator.
2295  */
2296 STATIC regnode *
2297 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2298 {
2299     register regnode *ret;
2300     register regnode *chain = NULL;
2301     register regnode *latest;
2302     I32 flags = 0, c = 0;
2303
2304     if (first) 
2305         ret = NULL;
2306     else {
2307         if (!SIZE_ONLY && RExC_extralen) 
2308             ret = reganode(pRExC_state, BRANCHJ,0);
2309         else
2310             ret = reg_node(pRExC_state, BRANCH);
2311     }
2312         
2313     if (!first && SIZE_ONLY) 
2314         RExC_extralen += 1;                     /* BRANCHJ */
2315     
2316     *flagp = WORST;                     /* Tentatively. */
2317
2318     RExC_parse--;
2319     nextchar(pRExC_state);
2320     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2321         flags &= ~TRYAGAIN;
2322         latest = regpiece(pRExC_state, &flags);
2323         if (latest == NULL) {
2324             if (flags & TRYAGAIN)
2325                 continue;
2326             return(NULL);
2327         }
2328         else if (ret == NULL)
2329             ret = latest;
2330         *flagp |= flags&HASWIDTH;
2331         if (chain == NULL)      /* First piece. */
2332             *flagp |= flags&SPSTART;
2333         else {
2334             RExC_naughty++;
2335             regtail(pRExC_state, chain, latest);
2336         }
2337         chain = latest;
2338         c++;
2339     }
2340     if (chain == NULL) {        /* Loop ran zero times. */
2341         chain = reg_node(pRExC_state, NOTHING);
2342         if (ret == NULL)
2343             ret = chain;
2344     }
2345     if (c == 1) {
2346         *flagp |= flags&SIMPLE;
2347     }
2348
2349     return(ret);
2350 }
2351
2352 /*
2353  - regpiece - something followed by possible [*+?]
2354  *
2355  * Note that the branching code sequences used for ? and the general cases
2356  * of * and + are somewhat optimized:  they use the same NOTHING node as
2357  * both the endmarker for their branch list and the body of the last branch.
2358  * It might seem that this node could be dispensed with entirely, but the
2359  * endmarker role is not redundant.
2360  */
2361 STATIC regnode *
2362 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2363 {
2364     register regnode *ret;
2365     register char op;
2366     register char *next;
2367     I32 flags;
2368     char *origparse = RExC_parse;
2369     char *maxpos;
2370     I32 min;
2371     I32 max = REG_INFTY;
2372
2373     ret = regatom(pRExC_state, &flags);
2374     if (ret == NULL) {
2375         if (flags & TRYAGAIN)
2376             *flagp |= TRYAGAIN;
2377         return(NULL);
2378     }
2379
2380     op = *RExC_parse;
2381
2382     if (op == '{' && regcurly(RExC_parse)) {
2383         next = RExC_parse + 1;
2384         maxpos = Nullch;
2385         while (isDIGIT(*next) || *next == ',') {
2386             if (*next == ',') {
2387                 if (maxpos)
2388                     break;
2389                 else
2390                     maxpos = next;
2391             }
2392             next++;
2393         }
2394         if (*next == '}') {             /* got one */
2395             if (!maxpos)
2396                 maxpos = next;
2397             RExC_parse++;
2398             min = atoi(RExC_parse);
2399             if (*maxpos == ',')
2400                 maxpos++;
2401             else
2402                 maxpos = RExC_parse;
2403             max = atoi(maxpos);
2404             if (!max && *maxpos != '0')
2405                 max = REG_INFTY;                /* meaning "infinity" */
2406             else if (max >= REG_INFTY)
2407                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2408             RExC_parse = next;
2409             nextchar(pRExC_state);
2410
2411         do_curly:
2412             if ((flags&SIMPLE)) {
2413                 RExC_naughty += 2 + RExC_naughty / 2;
2414                 reginsert(pRExC_state, CURLY, ret);
2415             }
2416             else {
2417                 regnode *w = reg_node(pRExC_state, WHILEM);
2418
2419                 w->flags = 0;
2420                 regtail(pRExC_state, ret, w);
2421                 if (!SIZE_ONLY && RExC_extralen) {
2422                     reginsert(pRExC_state, LONGJMP,ret);
2423                     reginsert(pRExC_state, NOTHING,ret);
2424                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2425                 }
2426                 reginsert(pRExC_state, CURLYX,ret);
2427                 if (!SIZE_ONLY && RExC_extralen)
2428                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2429                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2430                 if (SIZE_ONLY)
2431                     RExC_whilem_seen++, RExC_extralen += 3;
2432                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2433             }
2434             ret->flags = 0;
2435
2436             if (min > 0)
2437                 *flagp = WORST;
2438             if (max > 0)
2439                 *flagp |= HASWIDTH;
2440             if (max && max < min)
2441                 vFAIL("Can't do {n,m} with n > m");
2442             if (!SIZE_ONLY) {
2443                 ARG1_SET(ret, min);
2444                 ARG2_SET(ret, max);
2445             }
2446
2447             goto nest_check;
2448         }
2449     }
2450
2451     if (!ISMULT1(op)) {
2452         *flagp = flags;
2453         return(ret);
2454     }
2455
2456 #if 0                           /* Now runtime fix should be reliable. */
2457
2458     /* if this is reinstated, don't forget to put this back into perldiag:
2459
2460             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2461
2462            (F) The part of the regexp subject to either the * or + quantifier
2463            could match an empty string. The {#} shows in the regular
2464            expression about where the problem was discovered.
2465
2466     */
2467
2468     if (!(flags&HASWIDTH) && op != '?')
2469       vFAIL("Regexp *+ operand could be empty");
2470 #endif 
2471
2472     nextchar(pRExC_state);
2473
2474     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2475
2476     if (op == '*' && (flags&SIMPLE)) {
2477         reginsert(pRExC_state, STAR, ret);
2478         ret->flags = 0;
2479         RExC_naughty += 4;
2480     }
2481     else if (op == '*') {
2482         min = 0;
2483         goto do_curly;
2484     }
2485     else if (op == '+' && (flags&SIMPLE)) {
2486         reginsert(pRExC_state, PLUS, ret);
2487         ret->flags = 0;
2488         RExC_naughty += 3;
2489     }
2490     else if (op == '+') {
2491         min = 1;
2492         goto do_curly;
2493     }
2494     else if (op == '?') {
2495         min = 0; max = 1;
2496         goto do_curly;
2497     }
2498   nest_check:
2499     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2500         vWARN3(RExC_parse,
2501                "%.*s matches null string many times",
2502                RExC_parse - origparse,
2503                origparse);
2504     }
2505
2506     if (*RExC_parse == '?') {
2507         nextchar(pRExC_state);
2508         reginsert(pRExC_state, MINMOD, ret);
2509         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2510     }
2511     if (ISMULT2(RExC_parse)) {
2512         RExC_parse++;
2513         vFAIL("Nested quantifiers");
2514     }
2515
2516     return(ret);
2517 }
2518
2519 /*
2520  - regatom - the lowest level
2521  *
2522  * Optimization:  gobbles an entire sequence of ordinary characters so that
2523  * it can turn them into a single node, which is smaller to store and
2524  * faster to run.  Backslashed characters are exceptions, each becoming a
2525  * separate node; the code is simpler that way and it's not worth fixing.
2526  *
2527  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2528 STATIC regnode *
2529 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2530 {
2531     register regnode *ret = 0;
2532     I32 flags;
2533
2534     *flagp = WORST;             /* Tentatively. */
2535
2536 tryagain:
2537     switch (*RExC_parse) {
2538     case '^':
2539         RExC_seen_zerolen++;
2540         nextchar(pRExC_state);
2541         if (RExC_flags16 & PMf_MULTILINE)
2542             ret = reg_node(pRExC_state, MBOL);
2543         else if (RExC_flags16 & PMf_SINGLELINE)
2544             ret = reg_node(pRExC_state, SBOL);
2545         else
2546             ret = reg_node(pRExC_state, BOL);
2547         break;
2548     case '$':
2549         nextchar(pRExC_state);
2550         if (*RExC_parse) 
2551             RExC_seen_zerolen++;
2552         if (RExC_flags16 & PMf_MULTILINE)
2553             ret = reg_node(pRExC_state, MEOL);
2554         else if (RExC_flags16 & PMf_SINGLELINE)
2555             ret = reg_node(pRExC_state, SEOL);
2556         else
2557             ret = reg_node(pRExC_state, EOL);
2558         break;
2559     case '.':
2560         nextchar(pRExC_state);
2561         if (RExC_flags16 & PMf_SINGLELINE)
2562             ret = reg_node(pRExC_state, SANY);
2563         else
2564             ret = reg_node(pRExC_state, REG_ANY);
2565         *flagp |= HASWIDTH|SIMPLE;
2566         RExC_naughty++;
2567         break;
2568     case '[':
2569     {
2570         char *oregcomp_parse = ++RExC_parse;
2571         ret = regclass(pRExC_state);
2572         if (*RExC_parse != ']') {
2573             RExC_parse = oregcomp_parse;
2574             vFAIL("Unmatched [");
2575         }
2576         nextchar(pRExC_state);
2577         *flagp |= HASWIDTH|SIMPLE;
2578         break;
2579     }
2580     case '(':
2581         nextchar(pRExC_state);
2582         ret = reg(pRExC_state, 1, &flags);
2583         if (ret == NULL) {
2584                 if (flags & TRYAGAIN) {
2585                     if (RExC_parse == RExC_end) {
2586                          /* Make parent create an empty node if needed. */
2587                         *flagp |= TRYAGAIN;
2588                         return(NULL);
2589                     }
2590                     goto tryagain;
2591                 }
2592                 return(NULL);
2593         }
2594         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2595         break;
2596     case '|':
2597     case ')':
2598         if (flags & TRYAGAIN) {
2599             *flagp |= TRYAGAIN;
2600             return NULL;
2601         }
2602         vFAIL("Internal urp");
2603                                 /* Supposed to be caught earlier. */
2604         break;
2605     case '{':
2606         if (!regcurly(RExC_parse)) {
2607             RExC_parse++;
2608             goto defchar;
2609         }
2610         /* FALL THROUGH */
2611     case '?':
2612     case '+':
2613     case '*':
2614         RExC_parse++;
2615         vFAIL("Quantifier follows nothing");
2616         break;
2617     case '\\':
2618         switch (*++RExC_parse) {
2619         case 'A':
2620             RExC_seen_zerolen++;
2621             ret = reg_node(pRExC_state, SBOL);
2622             *flagp |= SIMPLE;
2623             nextchar(pRExC_state);
2624             break;
2625         case 'G':
2626             ret = reg_node(pRExC_state, GPOS);
2627             RExC_seen |= REG_SEEN_GPOS;
2628             *flagp |= SIMPLE;
2629             nextchar(pRExC_state);
2630             break;
2631         case 'Z':
2632             ret = reg_node(pRExC_state, SEOL);
2633             *flagp |= SIMPLE;
2634             nextchar(pRExC_state);
2635             break;
2636         case 'z':
2637             ret = reg_node(pRExC_state, EOS);
2638             *flagp |= SIMPLE;
2639             RExC_seen_zerolen++;                /* Do not optimize RE away */
2640             nextchar(pRExC_state);
2641             break;
2642         case 'C':
2643             ret = reg_node(pRExC_state, SANY);
2644             *flagp |= HASWIDTH|SIMPLE;
2645             nextchar(pRExC_state);
2646             break;
2647         case 'X':
2648             ret = reg_node(pRExC_state, CLUMP);
2649             *flagp |= HASWIDTH;
2650             nextchar(pRExC_state);
2651             if (UTF && !PL_utf8_mark)
2652                 is_utf8_mark((U8*)"~");         /* preload table */
2653             break;
2654         case 'w':
2655             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2656             *flagp |= HASWIDTH|SIMPLE;
2657             nextchar(pRExC_state);
2658             if (UTF && !PL_utf8_alnum)
2659                 is_utf8_alnum((U8*)"a");        /* preload table */
2660             break;
2661         case 'W':
2662             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2663             *flagp |= HASWIDTH|SIMPLE;
2664             nextchar(pRExC_state);
2665             if (UTF && !PL_utf8_alnum)
2666                 is_utf8_alnum((U8*)"a");        /* preload table */
2667             break;
2668         case 'b':
2669             RExC_seen_zerolen++;
2670             RExC_seen |= REG_SEEN_LOOKBEHIND;
2671             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2672             *flagp |= SIMPLE;
2673             nextchar(pRExC_state);
2674             if (UTF && !PL_utf8_alnum)
2675                 is_utf8_alnum((U8*)"a");        /* preload table */
2676             break;
2677         case 'B':
2678             RExC_seen_zerolen++;
2679             RExC_seen |= REG_SEEN_LOOKBEHIND;
2680             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2681             *flagp |= SIMPLE;
2682             nextchar(pRExC_state);
2683             if (UTF && !PL_utf8_alnum)
2684                 is_utf8_alnum((U8*)"a");        /* preload table */
2685             break;
2686         case 's':
2687             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2688             *flagp |= HASWIDTH|SIMPLE;
2689             nextchar(pRExC_state);
2690             if (UTF && !PL_utf8_space)
2691                 is_utf8_space((U8*)" ");        /* preload table */
2692             break;
2693         case 'S':
2694             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
2695             *flagp |= HASWIDTH|SIMPLE;
2696             nextchar(pRExC_state);
2697             if (UTF && !PL_utf8_space)
2698                 is_utf8_space((U8*)" ");        /* preload table */
2699             break;
2700         case 'd':
2701             ret = reg_node(pRExC_state, DIGIT);
2702             *flagp |= HASWIDTH|SIMPLE;
2703             nextchar(pRExC_state);
2704             if (UTF && !PL_utf8_digit)
2705                 is_utf8_digit((U8*)"1");        /* preload table */
2706             break;
2707         case 'D':
2708             ret = reg_node(pRExC_state, NDIGIT);
2709             *flagp |= HASWIDTH|SIMPLE;
2710             nextchar(pRExC_state);
2711             if (UTF && !PL_utf8_digit)
2712                 is_utf8_digit((U8*)"1");        /* preload table */
2713             break;
2714         case 'p':
2715         case 'P':
2716             {   /* a lovely hack--pretend we saw [\pX] instead */
2717                 char* oldregxend = RExC_end;
2718
2719                 if (RExC_parse[1] == '{') {
2720                     RExC_end = strchr(RExC_parse, '}');
2721                     if (!RExC_end) {
2722                         RExC_parse += 2;
2723                         RExC_end = oldregxend;
2724                         vFAIL("Missing right brace on \\p{}");
2725                     }
2726                     RExC_end++;
2727                 }
2728                 else
2729                     RExC_end = RExC_parse + 2;
2730                 RExC_parse--;
2731
2732                 ret = regclass(pRExC_state);
2733
2734                 RExC_end = oldregxend;
2735                 RExC_parse--;
2736                 nextchar(pRExC_state);
2737                 *flagp |= HASWIDTH|SIMPLE;
2738             }
2739             break;
2740         case 'n':
2741         case 'r':
2742         case 't':
2743         case 'f':
2744         case 'e':
2745         case 'a':
2746         case 'x':
2747         case 'c':
2748         case '0':
2749             goto defchar;
2750         case '1': case '2': case '3': case '4':
2751         case '5': case '6': case '7': case '8': case '9':
2752             {
2753                 I32 num = atoi(RExC_parse);
2754
2755                 if (num > 9 && num >= RExC_npar)
2756                     goto defchar;
2757                 else {
2758                     while (isDIGIT(*RExC_parse))
2759                         RExC_parse++;
2760
2761                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2762                         vFAIL("Reference to nonexistent group");
2763                     RExC_sawback = 1;
2764                     ret = reganode(pRExC_state, FOLD
2765                                    ? (LOC ? REFFL : REFF)
2766                                    : REF, num);
2767                     *flagp |= HASWIDTH;
2768                     RExC_parse--;
2769                     nextchar(pRExC_state);
2770                 }
2771             }
2772             break;
2773         case '\0':
2774             if (RExC_parse >= RExC_end)
2775                 FAIL("Trailing \\");
2776             /* FALL THROUGH */
2777         default:
2778             /* Do not generate `unrecognized' warnings here, we fall
2779                back into the quick-grab loop below */
2780             goto defchar;
2781         }
2782         break;
2783
2784     case '#':
2785         if (RExC_flags16 & PMf_EXTENDED) {
2786             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2787             if (RExC_parse < RExC_end)
2788                 goto tryagain;
2789         }
2790         /* FALL THROUGH */
2791
2792     default: {
2793             register STRLEN len;
2794             register UV ender;
2795             register char *p;
2796             char *oldp, *s;
2797             STRLEN numlen;
2798
2799             RExC_parse++;
2800
2801         defchar:
2802             ret = reg_node(pRExC_state, FOLD
2803                           ? (LOC ? EXACTFL : EXACTF)
2804                           : EXACT);
2805             s = STRING(ret);
2806             for (len = 0, p = RExC_parse - 1;
2807               len < 127 && p < RExC_end;
2808               len++)
2809             {
2810                 oldp = p;
2811
2812                 if (RExC_flags16 & PMf_EXTENDED)
2813                     p = regwhite(p, RExC_end);
2814                 switch (*p) {
2815                 case '^':
2816                 case '$':
2817                 case '.':
2818                 case '[':
2819                 case '(':
2820                 case ')':
2821                 case '|':
2822                     goto loopdone;
2823                 case '\\':
2824                     switch (*++p) {
2825                     case 'A':
2826                     case 'G':
2827                     case 'Z':
2828                     case 'z':
2829                     case 'w':
2830                     case 'W':
2831                     case 'b':
2832                     case 'B':
2833                     case 's':
2834                     case 'S':
2835                     case 'd':
2836                     case 'D':
2837                     case 'p':
2838                     case 'P':
2839                         --p;
2840                         goto loopdone;
2841                     case 'n':
2842                         ender = '\n';
2843                         p++;
2844                         break;
2845                     case 'r':
2846                         ender = '\r';
2847                         p++;
2848                         break;
2849                     case 't':
2850                         ender = '\t';
2851                         p++;
2852                         break;
2853                     case 'f':
2854                         ender = '\f';
2855                         p++;
2856                         break;
2857                     case 'e':
2858 #ifdef ASCIIish
2859                           ender = '\033';
2860 #else
2861                           ender = '\047';
2862 #endif
2863                         p++;
2864                         break;
2865                     case 'a':
2866 #ifdef ASCIIish
2867                           ender = '\007';
2868 #else
2869                           ender = '\057';
2870 #endif
2871                         p++;
2872                         break;
2873                     case 'x':
2874                         if (*++p == '{') {
2875                             char* e = strchr(p, '}');
2876          
2877                             if (!e) {
2878                                 RExC_parse = p + 1;
2879                                 vFAIL("Missing right brace on \\x{}");
2880                             }
2881                             else {
2882                                 numlen = 1;     /* allow underscores */
2883                                 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
2884                                 /* numlen is generous */
2885                                 if (numlen + len >= 127) {
2886                                     p--;
2887                                     goto loopdone;
2888                                 }
2889                                 p = e + 1;
2890                             }
2891                         }
2892                         else {
2893                             numlen = 0;         /* disallow underscores */
2894                             ender = (UV)scan_hex(p, 2, &numlen);
2895                             p += numlen;
2896                         }
2897                         break;
2898                     case 'c':
2899                         p++;
2900                         ender = UCHARAT(p++);
2901                         ender = toCTRL(ender);
2902                         break;
2903                     case '0': case '1': case '2': case '3':case '4':
2904                     case '5': case '6': case '7': case '8':case '9':
2905                         if (*p == '0' ||
2906                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
2907                             numlen = 0;         /* disallow underscores */
2908                             ender = (UV)scan_oct(p, 3, &numlen);
2909                             p += numlen;
2910                         }
2911                         else {
2912                             --p;
2913                             goto loopdone;
2914                         }
2915                         break;
2916                     case '\0':
2917                         if (p >= RExC_end)
2918                             FAIL("Trailing \\");
2919                         /* FALL THROUGH */
2920                     default:
2921                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
2922                             vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
2923                         goto normal_default;
2924                     }
2925                     break;
2926                 default:
2927                   normal_default:
2928                     if ((*p & 0xc0) == 0xc0 && UTF) {
2929                         ender = utf8_to_uv((U8*)p, RExC_end - p,
2930                                                &numlen, 0);
2931                         p += numlen;
2932                     }
2933                     else
2934                         ender = *p++;
2935                     break;
2936                 }
2937                 if (RExC_flags16 & PMf_EXTENDED)
2938                     p = regwhite(p, RExC_end);
2939                 if (UTF && FOLD) {
2940                     if (LOC)
2941                         ender = toLOWER_LC_uni(ender);
2942                     else
2943                         ender = toLOWER_uni(ender);
2944                 }
2945                 if (ISMULT2(p)) { /* Back off on ?+*. */
2946                     if (len)
2947                         p = oldp;
2948                     else if (ender >= 0x80 && UTF) {
2949                         reguni(pRExC_state, ender, s, &numlen);
2950                         s += numlen;
2951                         len += numlen;
2952                     }
2953                     else {
2954                         len++;
2955                         REGC(ender, s++);
2956                     }
2957                     break;
2958                 }
2959                 if (ender >= 0x80 && UTF) {
2960                     reguni(pRExC_state, ender, s, &numlen);
2961                     s += numlen;
2962                     len += numlen - 1;
2963                 }
2964                 else
2965                     REGC(ender, s++);
2966             }
2967         loopdone:
2968             RExC_parse = p - 1;
2969             nextchar(pRExC_state);
2970             {
2971                 /* len is STRLEN which is unsigned, need to copy to signed */
2972                 IV iv = len;
2973                 if (iv < 0)
2974                     vFAIL("Internal disaster");
2975             }
2976             if (len > 0)
2977                 *flagp |= HASWIDTH;
2978             if (len == 1)
2979                 *flagp |= SIMPLE;
2980             if (!SIZE_ONLY)
2981                 STR_LEN(ret) = len;
2982             if (SIZE_ONLY)
2983                 RExC_size += STR_SZ(len);
2984             else
2985                 RExC_emit += STR_SZ(len);
2986         }
2987         break;
2988     }
2989
2990     return(ret);
2991 }
2992
2993 STATIC char *
2994 S_regwhite(pTHX_ char *p, char *e)
2995 {
2996     while (p < e) {
2997         if (isSPACE(*p))
2998             ++p;
2999         else if (*p == '#') {
3000             do {
3001                 p++;
3002             } while (p < e && *p != '\n');
3003         }
3004         else
3005             break;
3006     }
3007     return p;
3008 }
3009
3010 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3011    Character classes ([:foo:]) can also be negated ([:^foo:]).
3012    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3013    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3014    but trigger warnings because they are currently unimplemented. */
3015 STATIC I32
3016 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3017 {
3018     char *posixcc = 0;
3019     I32 namedclass = OOB_NAMEDCLASS;
3020
3021     if (value == '[' && RExC_parse + 1 < RExC_end &&
3022         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3023         (*RExC_parse == ':' ||
3024          *RExC_parse == '=' ||
3025          *RExC_parse == '.')) {
3026         char  c = *RExC_parse;
3027         char* s = RExC_parse++;
3028             
3029         while (RExC_parse < RExC_end && *RExC_parse != c)
3030             RExC_parse++;
3031         if (RExC_parse == RExC_end)
3032             /* Grandfather lone [:, [=, [. */
3033             RExC_parse = s;
3034         else {
3035             char* t = RExC_parse++; /* skip over the c */
3036
3037             if (*RExC_parse == ']') {
3038                 RExC_parse++; /* skip over the ending ] */
3039                 posixcc = s + 1;
3040                 if (*s == ':') {
3041                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3042                     I32 skip = 5; /* the most common skip */
3043
3044                     switch (*posixcc) {
3045                     case 'a':
3046                         if (strnEQ(posixcc, "alnum", 5))
3047                             namedclass =
3048                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3049                         else if (strnEQ(posixcc, "alpha", 5))
3050                             namedclass =
3051                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3052                         else if (strnEQ(posixcc, "ascii", 5))
3053                             namedclass =
3054                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3055                         break;
3056                     case 'b':
3057                         if (strnEQ(posixcc, "blank", 5))
3058                             namedclass =
3059                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3060                         break;
3061                     case 'c':
3062                         if (strnEQ(posixcc, "cntrl", 5))
3063                             namedclass =
3064                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3065                         break;
3066                     case 'd':
3067                         if (strnEQ(posixcc, "digit", 5))
3068                             namedclass =
3069                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3070                         break;
3071                     case 'g':
3072                         if (strnEQ(posixcc, "graph", 5))
3073                             namedclass =
3074                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3075                         break;
3076                     case 'l':
3077                         if (strnEQ(posixcc, "lower", 5))
3078                             namedclass =
3079                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3080                         break;
3081                     case 'p':
3082                         if (strnEQ(posixcc, "print", 5))
3083                             namedclass =
3084                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3085                         else if (strnEQ(posixcc, "punct", 5))
3086                             namedclass =
3087                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3088                         break;
3089                     case 's':
3090                         if (strnEQ(posixcc, "space", 5))
3091                             namedclass =
3092                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3093                         break;
3094                     case 'u':
3095                         if (strnEQ(posixcc, "upper", 5))
3096                             namedclass =
3097                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3098                         break;
3099                     case 'w': /* this is not POSIX, this is the Perl \w */
3100                         if (strnEQ(posixcc, "word", 4)) {
3101                             namedclass =
3102                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3103                             skip = 4;
3104                         }
3105                         break;
3106                     case 'x':
3107                         if (strnEQ(posixcc, "xdigit", 6)) {
3108                             namedclass =
3109                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3110                             skip = 6;
3111                         }
3112                         break;
3113                     }
3114                     if (namedclass == OOB_NAMEDCLASS ||
3115                         posixcc[skip] != ':' ||
3116                         posixcc[skip+1] != ']')
3117                     {
3118                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3119                                       t - s - 1, s + 1);
3120                     }
3121                 } else if (!SIZE_ONLY) {
3122                     /* [[=foo=]] and [[.foo.]] are still future. */
3123
3124                     /* adjust RExC_parse so the warning shows after
3125                        the class closes */
3126                     while (*RExC_parse && *RExC_parse != ']')
3127                         RExC_parse++;
3128                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3129                 }
3130             } else {
3131                 /* Maternal grandfather:
3132                  * "[:" ending in ":" but not in ":]" */
3133                 RExC_parse = s;
3134             }
3135         }
3136     }
3137
3138     return namedclass;
3139 }
3140
3141 STATIC void
3142 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3143 {
3144     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3145         (*RExC_parse == ':' ||
3146          *RExC_parse == '=' ||
3147          *RExC_parse == '.')) {
3148         char *s = RExC_parse;
3149         char  c = *s++;
3150
3151         while(*s && isALNUM(*s))
3152             s++;
3153         if (*s && c == *s && s[1] == ']') {
3154             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3155
3156             /* [[=foo=]] and [[.foo.]] are still future. */
3157             if (c == '=' || c == '.')
3158             {
3159                 /* adjust RExC_parse so the error shows after
3160                    the class closes */
3161                 while (*RExC_parse && *RExC_parse++ != ']')
3162                     ;
3163                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3164             }
3165         }
3166     }
3167 }
3168
3169 STATIC regnode *
3170 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3171 {
3172     register UV value;
3173     register IV lastvalue = OOB_UNICODE;
3174     register IV range = 0;
3175     register regnode *ret;
3176     STRLEN numlen;
3177     IV namedclass;
3178     char *rangebegin;
3179     bool need_class = 0;
3180     SV *listsv;
3181     register char *e;
3182     UV n;
3183     bool dont_optimize_invert = FALSE;
3184
3185     ret = reganode(pRExC_state, ANYOF, 0);
3186
3187     if (!SIZE_ONLY)
3188         ANYOF_FLAGS(ret) = 0;
3189
3190     if (*RExC_parse == '^') {   /* Complement of range. */
3191         RExC_naughty++;
3192         RExC_parse++;
3193         if (!SIZE_ONLY)
3194             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3195     }
3196
3197     if (SIZE_ONLY)
3198         RExC_size += ANYOF_SKIP;
3199     else {
3200         RExC_emit += ANYOF_SKIP;
3201         if (FOLD)
3202             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3203         if (LOC)
3204             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3205         ANYOF_BITMAP_ZERO(ret);
3206         listsv = newSVpvn("# comment\n", 10);
3207     }
3208
3209     if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3210         checkposixcc(pRExC_state);
3211
3212     if (*RExC_parse == ']' || *RExC_parse == '-')
3213         goto charclassloop;             /* allow 1st char to be ] or - */
3214
3215     while (RExC_parse < RExC_end && *RExC_parse != ']') {
3216
3217     charclassloop:
3218
3219         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3220
3221         if (!range)
3222             rangebegin = RExC_parse;
3223         if (UTF) {
3224             value = utf8_to_uv((U8*)RExC_parse,
3225                                RExC_end - RExC_parse,
3226                                &numlen, 0);
3227             RExC_parse += numlen;
3228         }
3229         else
3230             value = UCHARAT(RExC_parse++);
3231         if (value == '[')
3232             namedclass = regpposixcc(pRExC_state, value);
3233         else if (value == '\\') {
3234             if (UTF) {
3235                 value = utf8_to_uv((U8*)RExC_parse,
3236                                    RExC_end - RExC_parse,
3237                                    &numlen, 0);
3238                 RExC_parse += numlen;
3239             }
3240             else
3241                 value = UCHARAT(RExC_parse++);
3242             /* Some compilers cannot handle switching on 64-bit integer
3243              * values, therefore value cannot be an UV.  Yes, this will
3244              * be a problem later if we want switch on Unicode.
3245              * A similar issue a little bit later when switching on
3246              * namedclass. --jhi */
3247             switch ((I32)value) {
3248             case 'w':   namedclass = ANYOF_ALNUM;       break;
3249             case 'W':   namedclass = ANYOF_NALNUM;      break;
3250             case 's':   namedclass = ANYOF_SPACE;       break;
3251             case 'S':   namedclass = ANYOF_NSPACE;      break;
3252             case 'd':   namedclass = ANYOF_DIGIT;       break;
3253             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3254             case 'p':
3255             case 'P':
3256                 if (*RExC_parse == '{') {
3257                     e = strchr(RExC_parse++, '}');
3258                     if (!e)
3259                         vFAIL("Missing right brace on \\p{}");
3260                     n = e - RExC_parse;
3261                 }
3262                 else {
3263                     e = RExC_parse;
3264                     n = 1;
3265                 }
3266                 if (!SIZE_ONLY) {
3267                     if (value == 'p')
3268                         Perl_sv_catpvf(aTHX_ listsv,
3269                                        "+utf8::%.*s\n", (int)n, RExC_parse);
3270                     else
3271                         Perl_sv_catpvf(aTHX_ listsv,
3272                                        "!utf8::%.*s\n", (int)n, RExC_parse);
3273                 }
3274                 RExC_parse = e + 1;
3275                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3276                 continue;
3277             case 'n':   value = '\n';                   break;
3278             case 'r':   value = '\r';                   break;
3279             case 't':   value = '\t';                   break;
3280             case 'f':   value = '\f';                   break;
3281             case 'b':   value = '\b';                   break;
3282 #ifdef ASCIIish
3283             case 'e':   value = '\033';                 break;
3284             case 'a':   value = '\007';                 break;
3285 #else
3286             case 'e':   value = '\047';                 break;
3287             case 'a':   value = '\057';                 break;
3288 #endif
3289             case 'x':
3290                 if (*RExC_parse == '{') {
3291                     e = strchr(RExC_parse++, '}');
3292                     if (!e) 
3293                         vFAIL("Missing right brace on \\x{}");
3294                     numlen = 1;         /* allow underscores */
3295                     value = (UV)scan_hex(RExC_parse,
3296                                          e - RExC_parse,
3297                                          &numlen);
3298                     RExC_parse = e + 1;
3299                 }
3300                 else {
3301                     numlen = 0;         /* disallow underscores */
3302                     value = (UV)scan_hex(RExC_parse, 2, &numlen);
3303                     RExC_parse += numlen;
3304                 }
3305                 break;
3306             case 'c':
3307                 value = UCHARAT(RExC_parse++);
3308                 value = toCTRL(value);
3309                 break;
3310             case '0': case '1': case '2': case '3': case '4':
3311             case '5': case '6': case '7': case '8': case '9':
3312                 numlen = 0;             /* disallow underscores */
3313                 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3314                 RExC_parse += numlen;
3315                 break;
3316             default:
3317                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3318                     vWARN2(RExC_parse,
3319                            "Unrecognized escape \\%c in character class passed through",
3320                            (int)value);
3321                 break;
3322             }
3323         } /* end of \blah */
3324
3325         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3326
3327             if (!SIZE_ONLY && !need_class)
3328                 ANYOF_CLASS_ZERO(ret);
3329
3330             need_class = 1;
3331
3332             /* a bad range like a-\d, a-[:digit:] ? */
3333             if (range) {
3334                 if (!SIZE_ONLY) {
3335                     if (ckWARN(WARN_REGEXP))
3336                         vWARN4(RExC_parse,
3337                                "False [] range \"%*.*s\"",
3338                                RExC_parse - rangebegin,
3339                                RExC_parse - rangebegin,
3340                                rangebegin);
3341                     if (lastvalue < 256) {
3342                         ANYOF_BITMAP_SET(ret, lastvalue);
3343                         ANYOF_BITMAP_SET(ret, '-');
3344                     }
3345                     else {
3346                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3347                         Perl_sv_catpvf(aTHX_ listsv,
3348                                        /* 0x002D is Unicode for '-' */
3349                                        "%04"UVxf"\n002D\n", (UV)lastvalue);
3350                     }
3351                 }
3352
3353                 range = 0; /* this was not a true range */
3354             }
3355
3356             if (!SIZE_ONLY) {
3357                 /* Possible truncation here but in some 64-bit environments
3358                  * the compiler gets heartburn about switch on 64-bit values.
3359                  * A similar issue a little earlier when switching on value.
3360                  * --jhi */
3361                 switch ((I32)namedclass) {
3362                 case ANYOF_ALNUM:
3363                     if (LOC)
3364                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3365                     else {
3366                         for (value = 0; value < 256; value++)
3367                             if (isALNUM(value))
3368                                 ANYOF_BITMAP_SET(ret, value);
3369                     }
3370                     dont_optimize_invert = TRUE;
3371                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3372                     break;
3373                 case ANYOF_NALNUM:
3374                     if (LOC)
3375                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3376                     else {
3377                         for (value = 0; value < 256; value++)
3378                             if (!isALNUM(value))
3379                                 ANYOF_BITMAP_SET(ret, value);
3380                     }
3381                     dont_optimize_invert = TRUE;
3382                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3383                     break;
3384                 case ANYOF_ALNUMC:
3385                     if (LOC)
3386                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3387                     else {
3388                         for (value = 0; value < 256; value++)
3389                             if (isALNUMC(value))
3390                                 ANYOF_BITMAP_SET(ret, value);
3391                     }
3392                     dont_optimize_invert = TRUE;
3393                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3394                     break;
3395                 case ANYOF_NALNUMC:
3396                     if (LOC)
3397                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3398                     else {
3399                         for (value = 0; value < 256; value++)
3400                             if (!isALNUMC(value))
3401                                 ANYOF_BITMAP_SET(ret, value);
3402                     }
3403                     dont_optimize_invert = TRUE;
3404                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3405                     break;
3406                 case ANYOF_ALPHA:
3407                     if (LOC)
3408                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3409                     else {
3410                         for (value = 0; value < 256; value++)
3411                             if (isALPHA(value))
3412                                 ANYOF_BITMAP_SET(ret, value);
3413                     }
3414                     dont_optimize_invert = TRUE;
3415                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3416                     break;
3417                 case ANYOF_NALPHA:
3418                     if (LOC)
3419                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3420                     else {
3421                         for (value = 0; value < 256; value++)
3422                             if (!isALPHA(value))
3423                                 ANYOF_BITMAP_SET(ret, value);
3424                     }
3425                     dont_optimize_invert = TRUE;
3426                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3427                     break;
3428                 case ANYOF_ASCII:
3429                     if (LOC)
3430                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3431                     else {
3432 #ifdef ASCIIish
3433                         for (value = 0; value < 128; value++)
3434                             ANYOF_BITMAP_SET(ret, value);
3435 #else  /* EBCDIC */
3436                         for (value = 0; value < 256; value++)
3437                             if (isASCII(value))
3438                                 ANYOF_BITMAP_SET(ret, value);
3439 #endif /* EBCDIC */
3440                     }
3441                     dont_optimize_invert = TRUE;
3442                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3443                     break;
3444                 case ANYOF_NASCII:
3445                     if (LOC)
3446                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3447                     else {
3448 #ifdef ASCIIish
3449                         for (value = 128; value < 256; value++)
3450                             ANYOF_BITMAP_SET(ret, value);
3451 #else  /* EBCDIC */
3452                         for (value = 0; value < 256; value++)
3453                             if (!isASCII(value))
3454                                 ANYOF_BITMAP_SET(ret, value);
3455 #endif /* EBCDIC */
3456                     }
3457                     dont_optimize_invert = TRUE;
3458                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3459                     break;
3460                 case ANYOF_BLANK:
3461                     if (LOC)
3462                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3463                     else {
3464                         for (value = 0; value < 256; value++)
3465                             if (isBLANK(value))
3466                                 ANYOF_BITMAP_SET(ret, value);
3467                     }
3468                     dont_optimize_invert = TRUE;
3469                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3470                     break;
3471                 case ANYOF_NBLANK:
3472                     if (LOC)
3473                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3474                     else {
3475                         for (value = 0; value < 256; value++)
3476                             if (!isBLANK(value))
3477                                 ANYOF_BITMAP_SET(ret, value);
3478                     }
3479                     dont_optimize_invert = TRUE;
3480                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3481                     break;
3482                 case ANYOF_CNTRL:
3483                     if (LOC)
3484                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3485                     else {
3486                         for (value = 0; value < 256; value++)
3487                             if (isCNTRL(value))
3488                                 ANYOF_BITMAP_SET(ret, value);
3489                     }
3490                     dont_optimize_invert = TRUE;
3491                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3492                     break;
3493                 case ANYOF_NCNTRL:
3494                     if (LOC)
3495                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3496                     else {
3497                         for (value = 0; value < 256; value++)
3498                             if (!isCNTRL(value))
3499                                 ANYOF_BITMAP_SET(ret, value);
3500                     }
3501                     dont_optimize_invert = TRUE;
3502                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3503                     break;
3504                 case ANYOF_DIGIT:
3505                     if (LOC)
3506                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3507                     else {
3508                         /* consecutive digits assumed */
3509                         for (value = '0'; value <= '9'; value++)
3510                             ANYOF_BITMAP_SET(ret, value);
3511                     }
3512                     dont_optimize_invert = TRUE;
3513                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3514                     break;
3515                 case ANYOF_NDIGIT:
3516                     if (LOC)
3517                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3518                     else {
3519                         /* consecutive digits assumed */
3520                         for (value = 0; value < '0'; value++)
3521                             ANYOF_BITMAP_SET(ret, value);
3522                         for (value = '9' + 1; value < 256; value++)
3523                             ANYOF_BITMAP_SET(ret, value);
3524                     }
3525                     dont_optimize_invert = TRUE;
3526                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3527                     break;
3528                 case ANYOF_GRAPH:
3529                     if (LOC)
3530                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3531                     else {
3532                         for (value = 0; value < 256; value++)
3533                             if (isGRAPH(value))
3534                                 ANYOF_BITMAP_SET(ret, value);
3535                     }
3536                     dont_optimize_invert = TRUE;
3537                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3538                     break;
3539                 case ANYOF_NGRAPH:
3540                     if (LOC)
3541                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3542                     else {
3543                         for (value = 0; value < 256; value++)
3544                             if (!isGRAPH(value))
3545                                 ANYOF_BITMAP_SET(ret, value);
3546                     }
3547                     dont_optimize_invert = TRUE;
3548                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3549                     break;
3550                 case ANYOF_LOWER:
3551                     if (LOC)
3552                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3553                     else {
3554                         for (value = 0; value < 256; value++)
3555                             if (isLOWER(value))
3556                                 ANYOF_BITMAP_SET(ret, value);
3557                     }
3558                     dont_optimize_invert = TRUE;
3559                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3560                     break;
3561                 case ANYOF_NLOWER:
3562                     if (LOC)
3563                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3564                     else {
3565                         for (value = 0; value < 256; value++)
3566                             if (!isLOWER(value))
3567                                 ANYOF_BITMAP_SET(ret, value);
3568                     }
3569                     dont_optimize_invert = TRUE;
3570                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3571                     break;
3572                 case ANYOF_PRINT:
3573                     if (LOC)
3574                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3575                     else {
3576                         for (value = 0; value < 256; value++)
3577                             if (isPRINT(value))
3578                                 ANYOF_BITMAP_SET(ret, value);
3579                     }
3580                     dont_optimize_invert = TRUE;
3581                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3582                     break;
3583                 case ANYOF_NPRINT:
3584                     if (LOC)
3585                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3586                     else {
3587                         for (value = 0; value < 256; value++)
3588                             if (!isPRINT(value))
3589                                 ANYOF_BITMAP_SET(ret, value);
3590                     }
3591                     dont_optimize_invert = TRUE;
3592                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3593                     break;
3594                 case ANYOF_PSXSPC:
3595                     if (LOC)
3596                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3597                     else {
3598                         for (value = 0; value < 256; value++)
3599                             if (isPSXSPC(value))
3600                                 ANYOF_BITMAP_SET(ret, value);
3601                     }
3602                     dont_optimize_invert = TRUE;
3603                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3604                     break;
3605                 case ANYOF_NPSXSPC:
3606                     if (LOC)
3607                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3608                     else {
3609                         for (value = 0; value < 256; value++)
3610                             if (!isPSXSPC(value))
3611                                 ANYOF_BITMAP_SET(ret, value);
3612                     }
3613                     dont_optimize_invert = TRUE;
3614                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3615                     break;
3616                 case ANYOF_PUNCT:
3617                     if (LOC)
3618                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3619                     else {
3620                         for (value = 0; value < 256; value++)
3621                             if (isPUNCT(value))
3622                                 ANYOF_BITMAP_SET(ret, value);
3623                     }
3624                     dont_optimize_invert = TRUE;
3625                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3626                     break;
3627                 case ANYOF_NPUNCT:
3628                     if (LOC)
3629                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3630                     else {
3631                         for (value = 0; value < 256; value++)
3632                             if (!isPUNCT(value))
3633                                 ANYOF_BITMAP_SET(ret, value);
3634                     }
3635                     dont_optimize_invert = TRUE;
3636                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3637                     break;
3638                 case ANYOF_SPACE:
3639                     if (LOC)
3640                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3641                     else {
3642                         for (value = 0; value < 256; value++)
3643                             if (isSPACE(value))
3644                                 ANYOF_BITMAP_SET(ret, value);
3645                     }
3646                     dont_optimize_invert = TRUE;
3647                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3648                     break;
3649                 case ANYOF_NSPACE:
3650                     if (LOC)
3651                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3652                     else {
3653                         for (value = 0; value < 256; value++)
3654                             if (!isSPACE(value))
3655                                 ANYOF_BITMAP_SET(ret, value);
3656                     }
3657                     dont_optimize_invert = TRUE;
3658                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3659                     break;
3660                 case ANYOF_UPPER:
3661                     if (LOC)
3662                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3663                     else {
3664                         for (value = 0; value < 256; value++)
3665                             if (isUPPER(value))
3666                                 ANYOF_BITMAP_SET(ret, value);
3667                     }
3668                     dont_optimize_invert = TRUE;
3669                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3670                     break;
3671                 case ANYOF_NUPPER:
3672                     if (LOC)
3673                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3674                     else {
3675                         for (value = 0; value < 256; value++)
3676                             if (!isUPPER(value))
3677                                 ANYOF_BITMAP_SET(ret, value);
3678                     }
3679                     dont_optimize_invert = TRUE;
3680                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3681                     break;
3682                 case ANYOF_XDIGIT:
3683                     if (LOC)
3684                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3685                     else {
3686                         for (value = 0; value < 256; value++)
3687                             if (isXDIGIT(value))
3688                                 ANYOF_BITMAP_SET(ret, value);
3689                     }
3690                     dont_optimize_invert = TRUE;
3691                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3692                     break;
3693                 case ANYOF_NXDIGIT:
3694                     if (LOC)
3695                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3696                     else {
3697                         for (value = 0; value < 256; value++)
3698                             if (!isXDIGIT(value))
3699                                 ANYOF_BITMAP_SET(ret, value);
3700                     }
3701                     dont_optimize_invert = TRUE;
3702                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3703                     break;
3704                 default:
3705                     vFAIL("Invalid [::] class");
3706                     break;
3707                 }
3708                 if (LOC)
3709                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3710                 continue;
3711             }
3712         } /* end of namedclass \blah */
3713
3714         if (range) {
3715             if (lastvalue > value) /* b-a */ {
3716                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3717                               RExC_parse - rangebegin,
3718                               RExC_parse - rangebegin,
3719                               rangebegin);
3720             }
3721             range = 0; /* not a true range */
3722         }
3723         else {
3724             lastvalue = value; /* save the beginning of the range */
3725             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3726                 RExC_parse[1] != ']') {
3727                 RExC_parse++;
3728
3729                 /* a bad range like \w-, [:word:]- ? */
3730                 if (namedclass > OOB_NAMEDCLASS) {
3731                     if (ckWARN(WARN_REGEXP))
3732                         vWARN4(RExC_parse,
3733                                "False [] range \"%*.*s\"",
3734                                RExC_parse - rangebegin,
3735                                RExC_parse - rangebegin,
3736                                rangebegin);
3737                     if (!SIZE_ONLY)
3738                         ANYOF_BITMAP_SET(ret, '-');
3739                 } else
3740                     range = 1;  /* yeah, it's a range! */
3741                 continue;       /* but do it the next time */
3742             }
3743         }
3744
3745         /* now is the next time */
3746         if (!SIZE_ONLY) {
3747             if (lastvalue < 256 && value < 256) {
3748 #ifndef ASCIIish /* EBCDIC, for example. */
3749                 if ((isLOWER(lastvalue) && isLOWER(value)) ||
3750                     (isUPPER(lastvalue) && isUPPER(value)))
3751                 {
3752                     IV i;
3753                     if (isLOWER(lastvalue)) {
3754                         for (i = lastvalue; i <= value; i++)
3755                             if (isLOWER(i))
3756                                 ANYOF_BITMAP_SET(ret, i);
3757                     } else {
3758                         for (i = lastvalue; i <= value; i++)
3759                             if (isUPPER(i))
3760                                 ANYOF_BITMAP_SET(ret, i);
3761                     }
3762                 }
3763                 else
3764 #endif
3765                     for ( ; lastvalue <= value; lastvalue++)
3766                         ANYOF_BITMAP_SET(ret, lastvalue);
3767             } else {
3768                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3769                 if (lastvalue < value)
3770                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3771                                    (UV)lastvalue, (UV)value);
3772                 else
3773                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3774                                    (UV)value);
3775             }
3776         }
3777
3778         range = 0; /* this range (if it was one) is done now */
3779     }
3780
3781     if (need_class) {
3782         if (SIZE_ONLY)
3783             RExC_size += ANYOF_CLASS_ADD_SKIP;
3784         else
3785             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3786     }
3787
3788     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3789     if (!SIZE_ONLY &&
3790         (ANYOF_FLAGS(ret) &
3791          /* If the only flag is folding (plus possibly inversion). */
3792          (ANYOF_FLAGS_ALL ^ ANYOF_INVERT) == ANYOF_FOLD)) {
3793         for (value = 0; value < 256; ++value) {
3794             if (ANYOF_BITMAP_TEST(ret, value)) {
3795                 IV fold = PL_fold[value];
3796
3797                 if (fold != value)
3798                     ANYOF_BITMAP_SET(ret, fold);
3799             }
3800         }
3801         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3802     }
3803
3804     /* optimize inverted simple patterns (e.g. [^a-z]) */
3805     if (!SIZE_ONLY && !dont_optimize_invert &&
3806         /* If the only flag is inversion. */
3807         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3808         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3809             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3810         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3811     }
3812
3813     if (!SIZE_ONLY) { 
3814         AV *av = newAV();
3815         SV *rv;
3816
3817         av_store(av, 0, listsv);
3818         av_store(av, 1, NULL);
3819         rv = newRV_noinc((SV*)av);
3820         n = add_data(pRExC_state, 1, "s");
3821         RExC_rx->data->data[n] = (void*)rv;
3822         ARG_SET(ret, n);
3823     }
3824
3825     return ret;
3826 }
3827
3828 STATIC char*
3829 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3830 {
3831     char* retval = RExC_parse++;
3832
3833     for (;;) {
3834         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3835                 RExC_parse[2] == '#') {
3836             while (*RExC_parse && *RExC_parse != ')')
3837                 RExC_parse++;
3838             RExC_parse++;
3839             continue;
3840         }
3841         if (RExC_flags16 & PMf_EXTENDED) {
3842             if (isSPACE(*RExC_parse)) {
3843                 RExC_parse++;
3844                 continue;
3845             }
3846             else if (*RExC_parse == '#') {
3847                 while (*RExC_parse && *RExC_parse != '\n')
3848                     RExC_parse++;
3849                 RExC_parse++;
3850                 continue;
3851             }
3852         }
3853         return retval;
3854     }
3855 }
3856
3857 /*
3858 - reg_node - emit a node
3859 */
3860 STATIC regnode *                        /* Location. */
3861 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3862 {
3863     register regnode *ret;
3864     register regnode *ptr;
3865
3866     ret = RExC_emit;
3867     if (SIZE_ONLY) {
3868         SIZE_ALIGN(RExC_size);
3869         RExC_size += 1;
3870         return(ret);
3871     }
3872
3873     NODE_ALIGN_FILL(ret);
3874     ptr = ret;
3875     FILL_ADVANCE_NODE(ptr, op);
3876     RExC_emit = ptr;
3877
3878     return(ret);
3879 }
3880
3881 /*
3882 - reganode - emit a node with an argument
3883 */
3884 STATIC regnode *                        /* Location. */
3885 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
3886 {
3887     register regnode *ret;
3888     register regnode *ptr;
3889
3890     ret = RExC_emit;
3891     if (SIZE_ONLY) {
3892         SIZE_ALIGN(RExC_size);
3893         RExC_size += 2;
3894         return(ret);
3895     }
3896
3897     NODE_ALIGN_FILL(ret);
3898     ptr = ret;
3899     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
3900     RExC_emit = ptr;
3901
3902     return(ret);
3903 }
3904
3905 /*
3906 - reguni - emit (if appropriate) a Unicode character
3907 */
3908 STATIC void
3909 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
3910 {
3911     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uv_to_utf8((U8*)s, uv) - (U8*)s);
3912 }
3913
3914 /*
3915 - reginsert - insert an operator in front of already-emitted operand
3916 *
3917 * Means relocating the operand.
3918 */
3919 STATIC void
3920 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
3921 {
3922     register regnode *src;
3923     register regnode *dst;
3924     register regnode *place;
3925     register int offset = regarglen[(U8)op];
3926     
3927 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
3928
3929     if (SIZE_ONLY) {
3930         RExC_size += NODE_STEP_REGNODE + offset;
3931         return;
3932     }
3933
3934     src = RExC_emit;
3935     RExC_emit += NODE_STEP_REGNODE + offset;
3936     dst = RExC_emit;
3937     while (src > opnd)
3938         StructCopy(--src, --dst, regnode);
3939
3940     place = opnd;               /* Op node, where operand used to be. */
3941     src = NEXTOPER(place);
3942     FILL_ADVANCE_NODE(place, op);
3943     Zero(src, offset, regnode);
3944 }
3945
3946 /*
3947 - regtail - set the next-pointer at the end of a node chain of p to val.
3948 */
3949 STATIC void
3950 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
3951 {
3952     register regnode *scan;
3953     register regnode *temp;
3954
3955     if (SIZE_ONLY)
3956         return;
3957
3958     /* Find last node. */
3959     scan = p;
3960     for (;;) {
3961         temp = regnext(scan);
3962         if (temp == NULL)
3963             break;
3964         scan = temp;
3965     }
3966
3967     if (reg_off_by_arg[OP(scan)]) {
3968         ARG_SET(scan, val - scan);
3969     }
3970     else {
3971         NEXT_OFF(scan) = val - scan;
3972     }
3973 }
3974
3975 /*
3976 - regoptail - regtail on operand of first argument; nop if operandless
3977 */
3978 STATIC void
3979 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
3980 {
3981     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
3982     if (p == NULL || SIZE_ONLY)
3983         return;
3984     if (PL_regkind[(U8)OP(p)] == BRANCH) {
3985         regtail(pRExC_state, NEXTOPER(p), val);
3986     }
3987     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
3988         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
3989     }
3990     else
3991         return;
3992 }
3993
3994 /*
3995  - regcurly - a little FSA that accepts {\d+,?\d*}
3996  */
3997 STATIC I32
3998 S_regcurly(pTHX_ register char *s)
3999 {
4000     if (*s++ != '{')
4001         return FALSE;
4002     if (!isDIGIT(*s))
4003         return FALSE;
4004     while (isDIGIT(*s))
4005         s++;
4006     if (*s == ',')
4007         s++;
4008     while (isDIGIT(*s))
4009         s++;
4010     if (*s != '}')
4011         return FALSE;
4012     return TRUE;
4013 }
4014
4015
4016 STATIC regnode *
4017 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4018 {
4019 #ifdef DEBUGGING
4020     register U8 op = EXACT;     /* Arbitrary non-END op. */
4021     register regnode *next;
4022
4023     while (op != END && (!last || node < last)) {
4024         /* While that wasn't END last time... */
4025
4026         NODE_ALIGN(node);
4027         op = OP(node);
4028         if (op == CLOSE)
4029             l--;        
4030         next = regnext(node);
4031         /* Where, what. */
4032         if (OP(node) == OPTIMIZED)
4033             goto after_print;
4034         regprop(sv, node);
4035         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4036                       (int)(2*l + 1), "", SvPVX(sv));
4037         if (next == NULL)               /* Next ptr. */
4038             PerlIO_printf(Perl_debug_log, "(0)");
4039         else 
4040             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4041         (void)PerlIO_putc(Perl_debug_log, '\n');
4042       after_print:
4043         if (PL_regkind[(U8)op] == BRANCHJ) {
4044             register regnode *nnode = (OP(next) == LONGJMP 
4045                                        ? regnext(next) 
4046                                        : next);
4047             if (last && nnode > last)
4048                 nnode = last;
4049             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4050         }
4051         else if (PL_regkind[(U8)op] == BRANCH) {
4052             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4053         }
4054         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4055             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4056                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4057         }
4058         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4059             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4060                              next, sv, l + 1);
4061         }
4062         else if ( op == PLUS || op == STAR) {
4063             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4064         }
4065         else if (op == ANYOF) {
4066             node = NEXTOPER(node);
4067             node += ANYOF_SKIP;
4068         }
4069         else if (PL_regkind[(U8)op] == EXACT) {
4070             /* Literal string, where present. */
4071             node += NODE_SZ_STR(node) - 1;
4072             node = NEXTOPER(node);
4073         }
4074         else {
4075             node = NEXTOPER(node);
4076             node += regarglen[(U8)op];
4077         }
4078         if (op == CURLYX || op == OPEN)
4079             l++;
4080         else if (op == WHILEM)
4081             l--;
4082     }
4083 #endif  /* DEBUGGING */
4084     return node;
4085 }
4086
4087 /*
4088  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4089  */
4090 void
4091 Perl_regdump(pTHX_ regexp *r)
4092 {
4093 #ifdef DEBUGGING
4094     SV *sv = sv_newmortal();
4095
4096     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4097
4098     /* Header fields of interest. */
4099     if (r->anchored_substr)
4100         PerlIO_printf(Perl_debug_log,
4101                       "anchored `%s%.*s%s'%s at %"IVdf" ", 
4102                       PL_colors[0],
4103                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4104                       SvPVX(r->anchored_substr), 
4105                       PL_colors[1],
4106                       SvTAIL(r->anchored_substr) ? "$" : "",
4107                       (IV)r->anchored_offset);
4108     if (r->float_substr)
4109         PerlIO_printf(Perl_debug_log,
4110                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ", 
4111                       PL_colors[0],
4112                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)), 
4113                       SvPVX(r->float_substr),
4114                       PL_colors[1],
4115                       SvTAIL(r->float_substr) ? "$" : "",
4116                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4117     if (r->check_substr)
4118         PerlIO_printf(Perl_debug_log, 
4119                       r->check_substr == r->float_substr 
4120                       ? "(checking floating" : "(checking anchored");
4121     if (r->reganch & ROPT_NOSCAN)
4122         PerlIO_printf(Perl_debug_log, " noscan");
4123     if (r->reganch & ROPT_CHECK_ALL)
4124         PerlIO_printf(Perl_debug_log, " isall");
4125     if (r->check_substr)
4126         PerlIO_printf(Perl_debug_log, ") ");
4127
4128     if (r->regstclass) {
4129         regprop(sv, r->regstclass);
4130         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4131     }
4132     if (r->reganch & ROPT_ANCH) {
4133         PerlIO_printf(Perl_debug_log, "anchored");
4134         if (r->reganch & ROPT_ANCH_BOL)
4135             PerlIO_printf(Perl_debug_log, "(BOL)");
4136         if (r->reganch & ROPT_ANCH_MBOL)
4137             PerlIO_printf(Perl_debug_log, "(MBOL)");
4138         if (r->reganch & ROPT_ANCH_SBOL)
4139             PerlIO_printf(Perl_debug_log, "(SBOL)");
4140         if (r->reganch & ROPT_ANCH_GPOS)
4141             PerlIO_printf(Perl_debug_log, "(GPOS)");
4142         PerlIO_putc(Perl_debug_log, ' ');
4143     }
4144     if (r->reganch & ROPT_GPOS_SEEN)
4145         PerlIO_printf(Perl_debug_log, "GPOS ");
4146     if (r->reganch & ROPT_SKIP)
4147         PerlIO_printf(Perl_debug_log, "plus ");
4148     if (r->reganch & ROPT_IMPLICIT)
4149         PerlIO_printf(Perl_debug_log, "implicit ");
4150     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4151     if (r->reganch & ROPT_EVAL_SEEN)
4152         PerlIO_printf(Perl_debug_log, "with eval ");
4153     PerlIO_printf(Perl_debug_log, "\n");
4154 #endif  /* DEBUGGING */
4155 }
4156
4157 STATIC void
4158 S_put_byte(pTHX_ SV *sv, int c)
4159 {
4160     if (isCNTRL(c) || c == 127 || c == 255 || !isPRINT(c))
4161         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4162     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4163         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4164     else
4165         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4166 }
4167
4168 /*
4169 - regprop - printable representation of opcode
4170 */
4171 void
4172 Perl_regprop(pTHX_ SV *sv, regnode *o)
4173 {
4174 #ifdef DEBUGGING
4175     register int k;
4176
4177     sv_setpvn(sv, "", 0);
4178     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4179         /* It would be nice to FAIL() here, but this may be called from
4180            regexec.c, and it would be hard to supply pRExC_state. */
4181         Perl_croak(aTHX_ "Corrupted regexp opcode");
4182     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4183
4184     k = PL_regkind[(U8)OP(o)];
4185
4186     if (k == EXACT)
4187         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4188                        STR_LEN(o), STRING(o), PL_colors[1]);
4189     else if (k == CURLY) {
4190         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4191             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4192         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4193     }
4194     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4195         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4196     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4197         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4198     else if (k == LOGICAL)
4199         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4200     else if (k == ANYOF) {
4201         int i, rangestart = -1;
4202         U8 flags = ANYOF_FLAGS(o);
4203         const char * const anyofs[] = { /* Should be syncronized with
4204                                          * ANYOF_ #xdefines in regcomp.h */
4205             "\\w",
4206             "\\W",
4207             "\\s",
4208             "\\S",
4209             "\\d",
4210             "\\D",
4211             "[:alnum:]",
4212             "[:^alnum:]",
4213             "[:alpha:]",
4214             "[:^alpha:]",
4215             "[:ascii:]",
4216             "[:^ascii:]",
4217             "[:ctrl:]",
4218             "[:^ctrl:]",
4219             "[:graph:]",
4220             "[:^graph:]",
4221             "[:lower:]",
4222             "[:^lower:]",
4223             "[:print:]",
4224             "[:^print:]",
4225             "[:punct:]",
4226             "[:^punct:]",
4227             "[:upper:]",
4228             "[:^upper:]",
4229             "[:xdigit:]",
4230             "[:^xdigit:]",
4231             "[:space:]",
4232             "[:^space:]",
4233             "[:blank:]",
4234             "[:^blank:]"
4235         };
4236
4237         if (flags & ANYOF_LOCALE)
4238             sv_catpv(sv, "{loc}");
4239         if (flags & ANYOF_FOLD)
4240             sv_catpv(sv, "{i}");
4241         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4242         if (flags & ANYOF_INVERT)
4243             sv_catpv(sv, "^");
4244         for (i = 0; i <= 256; i++) {
4245             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4246                 if (rangestart == -1)
4247                     rangestart = i;
4248             } else if (rangestart != -1) {
4249                 if (i <= rangestart + 3)
4250                     for (; rangestart < i; rangestart++)
4251                         put_byte(sv, rangestart);
4252                 else {
4253                     put_byte(sv, rangestart);
4254                     sv_catpv(sv, "-");
4255                     put_byte(sv, i - 1);
4256                 }
4257                 rangestart = -1;
4258             }
4259         }
4260
4261         if (o->flags & ANYOF_CLASS)
4262             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4263                 if (ANYOF_CLASS_TEST(o,i))
4264                     sv_catpv(sv, anyofs[i]);
4265
4266         if (flags & ANYOF_UNICODE)
4267             sv_catpv(sv, "{unicode}");
4268         else if (flags & ANYOF_UNICODE_ALL)
4269             sv_catpv(sv, "{all-unicode}");
4270
4271         {
4272             SV *lv;
4273             SV *sw = regclass_swash(o, FALSE, &lv);
4274             
4275             if (lv) {
4276                 if (sw) {
4277                     UV i;
4278                     U8 s[UTF8_MAXLEN+1];
4279                     
4280                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4281                         U8 *e = uv_to_utf8(s, i);
4282                         
4283                         if (i < 256 && swash_fetch(sw, s)) {
4284                             if (rangestart == -1)
4285                                 rangestart = i;
4286                         } else if (rangestart != -1) {
4287                             U8 *p;
4288                             
4289                             if (i <= rangestart + 3)
4290                                 for (; rangestart < i; rangestart++) {
4291                                     for(e = uv_to_utf8(s, rangestart), p = s; p < e; p++)
4292                                         put_byte(sv, *p);
4293                                 }
4294                             else {
4295                                 for (e = uv_to_utf8(s, rangestart), p = s; p < e; p++)
4296                                     put_byte(sv, *p);
4297                                 sv_catpv(sv, "-");
4298                                     for (e = uv_to_utf8(s, i - 1), p = s; p < e; p++)
4299                                         put_byte(sv, *p);
4300                                 }
4301                                 rangestart = -1;
4302                             }
4303                         }
4304                         
4305                     sv_catpv(sv, "..."); /* et cetera */
4306                 }
4307
4308                 {
4309                     char *s = savepv(SvPVX(lv));
4310                     char *origs = s;
4311                     
4312                     while(*s && *s != '\n') s++;
4313                     
4314                     if (*s == '\n') {
4315                         char *t = ++s;
4316                         
4317                         while (*s) {
4318                             if (*s == '\n')
4319                                 *s = ' ';
4320                             s++;
4321                         }
4322                         if (s[-1] == ' ')
4323                             s[-1] = 0;
4324                         
4325                         sv_catpv(sv, t);
4326                     }
4327                     
4328                     Safefree(origs);
4329                 }
4330             }
4331         }
4332
4333         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4334     }
4335     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4336         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4337 #endif  /* DEBUGGING */
4338 }
4339
4340 SV *
4341 Perl_re_intuit_string(pTHX_ regexp *prog)
4342 {                               /* Assume that RE_INTUIT is set */
4343     DEBUG_r(
4344         {   STRLEN n_a;
4345             char *s = SvPV(prog->check_substr,n_a);
4346
4347             if (!PL_colorset) reginitcolors();
4348             PerlIO_printf(Perl_debug_log,
4349                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4350                       PL_colors[4],PL_colors[5],PL_colors[0],
4351                       s,
4352                       PL_colors[1],
4353                       (strlen(s) > 60 ? "..." : ""));
4354         } );
4355
4356     return prog->check_substr;
4357 }
4358
4359 void
4360 Perl_pregfree(pTHX_ struct regexp *r)
4361 {
4362     DEBUG_r(if (!PL_colorset) reginitcolors());
4363
4364     if (!r || (--r->refcnt > 0))
4365         return;
4366     DEBUG_r(PerlIO_printf(Perl_debug_log,
4367                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4368                       PL_colors[4],PL_colors[5],PL_colors[0],
4369                       r->precomp,
4370                       PL_colors[1],
4371                       (strlen(r->precomp) > 60 ? "..." : "")));
4372
4373     if (r->precomp)
4374         Safefree(r->precomp);
4375     if (RX_MATCH_COPIED(r))
4376         Safefree(r->subbeg);
4377     if (r->substrs) {
4378         if (r->anchored_substr)
4379             SvREFCNT_dec(r->anchored_substr);
4380         if (r->float_substr)
4381             SvREFCNT_dec(r->float_substr);
4382         Safefree(r->substrs);
4383     }
4384     if (r->data) {
4385         int n = r->data->count;
4386         AV* new_comppad = NULL;
4387         AV* old_comppad;
4388         SV** old_curpad;
4389
4390         while (--n >= 0) {
4391             switch (r->data->what[n]) {
4392             case 's':
4393                 SvREFCNT_dec((SV*)r->data->data[n]);
4394                 break;
4395             case 'f':
4396                 Safefree(r->data->data[n]);
4397                 break;
4398             case 'p':
4399                 new_comppad = (AV*)r->data->data[n];
4400                 break;
4401             case 'o':
4402                 if (new_comppad == NULL)
4403                     Perl_croak(aTHX_ "panic: pregfree comppad");
4404                 old_comppad = PL_comppad;
4405                 old_curpad = PL_curpad;
4406                 /* Watch out for global destruction's random ordering. */
4407                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4408                     PL_comppad = new_comppad;
4409                     PL_curpad = AvARRAY(new_comppad);
4410                 }
4411                 else
4412                     PL_curpad = NULL;
4413                 op_free((OP_4tree*)r->data->data[n]);
4414                 PL_comppad = old_comppad;
4415                 PL_curpad = old_curpad;
4416                 SvREFCNT_dec((SV*)new_comppad);
4417                 new_comppad = NULL;
4418                 break;
4419             case 'n':
4420                 break;
4421             default:
4422                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4423             }
4424         }
4425         Safefree(r->data->what);
4426         Safefree(r->data);
4427     }
4428     Safefree(r->startp);
4429     Safefree(r->endp);
4430     Safefree(r);
4431 }
4432
4433 /*
4434  - regnext - dig the "next" pointer out of a node
4435  *
4436  * [Note, when REGALIGN is defined there are two places in regmatch()
4437  * that bypass this code for speed.]
4438  */
4439 regnode *
4440 Perl_regnext(pTHX_ register regnode *p)
4441 {
4442     register I32 offset;
4443
4444     if (p == &PL_regdummy)
4445         return(NULL);
4446
4447     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4448     if (offset == 0)
4449         return(NULL);
4450
4451     return(p+offset);
4452 }
4453
4454 STATIC void     
4455 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4456 {
4457     va_list args;
4458     STRLEN l1 = strlen(pat1);
4459     STRLEN l2 = strlen(pat2);
4460     char buf[512];
4461     SV *msv;
4462     char *message;
4463
4464     if (l1 > 510)
4465         l1 = 510;
4466     if (l1 + l2 > 510)
4467         l2 = 510 - l1;
4468     Copy(pat1, buf, l1 , char);
4469     Copy(pat2, buf + l1, l2 , char);
4470     buf[l1 + l2] = '\n';
4471     buf[l1 + l2 + 1] = '\0';
4472 #ifdef I_STDARG
4473     /* ANSI variant takes additional second argument */
4474     va_start(args, pat2);
4475 #else
4476     va_start(args);
4477 #endif
4478     msv = vmess(buf, &args);
4479     va_end(args);
4480     message = SvPV(msv,l1);
4481     if (l1 > 512)
4482         l1 = 512;
4483     Copy(message, buf, l1 , char);
4484     buf[l1] = '\0';                     /* Overwrite \n */
4485     Perl_croak(aTHX_ "%s", buf);
4486 }
4487
4488 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4489
4490 void
4491 Perl_save_re_context(pTHX)
4492 {                   
4493 #if 0
4494     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4495     SAVEI32(RExC_npar);         /* () count. */
4496     SAVEI32(RExC_size);         /* Code size. */
4497     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4498     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4499     SAVEI32(RExC_seen);         /* from regcomp.c */
4500     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4501     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4502     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4503     SAVEPPTR(RExC_end);         /* End of input for compile */
4504     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4505 #endif
4506
4507     SAVEI32(PL_reg_flags);              /* from regexec.c */
4508     SAVEPPTR(PL_bostr);
4509     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4510     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4511     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4512     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4513     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4514     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4515     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4516     SAVEI8(PL_regprev);                 /* char before regbol, \n if none */
4517     SAVEVPTR(PL_reg_start_tmp);         /* from regexec.c */
4518     PL_reg_start_tmp = 0;
4519     SAVEFREEPV(PL_reg_start_tmp);
4520     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4521     PL_reg_start_tmpl = 0;
4522     SAVEVPTR(PL_regdata);
4523     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4524     SAVEI32(PL_regnarrate);             /* from regexec.c */
4525     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4526     SAVEINT(PL_regindent);              /* from regexec.c */
4527     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4528     SAVEVPTR(PL_curcop);
4529     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4530     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4531     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4532     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4533     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4534     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4535     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4536     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4537     SAVEI32(PL_regnpar);                /* () count. */
4538 #ifdef DEBUGGING
4539     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */    
4540 #endif
4541 }
4542
4543 #ifdef PERL_OBJECT
4544 #include "XSUB.h"
4545 #undef this
4546 #define this pPerl
4547 #endif
4548
4549 static void
4550 clear_re(pTHXo_ void *r)
4551 {
4552     ReREFCNT_dec((regexp *)r);
4553 }