Re: [PATCH regcomp.[ch], regexp.h, pp_hot.c, t/op/re_tests] lazy $& and \Z fix
[p5sagit/p5-mst-13.2.git] / regcomp.c
1 /*    regcomp.c
2  */
3
4 /*
5  * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
6  */
7
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9  * confused with the original package (see point 3 below).  Thanks, Henry!
10  */
11
12 /* Additional note: this code is very heavily munged from Henry's version
13  * in places.  In some spots I've traded clarity for efficiency, so don't
14  * blame Henry for some of the lack of readability.
15  */
16
17 /* The names of the functions have been changed from regcomp and
18  * regexec to  pregcomp and pregexec in order to avoid conflicts
19  * with the POSIX routines of the same names.
20 */
21
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 #  ifndef PERL_IN_XSUB_RE
25 #    define PERL_IN_XSUB_RE
26 #  endif
27 /* need access to debugger hooks */
28 #  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
29 #    define DEBUGGING
30 #  endif
31 #endif
32
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 #  define Perl_pregcomp my_regcomp
36 #  define Perl_regdump my_regdump
37 #  define Perl_regprop my_regprop
38 #  define Perl_pregfree my_regfree
39 #  define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 #  define Perl_regnext my_regnext
42 #  define Perl_save_re_context my_save_re_context
43 #  define Perl_reginitcolors my_reginitcolors
44
45 #  define PERL_NO_GET_CONTEXT
46 #endif
47
48 /*SUPPRESS 112*/
49 /*
50  * pregcomp and pregexec -- regsub and regerror are not used in perl
51  *
52  *      Copyright (c) 1986 by University of Toronto.
53  *      Written by Henry Spencer.  Not derived from licensed software.
54  *
55  *      Permission is granted to anyone to use this software for any
56  *      purpose on any computer system, and to redistribute it freely,
57  *      subject to the following restrictions:
58  *
59  *      1. The author is not responsible for the consequences of use of
60  *              this software, no matter how awful, even if they arise
61  *              from defects in it.
62  *
63  *      2. The origin of this software must not be misrepresented, either
64  *              by explicit claim or by omission.
65  *
66  *      3. Altered versions must be plainly marked as such, and must not
67  *              be misrepresented as being the original software.
68  *
69  *
70  ****    Alterations to Henry's code are...
71  ****
72  ****    Copyright (c) 1991-2001, Larry Wall
73  ****
74  ****    You may distribute under the terms of either the GNU General Public
75  ****    License or the Artistic License, as specified in the README file.
76
77  *
78  * Beware that some of this code is subtly aware of the way operator
79  * precedence is structured in regular expressions.  Serious changes in
80  * regular-expression syntax might require a total rethink.
81  */
82 #include "EXTERN.h"
83 #define PERL_IN_REGCOMP_C
84 #include "perl.h"
85
86 #ifndef PERL_IN_XSUB_RE
87 #  include "INTERN.h"
88 #endif
89
90 #define REG_COMP_C
91 #include "regcomp.h"
92
93 #ifdef op
94 #undef op
95 #endif /* op */
96
97 #ifdef MSDOS
98 # if defined(BUGGY_MSC6)
99  /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
100  # pragma optimize("a",off)
101  /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
102  # pragma optimize("w",on )
103 # endif /* BUGGY_MSC6 */
104 #endif /* MSDOS */
105
106 #ifndef STATIC
107 #define STATIC  static
108 #endif
109
110 typedef struct RExC_state_t {
111     U16         flags16;                /* are we folding, multilining? */
112     char        *precomp;               /* uncompiled string. */
113     regexp      *rx;
114     char        *start;                 /* Start of input for compile */
115     char        *end;                   /* End of input for compile */
116     char        *parse;                 /* Input-scan pointer. */
117     I32         whilem_seen;            /* number of WHILEM in this expr */
118     regnode     *emit_start;            /* Start of emitted-code area */
119     regnode     *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
120     I32         naughty;                /* How bad is this pattern? */
121     I32         sawback;                /* Did we see \1, ...? */
122     U32         seen;
123     I32         size;                   /* Code size. */
124     I32         npar;                   /* () count. */
125     I32         extralen;
126     I32         seen_zerolen;
127     I32         seen_evals;
128     I32         utf8;
129 #if ADD_TO_REGEXEC
130     char        *starttry;              /* -Dr: where regtry was called. */
131 #define RExC_starttry   (pRExC_state->starttry)
132 #endif
133 } RExC_state_t;
134
135 #define RExC_flags16    (pRExC_state->flags16)
136 #define RExC_precomp    (pRExC_state->precomp)
137 #define RExC_rx         (pRExC_state->rx)
138 #define RExC_start      (pRExC_state->start)
139 #define RExC_end        (pRExC_state->end)
140 #define RExC_parse      (pRExC_state->parse)
141 #define RExC_whilem_seen        (pRExC_state->whilem_seen)
142 #define RExC_offsets    (pRExC_state->rx->offsets) /* I am not like the others */
143 #define RExC_emit       (pRExC_state->emit)
144 #define RExC_emit_start (pRExC_state->emit_start)
145 #define RExC_naughty    (pRExC_state->naughty)
146 #define RExC_sawback    (pRExC_state->sawback)
147 #define RExC_seen       (pRExC_state->seen)
148 #define RExC_size       (pRExC_state->size)
149 #define RExC_npar       (pRExC_state->npar)
150 #define RExC_extralen   (pRExC_state->extralen)
151 #define RExC_seen_zerolen       (pRExC_state->seen_zerolen)
152 #define RExC_seen_evals (pRExC_state->seen_evals)
153 #define RExC_utf8       (pRExC_state->utf8)
154
155 #define ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
156 #define ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
157         ((*s) == '{' && regcurly(s)))
158
159 #ifdef SPSTART
160 #undef SPSTART          /* dratted cpp namespace... */
161 #endif
162 /*
163  * Flags to be passed up and down.
164  */
165 #define WORST           0       /* Worst case. */
166 #define HASWIDTH        0x1     /* Known to match non-null strings. */
167 #define SIMPLE          0x2     /* Simple enough to be STAR/PLUS operand. */
168 #define SPSTART         0x4     /* Starts with * or +. */
169 #define TRYAGAIN        0x8     /* Weeded out a declaration. */
170
171 /* Length of a variant. */
172
173 typedef struct scan_data_t {
174     I32 len_min;
175     I32 len_delta;
176     I32 pos_min;
177     I32 pos_delta;
178     SV *last_found;
179     I32 last_end;                       /* min value, <0 unless valid. */
180     I32 last_start_min;
181     I32 last_start_max;
182     SV **longest;                       /* Either &l_fixed, or &l_float. */
183     SV *longest_fixed;
184     I32 offset_fixed;
185     SV *longest_float;
186     I32 offset_float_min;
187     I32 offset_float_max;
188     I32 flags;
189     I32 whilem_c;
190     I32 *last_closep;
191     struct regnode_charclass_class *start_class;
192 } scan_data_t;
193
194 /*
195  * Forward declarations for pregcomp()'s friends.
196  */
197
198 static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
199                                       0, 0, 0, 0, 0, 0};
200
201 #define SF_BEFORE_EOL           (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
202 #define SF_BEFORE_SEOL          0x1
203 #define SF_BEFORE_MEOL          0x2
204 #define SF_FIX_BEFORE_EOL       (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
205 #define SF_FL_BEFORE_EOL        (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
206
207 #ifdef NO_UNARY_PLUS
208 #  define SF_FIX_SHIFT_EOL      (0+2)
209 #  define SF_FL_SHIFT_EOL               (0+4)
210 #else
211 #  define SF_FIX_SHIFT_EOL      (+2)
212 #  define SF_FL_SHIFT_EOL               (+4)
213 #endif
214
215 #define SF_FIX_BEFORE_SEOL      (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
216 #define SF_FIX_BEFORE_MEOL      (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
217
218 #define SF_FL_BEFORE_SEOL       (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
219 #define SF_FL_BEFORE_MEOL       (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
220 #define SF_IS_INF               0x40
221 #define SF_HAS_PAR              0x80
222 #define SF_IN_PAR               0x100
223 #define SF_HAS_EVAL             0x200
224 #define SCF_DO_SUBSTR           0x400
225 #define SCF_DO_STCLASS_AND      0x0800
226 #define SCF_DO_STCLASS_OR       0x1000
227 #define SCF_DO_STCLASS          (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
228 #define SCF_WHILEM_VISITED_POS  0x2000
229
230 #define UTF RExC_utf8
231 #define LOC (RExC_flags16 & PMf_LOCALE)
232 #define FOLD (RExC_flags16 & PMf_FOLD)
233
234 #define OOB_UNICODE             12345678
235 #define OOB_NAMEDCLASS          -1
236
237 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
238 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
239
240
241 /* length of regex to show in messages that don't mark a position within */
242 #define RegexLengthToShowInErrorMessages 127
243
244 /*
245  * If MARKER[12] are adjusted, be sure to adjust the constants at the top
246  * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
247  * op/pragma/warn/regcomp.
248  */
249 #define MARKER1 "<-- HERE"    /* marker as it appears in the description */
250 #define MARKER2 " <-- HERE "  /* marker as it appears within the regex */
251
252 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
253
254 /*
255  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
256  * arg. Show regex, up to a maximum length. If it's too long, chop and add
257  * "...".
258  */
259 #define FAIL(msg)                                                             \
260     STMT_START {                                                             \
261         char *ellipses = "";                                                 \
262         IV len = RExC_end - RExC_precomp;                                \
263                                                                              \
264         if (!SIZE_ONLY)                                                      \
265             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
266                                                                              \
267         if (len > RegexLengthToShowInErrorMessages) {                        \
268             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
269             len = RegexLengthToShowInErrorMessages - 10;                     \
270             ellipses = "...";                                                \
271         }                                                                    \
272         Perl_croak(aTHX_ "%s in regex m/%.*s%s/",                            \
273                    msg, (int)len, RExC_precomp, ellipses);                  \
274     } STMT_END
275
276 /*
277  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
278  * args. Show regex, up to a maximum length. If it's too long, chop and add
279  * "...".
280  */
281 #define FAIL2(pat,msg)                                                        \
282     STMT_START {                                                             \
283         char *ellipses = "";                                                 \
284         IV len = RExC_end - RExC_precomp;                                \
285                                                                              \
286         if (!SIZE_ONLY)                                                      \
287             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
288                                                                              \
289         if (len > RegexLengthToShowInErrorMessages) {                        \
290             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
291             len = RegexLengthToShowInErrorMessages - 10;                     \
292             ellipses = "...";                                                \
293         }                                                                    \
294         S_re_croak2(aTHX_ pat, " in regex m/%.*s%s/",                        \
295                     msg, (int)len, RExC_precomp, ellipses);                \
296     } STMT_END
297
298
299 /*
300  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
301  */
302 #define Simple_vFAIL(m)                                                      \
303     STMT_START {                                                             \
304       IV offset = RExC_parse - RExC_precomp; \
305                                                                              \
306       Perl_croak(aTHX_ "%s" REPORT_LOCATION,               \
307                  m, (int)offset, RExC_precomp, RExC_precomp + offset);     \
308     } STMT_END
309
310 /*
311  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
312  */
313 #define vFAIL(m)                                                             \
314     STMT_START {                                                             \
315       if (!SIZE_ONLY)                                                        \
316             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
317       Simple_vFAIL(m);                                                       \
318     } STMT_END
319
320 /*
321  * Like Simple_vFAIL(), but accepts two arguments.
322  */
323 #define Simple_vFAIL2(m,a1)                                                  \
324     STMT_START {                                                             \
325       IV offset = RExC_parse - RExC_precomp; \
326                                                                              \
327       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1,       \
328                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
329     } STMT_END
330
331 /*
332  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
333  */
334 #define vFAIL2(m,a1)                                                         \
335     STMT_START {                                                             \
336       if (!SIZE_ONLY)                                                        \
337             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
338       Simple_vFAIL2(m, a1);                                                  \
339     } STMT_END
340
341
342 /*
343  * Like Simple_vFAIL(), but accepts three arguments.
344  */
345 #define Simple_vFAIL3(m, a1, a2)                                             \
346     STMT_START {                                                             \
347       IV offset = RExC_parse - RExC_precomp; \
348                                                                              \
349       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2,   \
350                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
351     } STMT_END
352
353 /*
354  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
355  */
356 #define vFAIL3(m,a1,a2)                                                      \
357     STMT_START {                                                             \
358       if (!SIZE_ONLY)                                                        \
359             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
360       Simple_vFAIL3(m, a1, a2);                                              \
361     } STMT_END
362
363 /*
364  * Like Simple_vFAIL(), but accepts four arguments.
365  */
366 #define Simple_vFAIL4(m, a1, a2, a3)                                         \
367     STMT_START {                                                             \
368       IV offset = RExC_parse - RExC_precomp; \
369                                                                              \
370       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3,\
371                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
372     } STMT_END
373
374 /*
375  * Like Simple_vFAIL(), but accepts five arguments.
376  */
377 #define Simple_vFAIL5(m, a1, a2, a3, a4)                                     \
378     STMT_START {                                                             \
379       IV offset = RExC_parse - RExC_precomp; \
380       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, a4,\
381                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
382     } STMT_END
383
384
385 #define vWARN(loc,m)                                                         \
386     STMT_START {                                                             \
387         IV offset = loc - RExC_precomp;          \
388         Perl_warner(aTHX_ WARN_REGEXP, "%s" REPORT_LOCATION,\
389                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
390     } STMT_END                                                               \
391
392 #define vWARNdep(loc,m)                                                         \
393     STMT_START {                                                             \
394         IV offset = loc - RExC_precomp;          \
395         int warn_cat = ckWARN(WARN_REGEXP) ? WARN_REGEXP : WARN_DEPRECATED;  \
396         Perl_warner(aTHX_ warn_cat, "%s" REPORT_LOCATION,\
397                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
398     } STMT_END                                                               \
399
400
401 #define vWARN2(loc, m, a1)                                                   \
402     STMT_START {                                                             \
403         IV offset = loc - RExC_precomp;          \
404         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
405                  a1,                                                         \
406                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
407     } STMT_END
408
409 #define vWARN3(loc, m, a1, a2)                                               \
410     STMT_START {                                                             \
411       IV offset = loc - RExC_precomp;        \
412         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,                    \
413                  a1, a2,                                                     \
414                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
415     } STMT_END
416
417 #define vWARN4(loc, m, a1, a2, a3)                                           \
418     STMT_START {                                                             \
419       IV offset = loc - RExC_precomp;            \
420         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
421                  a1, a2, a3,                                                 \
422                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
423     } STMT_END
424
425 /* used for the parse_flags section for (?c) -- japhy */
426 #define vWARN5(loc, m, a1, a2, a3, a4)                                       \
427   STMT_START {                                                   \
428       IV offset = loc - RExC_precomp;   \
429         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,      \
430                  a1, a2, a3, a4,                                 \
431                  (int)offset, RExC_precomp, RExC_precomp + offset);  \
432     } STMT_END
433
434
435 /* Allow for side effects in s */
436 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (void)(s);} STMT_END
437
438 /* Macros for recording node offsets.   20001227 mjd@plover.com 
439  * Nodes are numbered 1, 2, 3, 4.  Node #n's position is recorded in
440  * element 2*n-1 of the array.  Element #2n holds the byte length node #n.
441  * Element 0 holds the number n.
442  */
443
444 #define MJD_OFFSET_DEBUG(x)
445 /* #define MJD_OFFSET_DEBUG(x) fprintf x */
446
447
448 #  define Set_Node_Offset_To_R(node,byte)                           \
449    STMT_START {                                        \
450      if (! SIZE_ONLY) {                                  \
451        if((node) < 0) {                   \
452          Perl_croak(aTHX_ "value of node is %d in Offset macro", node); \
453        } else {                                                        \
454          RExC_offsets[2*(node)-1] = (byte);                               \
455        }                                                               \
456      }                                                                 \
457    } STMT_END
458
459 #  define Set_Node_Offset(node,byte) Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
460 #  define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
461
462 #  define Set_Node_Length_To_R(node,len)                            \
463    STMT_START {                                        \
464      if (! SIZE_ONLY) {                                  \
465        MJD_OFFSET_DEBUG((stderr, "** (%d) size of node %d is %d.\n", __LINE__, (node), (len))); \
466        if((node) < 0) {                   \
467          Perl_croak(aTHX_ "value of node is %d in Length macro", node); \
468        } else {                                                        \
469          RExC_offsets[2*(node)] = (len);                               \
470        }                                                               \
471      }                                                                 \
472    } STMT_END
473
474 #  define Set_Node_Length(node,len) Set_Node_Length_To_R((node)-RExC_emit_start, len)
475 #  define Set_Cur_Node_Length(len)  Set_Node_Length(RExC_emit, len)
476 #  define Set_Node_Cur_Length(node)   Set_Node_Length(node, RExC_parse - parse_start)
477
478 /* Get offsets and lengths */
479 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
480 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
481
482 static void clear_re(pTHX_ void *r);
483
484 /* Mark that we cannot extend a found fixed substring at this point.
485    Updata the longest found anchored substring and the longest found
486    floating substrings if needed. */
487
488 STATIC void
489 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
490 {
491     STRLEN l = CHR_SVLEN(data->last_found);
492     STRLEN old_l = CHR_SVLEN(*data->longest);
493
494     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
495         sv_setsv(*data->longest, data->last_found);
496         if (*data->longest == data->longest_fixed) {
497             data->offset_fixed = l ? data->last_start_min : data->pos_min;
498             if (data->flags & SF_BEFORE_EOL)
499                 data->flags
500                     |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
501             else
502                 data->flags &= ~SF_FIX_BEFORE_EOL;
503         }
504         else {
505             data->offset_float_min = l ? data->last_start_min : data->pos_min;
506             data->offset_float_max = (l
507                                       ? data->last_start_max
508                                       : data->pos_min + data->pos_delta);
509             if (data->flags & SF_BEFORE_EOL)
510                 data->flags
511                     |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
512             else
513                 data->flags &= ~SF_FL_BEFORE_EOL;
514         }
515     }
516     SvCUR_set(data->last_found, 0);
517     data->last_end = -1;
518     data->flags &= ~SF_BEFORE_EOL;
519 }
520
521 /* Can match anything (initialization) */
522 STATIC void
523 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
524 {
525     int value;
526
527     ANYOF_CLASS_ZERO(cl);
528     for (value = 0; value < 256; ++value)
529         ANYOF_BITMAP_SET(cl, value);
530     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
531     if (LOC)
532         cl->flags |= ANYOF_LOCALE;
533 }
534
535 /* Can match anything (initialization) */
536 STATIC int
537 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
538 {
539     int value;
540
541     for (value = 0; value <= ANYOF_MAX; value += 2)
542         if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
543             return 1;
544     if (!(cl->flags & ANYOF_UNICODE_ALL))
545         return 0;
546     for (value = 0; value < 256; ++value)
547         if (!ANYOF_BITMAP_TEST(cl, value))
548             return 0;
549     return 1;
550 }
551
552 /* Can match anything (initialization) */
553 STATIC void
554 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
555 {
556     Zero(cl, 1, struct regnode_charclass_class);
557     cl->type = ANYOF;
558     cl_anything(pRExC_state, cl);
559 }
560
561 STATIC void
562 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
563 {
564     Zero(cl, 1, struct regnode_charclass_class);
565     cl->type = ANYOF;
566     cl_anything(pRExC_state, cl);
567     if (LOC)
568         cl->flags |= ANYOF_LOCALE;
569 }
570
571 /* 'And' a given class with another one.  Can create false positives */
572 /* We assume that cl is not inverted */
573 STATIC void
574 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
575          struct regnode_charclass_class *and_with)
576 {
577     if (!(and_with->flags & ANYOF_CLASS)
578         && !(cl->flags & ANYOF_CLASS)
579         && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
580         && !(and_with->flags & ANYOF_FOLD)
581         && !(cl->flags & ANYOF_FOLD)) {
582         int i;
583
584         if (and_with->flags & ANYOF_INVERT)
585             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
586                 cl->bitmap[i] &= ~and_with->bitmap[i];
587         else
588             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
589                 cl->bitmap[i] &= and_with->bitmap[i];
590     } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
591     if (!(and_with->flags & ANYOF_EOS))
592         cl->flags &= ~ANYOF_EOS;
593
594     if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
595         cl->flags &= ~ANYOF_UNICODE_ALL;
596         cl->flags |= ANYOF_UNICODE;
597         ARG_SET(cl, ARG(and_with));
598     }
599     if (!(and_with->flags & ANYOF_UNICODE_ALL))
600         cl->flags &= ~ANYOF_UNICODE_ALL;
601     if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
602         cl->flags &= ~ANYOF_UNICODE;
603 }
604
605 /* 'OR' a given class with another one.  Can create false positives */
606 /* We assume that cl is not inverted */
607 STATIC void
608 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
609 {
610     if (or_with->flags & ANYOF_INVERT) {
611         /* We do not use
612          * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
613          *   <= (B1 | !B2) | (CL1 | !CL2)
614          * which is wasteful if CL2 is small, but we ignore CL2:
615          *   (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
616          * XXXX Can we handle case-fold?  Unclear:
617          *   (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
618          *   (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
619          */
620         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
621              && !(or_with->flags & ANYOF_FOLD)
622              && !(cl->flags & ANYOF_FOLD) ) {
623             int i;
624
625             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
626                 cl->bitmap[i] |= ~or_with->bitmap[i];
627         } /* XXXX: logic is complicated otherwise */
628         else {
629             cl_anything(pRExC_state, cl);
630         }
631     } else {
632         /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
633         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
634              && (!(or_with->flags & ANYOF_FOLD)
635                  || (cl->flags & ANYOF_FOLD)) ) {
636             int i;
637
638             /* OR char bitmap and class bitmap separately */
639             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
640                 cl->bitmap[i] |= or_with->bitmap[i];
641             if (or_with->flags & ANYOF_CLASS) {
642                 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
643                     cl->classflags[i] |= or_with->classflags[i];
644                 cl->flags |= ANYOF_CLASS;
645             }
646         }
647         else { /* XXXX: logic is complicated, leave it along for a moment. */
648             cl_anything(pRExC_state, cl);
649         }
650     }
651     if (or_with->flags & ANYOF_EOS)
652         cl->flags |= ANYOF_EOS;
653
654     if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
655         ARG(cl) != ARG(or_with)) {
656         cl->flags |= ANYOF_UNICODE_ALL;
657         cl->flags &= ~ANYOF_UNICODE;
658     }
659     if (or_with->flags & ANYOF_UNICODE_ALL) {
660         cl->flags |= ANYOF_UNICODE_ALL;
661         cl->flags &= ~ANYOF_UNICODE;
662     }
663 }
664
665 /* REx optimizer.  Converts nodes into quickier variants "in place".
666    Finds fixed substrings.  */
667
668 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
669    to the position after last scanned or to NULL. */
670
671 STATIC I32
672 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
673                         /* scanp: Start here (read-write). */
674                         /* deltap: Write maxlen-minlen here. */
675                         /* last: Stop before this one. */
676 {
677     I32 min = 0, pars = 0, code;
678     regnode *scan = *scanp, *next;
679     I32 delta = 0;
680     int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
681     int is_inf_internal = 0;            /* The studied chunk is infinite */
682     I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
683     scan_data_t data_fake;
684     struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
685
686     while (scan && OP(scan) != END && scan < last) {
687         /* Peephole optimizer: */
688
689         if (PL_regkind[(U8)OP(scan)] == EXACT) {
690             /* Merge several consecutive EXACTish nodes into one. */
691             regnode *n = regnext(scan);
692             U32 stringok = 1;
693 #ifdef DEBUGGING
694             regnode *stop = scan;
695 #endif
696
697             next = scan + NODE_SZ_STR(scan);
698             /* Skip NOTHING, merge EXACT*. */
699             while (n &&
700                    ( PL_regkind[(U8)OP(n)] == NOTHING ||
701                      (stringok && (OP(n) == OP(scan))))
702                    && NEXT_OFF(n)
703                    && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
704                 if (OP(n) == TAIL || n > next)
705                     stringok = 0;
706                 if (PL_regkind[(U8)OP(n)] == NOTHING) {
707                     NEXT_OFF(scan) += NEXT_OFF(n);
708                     next = n + NODE_STEP_REGNODE;
709 #ifdef DEBUGGING
710                     if (stringok)
711                         stop = n;
712 #endif
713                     n = regnext(n);
714                 }
715                 else if (stringok) {
716                     int oldl = STR_LEN(scan);
717                     regnode *nnext = regnext(n);
718
719                     if (oldl + STR_LEN(n) > U8_MAX)
720                         break;
721                     NEXT_OFF(scan) += NEXT_OFF(n);
722                     STR_LEN(scan) += STR_LEN(n);
723                     next = n + NODE_SZ_STR(n);
724                     /* Now we can overwrite *n : */
725                     Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
726 #ifdef DEBUGGING
727                     stop = next - 1;
728 #endif
729                     n = nnext;
730                 }
731             }
732 #ifdef DEBUGGING
733             /* Allow dumping */
734             n = scan + NODE_SZ_STR(scan);
735             while (n <= stop) {
736                 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
737                     OP(n) = OPTIMIZED;
738                     NEXT_OFF(n) = 0;
739                 }
740                 n++;
741             }
742 #endif
743         }
744         /* Follow the next-chain of the current node and optimize
745            away all the NOTHINGs from it.  */
746         if (OP(scan) != CURLYX) {
747             int max = (reg_off_by_arg[OP(scan)]
748                        ? I32_MAX
749                        /* I32 may be smaller than U16 on CRAYs! */
750                        : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
751             int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
752             int noff;
753             regnode *n = scan;
754         
755             /* Skip NOTHING and LONGJMP. */
756             while ((n = regnext(n))
757                    && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
758                        || ((OP(n) == LONGJMP) && (noff = ARG(n))))
759                    && off + noff < max)
760                 off += noff;
761             if (reg_off_by_arg[OP(scan)])
762                 ARG(scan) = off;
763             else
764                 NEXT_OFF(scan) = off;
765         }
766         /* The principal pseudo-switch.  Cannot be a switch, since we
767            look into several different things.  */
768         if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
769                    || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
770             next = regnext(scan);
771             code = OP(scan);
772         
773             if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
774                 I32 max1 = 0, min1 = I32_MAX, num = 0;
775                 struct regnode_charclass_class accum;
776                 
777                 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
778                     scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
779                 if (flags & SCF_DO_STCLASS)
780                     cl_init_zero(pRExC_state, &accum);
781                 while (OP(scan) == code) {
782                     I32 deltanext, minnext, f = 0, fake;
783                     struct regnode_charclass_class this_class;
784
785                     num++;
786                     data_fake.flags = 0;
787                     if (data) {         
788                         data_fake.whilem_c = data->whilem_c;
789                         data_fake.last_closep = data->last_closep;
790                     }
791                     else
792                         data_fake.last_closep = &fake;
793                     next = regnext(scan);
794                     scan = NEXTOPER(scan);
795                     if (code != BRANCH)
796                         scan = NEXTOPER(scan);
797                     if (flags & SCF_DO_STCLASS) {
798                         cl_init(pRExC_state, &this_class);
799                         data_fake.start_class = &this_class;
800                         f = SCF_DO_STCLASS_AND;
801                     }           
802                     if (flags & SCF_WHILEM_VISITED_POS)
803                         f |= SCF_WHILEM_VISITED_POS;
804                     /* we suppose the run is continuous, last=next...*/
805                     minnext = study_chunk(pRExC_state, &scan, &deltanext,
806                                           next, &data_fake, f);
807                     if (min1 > minnext)
808                         min1 = minnext;
809                     if (max1 < minnext + deltanext)
810                         max1 = minnext + deltanext;
811                     if (deltanext == I32_MAX)
812                         is_inf = is_inf_internal = 1;
813                     scan = next;
814                     if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
815                         pars++;
816                     if (data && (data_fake.flags & SF_HAS_EVAL))
817                         data->flags |= SF_HAS_EVAL;
818                     if (data)
819                         data->whilem_c = data_fake.whilem_c;
820                     if (flags & SCF_DO_STCLASS)
821                         cl_or(pRExC_state, &accum, &this_class);
822                     if (code == SUSPEND)
823                         break;
824                 }
825                 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
826                     min1 = 0;
827                 if (flags & SCF_DO_SUBSTR) {
828                     data->pos_min += min1;
829                     data->pos_delta += max1 - min1;
830                     if (max1 != min1 || is_inf)
831                         data->longest = &(data->longest_float);
832                 }
833                 min += min1;
834                 delta += max1 - min1;
835                 if (flags & SCF_DO_STCLASS_OR) {
836                     cl_or(pRExC_state, data->start_class, &accum);
837                     if (min1) {
838                         cl_and(data->start_class, &and_with);
839                         flags &= ~SCF_DO_STCLASS;
840                     }
841                 }
842                 else if (flags & SCF_DO_STCLASS_AND) {
843                     if (min1) {
844                         cl_and(data->start_class, &accum);
845                         flags &= ~SCF_DO_STCLASS;
846                     }
847                     else {
848                         /* Switch to OR mode: cache the old value of
849                          * data->start_class */
850                         StructCopy(data->start_class, &and_with,
851                                    struct regnode_charclass_class);
852                         flags &= ~SCF_DO_STCLASS_AND;
853                         StructCopy(&accum, data->start_class,
854                                    struct regnode_charclass_class);
855                         flags |= SCF_DO_STCLASS_OR;
856                         data->start_class->flags |= ANYOF_EOS;
857                     }
858                 }
859             }
860             else if (code == BRANCHJ)   /* single branch is optimized. */
861                 scan = NEXTOPER(NEXTOPER(scan));
862             else                        /* single branch is optimized. */
863                 scan = NEXTOPER(scan);
864             continue;
865         }
866         else if (OP(scan) == EXACT) {
867             I32 l = STR_LEN(scan);
868             UV uc = *((U8*)STRING(scan));
869             if (UTF) {
870                 U8 *s = (U8*)STRING(scan);
871                 l = utf8_length(s, s + l);
872                 uc = utf8_to_uvchr(s, NULL);
873             }
874             min += l;
875             if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
876                 /* The code below prefers earlier match for fixed
877                    offset, later match for variable offset.  */
878                 if (data->last_end == -1) { /* Update the start info. */
879                     data->last_start_min = data->pos_min;
880                     data->last_start_max = is_inf
881                         ? I32_MAX : data->pos_min + data->pos_delta;
882                 }
883                 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
884                 data->last_end = data->pos_min + l;
885                 data->pos_min += l; /* As in the first entry. */
886                 data->flags &= ~SF_BEFORE_EOL;
887             }
888             if (flags & SCF_DO_STCLASS_AND) {
889                 /* Check whether it is compatible with what we know already! */
890                 int compat = 1;
891
892                 if (uc >= 0x100 ||
893                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
894                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
895                     && (!(data->start_class->flags & ANYOF_FOLD)
896                         || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
897                     )
898                     compat = 0;
899                 ANYOF_CLASS_ZERO(data->start_class);
900                 ANYOF_BITMAP_ZERO(data->start_class);
901                 if (compat)
902                     ANYOF_BITMAP_SET(data->start_class, uc);
903                 data->start_class->flags &= ~ANYOF_EOS;
904                 if (uc < 0x100)
905                   data->start_class->flags &= ~ANYOF_UNICODE_ALL;
906             }
907             else if (flags & SCF_DO_STCLASS_OR) {
908                 /* false positive possible if the class is case-folded */
909                 if (uc < 0x100)
910                     ANYOF_BITMAP_SET(data->start_class, uc);
911                 else
912                     data->start_class->flags |= ANYOF_UNICODE_ALL;
913                 data->start_class->flags &= ~ANYOF_EOS;
914                 cl_and(data->start_class, &and_with);
915             }
916             flags &= ~SCF_DO_STCLASS;
917         }
918         else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
919             I32 l = STR_LEN(scan);
920             UV uc = *((U8*)STRING(scan));
921
922             /* Search for fixed substrings supports EXACT only. */
923             if (flags & SCF_DO_SUBSTR)
924                 scan_commit(pRExC_state, data);
925             if (UTF) {
926                 U8 *s = (U8 *)STRING(scan);
927                 l = utf8_length(s, s + l);
928                 uc = utf8_to_uvchr(s, NULL);
929             }
930             min += l;
931             if (data && (flags & SCF_DO_SUBSTR))
932                 data->pos_min += l;
933             if (flags & SCF_DO_STCLASS_AND) {
934                 /* Check whether it is compatible with what we know already! */
935                 int compat = 1;
936
937                 if (uc >= 0x100 ||
938                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
939                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
940                      && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
941                     compat = 0;
942                 ANYOF_CLASS_ZERO(data->start_class);
943                 ANYOF_BITMAP_ZERO(data->start_class);
944                 if (compat) {
945                     ANYOF_BITMAP_SET(data->start_class, uc);
946                     data->start_class->flags &= ~ANYOF_EOS;
947                     data->start_class->flags |= ANYOF_FOLD;
948                     if (OP(scan) == EXACTFL)
949                         data->start_class->flags |= ANYOF_LOCALE;
950                 }
951             }
952             else if (flags & SCF_DO_STCLASS_OR) {
953                 if (data->start_class->flags & ANYOF_FOLD) {
954                     /* false positive possible if the class is case-folded.
955                        Assume that the locale settings are the same... */
956                     if (uc < 0x100)
957                         ANYOF_BITMAP_SET(data->start_class, uc);
958                     data->start_class->flags &= ~ANYOF_EOS;
959                 }
960                 cl_and(data->start_class, &and_with);
961             }
962             flags &= ~SCF_DO_STCLASS;
963         }
964         else if (strchr((char*)PL_varies,OP(scan))) {
965             I32 mincount, maxcount, minnext, deltanext, fl = 0;
966             I32 f = flags, pos_before = 0;
967             regnode *oscan = scan;
968             struct regnode_charclass_class this_class;
969             struct regnode_charclass_class *oclass = NULL;
970             I32 next_is_eval = 0;
971
972             switch (PL_regkind[(U8)OP(scan)]) {
973             case WHILEM:                /* End of (?:...)* . */
974                 scan = NEXTOPER(scan);
975                 goto finish;
976             case PLUS:
977                 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
978                     next = NEXTOPER(scan);
979                     if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
980                         mincount = 1;
981                         maxcount = REG_INFTY;
982                         next = regnext(scan);
983                         scan = NEXTOPER(scan);
984                         goto do_curly;
985                     }
986                 }
987                 if (flags & SCF_DO_SUBSTR)
988                     data->pos_min++;
989                 min++;
990                 /* Fall through. */
991             case STAR:
992                 if (flags & SCF_DO_STCLASS) {
993                     mincount = 0;
994                     maxcount = REG_INFTY;
995                     next = regnext(scan);
996                     scan = NEXTOPER(scan);
997                     goto do_curly;
998                 }
999                 is_inf = is_inf_internal = 1;
1000                 scan = regnext(scan);
1001                 if (flags & SCF_DO_SUBSTR) {
1002                     scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
1003                     data->longest = &(data->longest_float);
1004                 }
1005                 goto optimize_curly_tail;
1006             case CURLY:
1007                 mincount = ARG1(scan);
1008                 maxcount = ARG2(scan);
1009                 next = regnext(scan);
1010                 if (OP(scan) == CURLYX) {
1011                     I32 lp = (data ? *(data->last_closep) : 0);
1012
1013                     scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
1014                 }
1015                 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
1016                 next_is_eval = (OP(scan) == EVAL);
1017               do_curly:
1018                 if (flags & SCF_DO_SUBSTR) {
1019                     if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
1020                     pos_before = data->pos_min;
1021                 }
1022                 if (data) {
1023                     fl = data->flags;
1024                     data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
1025                     if (is_inf)
1026                         data->flags |= SF_IS_INF;
1027                 }
1028                 if (flags & SCF_DO_STCLASS) {
1029                     cl_init(pRExC_state, &this_class);
1030                     oclass = data->start_class;
1031                     data->start_class = &this_class;
1032                     f |= SCF_DO_STCLASS_AND;
1033                     f &= ~SCF_DO_STCLASS_OR;
1034                 }
1035                 /* These are the cases when once a subexpression
1036                    fails at a particular position, it cannot succeed
1037                    even after backtracking at the enclosing scope.
1038                 
1039                    XXXX what if minimal match and we are at the
1040                         initial run of {n,m}? */
1041                 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
1042                     f &= ~SCF_WHILEM_VISITED_POS;
1043
1044                 /* This will finish on WHILEM, setting scan, or on NULL: */
1045                 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data,
1046                                       mincount == 0
1047                                         ? (f & ~SCF_DO_SUBSTR) : f);
1048
1049                 if (flags & SCF_DO_STCLASS)
1050                     data->start_class = oclass;
1051                 if (mincount == 0 || minnext == 0) {
1052                     if (flags & SCF_DO_STCLASS_OR) {
1053                         cl_or(pRExC_state, data->start_class, &this_class);
1054                     }
1055                     else if (flags & SCF_DO_STCLASS_AND) {
1056                         /* Switch to OR mode: cache the old value of
1057                          * data->start_class */
1058                         StructCopy(data->start_class, &and_with,
1059                                    struct regnode_charclass_class);
1060                         flags &= ~SCF_DO_STCLASS_AND;
1061                         StructCopy(&this_class, data->start_class,
1062                                    struct regnode_charclass_class);
1063                         flags |= SCF_DO_STCLASS_OR;
1064                         data->start_class->flags |= ANYOF_EOS;
1065                     }
1066                 } else {                /* Non-zero len */
1067                     if (flags & SCF_DO_STCLASS_OR) {
1068                         cl_or(pRExC_state, data->start_class, &this_class);
1069                         cl_and(data->start_class, &and_with);
1070                     }
1071                     else if (flags & SCF_DO_STCLASS_AND)
1072                         cl_and(data->start_class, &this_class);
1073                     flags &= ~SCF_DO_STCLASS;
1074                 }
1075                 if (!scan)              /* It was not CURLYX, but CURLY. */
1076                     scan = next;
1077                 if (ckWARN(WARN_REGEXP)
1078                        /* ? quantifier ok, except for (?{ ... }) */
1079                     && (next_is_eval || !(mincount == 0 && maxcount == 1))
1080                     && (minnext == 0) && (deltanext == 0)
1081                     && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1082                     && maxcount <= REG_INFTY/3) /* Complement check for big count */
1083                 {
1084                     vWARN(RExC_parse,
1085                           "Quantifier unexpected on zero-length expression");
1086                 }
1087
1088                 min += minnext * mincount;
1089                 is_inf_internal |= ((maxcount == REG_INFTY
1090                                      && (minnext + deltanext) > 0)
1091                                     || deltanext == I32_MAX);
1092                 is_inf |= is_inf_internal;
1093                 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1094
1095                 /* Try powerful optimization CURLYX => CURLYN. */
1096                 if (  OP(oscan) == CURLYX && data
1097                       && data->flags & SF_IN_PAR
1098                       && !(data->flags & SF_HAS_EVAL)
1099                       && !deltanext && minnext == 1 ) {
1100                     /* Try to optimize to CURLYN.  */
1101                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1102                     regnode *nxt1 = nxt;
1103 #ifdef DEBUGGING
1104                     regnode *nxt2;
1105 #endif
1106
1107                     /* Skip open. */
1108                     nxt = regnext(nxt);
1109                     if (!strchr((char*)PL_simple,OP(nxt))
1110                         && !(PL_regkind[(U8)OP(nxt)] == EXACT
1111                              && STR_LEN(nxt) == 1))
1112                         goto nogo;
1113 #ifdef DEBUGGING
1114                     nxt2 = nxt;
1115 #endif
1116                     nxt = regnext(nxt);
1117                     if (OP(nxt) != CLOSE)
1118                         goto nogo;
1119                     /* Now we know that nxt2 is the only contents: */
1120                     oscan->flags = ARG(nxt);
1121                     OP(oscan) = CURLYN;
1122                     OP(nxt1) = NOTHING; /* was OPEN. */
1123 #ifdef DEBUGGING
1124                     OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1125                     NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
1126                     NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
1127                     OP(nxt) = OPTIMIZED;        /* was CLOSE. */
1128                     OP(nxt + 1) = OPTIMIZED; /* was count. */
1129                     NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
1130 #endif
1131                 }
1132               nogo:
1133
1134                 /* Try optimization CURLYX => CURLYM. */
1135                 if (  OP(oscan) == CURLYX && data
1136                       && !(data->flags & SF_HAS_PAR)
1137                       && !(data->flags & SF_HAS_EVAL)
1138                       && !deltanext  ) {
1139                     /* XXXX How to optimize if data == 0? */
1140                     /* Optimize to a simpler form.  */
1141                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
1142                     regnode *nxt2;
1143
1144                     OP(oscan) = CURLYM;
1145                     while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
1146                             && (OP(nxt2) != WHILEM))
1147                         nxt = nxt2;
1148                     OP(nxt2)  = SUCCEED; /* Whas WHILEM */
1149                     /* Need to optimize away parenths. */
1150                     if (data->flags & SF_IN_PAR) {
1151                         /* Set the parenth number.  */
1152                         regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
1153
1154                         if (OP(nxt) != CLOSE)
1155                             FAIL("Panic opt close");
1156                         oscan->flags = ARG(nxt);
1157                         OP(nxt1) = OPTIMIZED;   /* was OPEN. */
1158                         OP(nxt) = OPTIMIZED;    /* was CLOSE. */
1159 #ifdef DEBUGGING
1160                         OP(nxt1 + 1) = OPTIMIZED; /* was count. */
1161                         OP(nxt + 1) = OPTIMIZED; /* was count. */
1162                         NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
1163                         NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
1164 #endif
1165 #if 0
1166                         while ( nxt1 && (OP(nxt1) != WHILEM)) {
1167                             regnode *nnxt = regnext(nxt1);
1168                         
1169                             if (nnxt == nxt) {
1170                                 if (reg_off_by_arg[OP(nxt1)])
1171                                     ARG_SET(nxt1, nxt2 - nxt1);
1172                                 else if (nxt2 - nxt1 < U16_MAX)
1173                                     NEXT_OFF(nxt1) = nxt2 - nxt1;
1174                                 else
1175                                     OP(nxt) = NOTHING;  /* Cannot beautify */
1176                             }
1177                             nxt1 = nnxt;
1178                         }
1179 #endif
1180                         /* Optimize again: */
1181                         study_chunk(pRExC_state, &nxt1, &deltanext, nxt,
1182                                     NULL, 0);
1183                     }
1184                     else
1185                         oscan->flags = 0;
1186                 }
1187                 else if ((OP(oscan) == CURLYX)
1188                          && (flags & SCF_WHILEM_VISITED_POS)
1189                          /* See the comment on a similar expression above.
1190                             However, this time it not a subexpression
1191                             we care about, but the expression itself. */
1192                          && (maxcount == REG_INFTY)
1193                          && data && ++data->whilem_c < 16) {
1194                     /* This stays as CURLYX, we can put the count/of pair. */
1195                     /* Find WHILEM (as in regexec.c) */
1196                     regnode *nxt = oscan + NEXT_OFF(oscan);
1197
1198                     if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
1199                         nxt += ARG(nxt);
1200                     PREVOPER(nxt)->flags = data->whilem_c
1201                         | (RExC_whilem_seen << 4); /* On WHILEM */
1202                 }
1203                 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
1204                     pars++;
1205                 if (flags & SCF_DO_SUBSTR) {
1206                     SV *last_str = Nullsv;
1207                     int counted = mincount != 0;
1208
1209                     if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
1210                         I32 b = pos_before >= data->last_start_min
1211                             ? pos_before : data->last_start_min;
1212                         STRLEN l;
1213                         char *s = SvPV(data->last_found, l);
1214                         I32 old = b - data->last_start_min;
1215
1216                         if (UTF)
1217                             old = utf8_hop((U8*)s, old) - (U8*)s;
1218                         
1219                         l -= old;
1220                         /* Get the added string: */
1221                         last_str = newSVpvn(s  + old, l);
1222                         if (deltanext == 0 && pos_before == b) {
1223                             /* What was added is a constant string */
1224                             if (mincount > 1) {
1225                                 SvGROW(last_str, (mincount * l) + 1);
1226                                 repeatcpy(SvPVX(last_str) + l,
1227                                           SvPVX(last_str), l, mincount - 1);
1228                                 SvCUR(last_str) *= mincount;
1229                                 /* Add additional parts. */
1230                                 SvCUR_set(data->last_found,
1231                                           SvCUR(data->last_found) - l);
1232                                 sv_catsv(data->last_found, last_str);
1233                                 data->last_end += l * (mincount - 1);
1234                             }
1235                         } else {
1236                             /* start offset must point into the last copy */
1237                             data->last_start_min += minnext * (mincount - 1);
1238                             data->last_start_max += is_inf ? 0 : (maxcount - 1)
1239                                 * (minnext + data->pos_delta);
1240                         }
1241                     }
1242                     /* It is counted once already... */
1243                     data->pos_min += minnext * (mincount - counted);
1244                     data->pos_delta += - counted * deltanext +
1245                         (minnext + deltanext) * maxcount - minnext * mincount;
1246                     if (mincount != maxcount) {
1247                          /* Cannot extend fixed substrings found inside
1248                             the group.  */
1249                         scan_commit(pRExC_state,data);
1250                         if (mincount && last_str) {
1251                             sv_setsv(data->last_found, last_str);
1252                             data->last_end = data->pos_min;
1253                             data->last_start_min =
1254                                 data->pos_min - CHR_SVLEN(last_str);
1255                             data->last_start_max = is_inf
1256                                 ? I32_MAX
1257                                 : data->pos_min + data->pos_delta
1258                                 - CHR_SVLEN(last_str);
1259                         }
1260                         data->longest = &(data->longest_float);
1261                     }
1262                     SvREFCNT_dec(last_str);
1263                 }
1264                 if (data && (fl & SF_HAS_EVAL))
1265                     data->flags |= SF_HAS_EVAL;
1266               optimize_curly_tail:
1267                 if (OP(oscan) != CURLYX) {
1268                     while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
1269                            && NEXT_OFF(next))
1270                         NEXT_OFF(oscan) += NEXT_OFF(next);
1271                 }
1272                 continue;
1273             default:                    /* REF and CLUMP only? */
1274                 if (flags & SCF_DO_SUBSTR) {
1275                     scan_commit(pRExC_state,data);      /* Cannot expect anything... */
1276                     data->longest = &(data->longest_float);
1277                 }
1278                 is_inf = is_inf_internal = 1;
1279                 if (flags & SCF_DO_STCLASS_OR)
1280                     cl_anything(pRExC_state, data->start_class);
1281                 flags &= ~SCF_DO_STCLASS;
1282                 break;
1283             }
1284         }
1285         else if (strchr((char*)PL_simple,OP(scan))) {
1286             int value = 0;
1287
1288             if (flags & SCF_DO_SUBSTR) {
1289                 scan_commit(pRExC_state,data);
1290                 data->pos_min++;
1291             }
1292             min++;
1293             if (flags & SCF_DO_STCLASS) {
1294                 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
1295
1296                 /* Some of the logic below assumes that switching
1297                    locale on will only add false positives. */
1298                 switch (PL_regkind[(U8)OP(scan)]) {
1299                 case SANY:
1300                 default:
1301                   do_default:
1302                     /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
1303                     if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1304                         cl_anything(pRExC_state, data->start_class);
1305                     break;
1306                 case REG_ANY:
1307                     if (OP(scan) == SANY)
1308                         goto do_default;
1309                     if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
1310                         value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
1311                                  || (data->start_class->flags & ANYOF_CLASS));
1312                         cl_anything(pRExC_state, data->start_class);
1313                     }
1314                     if (flags & SCF_DO_STCLASS_AND || !value)
1315                         ANYOF_BITMAP_CLEAR(data->start_class,'\n');
1316                     break;
1317                 case ANYOF:
1318                     if (flags & SCF_DO_STCLASS_AND)
1319                         cl_and(data->start_class,
1320                                (struct regnode_charclass_class*)scan);
1321                     else
1322                         cl_or(pRExC_state, data->start_class,
1323                               (struct regnode_charclass_class*)scan);
1324                     break;
1325                 case ALNUM:
1326                     if (flags & SCF_DO_STCLASS_AND) {
1327                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1328                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1329                             for (value = 0; value < 256; value++)
1330                                 if (!isALNUM(value))
1331                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1332                         }
1333                     }
1334                     else {
1335                         if (data->start_class->flags & ANYOF_LOCALE)
1336                             ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1337                         else {
1338                             for (value = 0; value < 256; value++)
1339                                 if (isALNUM(value))
1340                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1341                         }
1342                     }
1343                     break;
1344                 case ALNUML:
1345                     if (flags & SCF_DO_STCLASS_AND) {
1346                         if (data->start_class->flags & ANYOF_LOCALE)
1347                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
1348                     }
1349                     else {
1350                         ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
1351                         data->start_class->flags |= ANYOF_LOCALE;
1352                     }
1353                     break;
1354                 case NALNUM:
1355                     if (flags & SCF_DO_STCLASS_AND) {
1356                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1357                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1358                             for (value = 0; value < 256; value++)
1359                                 if (isALNUM(value))
1360                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1361                         }
1362                     }
1363                     else {
1364                         if (data->start_class->flags & ANYOF_LOCALE)
1365                             ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1366                         else {
1367                             for (value = 0; value < 256; value++)
1368                                 if (!isALNUM(value))
1369                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1370                         }
1371                     }
1372                     break;
1373                 case NALNUML:
1374                     if (flags & SCF_DO_STCLASS_AND) {
1375                         if (data->start_class->flags & ANYOF_LOCALE)
1376                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
1377                     }
1378                     else {
1379                         data->start_class->flags |= ANYOF_LOCALE;
1380                         ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
1381                     }
1382                     break;
1383                 case SPACE:
1384                     if (flags & SCF_DO_STCLASS_AND) {
1385                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1386                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1387                             for (value = 0; value < 256; value++)
1388                                 if (!isSPACE(value))
1389                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1390                         }
1391                     }
1392                     else {
1393                         if (data->start_class->flags & ANYOF_LOCALE)
1394                             ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1395                         else {
1396                             for (value = 0; value < 256; value++)
1397                                 if (isSPACE(value))
1398                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1399                         }
1400                     }
1401                     break;
1402                 case SPACEL:
1403                     if (flags & SCF_DO_STCLASS_AND) {
1404                         if (data->start_class->flags & ANYOF_LOCALE)
1405                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
1406                     }
1407                     else {
1408                         data->start_class->flags |= ANYOF_LOCALE;
1409                         ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
1410                     }
1411                     break;
1412                 case NSPACE:
1413                     if (flags & SCF_DO_STCLASS_AND) {
1414                         if (!(data->start_class->flags & ANYOF_LOCALE)) {
1415                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1416                             for (value = 0; value < 256; value++)
1417                                 if (isSPACE(value))
1418                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1419                         }
1420                     }
1421                     else {
1422                         if (data->start_class->flags & ANYOF_LOCALE)
1423                             ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1424                         else {
1425                             for (value = 0; value < 256; value++)
1426                                 if (!isSPACE(value))
1427                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1428                         }
1429                     }
1430                     break;
1431                 case NSPACEL:
1432                     if (flags & SCF_DO_STCLASS_AND) {
1433                         if (data->start_class->flags & ANYOF_LOCALE) {
1434                             ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
1435                             for (value = 0; value < 256; value++)
1436                                 if (!isSPACE(value))
1437                                     ANYOF_BITMAP_CLEAR(data->start_class, value);
1438                         }
1439                     }
1440                     else {
1441                         data->start_class->flags |= ANYOF_LOCALE;
1442                         ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
1443                     }
1444                     break;
1445                 case DIGIT:
1446                     if (flags & SCF_DO_STCLASS_AND) {
1447                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
1448                         for (value = 0; value < 256; value++)
1449                             if (!isDIGIT(value))
1450                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1451                     }
1452                     else {
1453                         if (data->start_class->flags & ANYOF_LOCALE)
1454                             ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
1455                         else {
1456                             for (value = 0; value < 256; value++)
1457                                 if (isDIGIT(value))
1458                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1459                         }
1460                     }
1461                     break;
1462                 case NDIGIT:
1463                     if (flags & SCF_DO_STCLASS_AND) {
1464                         ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
1465                         for (value = 0; value < 256; value++)
1466                             if (isDIGIT(value))
1467                                 ANYOF_BITMAP_CLEAR(data->start_class, value);
1468                     }
1469                     else {
1470                         if (data->start_class->flags & ANYOF_LOCALE)
1471                             ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
1472                         else {
1473                             for (value = 0; value < 256; value++)
1474                                 if (!isDIGIT(value))
1475                                     ANYOF_BITMAP_SET(data->start_class, value);                 
1476                         }
1477                     }
1478                     break;
1479                 }
1480                 if (flags & SCF_DO_STCLASS_OR)
1481                     cl_and(data->start_class, &and_with);
1482                 flags &= ~SCF_DO_STCLASS;
1483             }
1484         }
1485         else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
1486             data->flags |= (OP(scan) == MEOL
1487                             ? SF_BEFORE_MEOL
1488                             : SF_BEFORE_SEOL);
1489         }
1490         else if (  PL_regkind[(U8)OP(scan)] == BRANCHJ
1491                  /* Lookbehind, or need to calculate parens/evals/stclass: */
1492                    && (scan->flags || data || (flags & SCF_DO_STCLASS))
1493                    && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
1494             /* Lookahead/lookbehind */
1495             I32 deltanext, minnext, fake = 0;
1496             regnode *nscan;
1497             struct regnode_charclass_class intrnl;
1498             int f = 0;
1499
1500             data_fake.flags = 0;
1501             if (data) {         
1502                 data_fake.whilem_c = data->whilem_c;
1503                 data_fake.last_closep = data->last_closep;
1504             }
1505             else
1506                 data_fake.last_closep = &fake;
1507             if ( flags & SCF_DO_STCLASS && !scan->flags
1508                  && OP(scan) == IFMATCH ) { /* Lookahead */
1509                 cl_init(pRExC_state, &intrnl);
1510                 data_fake.start_class = &intrnl;
1511                 f |= SCF_DO_STCLASS_AND;
1512             }
1513             if (flags & SCF_WHILEM_VISITED_POS)
1514                 f |= SCF_WHILEM_VISITED_POS;
1515             next = regnext(scan);
1516             nscan = NEXTOPER(NEXTOPER(scan));
1517             minnext = study_chunk(pRExC_state, &nscan, &deltanext, last, &data_fake, f);
1518             if (scan->flags) {
1519                 if (deltanext) {
1520                     vFAIL("Variable length lookbehind not implemented");
1521                 }
1522                 else if (minnext > U8_MAX) {
1523                     vFAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
1524                 }
1525                 scan->flags = minnext;
1526             }
1527             if (data && data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
1528                 pars++;
1529             if (data && (data_fake.flags & SF_HAS_EVAL))
1530                 data->flags |= SF_HAS_EVAL;
1531             if (data)
1532                 data->whilem_c = data_fake.whilem_c;
1533             if (f & SCF_DO_STCLASS_AND) {
1534                 int was = (data->start_class->flags & ANYOF_EOS);
1535
1536                 cl_and(data->start_class, &intrnl);
1537                 if (was)
1538                     data->start_class->flags |= ANYOF_EOS;
1539             }
1540         }
1541         else if (OP(scan) == OPEN) {
1542             pars++;
1543         }
1544         else if (OP(scan) == CLOSE) {
1545             if (ARG(scan) == is_par) {
1546                 next = regnext(scan);
1547
1548                 if ( next && (OP(next) != WHILEM) && next < last)
1549                     is_par = 0;         /* Disable optimization */
1550             }
1551             if (data)
1552                 *(data->last_closep) = ARG(scan);
1553         }
1554         else if (OP(scan) == EVAL) {
1555                 if (data)
1556                     data->flags |= SF_HAS_EVAL;
1557         }
1558         else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
1559                 if (flags & SCF_DO_SUBSTR) {
1560                     scan_commit(pRExC_state,data);
1561                     data->longest = &(data->longest_float);
1562                 }
1563                 is_inf = is_inf_internal = 1;
1564                 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
1565                     cl_anything(pRExC_state, data->start_class);
1566                 flags &= ~SCF_DO_STCLASS;
1567         }
1568         /* Else: zero-length, ignore. */
1569         scan = regnext(scan);
1570     }
1571
1572   finish:
1573     *scanp = scan;
1574     *deltap = is_inf_internal ? I32_MAX : delta;
1575     if (flags & SCF_DO_SUBSTR && is_inf)
1576         data->pos_delta = I32_MAX - data->pos_min;
1577     if (is_par > U8_MAX)
1578         is_par = 0;
1579     if (is_par && pars==1 && data) {
1580         data->flags |= SF_IN_PAR;
1581         data->flags &= ~SF_HAS_PAR;
1582     }
1583     else if (pars && data) {
1584         data->flags |= SF_HAS_PAR;
1585         data->flags &= ~SF_IN_PAR;
1586     }
1587     if (flags & SCF_DO_STCLASS_OR)
1588         cl_and(data->start_class, &and_with);
1589     return min;
1590 }
1591
1592 STATIC I32
1593 S_add_data(pTHX_ RExC_state_t *pRExC_state, I32 n, char *s)
1594 {
1595     if (RExC_rx->data) {
1596         Renewc(RExC_rx->data,
1597                sizeof(*RExC_rx->data) + sizeof(void*) * (RExC_rx->data->count + n - 1),
1598                char, struct reg_data);
1599         Renew(RExC_rx->data->what, RExC_rx->data->count + n, U8);
1600         RExC_rx->data->count += n;
1601     }
1602     else {
1603         Newc(1207, RExC_rx->data, sizeof(*RExC_rx->data) + sizeof(void*) * (n - 1),
1604              char, struct reg_data);
1605         New(1208, RExC_rx->data->what, n, U8);
1606         RExC_rx->data->count = n;
1607     }
1608     Copy(s, RExC_rx->data->what + RExC_rx->data->count - n, n, U8);
1609     return RExC_rx->data->count - n;
1610 }
1611
1612 void
1613 Perl_reginitcolors(pTHX)
1614 {
1615     int i = 0;
1616     char *s = PerlEnv_getenv("PERL_RE_COLORS");
1617         
1618     if (s) {
1619         PL_colors[0] = s = savepv(s);
1620         while (++i < 6) {
1621             s = strchr(s, '\t');
1622             if (s) {
1623                 *s = '\0';
1624                 PL_colors[i] = ++s;
1625             }
1626             else
1627                 PL_colors[i] = s = "";
1628         }
1629     } else {
1630         while (i < 6)
1631             PL_colors[i++] = "";
1632     }
1633     PL_colorset = 1;
1634 }
1635
1636
1637 /*
1638  - pregcomp - compile a regular expression into internal code
1639  *
1640  * We can't allocate space until we know how big the compiled form will be,
1641  * but we can't compile it (and thus know how big it is) until we've got a
1642  * place to put the code.  So we cheat:  we compile it twice, once with code
1643  * generation turned off and size counting turned on, and once "for real".
1644  * This also means that we don't allocate space until we are sure that the
1645  * thing really will compile successfully, and we never have to move the
1646  * code and thus invalidate pointers into it.  (Note that it has to be in
1647  * one piece because free() must be able to free it all.) [NB: not true in perl]
1648  *
1649  * Beware that the optimization-preparation code in here knows about some
1650  * of the structure of the compiled regexp.  [I'll say.]
1651  */
1652 regexp *
1653 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
1654 {
1655     register regexp *r;
1656     regnode *scan;
1657     regnode *first;
1658     I32 flags;
1659     I32 minlen = 0;
1660     I32 sawplus = 0;
1661     I32 sawopen = 0;
1662     scan_data_t data;
1663     RExC_state_t RExC_state;
1664     RExC_state_t *pRExC_state = &RExC_state;
1665
1666     if (exp == NULL)
1667         FAIL("NULL regexp argument");
1668
1669     /* XXXX This looks very suspicious... */
1670     if (pm->op_pmdynflags & PMdf_CMP_UTF8)
1671         RExC_utf8 = 1;
1672     else
1673         RExC_utf8 = 0;
1674
1675     RExC_precomp = exp;
1676     DEBUG_r(if (!PL_colorset) reginitcolors());
1677     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling REx%s `%s%*s%s'\n",
1678                       PL_colors[4],PL_colors[5],PL_colors[0],
1679                       (int)(xend - exp), RExC_precomp, PL_colors[1]));
1680     RExC_flags16 = pm->op_pmflags;
1681     RExC_sawback = 0;
1682
1683     RExC_seen = 0;
1684     RExC_seen_zerolen = *exp == '^' ? -1 : 0;
1685     RExC_seen_evals = 0;
1686     RExC_extralen = 0;
1687
1688     /* First pass: determine size, legality. */
1689     RExC_parse = exp;
1690     RExC_start = exp;
1691     RExC_end = xend;
1692     RExC_naughty = 0;
1693     RExC_npar = 1;
1694     RExC_size = 0L;
1695     RExC_emit = &PL_regdummy;
1696     RExC_whilem_seen = 0;
1697 #if 0 /* REGC() is (currently) a NOP at the first pass.
1698        * Clever compilers notice this and complain. --jhi */
1699     REGC((U8)REG_MAGIC, (char*)RExC_emit);
1700 #endif
1701     if (reg(pRExC_state, 0, &flags) == NULL) {
1702         RExC_precomp = Nullch;
1703         return(NULL);
1704     }
1705     DEBUG_r(PerlIO_printf(Perl_debug_log, "size %"IVdf" ", (IV)RExC_size));
1706
1707     /* Small enough for pointer-storage convention?
1708        If extralen==0, this means that we will not need long jumps. */
1709     if (RExC_size >= 0x10000L && RExC_extralen)
1710         RExC_size += RExC_extralen;
1711     else
1712         RExC_extralen = 0;
1713     if (RExC_whilem_seen > 15)
1714         RExC_whilem_seen = 15;
1715
1716     /* Allocate space and initialize. */
1717     Newc(1001, r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode),
1718          char, regexp);
1719     if (r == NULL)
1720         FAIL("Regexp out of space");
1721
1722 #ifdef DEBUGGING
1723     /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
1724     Zero(r, sizeof(regexp) + (unsigned)RExC_size * sizeof(regnode), char);
1725 #endif
1726     r->refcnt = 1;
1727     r->prelen = xend - exp;
1728     r->precomp = savepvn(RExC_precomp, r->prelen);
1729     r->subbeg = NULL;
1730     r->reganch = pm->op_pmflags & PMf_COMPILETIME;
1731     r->nparens = RExC_npar - 1; /* set early to validate backrefs */
1732
1733     r->substrs = 0;                     /* Useful during FAIL. */
1734     r->startp = 0;                      /* Useful during FAIL. */
1735     r->endp = 0;                        /* Useful during FAIL. */
1736
1737     Newz(1304, r->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
1738     if (r->offsets) {
1739       r->offsets[0] = RExC_size; 
1740     }
1741     DEBUG_r(PerlIO_printf(Perl_debug_log, 
1742                           "%s %"UVuf" bytes for offset annotations.\n", 
1743                           r->offsets ? "Got" : "Couldn't get", 
1744                           (UV)((2*RExC_size+1) * sizeof(U32))));
1745
1746     RExC_rx = r;
1747
1748     /* Second pass: emit code. */
1749     RExC_flags16 = pm->op_pmflags;      /* don't let top level (?i) bleed */
1750     RExC_parse = exp;
1751     RExC_end = xend;
1752     RExC_naughty = 0;
1753     RExC_npar = 1;
1754     RExC_emit_start = r->program;
1755     RExC_emit = r->program;
1756     /* Store the count of eval-groups for security checks: */
1757     RExC_emit->next_off = ((RExC_seen_evals > U16_MAX) ? U16_MAX : RExC_seen_evals);
1758     REGC((U8)REG_MAGIC, (char*) RExC_emit++);
1759     r->data = 0;
1760     if (reg(pRExC_state, 0, &flags) == NULL)
1761         return(NULL);
1762
1763     /* Dig out information for optimizations. */
1764     r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
1765     pm->op_pmflags = RExC_flags16;
1766     if (UTF)
1767         r->reganch |= ROPT_UTF8;        /* Unicode in it? */
1768     r->regstclass = NULL;
1769     if (RExC_naughty >= 10)     /* Probably an expensive pattern. */
1770         r->reganch |= ROPT_NAUGHTY;
1771     scan = r->program + 1;              /* First BRANCH. */
1772
1773     /* XXXX To minimize changes to RE engine we always allocate
1774        3-units-long substrs field. */
1775     Newz(1004, r->substrs, 1, struct reg_substr_data);
1776
1777     StructCopy(&zero_scan_data, &data, scan_data_t);
1778     /* XXXX Should not we check for something else?  Usually it is OPEN1... */
1779     if (OP(scan) != BRANCH) {   /* Only one top-level choice. */
1780         I32 fake;
1781         STRLEN longest_float_length, longest_fixed_length;
1782         struct regnode_charclass_class ch_class;
1783         int stclass_flag;
1784         I32 last_close = 0;
1785
1786         first = scan;
1787         /* Skip introductions and multiplicators >= 1. */
1788         while ((OP(first) == OPEN && (sawopen = 1)) ||
1789                /* An OR of *one* alternative - should not happen now. */
1790             (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
1791             (OP(first) == PLUS) ||
1792             (OP(first) == MINMOD) ||
1793                /* An {n,m} with n>0 */
1794             (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
1795                 if (OP(first) == PLUS)
1796                     sawplus = 1;
1797                 else
1798                     first += regarglen[(U8)OP(first)];
1799                 first = NEXTOPER(first);
1800         }
1801
1802         /* Starting-point info. */
1803       again:
1804         if (PL_regkind[(U8)OP(first)] == EXACT) {
1805             if (OP(first) == EXACT)
1806                 ;       /* Empty, get anchored substr later. */
1807             else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
1808                 r->regstclass = first;
1809         }
1810         else if (strchr((char*)PL_simple,OP(first)))
1811             r->regstclass = first;
1812         else if (PL_regkind[(U8)OP(first)] == BOUND ||
1813                  PL_regkind[(U8)OP(first)] == NBOUND)
1814             r->regstclass = first;
1815         else if (PL_regkind[(U8)OP(first)] == BOL) {
1816             r->reganch |= (OP(first) == MBOL
1817                            ? ROPT_ANCH_MBOL
1818                            : (OP(first) == SBOL
1819                               ? ROPT_ANCH_SBOL
1820                               : ROPT_ANCH_BOL));
1821             first = NEXTOPER(first);
1822             goto again;
1823         }
1824         else if (OP(first) == GPOS) {
1825             r->reganch |= ROPT_ANCH_GPOS;
1826             first = NEXTOPER(first);
1827             goto again;
1828         }
1829         else if ((OP(first) == STAR &&
1830             PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
1831             !(r->reganch & ROPT_ANCH) )
1832         {
1833             /* turn .* into ^.* with an implied $*=1 */
1834             int type = OP(NEXTOPER(first));
1835
1836             if (type == REG_ANY)
1837                 type = ROPT_ANCH_MBOL;
1838             else
1839                 type = ROPT_ANCH_SBOL;
1840
1841             r->reganch |= type | ROPT_IMPLICIT;
1842             first = NEXTOPER(first);
1843             goto again;
1844         }
1845         if (sawplus && (!sawopen || !RExC_sawback)
1846             && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
1847             /* x+ must match at the 1st pos of run of x's */
1848             r->reganch |= ROPT_SKIP;
1849
1850         /* Scan is after the zeroth branch, first is atomic matcher. */
1851         DEBUG_r(PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
1852                               (IV)(first - scan + 1)));
1853         /*
1854         * If there's something expensive in the r.e., find the
1855         * longest literal string that must appear and make it the
1856         * regmust.  Resolve ties in favor of later strings, since
1857         * the regstart check works with the beginning of the r.e.
1858         * and avoiding duplication strengthens checking.  Not a
1859         * strong reason, but sufficient in the absence of others.
1860         * [Now we resolve ties in favor of the earlier string if
1861         * it happens that c_offset_min has been invalidated, since the
1862         * earlier string may buy us something the later one won't.]
1863         */
1864         minlen = 0;
1865
1866         data.longest_fixed = newSVpvn("",0);
1867         data.longest_float = newSVpvn("",0);
1868         data.last_found = newSVpvn("",0);
1869         data.longest = &(data.longest_fixed);
1870         first = scan;
1871         if (!r->regstclass) {
1872             cl_init(pRExC_state, &ch_class);
1873             data.start_class = &ch_class;
1874             stclass_flag = SCF_DO_STCLASS_AND;
1875         } else                          /* XXXX Check for BOUND? */
1876             stclass_flag = 0;
1877         data.last_closep = &last_close;
1878
1879         minlen = study_chunk(pRExC_state, &first, &fake, scan + RExC_size, /* Up to end */
1880                              &data, SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag);
1881         if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
1882              && data.last_start_min == 0 && data.last_end > 0
1883              && !RExC_seen_zerolen
1884              && (!(RExC_seen & REG_SEEN_GPOS) || (r->reganch & ROPT_ANCH_GPOS)))
1885             r->reganch |= ROPT_CHECK_ALL;
1886         scan_commit(pRExC_state, &data);
1887         SvREFCNT_dec(data.last_found);
1888
1889         longest_float_length = CHR_SVLEN(data.longest_float);
1890         if (longest_float_length
1891             || (data.flags & SF_FL_BEFORE_EOL
1892                 && (!(data.flags & SF_FL_BEFORE_MEOL)
1893                     || (RExC_flags16 & PMf_MULTILINE)))) {
1894             int t;
1895
1896             if (SvCUR(data.longest_fixed)                       /* ok to leave SvCUR */
1897                 && data.offset_fixed == data.offset_float_min
1898                 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
1899                     goto remove_float;          /* As in (a)+. */
1900
1901             r->float_substr = data.longest_float;
1902             r->float_min_offset = data.offset_float_min;
1903             r->float_max_offset = data.offset_float_max;
1904             t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
1905                        && (!(data.flags & SF_FL_BEFORE_MEOL)
1906                            || (RExC_flags16 & PMf_MULTILINE)));
1907             fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
1908         }
1909         else {
1910           remove_float:
1911             r->float_substr = Nullsv;
1912             SvREFCNT_dec(data.longest_float);
1913             longest_float_length = 0;
1914         }
1915
1916         longest_fixed_length = CHR_SVLEN(data.longest_fixed);
1917         if (longest_fixed_length
1918             || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
1919                 && (!(data.flags & SF_FIX_BEFORE_MEOL)
1920                     || (RExC_flags16 & PMf_MULTILINE)))) {
1921             int t;
1922
1923             r->anchored_substr = data.longest_fixed;
1924             r->anchored_offset = data.offset_fixed;
1925             t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
1926                  && (!(data.flags & SF_FIX_BEFORE_MEOL)
1927                      || (RExC_flags16 & PMf_MULTILINE)));
1928             fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
1929         }
1930         else {
1931             r->anchored_substr = Nullsv;
1932             SvREFCNT_dec(data.longest_fixed);
1933             longest_fixed_length = 0;
1934         }
1935         if (r->regstclass
1936             && (OP(r->regstclass) == REG_ANY || OP(r->regstclass) == SANY))
1937             r->regstclass = NULL;
1938         if ((!r->anchored_substr || r->anchored_offset) && stclass_flag
1939             && !(data.start_class->flags & ANYOF_EOS)
1940             && !cl_is_anything(data.start_class)) {
1941             I32 n = add_data(pRExC_state, 1, "f");
1942
1943             New(1006, RExC_rx->data->data[n], 1,
1944                 struct regnode_charclass_class);
1945             StructCopy(data.start_class,
1946                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1947                        struct regnode_charclass_class);
1948             r->regstclass = (regnode*)RExC_rx->data->data[n];
1949             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
1950             PL_regdata = r->data; /* for regprop() */
1951             DEBUG_r({ SV *sv = sv_newmortal();
1952                       regprop(sv, (regnode*)data.start_class);
1953                       PerlIO_printf(Perl_debug_log,
1954                                     "synthetic stclass `%s'.\n",
1955                                     SvPVX(sv));});
1956         }
1957
1958         /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
1959         if (longest_fixed_length > longest_float_length) {
1960             r->check_substr = r->anchored_substr;
1961             r->check_offset_min = r->check_offset_max = r->anchored_offset;
1962             if (r->reganch & ROPT_ANCH_SINGLE)
1963                 r->reganch |= ROPT_NOSCAN;
1964         }
1965         else {
1966             r->check_substr = r->float_substr;
1967             r->check_offset_min = data.offset_float_min;
1968             r->check_offset_max = data.offset_float_max;
1969         }
1970         /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
1971            This should be changed ASAP!  */
1972         if (r->check_substr && !(r->reganch & ROPT_ANCH_GPOS)) {
1973             r->reganch |= RE_USE_INTUIT;
1974             if (SvTAIL(r->check_substr))
1975                 r->reganch |= RE_INTUIT_TAIL;
1976         }
1977     }
1978     else {
1979         /* Several toplevels. Best we can is to set minlen. */
1980         I32 fake;
1981         struct regnode_charclass_class ch_class;
1982         I32 last_close = 0;
1983         
1984         DEBUG_r(PerlIO_printf(Perl_debug_log, "\n"));
1985         scan = r->program + 1;
1986         cl_init(pRExC_state, &ch_class);
1987         data.start_class = &ch_class;
1988         data.last_closep = &last_close;
1989         minlen = study_chunk(pRExC_state, &scan, &fake, scan + RExC_size, &data, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS);
1990         r->check_substr = r->anchored_substr = r->float_substr = Nullsv;
1991         if (!(data.start_class->flags & ANYOF_EOS)
1992             && !cl_is_anything(data.start_class)) {
1993             I32 n = add_data(pRExC_state, 1, "f");
1994
1995             New(1006, RExC_rx->data->data[n], 1,
1996                 struct regnode_charclass_class);
1997             StructCopy(data.start_class,
1998                        (struct regnode_charclass_class*)RExC_rx->data->data[n],
1999                        struct regnode_charclass_class);
2000             r->regstclass = (regnode*)RExC_rx->data->data[n];
2001             r->reganch &= ~ROPT_SKIP;   /* Used in find_byclass(). */
2002             DEBUG_r({ SV* sv = sv_newmortal();
2003                       regprop(sv, (regnode*)data.start_class);
2004                       PerlIO_printf(Perl_debug_log,
2005                                     "synthetic stclass `%s'.\n",
2006                                     SvPVX(sv));});
2007         }
2008     }
2009
2010     r->minlen = minlen;
2011     if (RExC_seen & REG_SEEN_GPOS)
2012         r->reganch |= ROPT_GPOS_SEEN;
2013     if (RExC_seen & REG_SEEN_LOOKBEHIND)
2014         r->reganch |= ROPT_LOOKBEHIND_SEEN;
2015     if (RExC_seen & REG_SEEN_EVAL)
2016         r->reganch |= ROPT_EVAL_SEEN;
2017     if (RExC_seen & REG_SEEN_CANY)
2018         r->reganch |= ROPT_CANY_SEEN;
2019     if (RExC_seen & REG_SEEN_SEOL)
2020         r->reganch |= ROPT_SEOL_SEEN;
2021     Newz(1002, r->startp, RExC_npar, I32);
2022     Newz(1002, r->endp, RExC_npar, I32);
2023     PL_regdata = r->data; /* for regprop() */
2024     DEBUG_r(regdump(r));
2025     return(r);
2026 }
2027
2028 /*
2029  - reg - regular expression, i.e. main body or parenthesized thing
2030  *
2031  * Caller must absorb opening parenthesis.
2032  *
2033  * Combining parenthesis handling with the base level of regular expression
2034  * is a trifle forced, but the need to tie the tails of the branches to what
2035  * follows makes it hard to avoid.
2036  */
2037 STATIC regnode *
2038 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2039     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2040 {
2041     register regnode *ret;              /* Will be the head of the group. */
2042     register regnode *br;
2043     register regnode *lastbr;
2044     register regnode *ender = 0;
2045     register I32 parno = 0;
2046     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2047
2048     /* for (?g), (?gc), and (?o) warnings; warning
2049        about (?c) will warn about (?g) -- japhy    */
2050
2051     I32 wastedflags = 0x00,
2052         wasted_o    = 0x01,
2053         wasted_g    = 0x02,
2054         wasted_gc   = 0x02 | 0x04,
2055         wasted_c    = 0x04;
2056
2057     char * parse_start = RExC_parse; /* MJD */
2058     char *oregcomp_parse = RExC_parse;
2059     char c;
2060
2061     *flagp = 0;                         /* Tentatively. */
2062
2063
2064     /* Make an OPEN node, if parenthesized. */
2065     if (paren) {
2066         if (*RExC_parse == '?') { /* (?...) */
2067             U16 posflags = 0, negflags = 0;
2068             U16 *flagsp = &posflags;
2069             int logical = 0;
2070             char *seqstart = RExC_parse;
2071
2072             RExC_parse++;
2073             paren = *RExC_parse++;
2074             ret = NULL;                 /* For look-ahead/behind. */
2075             switch (paren) {
2076             case '<':           /* (?<...) */
2077                 RExC_seen |= REG_SEEN_LOOKBEHIND;
2078                 if (*RExC_parse == '!')
2079                     paren = ',';
2080                 if (*RExC_parse != '=' && *RExC_parse != '!')
2081                     goto unknown;
2082                 RExC_parse++;
2083             case '=':           /* (?=...) */
2084             case '!':           /* (?!...) */
2085                 RExC_seen_zerolen++;
2086             case ':':           /* (?:...) */
2087             case '>':           /* (?>...) */
2088                 break;
2089             case '$':           /* (?$...) */
2090             case '@':           /* (?@...) */
2091                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2092                 break;
2093             case '#':           /* (?#...) */
2094                 while (*RExC_parse && *RExC_parse != ')')
2095                     RExC_parse++;
2096                 if (*RExC_parse != ')')
2097                     FAIL("Sequence (?#... not terminated");
2098                 nextchar(pRExC_state);
2099                 *flagp = TRYAGAIN;
2100                 return NULL;
2101             case 'p':           /* (?p...) */
2102                 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
2103                     vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
2104                 /* FALL THROUGH*/
2105             case '?':           /* (??...) */
2106                 logical = 1;
2107                 paren = *RExC_parse++;
2108                 /* FALL THROUGH */
2109             case '{':           /* (?{...}) */
2110             {
2111                 I32 count = 1, n = 0;
2112                 char c;
2113                 char *s = RExC_parse;
2114                 SV *sv;
2115                 OP_4tree *sop, *rop;
2116
2117                 RExC_seen_zerolen++;
2118                 RExC_seen |= REG_SEEN_EVAL;
2119                 while (count && (c = *RExC_parse)) {
2120                     if (c == '\\' && RExC_parse[1])
2121                         RExC_parse++;
2122                     else if (c == '{')
2123                         count++;
2124                     else if (c == '}')
2125                         count--;
2126                     RExC_parse++;
2127                 }
2128                 if (*RExC_parse != ')')
2129                 {
2130                     RExC_parse = s;             
2131                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2132                 }
2133                 if (!SIZE_ONLY) {
2134                     AV *av;
2135                 
2136                     if (RExC_parse - 1 - s)
2137                         sv = newSVpvn(s, RExC_parse - 1 - s);
2138                     else
2139                         sv = newSVpvn("", 0);
2140
2141                     ENTER;
2142                     Perl_save_re_context(aTHX);
2143                     rop = sv_compile_2op(sv, &sop, "re", &av);
2144                     sop->op_private |= OPpREFCOUNTED;
2145                     /* re_dup will OpREFCNT_inc */
2146                     OpREFCNT_set(sop, 1);
2147                     LEAVE;
2148
2149                     n = add_data(pRExC_state, 3, "nop");
2150                     RExC_rx->data->data[n] = (void*)rop;
2151                     RExC_rx->data->data[n+1] = (void*)sop;
2152                     RExC_rx->data->data[n+2] = (void*)av;
2153                     SvREFCNT_dec(sv);
2154                 }
2155                 else {                                          /* First pass */
2156                     if (PL_reginterp_cnt < ++RExC_seen_evals
2157                         && PL_curcop != &PL_compiling)
2158                         /* No compiled RE interpolated, has runtime
2159                            components ===> unsafe.  */
2160                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2161                     if (PL_tainting && PL_tainted)
2162                         FAIL("Eval-group in insecure regular expression");
2163                 }
2164                 
2165                 nextchar(pRExC_state);
2166                 if (logical) {
2167                     ret = reg_node(pRExC_state, LOGICAL);
2168                     if (!SIZE_ONLY)
2169                         ret->flags = 2;
2170                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2171                     /* deal with the length of this later - MJD */
2172                     return ret;
2173                 }
2174                 return reganode(pRExC_state, EVAL, n);
2175             }
2176             case '(':           /* (?(?{...})...) and (?(?=...)...) */
2177             {
2178                 if (RExC_parse[0] == '?') {        /* (?(?...)) */
2179                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2180                         || RExC_parse[1] == '<'
2181                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2182                         I32 flag;
2183                         
2184                         ret = reg_node(pRExC_state, LOGICAL);
2185                         if (!SIZE_ONLY)
2186                             ret->flags = 1;
2187                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2188                         goto insert_if;
2189                     }
2190                 }
2191                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2192                     /* (?(1)...) */
2193                     parno = atoi(RExC_parse++);
2194
2195                     while (isDIGIT(*RExC_parse))
2196                         RExC_parse++;
2197                     ret = reganode(pRExC_state, GROUPP, parno);
2198                     
2199                     if ((c = *nextchar(pRExC_state)) != ')')
2200                         vFAIL("Switch condition not recognized");
2201                   insert_if:
2202                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2203                     br = regbranch(pRExC_state, &flags, 1);
2204                     if (br == NULL)
2205                         br = reganode(pRExC_state, LONGJMP, 0);
2206                     else
2207                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2208                     c = *nextchar(pRExC_state);
2209                     if (flags&HASWIDTH)
2210                         *flagp |= HASWIDTH;
2211                     if (c == '|') {
2212                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2213                         regbranch(pRExC_state, &flags, 1);
2214                         regtail(pRExC_state, ret, lastbr);
2215                         if (flags&HASWIDTH)
2216                             *flagp |= HASWIDTH;
2217                         c = *nextchar(pRExC_state);
2218                     }
2219                     else
2220                         lastbr = NULL;
2221                     if (c != ')')
2222                         vFAIL("Switch (?(condition)... contains too many branches");
2223                     ender = reg_node(pRExC_state, TAIL);
2224                     regtail(pRExC_state, br, ender);
2225                     if (lastbr) {
2226                         regtail(pRExC_state, lastbr, ender);
2227                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2228                     }
2229                     else
2230                         regtail(pRExC_state, ret, ender);
2231                     return ret;
2232                 }
2233                 else {
2234                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2235                 }
2236             }
2237             case 0:
2238                 RExC_parse--; /* for vFAIL to print correctly */
2239                 vFAIL("Sequence (? incomplete");
2240                 break;
2241             default:
2242                 --RExC_parse;
2243               parse_flags:      /* (?i) */
2244                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2245                     /* (?g), (?gc) and (?o) are useless here
2246                        and must be globally applied -- japhy */
2247
2248                     if (*RExC_parse == 'o' || *RExC_parse == 'g') {
2249                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2250                             I32 wflagbit = *RExC_parse == 'o' ? wasted_o : wasted_g;
2251                             if (! (wastedflags & wflagbit) ) {
2252                                 wastedflags |= wflagbit;
2253                                 vWARN5(
2254                                     RExC_parse + 1,
2255                                     "Useless (%s%c) - %suse /%c modifier",
2256                                     flagsp == &negflags ? "?-" : "?",
2257                                     *RExC_parse,
2258                                     flagsp == &negflags ? "don't " : "",
2259                                     *RExC_parse
2260                                 );
2261                             }
2262                         }
2263                     }
2264                     else if (*RExC_parse == 'c') {
2265                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2266                             if (! (wastedflags & wasted_c) ) {
2267                                 wastedflags |= wasted_gc;
2268                                 vWARN3(
2269                                     RExC_parse + 1,
2270                                     "Useless (%sc) - %suse /gc modifier",
2271                                     flagsp == &negflags ? "?-" : "?",
2272                                     flagsp == &negflags ? "don't " : ""
2273                                 );
2274                             }
2275                         }
2276                     }
2277                     else { pmflag(flagsp, *RExC_parse); }
2278
2279                     ++RExC_parse;
2280                 }
2281                 if (*RExC_parse == '-') {
2282                     flagsp = &negflags;
2283                     wastedflags = 0;  /* reset so (?g-c) warns twice */
2284                     ++RExC_parse;
2285                     goto parse_flags;
2286                 }
2287                 RExC_flags16 |= posflags;
2288                 RExC_flags16 &= ~negflags;
2289                 if (*RExC_parse == ':') {
2290                     RExC_parse++;
2291                     paren = ':';
2292                     break;
2293                 }               
2294               unknown:
2295                 if (*RExC_parse != ')') {
2296                     RExC_parse++;
2297                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2298                 }
2299                 nextchar(pRExC_state);
2300                 *flagp = TRYAGAIN;
2301                 return NULL;
2302             }
2303         }
2304         else {                  /* (...) */
2305             parno = RExC_npar;
2306             RExC_npar++;
2307             ret = reganode(pRExC_state, OPEN, parno);
2308             Set_Node_Length(ret, 1); /* MJD */
2309             Set_Node_Offset(ret, RExC_parse); /* MJD */
2310             open = 1;
2311         }
2312     }
2313     else                        /* ! paren */
2314         ret = NULL;
2315
2316     /* Pick up the branches, linking them together. */
2317     parse_start = RExC_parse;   /* MJD */
2318     br = regbranch(pRExC_state, &flags, 1);
2319     /*     branch_len = (paren != 0); */
2320     
2321     if (br == NULL)
2322         return(NULL);
2323     if (*RExC_parse == '|') {
2324         if (!SIZE_ONLY && RExC_extralen) {
2325             reginsert(pRExC_state, BRANCHJ, br);
2326         }
2327         else {                  /* MJD */
2328             reginsert(pRExC_state, BRANCH, br);
2329             Set_Node_Length(br, paren != 0);
2330             Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2331         }
2332         have_branch = 1;
2333         if (SIZE_ONLY)
2334             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2335     }
2336     else if (paren == ':') {
2337         *flagp |= flags&SIMPLE;
2338     }
2339     if (open) {                         /* Starts with OPEN. */
2340         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2341     }
2342     else if (paren != '?')              /* Not Conditional */
2343         ret = br;
2344     if (flags&HASWIDTH)
2345         *flagp |= HASWIDTH;
2346     *flagp |= flags&SPSTART;
2347     lastbr = br;
2348     while (*RExC_parse == '|') {
2349         if (!SIZE_ONLY && RExC_extralen) {
2350             ender = reganode(pRExC_state, LONGJMP,0);
2351             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2352         }
2353         if (SIZE_ONLY)
2354             RExC_extralen += 2;         /* Account for LONGJMP. */
2355         nextchar(pRExC_state);
2356         br = regbranch(pRExC_state, &flags, 0);
2357         
2358         if (br == NULL)
2359             return(NULL);
2360         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2361         lastbr = br;
2362         if (flags&HASWIDTH)
2363             *flagp |= HASWIDTH;
2364         *flagp |= flags&SPSTART;
2365     }
2366
2367     if (have_branch || paren != ':') {
2368         /* Make a closing node, and hook it on the end. */
2369         switch (paren) {
2370         case ':':
2371             ender = reg_node(pRExC_state, TAIL);
2372             break;
2373         case 1:
2374             ender = reganode(pRExC_state, CLOSE, parno);
2375             Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2376             Set_Node_Length(ender,1); /* MJD */
2377             break;
2378         case '<':
2379         case ',':
2380         case '=':
2381         case '!':
2382             *flagp &= ~HASWIDTH;
2383             /* FALL THROUGH */
2384         case '>':
2385             ender = reg_node(pRExC_state, SUCCEED);
2386             break;
2387         case 0:
2388             ender = reg_node(pRExC_state, END);
2389             break;
2390         }
2391         regtail(pRExC_state, lastbr, ender);
2392
2393         if (have_branch) {
2394             /* Hook the tails of the branches to the closing node. */
2395             for (br = ret; br != NULL; br = regnext(br)) {
2396                 regoptail(pRExC_state, br, ender);
2397             }
2398         }
2399     }
2400
2401     {
2402         char *p;
2403         static char parens[] = "=!<,>";
2404
2405         if (paren && (p = strchr(parens, paren))) {
2406             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2407             int flag = (p - parens) > 1;
2408
2409             if (paren == '>')
2410                 node = SUSPEND, flag = 0;
2411             reginsert(pRExC_state, node,ret);
2412             ret->flags = flag;
2413             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2414         }
2415     }
2416
2417     /* Check for proper termination. */
2418     if (paren) {
2419         RExC_flags16 = oregflags;
2420         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2421             RExC_parse = oregcomp_parse;
2422             vFAIL("Unmatched (");
2423         }
2424     }
2425     else if (!paren && RExC_parse < RExC_end) {
2426         if (*RExC_parse == ')') {
2427             RExC_parse++;
2428             vFAIL("Unmatched )");
2429         }
2430         else
2431             FAIL("Junk on end of regexp");      /* "Can't happen". */
2432         /* NOTREACHED */
2433     }
2434
2435     return(ret);
2436 }
2437
2438 /*
2439  - regbranch - one alternative of an | operator
2440  *
2441  * Implements the concatenation operator.
2442  */
2443 STATIC regnode *
2444 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2445 {
2446     register regnode *ret;
2447     register regnode *chain = NULL;
2448     register regnode *latest;
2449     I32 flags = 0, c = 0;
2450
2451     if (first)
2452         ret = NULL;
2453     else {
2454         if (!SIZE_ONLY && RExC_extralen)
2455             ret = reganode(pRExC_state, BRANCHJ,0);
2456         else {
2457             ret = reg_node(pRExC_state, BRANCH);
2458             Set_Node_Length(ret, 1);
2459         }
2460     }
2461         
2462     if (!first && SIZE_ONLY)
2463         RExC_extralen += 1;                     /* BRANCHJ */
2464
2465     *flagp = WORST;                     /* Tentatively. */
2466
2467     RExC_parse--;
2468     nextchar(pRExC_state);
2469     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2470         flags &= ~TRYAGAIN;
2471         latest = regpiece(pRExC_state, &flags);
2472         if (latest == NULL) {
2473             if (flags & TRYAGAIN)
2474                 continue;
2475             return(NULL);
2476         }
2477         else if (ret == NULL)
2478             ret = latest;
2479         *flagp |= flags&HASWIDTH;
2480         if (chain == NULL)      /* First piece. */
2481             *flagp |= flags&SPSTART;
2482         else {
2483             RExC_naughty++;
2484             regtail(pRExC_state, chain, latest);
2485         }
2486         chain = latest;
2487         c++;
2488     }
2489     if (chain == NULL) {        /* Loop ran zero times. */
2490         chain = reg_node(pRExC_state, NOTHING);
2491         if (ret == NULL)
2492             ret = chain;
2493     }
2494     if (c == 1) {
2495         *flagp |= flags&SIMPLE;
2496     }
2497
2498     return(ret);
2499 }
2500
2501 /*
2502  - regpiece - something followed by possible [*+?]
2503  *
2504  * Note that the branching code sequences used for ? and the general cases
2505  * of * and + are somewhat optimized:  they use the same NOTHING node as
2506  * both the endmarker for their branch list and the body of the last branch.
2507  * It might seem that this node could be dispensed with entirely, but the
2508  * endmarker role is not redundant.
2509  */
2510 STATIC regnode *
2511 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2512 {
2513     register regnode *ret;
2514     register char op;
2515     register char *next;
2516     I32 flags;
2517     char *origparse = RExC_parse;
2518     char *maxpos;
2519     I32 min;
2520     I32 max = REG_INFTY;
2521     char *parse_start;
2522
2523     ret = regatom(pRExC_state, &flags);
2524     if (ret == NULL) {
2525         if (flags & TRYAGAIN)
2526             *flagp |= TRYAGAIN;
2527         return(NULL);
2528     }
2529
2530     op = *RExC_parse;
2531
2532     if (op == '{' && regcurly(RExC_parse)) {
2533         parse_start = RExC_parse; /* MJD */
2534         next = RExC_parse + 1;
2535         maxpos = Nullch;
2536         while (isDIGIT(*next) || *next == ',') {
2537             if (*next == ',') {
2538                 if (maxpos)
2539                     break;
2540                 else
2541                     maxpos = next;
2542             }
2543             next++;
2544         }
2545         if (*next == '}') {             /* got one */
2546             if (!maxpos)
2547                 maxpos = next;
2548             RExC_parse++;
2549             min = atoi(RExC_parse);
2550             if (*maxpos == ',')
2551                 maxpos++;
2552             else
2553                 maxpos = RExC_parse;
2554             max = atoi(maxpos);
2555             if (!max && *maxpos != '0')
2556                 max = REG_INFTY;                /* meaning "infinity" */
2557             else if (max >= REG_INFTY)
2558                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2559             RExC_parse = next;
2560             nextchar(pRExC_state);
2561
2562         do_curly:
2563             if ((flags&SIMPLE)) {
2564                 RExC_naughty += 2 + RExC_naughty / 2;
2565                 reginsert(pRExC_state, CURLY, ret);
2566                 Set_Node_Offset(ret, parse_start+1); /* MJD */
2567                 Set_Node_Cur_Length(ret);
2568             }
2569             else {
2570                 regnode *w = reg_node(pRExC_state, WHILEM);
2571
2572                 w->flags = 0;
2573                 regtail(pRExC_state, ret, w);
2574                 if (!SIZE_ONLY && RExC_extralen) {
2575                     reginsert(pRExC_state, LONGJMP,ret);
2576                     reginsert(pRExC_state, NOTHING,ret);
2577                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2578                 }
2579                 reginsert(pRExC_state, CURLYX,ret);
2580                                 /* MJD hk */
2581                 Set_Node_Offset(ret, parse_start+1);
2582                 Set_Node_Length(ret, 
2583                                 op == '{' ? (RExC_parse - parse_start) : 1);
2584                 
2585                 if (!SIZE_ONLY && RExC_extralen)
2586                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2587                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2588                 if (SIZE_ONLY)
2589                     RExC_whilem_seen++, RExC_extralen += 3;
2590                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2591             }
2592             ret->flags = 0;
2593
2594             if (min > 0)
2595                 *flagp = WORST;
2596             if (max > 0)
2597                 *flagp |= HASWIDTH;
2598             if (max && max < min)
2599                 vFAIL("Can't do {n,m} with n > m");
2600             if (!SIZE_ONLY) {
2601                 ARG1_SET(ret, min);
2602                 ARG2_SET(ret, max);
2603             }
2604
2605             goto nest_check;
2606         }
2607     }
2608
2609     if (!ISMULT1(op)) {
2610         *flagp = flags;
2611         return(ret);
2612     }
2613
2614 #if 0                           /* Now runtime fix should be reliable. */
2615
2616     /* if this is reinstated, don't forget to put this back into perldiag:
2617
2618             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2619
2620            (F) The part of the regexp subject to either the * or + quantifier
2621            could match an empty string. The {#} shows in the regular
2622            expression about where the problem was discovered.
2623
2624     */
2625
2626     if (!(flags&HASWIDTH) && op != '?')
2627       vFAIL("Regexp *+ operand could be empty");
2628 #endif
2629
2630     parse_start = RExC_parse;
2631     nextchar(pRExC_state);
2632
2633     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2634
2635     if (op == '*' && (flags&SIMPLE)) {
2636         reginsert(pRExC_state, STAR, ret);
2637         ret->flags = 0;
2638         RExC_naughty += 4;
2639     }
2640     else if (op == '*') {
2641         min = 0;
2642         goto do_curly;
2643     }
2644     else if (op == '+' && (flags&SIMPLE)) {
2645         reginsert(pRExC_state, PLUS, ret);
2646         ret->flags = 0;
2647         RExC_naughty += 3;
2648     }
2649     else if (op == '+') {
2650         min = 1;
2651         goto do_curly;
2652     }
2653     else if (op == '?') {
2654         min = 0; max = 1;
2655         goto do_curly;
2656     }
2657   nest_check:
2658     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2659         vWARN3(RExC_parse,
2660                "%.*s matches null string many times",
2661                RExC_parse - origparse,
2662                origparse);
2663     }
2664
2665     if (*RExC_parse == '?') {
2666         nextchar(pRExC_state);
2667         reginsert(pRExC_state, MINMOD, ret);
2668         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2669     }
2670     if (ISMULT2(RExC_parse)) {
2671         RExC_parse++;
2672         vFAIL("Nested quantifiers");
2673     }
2674
2675     return(ret);
2676 }
2677
2678 /*
2679  - regatom - the lowest level
2680  *
2681  * Optimization:  gobbles an entire sequence of ordinary characters so that
2682  * it can turn them into a single node, which is smaller to store and
2683  * faster to run.  Backslashed characters are exceptions, each becoming a
2684  * separate node; the code is simpler that way and it's not worth fixing.
2685  *
2686  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2687 STATIC regnode *
2688 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2689 {
2690     register regnode *ret = 0;
2691     I32 flags;
2692     char *parse_start = 0;
2693
2694     *flagp = WORST;             /* Tentatively. */
2695
2696 tryagain:
2697     switch (*RExC_parse) {
2698     case '^':
2699         RExC_seen_zerolen++;
2700         nextchar(pRExC_state);
2701         if (RExC_flags16 & PMf_MULTILINE)
2702             ret = reg_node(pRExC_state, MBOL);
2703         else if (RExC_flags16 & PMf_SINGLELINE)
2704             ret = reg_node(pRExC_state, SBOL);
2705         else
2706             ret = reg_node(pRExC_state, BOL);
2707         Set_Node_Length(ret, 1); /* MJD */
2708         break;
2709     case '$':
2710         nextchar(pRExC_state);
2711         if (*RExC_parse)
2712             RExC_seen_zerolen++;
2713         if (RExC_flags16 & PMf_MULTILINE)
2714             ret = reg_node(pRExC_state, MEOL);
2715         else if (RExC_flags16 & PMf_SINGLELINE)
2716             ret = reg_node(pRExC_state, SEOL);
2717         else
2718             ret = reg_node(pRExC_state, EOL);
2719         Set_Node_Length(ret, 1); /* MJD */
2720         break;
2721     case '.':
2722         nextchar(pRExC_state);
2723         if (RExC_flags16 & PMf_SINGLELINE)
2724             ret = reg_node(pRExC_state, SANY);
2725         else
2726             ret = reg_node(pRExC_state, REG_ANY);
2727         *flagp |= HASWIDTH|SIMPLE;
2728         RExC_naughty++;
2729         Set_Node_Length(ret, 1); /* MJD */
2730         break;
2731     case '[':
2732     {
2733         char *oregcomp_parse = ++RExC_parse;
2734         ret = regclass(pRExC_state);
2735         if (*RExC_parse != ']') {
2736             RExC_parse = oregcomp_parse;
2737             vFAIL("Unmatched [");
2738         }
2739         nextchar(pRExC_state);
2740         *flagp |= HASWIDTH|SIMPLE;
2741         Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2742         break;
2743     }
2744     case '(':
2745         nextchar(pRExC_state);
2746         ret = reg(pRExC_state, 1, &flags);
2747         if (ret == NULL) {
2748                 if (flags & TRYAGAIN) {
2749                     if (RExC_parse == RExC_end) {
2750                          /* Make parent create an empty node if needed. */
2751                         *flagp |= TRYAGAIN;
2752                         return(NULL);
2753                     }
2754                     goto tryagain;
2755                 }
2756                 return(NULL);
2757         }
2758         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2759         break;
2760     case '|':
2761     case ')':
2762         if (flags & TRYAGAIN) {
2763             *flagp |= TRYAGAIN;
2764             return NULL;
2765         }
2766         vFAIL("Internal urp");
2767                                 /* Supposed to be caught earlier. */
2768         break;
2769     case '{':
2770         if (!regcurly(RExC_parse)) {
2771             RExC_parse++;
2772             goto defchar;
2773         }
2774         /* FALL THROUGH */
2775     case '?':
2776     case '+':
2777     case '*':
2778         RExC_parse++;
2779         vFAIL("Quantifier follows nothing");
2780         break;
2781     case '\\':
2782         switch (*++RExC_parse) {
2783         case 'A':
2784             RExC_seen_zerolen++;
2785             ret = reg_node(pRExC_state, SBOL);
2786             *flagp |= SIMPLE;
2787             nextchar(pRExC_state);
2788             Set_Node_Length(ret, 2); /* MJD */
2789             break;
2790         case 'G':
2791             ret = reg_node(pRExC_state, GPOS);
2792             RExC_seen |= REG_SEEN_GPOS;
2793             *flagp |= SIMPLE;
2794             nextchar(pRExC_state);
2795             Set_Node_Length(ret, 2); /* MJD */
2796             break;
2797         case 'Z':
2798             ret = reg_node(pRExC_state, SEOL);
2799             RExC_seen |= REG_SEEN_SEOL;
2800             *flagp |= SIMPLE;
2801             nextchar(pRExC_state);
2802             break;
2803         case 'z':
2804             ret = reg_node(pRExC_state, EOS);
2805             *flagp |= SIMPLE;
2806             RExC_seen_zerolen++;                /* Do not optimize RE away */
2807             nextchar(pRExC_state);
2808             Set_Node_Length(ret, 2); /* MJD */
2809             break;
2810         case 'C':
2811             ret = reg_node(pRExC_state, CANY);
2812             RExC_seen |= REG_SEEN_CANY;
2813             *flagp |= HASWIDTH|SIMPLE;
2814             nextchar(pRExC_state);
2815             Set_Node_Length(ret, 2); /* MJD */
2816             break;
2817         case 'X':
2818             ret = reg_node(pRExC_state, CLUMP);
2819             *flagp |= HASWIDTH;
2820             nextchar(pRExC_state);
2821             Set_Node_Length(ret, 2); /* MJD */
2822             break;
2823         case 'w':
2824             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2825             *flagp |= HASWIDTH|SIMPLE;
2826             nextchar(pRExC_state);
2827             Set_Node_Length(ret, 2); /* MJD */
2828             break;
2829         case 'W':
2830             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2831             *flagp |= HASWIDTH|SIMPLE;
2832             nextchar(pRExC_state);
2833             Set_Node_Length(ret, 2); /* MJD */
2834             break;
2835         case 'b':
2836             RExC_seen_zerolen++;
2837             RExC_seen |= REG_SEEN_LOOKBEHIND;
2838             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2839             *flagp |= SIMPLE;
2840             nextchar(pRExC_state);
2841             Set_Node_Length(ret, 2); /* MJD */
2842             break;
2843         case 'B':
2844             RExC_seen_zerolen++;
2845             RExC_seen |= REG_SEEN_LOOKBEHIND;
2846             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2847             *flagp |= SIMPLE;
2848             nextchar(pRExC_state);
2849             Set_Node_Length(ret, 2); /* MJD */
2850             break;
2851         case 's':
2852             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2853             *flagp |= HASWIDTH|SIMPLE;
2854             nextchar(pRExC_state);
2855             Set_Node_Length(ret, 2); /* MJD */
2856             break;
2857         case 'S':
2858             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
2859             *flagp |= HASWIDTH|SIMPLE;
2860             nextchar(pRExC_state);
2861             Set_Node_Length(ret, 2); /* MJD */
2862             break;
2863         case 'd':
2864             ret = reg_node(pRExC_state, DIGIT);
2865             *flagp |= HASWIDTH|SIMPLE;
2866             nextchar(pRExC_state);
2867             Set_Node_Length(ret, 2); /* MJD */
2868             break;
2869         case 'D':
2870             ret = reg_node(pRExC_state, NDIGIT);
2871             *flagp |= HASWIDTH|SIMPLE;
2872             nextchar(pRExC_state);
2873             Set_Node_Length(ret, 2); /* MJD */
2874             break;
2875         case 'p':
2876         case 'P':
2877             {   
2878                 char* oldregxend = RExC_end;
2879                 char* parse_start = RExC_parse;
2880
2881                 if (RExC_parse[1] == '{') {
2882                   /* a lovely hack--pretend we saw [\pX] instead */
2883                     RExC_end = strchr(RExC_parse, '}');
2884                     if (!RExC_end) {
2885                         U8 c = (U8)*RExC_parse;
2886                         RExC_parse += 2;
2887                         RExC_end = oldregxend;
2888                         vFAIL2("Missing right brace on \\%c{}", c);
2889                     }
2890                     RExC_end++;
2891                 }
2892                 else
2893                     RExC_end = RExC_parse + 2;
2894                 RExC_parse--;
2895
2896                 ret = regclass(pRExC_state);
2897
2898                 RExC_end = oldregxend;
2899                 RExC_parse--;
2900                 Set_Node_Cur_Length(ret); /* MJD */
2901                 nextchar(pRExC_state);
2902                 *flagp |= HASWIDTH|SIMPLE;
2903             }
2904             break;
2905         case 'n':
2906         case 'r':
2907         case 't':
2908         case 'f':
2909         case 'e':
2910         case 'a':
2911         case 'x':
2912         case 'c':
2913         case '0':
2914             goto defchar;
2915         case '1': case '2': case '3': case '4':
2916         case '5': case '6': case '7': case '8': case '9':
2917             {
2918                 I32 num = atoi(RExC_parse);
2919
2920                 if (num > 9 && num >= RExC_npar)
2921                     goto defchar;
2922                 else {
2923                     char * parse_start = RExC_parse - 1; /* MJD */
2924                     while (isDIGIT(*RExC_parse))
2925                         RExC_parse++;
2926
2927                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2928                         vFAIL("Reference to nonexistent group");
2929                     RExC_sawback = 1;
2930                     ret = reganode(pRExC_state, FOLD
2931                                    ? (LOC ? REFFL : REFF)
2932                                    : REF, num);
2933                     *flagp |= HASWIDTH;
2934                     
2935                     /* override incorrect value set in reganode MJD */
2936                     Set_Node_Offset(ret, parse_start+1); 
2937                     Set_Node_Cur_Length(ret); /* MJD */
2938                     RExC_parse--;
2939                     nextchar(pRExC_state);
2940                 }
2941             }
2942             break;
2943         case '\0':
2944             if (RExC_parse >= RExC_end)
2945                 FAIL("Trailing \\");
2946             /* FALL THROUGH */
2947         default:
2948             /* Do not generate `unrecognized' warnings here, we fall
2949                back into the quick-grab loop below */
2950             goto defchar;
2951         }
2952         break;
2953
2954     case '#':
2955         if (RExC_flags16 & PMf_EXTENDED) {
2956             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2957             if (RExC_parse < RExC_end)
2958                 goto tryagain;
2959         }
2960         /* FALL THROUGH */
2961
2962     default: {
2963             register STRLEN len;
2964             register UV ender;
2965             register char *p;
2966             char *oldp, *s;
2967             STRLEN numlen;
2968             STRLEN ulen;
2969             U8 tmpbuf[UTF8_MAXLEN*2+1];
2970
2971             parse_start = RExC_parse - 1;
2972
2973             RExC_parse++;
2974
2975         defchar:
2976             ret = reg_node(pRExC_state, FOLD
2977                           ? (LOC ? EXACTFL : EXACTF)
2978                           : EXACT);
2979             s = STRING(ret);
2980             for (len = 0, p = RExC_parse - 1;
2981               len < 127 && p < RExC_end;
2982               len++)
2983             {
2984                 oldp = p;
2985
2986                 if (RExC_flags16 & PMf_EXTENDED)
2987                     p = regwhite(p, RExC_end);
2988                 switch (*p) {
2989                 case '^':
2990                 case '$':
2991                 case '.':
2992                 case '[':
2993                 case '(':
2994                 case ')':
2995                 case '|':
2996                     goto loopdone;
2997                 case '\\':
2998                     switch (*++p) {
2999                     case 'A':
3000                     case 'G':
3001                     case 'Z':
3002                     case 'z':
3003                     case 'w':
3004                     case 'W':
3005                     case 'b':
3006                     case 'B':
3007                     case 's':
3008                     case 'S':
3009                     case 'd':
3010                     case 'D':
3011                     case 'p':
3012                     case 'P':
3013                         --p;
3014                         goto loopdone;
3015                     case 'n':
3016                         ender = '\n';
3017                         p++;
3018                         break;
3019                     case 'r':
3020                         ender = '\r';
3021                         p++;
3022                         break;
3023                     case 't':
3024                         ender = '\t';
3025                         p++;
3026                         break;
3027                     case 'f':
3028                         ender = '\f';
3029                         p++;
3030                         break;
3031                     case 'e':
3032                           ender = ASCII_TO_NATIVE('\033');
3033                         p++;
3034                         break;
3035                     case 'a':
3036                           ender = ASCII_TO_NATIVE('\007');
3037                         p++;
3038                         break;
3039                     case 'x':
3040                         if (*++p == '{') {
3041                             char* e = strchr(p, '}');
3042         
3043                             if (!e) {
3044                                 RExC_parse = p + 1;
3045                                 vFAIL("Missing right brace on \\x{}");
3046                             }
3047                             else {
3048                                 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3049                                     | PERL_SCAN_DISALLOW_PREFIX;
3050                                 numlen = e - p - 1;
3051                                 ender = grok_hex(p + 1, &numlen, &flags, NULL);
3052                                 if (ender > 0xff)
3053                                     RExC_utf8 = 1;
3054                                 /* numlen is generous */
3055                                 if (numlen + len >= 127) {
3056                                     p--;
3057                                     goto loopdone;
3058                                 }
3059                                 p = e + 1;
3060                             }
3061                         }
3062                         else {
3063                             I32 flags = PERL_SCAN_DISALLOW_PREFIX;
3064                             numlen = 2;
3065                             ender = grok_hex(p, &numlen, &flags, NULL);
3066                             p += numlen;
3067                         }
3068                         break;
3069                     case 'c':
3070                         p++;
3071                         ender = UCHARAT(p++);
3072                         ender = toCTRL(ender);
3073                         break;
3074                     case '0': case '1': case '2': case '3':case '4':
3075                     case '5': case '6': case '7': case '8':case '9':
3076                         if (*p == '0' ||
3077                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
3078                             I32 flags = 0;
3079                             numlen = 3;
3080                             ender = grok_oct(p, &numlen, &flags, NULL);
3081                             p += numlen;
3082                         }
3083                         else {
3084                             --p;
3085                             goto loopdone;
3086                         }
3087                         break;
3088                     case '\0':
3089                         if (p >= RExC_end)
3090                             FAIL("Trailing \\");
3091                         /* FALL THROUGH */
3092                     default:
3093                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3094                             vWARN2(p + 1, "Unrecognized escape \\%c passed through", UCHARAT(p));
3095                         goto normal_default;
3096                     }
3097                     break;
3098                 default:
3099                   normal_default:
3100                     if (UTF8_IS_START(*p) && UTF) {
3101                         ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3102                                                &numlen, 0);
3103                         p += numlen;
3104                     }
3105                     else
3106                         ender = *p++;
3107                     break;
3108                 }
3109                 if (RExC_flags16 & PMf_EXTENDED)
3110                     p = regwhite(p, RExC_end);
3111                 if (UTF && FOLD) {
3112                     toLOWER_uni(ender, tmpbuf, &ulen);
3113                     ender = utf8_to_uvchr(tmpbuf, 0);
3114                 }
3115                 if (ISMULT2(p)) { /* Back off on ?+*. */
3116                     if (len)
3117                         p = oldp;
3118                     else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3119                         reguni(pRExC_state, ender, s, &numlen);
3120                         s += numlen;
3121                         len += numlen;
3122                     }
3123                     else {
3124                         len++;
3125                         REGC(ender, s++);
3126                     }
3127                     break;
3128                 }
3129                 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3130                     reguni(pRExC_state, ender, s, &numlen);
3131                     s += numlen;
3132                     len += numlen - 1;
3133                 }
3134                 else
3135                     REGC(ender, s++);
3136             }
3137         loopdone:
3138             RExC_parse = p - 1;
3139             Set_Node_Cur_Length(ret); /* MJD */
3140             nextchar(pRExC_state);
3141             {
3142                 /* len is STRLEN which is unsigned, need to copy to signed */
3143                 IV iv = len;
3144                 if (iv < 0)
3145                     vFAIL("Internal disaster");
3146             }
3147             if (len > 0)
3148                 *flagp |= HASWIDTH;
3149             if (len == 1)
3150                 *flagp |= SIMPLE;
3151             if (!SIZE_ONLY)
3152                 STR_LEN(ret) = len;
3153             if (SIZE_ONLY)
3154                 RExC_size += STR_SZ(len);
3155             else
3156                 RExC_emit += STR_SZ(len);
3157         }
3158         break;
3159     }
3160
3161     if (PL_encoding && PL_regkind[(U8)OP(ret)] == EXACT && !RExC_utf8) {
3162          STRLEN oldlen = STR_LEN(ret);
3163          SV *sv        = sv_2mortal(newSVpvn(STRING(ret), oldlen));
3164          char *s       = Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
3165          STRLEN newlen = SvCUR(sv);
3166          if (!SIZE_ONLY) {
3167               DEBUG_r(PerlIO_printf(Perl_debug_log, "recode %*s to %*s\n",
3168                                     oldlen, STRING(ret), newlen, s));
3169               Copy(s, STRING(ret), newlen, char);
3170               STR_LEN(ret) += newlen - oldlen;
3171               RExC_emit += STR_SZ(newlen) - STR_SZ(oldlen);
3172          } else
3173               RExC_size += STR_SZ(newlen) - STR_SZ(oldlen);
3174          RExC_utf8 = 1;
3175     }
3176
3177     return(ret);
3178 }
3179
3180 STATIC char *
3181 S_regwhite(pTHX_ char *p, char *e)
3182 {
3183     while (p < e) {
3184         if (isSPACE(*p))
3185             ++p;
3186         else if (*p == '#') {
3187             do {
3188                 p++;
3189             } while (p < e && *p != '\n');
3190         }
3191         else
3192             break;
3193     }
3194     return p;
3195 }
3196
3197 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3198    Character classes ([:foo:]) can also be negated ([:^foo:]).
3199    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3200    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3201    but trigger failures because they are currently unimplemented. */
3202
3203 #define POSIXCC_DONE(c)   ((c) == ':')
3204 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
3205 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
3206
3207 STATIC I32
3208 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3209 {
3210     char *posixcc = 0;
3211     I32 namedclass = OOB_NAMEDCLASS;
3212
3213     if (value == '[' && RExC_parse + 1 < RExC_end &&
3214         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3215         POSIXCC(UCHARAT(RExC_parse))) {
3216         char  c = UCHARAT(RExC_parse);
3217         char* s = RExC_parse++;
3218         
3219         while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != c)
3220             RExC_parse++;
3221         if (RExC_parse == RExC_end)
3222             /* Grandfather lone [:, [=, [. */
3223             RExC_parse = s;
3224         else {
3225             char* t = RExC_parse++; /* skip over the c */
3226
3227             if (UCHARAT(RExC_parse) == ']') {
3228                 RExC_parse++; /* skip over the ending ] */
3229                 posixcc = s + 1;
3230                 if (*s == ':') {
3231                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3232                     I32 skip = 5; /* the most common skip */
3233
3234                     switch (*posixcc) {
3235                     case 'a':
3236                         if (strnEQ(posixcc, "alnum", 5))
3237                             namedclass =
3238                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3239                         else if (strnEQ(posixcc, "alpha", 5))
3240                             namedclass =
3241                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3242                         else if (strnEQ(posixcc, "ascii", 5))
3243                             namedclass =
3244                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3245                         break;
3246                     case 'b':
3247                         if (strnEQ(posixcc, "blank", 5))
3248                             namedclass =
3249                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3250                         break;
3251                     case 'c':
3252                         if (strnEQ(posixcc, "cntrl", 5))
3253                             namedclass =
3254                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3255                         break;
3256                     case 'd':
3257                         if (strnEQ(posixcc, "digit", 5))
3258                             namedclass =
3259                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3260                         break;
3261                     case 'g':
3262                         if (strnEQ(posixcc, "graph", 5))
3263                             namedclass =
3264                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3265                         break;
3266                     case 'l':
3267                         if (strnEQ(posixcc, "lower", 5))
3268                             namedclass =
3269                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3270                         break;
3271                     case 'p':
3272                         if (strnEQ(posixcc, "print", 5))
3273                             namedclass =
3274                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3275                         else if (strnEQ(posixcc, "punct", 5))
3276                             namedclass =
3277                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3278                         break;
3279                     case 's':
3280                         if (strnEQ(posixcc, "space", 5))
3281                             namedclass =
3282                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3283                         break;
3284                     case 'u':
3285                         if (strnEQ(posixcc, "upper", 5))
3286                             namedclass =
3287                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3288                         break;
3289                     case 'w': /* this is not POSIX, this is the Perl \w */
3290                         if (strnEQ(posixcc, "word", 4)) {
3291                             namedclass =
3292                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3293                             skip = 4;
3294                         }
3295                         break;
3296                     case 'x':
3297                         if (strnEQ(posixcc, "xdigit", 6)) {
3298                             namedclass =
3299                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3300                             skip = 6;
3301                         }
3302                         break;
3303                     }
3304                     if (namedclass == OOB_NAMEDCLASS ||
3305                         posixcc[skip] != ':' ||
3306                         posixcc[skip+1] != ']')
3307                     {
3308                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3309                                       t - s - 1, s + 1);
3310                     }
3311                 } else if (!SIZE_ONLY) {
3312                     /* [[=foo=]] and [[.foo.]] are still future. */
3313
3314                     /* adjust RExC_parse so the warning shows after
3315                        the class closes */
3316                     while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse) != ']')
3317                         RExC_parse++;
3318                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3319                 }
3320             } else {
3321                 /* Maternal grandfather:
3322                  * "[:" ending in ":" but not in ":]" */
3323                 RExC_parse = s;
3324             }
3325         }
3326     }
3327
3328     return namedclass;
3329 }
3330
3331 STATIC void
3332 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3333 {
3334     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3335         POSIXCC(UCHARAT(RExC_parse))) {
3336         char *s = RExC_parse;
3337         char  c = *s++;
3338
3339         while(*s && isALNUM(*s))
3340             s++;
3341         if (*s && c == *s && s[1] == ']') {
3342             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3343
3344             /* [[=foo=]] and [[.foo.]] are still future. */
3345             if (POSIXCC_NOTYET(c)) {
3346                 /* adjust RExC_parse so the error shows after
3347                    the class closes */
3348                 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse++) != ']')
3349                     ;
3350                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3351             }
3352         }
3353     }
3354 }
3355
3356 STATIC regnode *
3357 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3358 {
3359     register UV value;
3360     register UV nextvalue;
3361     register IV prevvalue = OOB_UNICODE;
3362     register IV range = 0;
3363     register regnode *ret;
3364     STRLEN numlen;
3365     IV namedclass;
3366     char *rangebegin = 0;
3367     bool need_class = 0;
3368     SV *listsv = Nullsv;
3369     register char *e;
3370     UV n;
3371     bool optimize_invert = TRUE;
3372
3373     ret = reganode(pRExC_state, ANYOF, 0);
3374
3375     if (!SIZE_ONLY)
3376         ANYOF_FLAGS(ret) = 0;
3377
3378     if (UCHARAT(RExC_parse) == '^') {   /* Complement of range. */
3379         RExC_naughty++;
3380         RExC_parse++;
3381         if (!SIZE_ONLY)
3382             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3383     }
3384
3385     if (SIZE_ONLY)
3386         RExC_size += ANYOF_SKIP;
3387     else {
3388         RExC_emit += ANYOF_SKIP;
3389         if (FOLD)
3390             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3391         if (LOC)
3392             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3393         ANYOF_BITMAP_ZERO(ret);
3394         listsv = newSVpvn("# comment\n", 10);
3395     }
3396
3397     nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
3398
3399     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && POSIXCC(nextvalue))
3400         checkposixcc(pRExC_state);
3401
3402     if (UCHARAT(RExC_parse) == ']' || UCHARAT(RExC_parse) == '-')
3403         goto charclassloop;             /* allow 1st char to be ] or - */
3404
3405     while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != ']') {
3406
3407     charclassloop:
3408
3409         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3410
3411         if (!range)
3412             rangebegin = RExC_parse;
3413         if (UTF) {
3414             value = utf8n_to_uvchr((U8*)RExC_parse,
3415                                    RExC_end - RExC_parse,
3416                                    &numlen, 0);
3417             RExC_parse += numlen;
3418         }
3419         else
3420             value = UCHARAT(RExC_parse++);
3421         nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
3422         if (value == '[' && POSIXCC(nextvalue))
3423             namedclass = regpposixcc(pRExC_state, value);
3424         else if (value == '\\') {
3425             if (UTF) {
3426                 value = utf8n_to_uvchr((U8*)RExC_parse,
3427                                    RExC_end - RExC_parse,
3428                                    &numlen, 0);
3429                 RExC_parse += numlen;
3430             }
3431             else
3432                 value = UCHARAT(RExC_parse++);
3433             /* Some compilers cannot handle switching on 64-bit integer
3434              * values, therefore value cannot be an UV.  Yes, this will
3435              * be a problem later if we want switch on Unicode.
3436              * A similar issue a little bit later when switching on
3437              * namedclass. --jhi */
3438             switch ((I32)value) {
3439             case 'w':   namedclass = ANYOF_ALNUM;       break;
3440             case 'W':   namedclass = ANYOF_NALNUM;      break;
3441             case 's':   namedclass = ANYOF_SPACE;       break;
3442             case 'S':   namedclass = ANYOF_NSPACE;      break;
3443             case 'd':   namedclass = ANYOF_DIGIT;       break;
3444             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3445             case 'p':
3446             case 'P':
3447                 if (*RExC_parse == '{') {
3448                     U8 c = (U8)value;
3449                     e = strchr(RExC_parse++, '}');
3450                     if (!e)
3451                         vFAIL2("Missing right brace on \\%c{}", c);
3452                     while (isSPACE(UCHARAT(RExC_parse)))
3453                         RExC_parse++;
3454                     if (e == RExC_parse)
3455                         vFAIL2("Empty \\%c{}", c);
3456                     n = e - RExC_parse;
3457                     while (isSPACE(UCHARAT(RExC_parse + n - 1)))
3458                         n--;
3459                 }
3460                 else {
3461                     e = RExC_parse;
3462                     n = 1;
3463                 }
3464                 if (!SIZE_ONLY) {
3465                     if (UCHARAT(RExC_parse) == '^') {
3466                          RExC_parse++;
3467                          n--;
3468                          value = value == 'p' ? 'P' : 'p'; /* toggle */
3469                          while (isSPACE(UCHARAT(RExC_parse))) {
3470                               RExC_parse++;
3471                               n--;
3472                          }
3473                     }
3474                     if (value == 'p')
3475                          Perl_sv_catpvf(aTHX_ listsv,
3476                                         "+utf8::%.*s\n", (int)n, RExC_parse);
3477                     else
3478                          Perl_sv_catpvf(aTHX_ listsv,
3479                                         "!utf8::%.*s\n", (int)n, RExC_parse);
3480                 }
3481                 RExC_parse = e + 1;
3482                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3483                 continue;
3484             case 'n':   value = '\n';                   break;
3485             case 'r':   value = '\r';                   break;
3486             case 't':   value = '\t';                   break;
3487             case 'f':   value = '\f';                   break;
3488             case 'b':   value = '\b';                   break;
3489             case 'e':   value = ASCII_TO_NATIVE('\033');break;
3490             case 'a':   value = ASCII_TO_NATIVE('\007');break;
3491             case 'x':
3492                 if (*RExC_parse == '{') {
3493                     I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3494                         | PERL_SCAN_DISALLOW_PREFIX;
3495                     e = strchr(RExC_parse++, '}');
3496                     if (!e)
3497                         vFAIL("Missing right brace on \\x{}");
3498
3499                     numlen = e - RExC_parse;
3500                     value = grok_hex(RExC_parse, &numlen, &flags, NULL);
3501                     RExC_parse = e + 1;
3502                 }
3503                 else {
3504                     I32 flags = PERL_SCAN_DISALLOW_PREFIX;
3505                     numlen = 2;
3506                     value = grok_hex(RExC_parse, &numlen, &flags, NULL);
3507                     RExC_parse += numlen;
3508                 }
3509                 break;
3510             case 'c':
3511                 value = UCHARAT(RExC_parse++);
3512                 value = toCTRL(value);
3513                 break;
3514             case '0': case '1': case '2': case '3': case '4':
3515             case '5': case '6': case '7': case '8': case '9':
3516             {
3517                 I32 flags = 0;
3518                 numlen = 3;
3519                 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
3520                 RExC_parse += numlen;
3521                 break;
3522             }
3523             default:
3524                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3525                     vWARN2(RExC_parse,
3526                            "Unrecognized escape \\%c in character class passed through",
3527                            (int)value);
3528                 break;
3529             }
3530         } /* end of \blah */
3531
3532         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3533
3534             if (!SIZE_ONLY && !need_class)
3535                 ANYOF_CLASS_ZERO(ret);
3536
3537             need_class = 1;
3538
3539             /* a bad range like a-\d, a-[:digit:] ? */
3540             if (range) {
3541                 if (!SIZE_ONLY) {
3542                     if (ckWARN(WARN_REGEXP))
3543                         vWARN4(RExC_parse,
3544                                "False [] range \"%*.*s\"",
3545                                RExC_parse - rangebegin,
3546                                RExC_parse - rangebegin,
3547                                rangebegin);
3548                     if (prevvalue < 256) {
3549                         ANYOF_BITMAP_SET(ret, prevvalue);
3550                         ANYOF_BITMAP_SET(ret, '-');
3551                     }
3552                     else {
3553                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3554                         Perl_sv_catpvf(aTHX_ listsv,
3555                                        "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3556                     }
3557                 }
3558
3559                 range = 0; /* this was not a true range */
3560             }
3561
3562             if (!SIZE_ONLY) {
3563                 if (namedclass > OOB_NAMEDCLASS)
3564                     optimize_invert = FALSE;
3565                 /* Possible truncation here but in some 64-bit environments
3566                  * the compiler gets heartburn about switch on 64-bit values.
3567                  * A similar issue a little earlier when switching on value.
3568                  * --jhi */
3569                 switch ((I32)namedclass) {
3570                 case ANYOF_ALNUM:
3571                     if (LOC)
3572                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3573                     else {
3574                         for (value = 0; value < 256; value++)
3575                             if (isALNUM(value))
3576                                 ANYOF_BITMAP_SET(ret, value);
3577                     }
3578                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3579                     break;
3580                 case ANYOF_NALNUM:
3581                     if (LOC)
3582                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3583                     else {
3584                         for (value = 0; value < 256; value++)
3585                             if (!isALNUM(value))
3586                                 ANYOF_BITMAP_SET(ret, value);
3587                     }
3588                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3589                     break;
3590                 case ANYOF_ALNUMC:
3591                     if (LOC)
3592                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3593                     else {
3594                         for (value = 0; value < 256; value++)
3595                             if (isALNUMC(value))
3596                                 ANYOF_BITMAP_SET(ret, value);
3597                     }
3598                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3599                     break;
3600                 case ANYOF_NALNUMC:
3601                     if (LOC)
3602                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3603                     else {
3604                         for (value = 0; value < 256; value++)
3605                             if (!isALNUMC(value))
3606                                 ANYOF_BITMAP_SET(ret, value);
3607                     }
3608                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3609                     break;
3610                 case ANYOF_ALPHA:
3611                     if (LOC)
3612                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3613                     else {
3614                         for (value = 0; value < 256; value++)
3615                             if (isALPHA(value))
3616                                 ANYOF_BITMAP_SET(ret, value);
3617                     }
3618                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3619                     break;
3620                 case ANYOF_NALPHA:
3621                     if (LOC)
3622                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3623                     else {
3624                         for (value = 0; value < 256; value++)
3625                             if (!isALPHA(value))
3626                                 ANYOF_BITMAP_SET(ret, value);
3627                     }
3628                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3629                     break;
3630                 case ANYOF_ASCII:
3631                     if (LOC)
3632                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3633                     else {
3634 #ifndef EBCDIC
3635                         for (value = 0; value < 128; value++)
3636                             ANYOF_BITMAP_SET(ret, value);
3637 #else  /* EBCDIC */
3638                         for (value = 0; value < 256; value++) {
3639                             if (isASCII(value))
3640                                 ANYOF_BITMAP_SET(ret, value);
3641                         }
3642 #endif /* EBCDIC */
3643                     }
3644                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3645                     break;
3646                 case ANYOF_NASCII:
3647                     if (LOC)
3648                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3649                     else {
3650 #ifndef EBCDIC
3651                         for (value = 128; value < 256; value++)
3652                             ANYOF_BITMAP_SET(ret, value);
3653 #else  /* EBCDIC */
3654                         for (value = 0; value < 256; value++) {
3655                             if (!isASCII(value))
3656                                 ANYOF_BITMAP_SET(ret, value);
3657                         }
3658 #endif /* EBCDIC */
3659                     }
3660                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3661                     break;
3662                 case ANYOF_BLANK:
3663                     if (LOC)
3664                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3665                     else {
3666                         for (value = 0; value < 256; value++)
3667                             if (isBLANK(value))
3668                                 ANYOF_BITMAP_SET(ret, value);
3669                     }
3670                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3671                     break;
3672                 case ANYOF_NBLANK:
3673                     if (LOC)
3674                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3675                     else {
3676                         for (value = 0; value < 256; value++)
3677                             if (!isBLANK(value))
3678                                 ANYOF_BITMAP_SET(ret, value);
3679                     }
3680                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3681                     break;
3682                 case ANYOF_CNTRL:
3683                     if (LOC)
3684                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3685                     else {
3686                         for (value = 0; value < 256; value++)
3687                             if (isCNTRL(value))
3688                                 ANYOF_BITMAP_SET(ret, value);
3689                     }
3690                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3691                     break;
3692                 case ANYOF_NCNTRL:
3693                     if (LOC)
3694                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3695                     else {
3696                         for (value = 0; value < 256; value++)
3697                             if (!isCNTRL(value))
3698                                 ANYOF_BITMAP_SET(ret, value);
3699                     }
3700                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3701                     break;
3702                 case ANYOF_DIGIT:
3703                     if (LOC)
3704                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3705                     else {
3706                         /* consecutive digits assumed */
3707                         for (value = '0'; value <= '9'; value++)
3708                             ANYOF_BITMAP_SET(ret, value);
3709                     }
3710                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3711                     break;
3712                 case ANYOF_NDIGIT:
3713                     if (LOC)
3714                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3715                     else {
3716                         /* consecutive digits assumed */
3717                         for (value = 0; value < '0'; value++)
3718                             ANYOF_BITMAP_SET(ret, value);
3719                         for (value = '9' + 1; value < 256; value++)
3720                             ANYOF_BITMAP_SET(ret, value);
3721                     }
3722                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3723                     break;
3724                 case ANYOF_GRAPH:
3725                     if (LOC)
3726                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3727                     else {
3728                         for (value = 0; value < 256; value++)
3729                             if (isGRAPH(value))
3730                                 ANYOF_BITMAP_SET(ret, value);
3731                     }
3732                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3733                     break;
3734                 case ANYOF_NGRAPH:
3735                     if (LOC)
3736                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3737                     else {
3738                         for (value = 0; value < 256; value++)
3739                             if (!isGRAPH(value))
3740                                 ANYOF_BITMAP_SET(ret, value);
3741                     }
3742                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3743                     break;
3744                 case ANYOF_LOWER:
3745                     if (LOC)
3746                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3747                     else {
3748                         for (value = 0; value < 256; value++)
3749                             if (isLOWER(value))
3750                                 ANYOF_BITMAP_SET(ret, value);
3751                     }
3752                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3753                     break;
3754                 case ANYOF_NLOWER:
3755                     if (LOC)
3756                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3757                     else {
3758                         for (value = 0; value < 256; value++)
3759                             if (!isLOWER(value))
3760                                 ANYOF_BITMAP_SET(ret, value);
3761                     }
3762                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3763                     break;
3764                 case ANYOF_PRINT:
3765                     if (LOC)
3766                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3767                     else {
3768                         for (value = 0; value < 256; value++)
3769                             if (isPRINT(value))
3770                                 ANYOF_BITMAP_SET(ret, value);
3771                     }
3772                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3773                     break;
3774                 case ANYOF_NPRINT:
3775                     if (LOC)
3776                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3777                     else {
3778                         for (value = 0; value < 256; value++)
3779                             if (!isPRINT(value))
3780                                 ANYOF_BITMAP_SET(ret, value);
3781                     }
3782                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3783                     break;
3784                 case ANYOF_PSXSPC:
3785                     if (LOC)
3786                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3787                     else {
3788                         for (value = 0; value < 256; value++)
3789                             if (isPSXSPC(value))
3790                                 ANYOF_BITMAP_SET(ret, value);
3791                     }
3792                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3793                     break;
3794                 case ANYOF_NPSXSPC:
3795                     if (LOC)
3796                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3797                     else {
3798                         for (value = 0; value < 256; value++)
3799                             if (!isPSXSPC(value))
3800                                 ANYOF_BITMAP_SET(ret, value);
3801                     }
3802                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3803                     break;
3804                 case ANYOF_PUNCT:
3805                     if (LOC)
3806                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3807                     else {
3808                         for (value = 0; value < 256; value++)
3809                             if (isPUNCT(value))
3810                                 ANYOF_BITMAP_SET(ret, value);
3811                     }
3812                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3813                     break;
3814                 case ANYOF_NPUNCT:
3815                     if (LOC)
3816                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3817                     else {
3818                         for (value = 0; value < 256; value++)
3819                             if (!isPUNCT(value))
3820                                 ANYOF_BITMAP_SET(ret, value);
3821                     }
3822                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3823                     break;
3824                 case ANYOF_SPACE:
3825                     if (LOC)
3826                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3827                     else {
3828                         for (value = 0; value < 256; value++)
3829                             if (isSPACE(value))
3830                                 ANYOF_BITMAP_SET(ret, value);
3831                     }
3832                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3833                     break;
3834                 case ANYOF_NSPACE:
3835                     if (LOC)
3836                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3837                     else {
3838                         for (value = 0; value < 256; value++)
3839                             if (!isSPACE(value))
3840                                 ANYOF_BITMAP_SET(ret, value);
3841                     }
3842                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3843                     break;
3844                 case ANYOF_UPPER:
3845                     if (LOC)
3846                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3847                     else {
3848                         for (value = 0; value < 256; value++)
3849                             if (isUPPER(value))
3850                                 ANYOF_BITMAP_SET(ret, value);
3851                     }
3852                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3853                     break;
3854                 case ANYOF_NUPPER:
3855                     if (LOC)
3856                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3857                     else {
3858                         for (value = 0; value < 256; value++)
3859                             if (!isUPPER(value))
3860                                 ANYOF_BITMAP_SET(ret, value);
3861                     }
3862                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3863                     break;
3864                 case ANYOF_XDIGIT:
3865                     if (LOC)
3866                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3867                     else {
3868                         for (value = 0; value < 256; value++)
3869                             if (isXDIGIT(value))
3870                                 ANYOF_BITMAP_SET(ret, value);
3871                     }
3872                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3873                     break;
3874                 case ANYOF_NXDIGIT:
3875                     if (LOC)
3876                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3877                     else {
3878                         for (value = 0; value < 256; value++)
3879                             if (!isXDIGIT(value))
3880                                 ANYOF_BITMAP_SET(ret, value);
3881                     }
3882                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3883                     break;
3884                 default:
3885                     vFAIL("Invalid [::] class");
3886                     break;
3887                 }
3888                 if (LOC)
3889                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3890                 continue;
3891             }
3892         } /* end of namedclass \blah */
3893
3894         if (range) {
3895             if (prevvalue > value) /* b-a */ {
3896                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3897                               RExC_parse - rangebegin,
3898                               RExC_parse - rangebegin,
3899                               rangebegin);
3900                 range = 0; /* not a valid range */
3901             }
3902         }
3903         else {
3904             prevvalue = value; /* save the beginning of the range */
3905             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3906                 RExC_parse[1] != ']') {
3907                 RExC_parse++;
3908
3909                 /* a bad range like \w-, [:word:]- ? */
3910                 if (namedclass > OOB_NAMEDCLASS) {
3911                     if (ckWARN(WARN_REGEXP))
3912                         vWARN4(RExC_parse,
3913                                "False [] range \"%*.*s\"",
3914                                RExC_parse - rangebegin,
3915                                RExC_parse - rangebegin,
3916                                rangebegin);
3917                     if (!SIZE_ONLY)
3918                         ANYOF_BITMAP_SET(ret, '-');
3919                 } else
3920                     range = 1;  /* yeah, it's a range! */
3921                 continue;       /* but do it the next time */
3922             }
3923         }
3924
3925         /* now is the next time */
3926         if (!SIZE_ONLY) {
3927             IV i;
3928
3929             if (prevvalue < 256) {
3930                 IV ceilvalue = value < 256 ? value : 255;
3931
3932 #ifdef EBCDIC
3933                 if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3934                     (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3935                 {
3936                     if (isLOWER(prevvalue)) {
3937                         for (i = prevvalue; i <= ceilvalue; i++)
3938                             if (isLOWER(i))
3939                                 ANYOF_BITMAP_SET(ret, i);
3940                     } else {
3941                         for (i = prevvalue; i <= ceilvalue; i++)
3942                             if (isUPPER(i))
3943                                 ANYOF_BITMAP_SET(ret, i);
3944                     }
3945                 }
3946                 else
3947 #endif
3948                     for (i = prevvalue; i <= ceilvalue; i++)
3949                         ANYOF_BITMAP_SET(ret, i);
3950           }
3951           if (value > 255) {
3952                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3953                 if (prevvalue < value)
3954                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3955                                    (UV)prevvalue, (UV)value);
3956                 else if (prevvalue == value)
3957                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3958                                    (UV)value);
3959             }
3960         }
3961
3962         range = 0; /* this range (if it was one) is done now */
3963     }
3964
3965     if (need_class) {
3966         ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3967         if (SIZE_ONLY)
3968             RExC_size += ANYOF_CLASS_ADD_SKIP;
3969         else
3970             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3971     }
3972
3973     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3974     if (!SIZE_ONLY &&
3975          /* If the only flag is folding (plus possibly inversion). */
3976         ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
3977        ) {
3978         for (value = 0; value < 256; ++value) {
3979             if (ANYOF_BITMAP_TEST(ret, value)) {
3980                 IV fold = PL_fold[value];
3981
3982                 if (fold != value)
3983                     ANYOF_BITMAP_SET(ret, fold);
3984             }
3985         }
3986         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3987     }
3988
3989     /* optimize inverted simple patterns (e.g. [^a-z]) */
3990     if (!SIZE_ONLY && optimize_invert &&
3991         /* If the only flag is inversion. */
3992         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3993         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3994             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3995         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3996     }
3997
3998     if (!SIZE_ONLY) {
3999         AV *av = newAV();
4000         SV *rv;
4001
4002         av_store(av, 0, listsv);
4003         av_store(av, 1, NULL);
4004         rv = newRV_noinc((SV*)av);
4005         n = add_data(pRExC_state, 1, "s");
4006         RExC_rx->data->data[n] = (void*)rv;
4007         ARG_SET(ret, n);
4008     }
4009
4010     return ret;
4011 }
4012
4013 STATIC char*
4014 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
4015 {
4016     char* retval = RExC_parse++;
4017
4018     for (;;) {
4019         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
4020                 RExC_parse[2] == '#') {
4021             while (*RExC_parse && *RExC_parse != ')')
4022                 RExC_parse++;
4023             RExC_parse++;
4024             continue;
4025         }
4026         if (RExC_flags16 & PMf_EXTENDED) {
4027             if (isSPACE(*RExC_parse)) {
4028                 RExC_parse++;
4029                 continue;
4030             }
4031             else if (*RExC_parse == '#') {
4032                 while (*RExC_parse && *RExC_parse != '\n')
4033                     RExC_parse++;
4034                 RExC_parse++;
4035                 continue;
4036             }
4037         }
4038         return retval;
4039     }
4040 }
4041
4042 /*
4043 - reg_node - emit a node
4044 */
4045 STATIC regnode *                        /* Location. */
4046 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
4047 {
4048     register regnode *ret;
4049     register regnode *ptr;
4050
4051     ret = RExC_emit;
4052     if (SIZE_ONLY) {
4053         SIZE_ALIGN(RExC_size);
4054         RExC_size += 1;
4055         return(ret);
4056     }
4057
4058     NODE_ALIGN_FILL(ret);
4059     ptr = ret;
4060     FILL_ADVANCE_NODE(ptr, op);
4061     if (RExC_offsets) {         /* MJD */
4062       MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n", 
4063               "reg_node", __LINE__, 
4064               reg_name[op],
4065               RExC_emit - RExC_emit_start > RExC_offsets[0] 
4066               ? "Overwriting end of array!\n" : "OK",
4067               RExC_emit - RExC_emit_start,
4068               RExC_parse - RExC_start,
4069               RExC_offsets[0])); 
4070       Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
4071     }
4072             
4073     RExC_emit = ptr;
4074
4075     return(ret);
4076 }
4077
4078 /*
4079 - reganode - emit a node with an argument
4080 */
4081 STATIC regnode *                        /* Location. */
4082 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
4083 {
4084     register regnode *ret;
4085     register regnode *ptr;
4086
4087     ret = RExC_emit;
4088     if (SIZE_ONLY) {
4089         SIZE_ALIGN(RExC_size);
4090         RExC_size += 2;
4091         return(ret);
4092     }
4093
4094     NODE_ALIGN_FILL(ret);
4095     ptr = ret;
4096     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
4097     if (RExC_offsets) {         /* MJD */
4098       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4099               "reganode",
4100               RExC_emit - RExC_emit_start > RExC_offsets[0] ? 
4101               "Overwriting end of array!\n" : "OK",
4102               RExC_emit - RExC_emit_start,
4103               RExC_parse - RExC_start,
4104               RExC_offsets[0])); 
4105       Set_Cur_Node_Offset;
4106     }
4107             
4108     RExC_emit = ptr;
4109
4110     return(ret);
4111 }
4112
4113 /*
4114 - reguni - emit (if appropriate) a Unicode character
4115 */
4116 STATIC void
4117 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4118 {
4119     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4120 }
4121
4122 /*
4123 - reginsert - insert an operator in front of already-emitted operand
4124 *
4125 * Means relocating the operand.
4126 */
4127 STATIC void
4128 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4129 {
4130     register regnode *src;
4131     register regnode *dst;
4132     register regnode *place;
4133     register int offset = regarglen[(U8)op];
4134
4135 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4136
4137     if (SIZE_ONLY) {
4138         RExC_size += NODE_STEP_REGNODE + offset;
4139         return;
4140     }
4141
4142     src = RExC_emit;
4143     RExC_emit += NODE_STEP_REGNODE + offset;
4144     dst = RExC_emit;
4145     while (src > opnd) {
4146         StructCopy(--src, --dst, regnode);
4147         if (RExC_offsets) {     /* MJD 20010112 */
4148           MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n", 
4149                   "reg_insert",
4150                   dst - RExC_emit_start > RExC_offsets[0] 
4151                   ? "Overwriting end of array!\n" : "OK",
4152                   src - RExC_emit_start,
4153                   dst - RExC_emit_start,
4154                   RExC_offsets[0])); 
4155           Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4156           Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4157         }
4158     }
4159     
4160
4161     place = opnd;               /* Op node, where operand used to be. */
4162     if (RExC_offsets) {         /* MJD */
4163       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4164               "reginsert",
4165               place - RExC_emit_start > RExC_offsets[0] 
4166               ? "Overwriting end of array!\n" : "OK",
4167               place - RExC_emit_start,
4168               RExC_parse - RExC_start,
4169               RExC_offsets[0])); 
4170       Set_Node_Offset(place, RExC_parse);
4171     }
4172     src = NEXTOPER(place);
4173     FILL_ADVANCE_NODE(place, op);
4174     Zero(src, offset, regnode);
4175 }
4176
4177 /*
4178 - regtail - set the next-pointer at the end of a node chain of p to val.
4179 */
4180 STATIC void
4181 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4182 {
4183     register regnode *scan;
4184     register regnode *temp;
4185
4186     if (SIZE_ONLY)
4187         return;
4188
4189     /* Find last node. */
4190     scan = p;
4191     for (;;) {
4192         temp = regnext(scan);
4193         if (temp == NULL)
4194             break;
4195         scan = temp;
4196     }
4197
4198     if (reg_off_by_arg[OP(scan)]) {
4199         ARG_SET(scan, val - scan);
4200     }
4201     else {
4202         NEXT_OFF(scan) = val - scan;
4203     }
4204 }
4205
4206 /*
4207 - regoptail - regtail on operand of first argument; nop if operandless
4208 */
4209 STATIC void
4210 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4211 {
4212     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4213     if (p == NULL || SIZE_ONLY)
4214         return;
4215     if (PL_regkind[(U8)OP(p)] == BRANCH) {
4216         regtail(pRExC_state, NEXTOPER(p), val);
4217     }
4218     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4219         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4220     }
4221     else
4222         return;
4223 }
4224
4225 /*
4226  - regcurly - a little FSA that accepts {\d+,?\d*}
4227  */
4228 STATIC I32
4229 S_regcurly(pTHX_ register char *s)
4230 {
4231     if (*s++ != '{')
4232         return FALSE;
4233     if (!isDIGIT(*s))
4234         return FALSE;
4235     while (isDIGIT(*s))
4236         s++;
4237     if (*s == ',')
4238         s++;
4239     while (isDIGIT(*s))
4240         s++;
4241     if (*s != '}')
4242         return FALSE;
4243     return TRUE;
4244 }
4245
4246
4247 #ifdef DEBUGGING
4248
4249 STATIC regnode *
4250 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4251 {
4252     register U8 op = EXACT;     /* Arbitrary non-END op. */
4253     register regnode *next;
4254
4255     while (op != END && (!last || node < last)) {
4256         /* While that wasn't END last time... */
4257
4258         NODE_ALIGN(node);
4259         op = OP(node);
4260         if (op == CLOSE)
4261             l--;        
4262         next = regnext(node);
4263         /* Where, what. */
4264         if (OP(node) == OPTIMIZED)
4265             goto after_print;
4266         regprop(sv, node);
4267         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4268                       (int)(2*l + 1), "", SvPVX(sv));
4269         if (next == NULL)               /* Next ptr. */
4270             PerlIO_printf(Perl_debug_log, "(0)");
4271         else
4272             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4273         (void)PerlIO_putc(Perl_debug_log, '\n');
4274       after_print:
4275         if (PL_regkind[(U8)op] == BRANCHJ) {
4276             register regnode *nnode = (OP(next) == LONGJMP
4277                                        ? regnext(next)
4278                                        : next);
4279             if (last && nnode > last)
4280                 nnode = last;
4281             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4282         }
4283         else if (PL_regkind[(U8)op] == BRANCH) {
4284             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4285         }
4286         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4287             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4288                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4289         }
4290         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4291             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4292                              next, sv, l + 1);
4293         }
4294         else if ( op == PLUS || op == STAR) {
4295             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4296         }
4297         else if (op == ANYOF) {
4298             /* arglen 1 + class block */
4299             node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4300                     ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4301             node = NEXTOPER(node);
4302         }
4303         else if (PL_regkind[(U8)op] == EXACT) {
4304             /* Literal string, where present. */
4305             node += NODE_SZ_STR(node) - 1;
4306             node = NEXTOPER(node);
4307         }
4308         else {
4309             node = NEXTOPER(node);
4310             node += regarglen[(U8)op];
4311         }
4312         if (op == CURLYX || op == OPEN)
4313             l++;
4314         else if (op == WHILEM)
4315             l--;
4316     }
4317     return node;
4318 }
4319
4320 #endif  /* DEBUGGING */
4321
4322 /*
4323  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4324  */
4325 void
4326 Perl_regdump(pTHX_ regexp *r)
4327 {
4328 #ifdef DEBUGGING
4329     SV *sv = sv_newmortal();
4330
4331     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4332
4333     /* Header fields of interest. */
4334     if (r->anchored_substr)
4335         PerlIO_printf(Perl_debug_log,
4336                       "anchored `%s%.*s%s'%s at %"IVdf" ",
4337                       PL_colors[0],
4338                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4339                       SvPVX(r->anchored_substr),
4340                       PL_colors[1],
4341                       SvTAIL(r->anchored_substr) ? "$" : "",
4342                       (IV)r->anchored_offset);
4343     if (r->float_substr)
4344         PerlIO_printf(Perl_debug_log,
4345                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4346                       PL_colors[0],
4347                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4348                       SvPVX(r->float_substr),
4349                       PL_colors[1],
4350                       SvTAIL(r->float_substr) ? "$" : "",
4351                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4352     if (r->check_substr)
4353         PerlIO_printf(Perl_debug_log,
4354                       r->check_substr == r->float_substr
4355                       ? "(checking floating" : "(checking anchored");
4356     if (r->reganch & ROPT_NOSCAN)
4357         PerlIO_printf(Perl_debug_log, " noscan");
4358     if (r->reganch & ROPT_CHECK_ALL)
4359         PerlIO_printf(Perl_debug_log, " isall");
4360     if (r->check_substr)
4361         PerlIO_printf(Perl_debug_log, ") ");
4362
4363     if (r->regstclass) {
4364         regprop(sv, r->regstclass);
4365         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4366     }
4367     if (r->reganch & ROPT_ANCH) {
4368         PerlIO_printf(Perl_debug_log, "anchored");
4369         if (r->reganch & ROPT_ANCH_BOL)
4370             PerlIO_printf(Perl_debug_log, "(BOL)");
4371         if (r->reganch & ROPT_ANCH_MBOL)
4372             PerlIO_printf(Perl_debug_log, "(MBOL)");
4373         if (r->reganch & ROPT_ANCH_SBOL)
4374             PerlIO_printf(Perl_debug_log, "(SBOL)");
4375         if (r->reganch & ROPT_ANCH_GPOS)
4376             PerlIO_printf(Perl_debug_log, "(GPOS)");
4377         PerlIO_putc(Perl_debug_log, ' ');
4378     }
4379     if (r->reganch & ROPT_GPOS_SEEN)
4380         PerlIO_printf(Perl_debug_log, "GPOS ");
4381     if (r->reganch & ROPT_SKIP)
4382         PerlIO_printf(Perl_debug_log, "plus ");
4383     if (r->reganch & ROPT_IMPLICIT)
4384         PerlIO_printf(Perl_debug_log, "implicit ");
4385     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4386     if (r->reganch & ROPT_EVAL_SEEN)
4387         PerlIO_printf(Perl_debug_log, "with eval ");
4388     PerlIO_printf(Perl_debug_log, "\n");
4389     if (r->offsets) {
4390       U32 i;
4391       U32 len = r->offsets[0];
4392       PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
4393       for (i = 1; i <= len; i++)
4394         PerlIO_printf(Perl_debug_log, "%"UVuf"[%"UVuf"] ", 
4395                       (UV)r->offsets[i*2-1], 
4396                       (UV)r->offsets[i*2]);
4397       PerlIO_printf(Perl_debug_log, "\n");
4398     }
4399 #endif  /* DEBUGGING */
4400 }
4401
4402 #ifdef DEBUGGING
4403
4404 STATIC void
4405 S_put_byte(pTHX_ SV *sv, int c)
4406 {
4407     if (isCNTRL(c) || c == 255 || !isPRINT(c))
4408         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4409     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4410         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4411     else
4412         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4413 }
4414
4415 #endif  /* DEBUGGING */
4416
4417 /*
4418 - regprop - printable representation of opcode
4419 */
4420 void
4421 Perl_regprop(pTHX_ SV *sv, regnode *o)
4422 {
4423 #ifdef DEBUGGING
4424     register int k;
4425
4426     sv_setpvn(sv, "", 0);
4427     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4428         /* It would be nice to FAIL() here, but this may be called from
4429            regexec.c, and it would be hard to supply pRExC_state. */
4430         Perl_croak(aTHX_ "Corrupted regexp opcode");
4431     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4432
4433     k = PL_regkind[(U8)OP(o)];
4434
4435     if (k == EXACT) {
4436         SV *dsv = sv_2mortal(newSVpvn("", 0));
4437         bool do_utf8 = DO_UTF8(sv);
4438         char *s    = do_utf8 ?
4439           pv_uni_display(dsv, (U8*)STRING(o), STR_LEN(o), 60, 0) :
4440           STRING(o);
4441         int len = do_utf8 ?
4442           strlen(s) :
4443           STR_LEN(o);
4444         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>",
4445                        PL_colors[0],
4446                        len, s,
4447                        PL_colors[1]);
4448     }
4449     else if (k == CURLY) {
4450         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4451             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4452         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4453     }
4454     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4455         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4456     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4457         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4458     else if (k == LOGICAL)
4459         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4460     else if (k == ANYOF) {
4461         int i, rangestart = -1;
4462         U8 flags = ANYOF_FLAGS(o);
4463         const char * const anyofs[] = { /* Should be syncronized with
4464                                          * ANYOF_ #xdefines in regcomp.h */
4465             "\\w",
4466             "\\W",
4467             "\\s",
4468             "\\S",
4469             "\\d",
4470             "\\D",
4471             "[:alnum:]",
4472             "[:^alnum:]",
4473             "[:alpha:]",
4474             "[:^alpha:]",
4475             "[:ascii:]",
4476             "[:^ascii:]",
4477             "[:ctrl:]",
4478             "[:^ctrl:]",
4479             "[:graph:]",
4480             "[:^graph:]",
4481             "[:lower:]",
4482             "[:^lower:]",
4483             "[:print:]",
4484             "[:^print:]",
4485             "[:punct:]",
4486             "[:^punct:]",
4487             "[:upper:]",
4488             "[:^upper:]",
4489             "[:xdigit:]",
4490             "[:^xdigit:]",
4491             "[:space:]",
4492             "[:^space:]",
4493             "[:blank:]",
4494             "[:^blank:]"
4495         };
4496
4497         if (flags & ANYOF_LOCALE)
4498             sv_catpv(sv, "{loc}");
4499         if (flags & ANYOF_FOLD)
4500             sv_catpv(sv, "{i}");
4501         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4502         if (flags & ANYOF_INVERT)
4503             sv_catpv(sv, "^");
4504         for (i = 0; i <= 256; i++) {
4505             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4506                 if (rangestart == -1)
4507                     rangestart = i;
4508             } else if (rangestart != -1) {
4509                 if (i <= rangestart + 3)
4510                     for (; rangestart < i; rangestart++)
4511                         put_byte(sv, rangestart);
4512                 else {
4513                     put_byte(sv, rangestart);
4514                     sv_catpv(sv, "-");
4515                     put_byte(sv, i - 1);
4516                 }
4517                 rangestart = -1;
4518             }
4519         }
4520
4521         if (o->flags & ANYOF_CLASS)
4522             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4523                 if (ANYOF_CLASS_TEST(o,i))
4524                     sv_catpv(sv, anyofs[i]);
4525
4526         if (flags & ANYOF_UNICODE)
4527             sv_catpv(sv, "{unicode}");
4528         else if (flags & ANYOF_UNICODE_ALL)
4529             sv_catpv(sv, "{unicode_all}");
4530
4531         {
4532             SV *lv;
4533             SV *sw = regclass_swash(o, FALSE, &lv);
4534         
4535             if (lv) {
4536                 if (sw) {
4537                     UV i;
4538                     U8 s[UTF8_MAXLEN+1];
4539                 
4540                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4541                         U8 *e = uvchr_to_utf8(s, i);
4542                         
4543                         if (i < 256 && swash_fetch(sw, s, TRUE)) {
4544                             if (rangestart == -1)
4545                                 rangestart = i;
4546                         } else if (rangestart != -1) {
4547                             U8 *p;
4548                         
4549                             if (i <= rangestart + 3)
4550                                 for (; rangestart < i; rangestart++) {
4551                                     for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4552                                         put_byte(sv, *p);
4553                                 }
4554                             else {
4555                                 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4556                                     put_byte(sv, *p);
4557                                 sv_catpv(sv, "-");
4558                                     for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4559                                         put_byte(sv, *p);
4560                                 }
4561                                 rangestart = -1;
4562                             }
4563                         }
4564                         
4565                     sv_catpv(sv, "..."); /* et cetera */
4566                 }
4567
4568                 {
4569                     char *s = savepv(SvPVX(lv));
4570                     char *origs = s;
4571                 
4572                     while(*s && *s != '\n') s++;
4573                 
4574                     if (*s == '\n') {
4575                         char *t = ++s;
4576                         
4577                         while (*s) {
4578                             if (*s == '\n')
4579                                 *s = ' ';
4580                             s++;
4581                         }
4582                         if (s[-1] == ' ')
4583                             s[-1] = 0;
4584                         
4585                         sv_catpv(sv, t);
4586                     }
4587                 
4588                     Safefree(origs);
4589                 }
4590             }
4591         }
4592
4593         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4594     }
4595     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4596         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4597 #endif  /* DEBUGGING */
4598 }
4599
4600 SV *
4601 Perl_re_intuit_string(pTHX_ regexp *prog)
4602 {                               /* Assume that RE_INTUIT is set */
4603     DEBUG_r(
4604         {   STRLEN n_a;
4605             char *s = SvPV(prog->check_substr,n_a);
4606
4607             if (!PL_colorset) reginitcolors();
4608             PerlIO_printf(Perl_debug_log,
4609                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4610                       PL_colors[4],PL_colors[5],PL_colors[0],
4611                       s,
4612                       PL_colors[1],
4613                       (strlen(s) > 60 ? "..." : ""));
4614         } );
4615
4616     return prog->check_substr;
4617 }
4618
4619 void
4620 Perl_pregfree(pTHX_ struct regexp *r)
4621 {
4622     DEBUG_r(if (!PL_colorset) reginitcolors());
4623
4624     if (!r || (--r->refcnt > 0))
4625         return;
4626     DEBUG_r(PerlIO_printf(Perl_debug_log,
4627                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4628                       PL_colors[4],PL_colors[5],PL_colors[0],
4629                       r->precomp,
4630                       PL_colors[1],
4631                       (strlen(r->precomp) > 60 ? "..." : "")));
4632
4633     if (r->precomp)
4634         Safefree(r->precomp);
4635     if (r->offsets)             /* 20010421 MJD */
4636         Safefree(r->offsets);
4637     if (RX_MATCH_COPIED(r))
4638         Safefree(r->subbeg);
4639     if (r->substrs) {
4640         if (r->anchored_substr)
4641             SvREFCNT_dec(r->anchored_substr);
4642         if (r->float_substr)
4643             SvREFCNT_dec(r->float_substr);
4644         Safefree(r->substrs);
4645     }
4646     if (r->data) {
4647         int n = r->data->count;
4648         AV* new_comppad = NULL;
4649         AV* old_comppad;
4650         SV** old_curpad;
4651
4652         while (--n >= 0) {
4653           /* If you add a ->what type here, update the comment in regcomp.h */
4654             switch (r->data->what[n]) {
4655             case 's':
4656                 SvREFCNT_dec((SV*)r->data->data[n]);
4657                 break;
4658             case 'f':
4659                 Safefree(r->data->data[n]);
4660                 break;
4661             case 'p':
4662                 new_comppad = (AV*)r->data->data[n];
4663                 break;
4664             case 'o':
4665                 if (new_comppad == NULL)
4666                     Perl_croak(aTHX_ "panic: pregfree comppad");
4667                 old_comppad = PL_comppad;
4668                 old_curpad = PL_curpad;
4669                 /* Watch out for global destruction's random ordering. */
4670                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4671                     PL_comppad = new_comppad;
4672                     PL_curpad = AvARRAY(new_comppad);
4673                 }
4674                 else
4675                     PL_curpad = NULL;
4676
4677                 if (!OpREFCNT_dec((OP_4tree*)r->data->data[n])) {
4678                     op_free((OP_4tree*)r->data->data[n]);
4679                 }
4680
4681                 PL_comppad = old_comppad;
4682                 PL_curpad = old_curpad;
4683                 SvREFCNT_dec((SV*)new_comppad);
4684                 new_comppad = NULL;
4685                 break;
4686             case 'n':
4687                 break;
4688             default:
4689                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4690             }
4691         }
4692         Safefree(r->data->what);
4693         Safefree(r->data);
4694     }
4695     Safefree(r->startp);
4696     Safefree(r->endp);
4697     Safefree(r);
4698 }
4699
4700 /*
4701  - regnext - dig the "next" pointer out of a node
4702  *
4703  * [Note, when REGALIGN is defined there are two places in regmatch()
4704  * that bypass this code for speed.]
4705  */
4706 regnode *
4707 Perl_regnext(pTHX_ register regnode *p)
4708 {
4709     register I32 offset;
4710
4711     if (p == &PL_regdummy)
4712         return(NULL);
4713
4714     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4715     if (offset == 0)
4716         return(NULL);
4717
4718     return(p+offset);
4719 }
4720
4721 STATIC void     
4722 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4723 {
4724     va_list args;
4725     STRLEN l1 = strlen(pat1);
4726     STRLEN l2 = strlen(pat2);
4727     char buf[512];
4728     SV *msv;
4729     char *message;
4730
4731     if (l1 > 510)
4732         l1 = 510;
4733     if (l1 + l2 > 510)
4734         l2 = 510 - l1;
4735     Copy(pat1, buf, l1 , char);
4736     Copy(pat2, buf + l1, l2 , char);
4737     buf[l1 + l2] = '\n';
4738     buf[l1 + l2 + 1] = '\0';
4739 #ifdef I_STDARG
4740     /* ANSI variant takes additional second argument */
4741     va_start(args, pat2);
4742 #else
4743     va_start(args);
4744 #endif
4745     msv = vmess(buf, &args);
4746     va_end(args);
4747     message = SvPV(msv,l1);
4748     if (l1 > 512)
4749         l1 = 512;
4750     Copy(message, buf, l1 , char);
4751     buf[l1] = '\0';                     /* Overwrite \n */
4752     Perl_croak(aTHX_ "%s", buf);
4753 }
4754
4755 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4756
4757 void
4758 Perl_save_re_context(pTHX)
4759 {
4760 #if 0
4761     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4762     SAVEI32(RExC_npar);         /* () count. */
4763     SAVEI32(RExC_size);         /* Code size. */
4764     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4765     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4766     SAVEI32(RExC_seen);         /* from regcomp.c */
4767     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4768     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4769     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4770     SAVEPPTR(RExC_end);         /* End of input for compile */
4771     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4772 #endif
4773
4774     SAVEI32(PL_reg_flags);              /* from regexec.c */
4775     SAVEPPTR(PL_bostr);
4776     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4777     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4778     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4779     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4780     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4781     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4782     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4783     SAVEGENERICPV(PL_reg_start_tmp);            /* from regexec.c */
4784     PL_reg_start_tmp = 0;
4785     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4786     PL_reg_start_tmpl = 0;
4787     SAVEVPTR(PL_regdata);
4788     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4789     SAVEI32(PL_regnarrate);             /* from regexec.c */
4790     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4791     SAVEINT(PL_regindent);              /* from regexec.c */
4792     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4793     SAVEVPTR(PL_curcop);
4794     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4795     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4796     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4797     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4798     SAVEI8(PL_reg_match_utf8);          /* from regexec.c */
4799     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4800     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4801     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4802     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4803     SAVEI32(PL_regnpar);                /* () count. */
4804     SAVEI32(PL_regsize);                /* from regexec.c */
4805 #ifdef DEBUGGING
4806     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */
4807 #endif
4808 }
4809
4810 static void
4811 clear_re(pTHX_ void *r)
4812 {
4813     ReREFCNT_dec((regexp *)r);
4814 }
4815