Re: [ID 20010809.023] perlre misleads when stating that (?i) should be at front of...
[p5sagit/p5-mst-13.2.git] / regcomp.c
1 /*    regcomp.c
2  */
3
4 /*
5  * "A fair jaw-cracker dwarf-language must be."  --Samwise Gamgee
6  */
7
8 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
9  * confused with the original package (see point 3 below).  Thanks, Henry!
10  */
11
12 /* Additional note: this code is very heavily munged from Henry's version
13  * in places.  In some spots I've traded clarity for efficiency, so don't
14  * blame Henry for some of the lack of readability.
15  */
16
17 /* The names of the functions have been changed from regcomp and
18  * regexec to  pregcomp and pregexec in order to avoid conflicts
19  * with the POSIX routines of the same names.
20 */
21
22 #ifdef PERL_EXT_RE_BUILD
23 /* need to replace pregcomp et al, so enable that */
24 #  ifndef PERL_IN_XSUB_RE
25 #    define PERL_IN_XSUB_RE
26 #  endif
27 /* need access to debugger hooks */
28 #  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
29 #    define DEBUGGING
30 #  endif
31 #endif
32
33 #ifdef PERL_IN_XSUB_RE
34 /* We *really* need to overwrite these symbols: */
35 #  define Perl_pregcomp my_regcomp
36 #  define Perl_regdump my_regdump
37 #  define Perl_regprop my_regprop
38 #  define Perl_pregfree my_regfree
39 #  define Perl_re_intuit_string my_re_intuit_string
40 /* *These* symbols are masked to allow static link. */
41 #  define Perl_regnext my_regnext
42 #  define Perl_save_re_context my_save_re_context
43 #  define Perl_reginitcolors my_reginitcolors
44
45 #  define PERL_NO_GET_CONTEXT
46 #endif
47
48 /*SUPPRESS 112*/
49 /*
50  * pregcomp and pregexec -- regsub and regerror are not used in perl
51  *
52  *      Copyright (c) 1986 by University of Toronto.
53  *      Written by Henry Spencer.  Not derived from licensed software.
54  *
55  *      Permission is granted to anyone to use this software for any
56  *      purpose on any computer system, and to redistribute it freely,
57  *      subject to the following restrictions:
58  *
59  *      1. The author is not responsible for the consequences of use of
60  *              this software, no matter how awful, even if they arise
61  *              from defects in it.
62  *
63  *      2. The origin of this software must not be misrepresented, either
64  *              by explicit claim or by omission.
65  *
66  *      3. Altered versions must be plainly marked as such, and must not
67  *              be misrepresented as being the original software.
68  *
69  *
70  ****    Alterations to Henry's code are...
71  ****
72  ****    Copyright (c) 1991-2001, Larry Wall
73  ****
74  ****    You may distribute under the terms of either the GNU General Public
75  ****    License or the Artistic License, as specified in the README file.
76
77  *
78  * Beware that some of this code is subtly aware of the way operator
79  * precedence is structured in regular expressions.  Serious changes in
80  * regular-expression syntax might require a total rethink.
81  */
82 #include "EXTERN.h"
83 #define PERL_IN_REGCOMP_C
84 #include "perl.h"
85
86 #ifdef PERL_IN_XSUB_RE
87 #  if defined(PERL_CAPI) || defined(PERL_OBJECT)
88 #    include "XSUB.h"
89 #  endif
90 #else
91 #  include "INTERN.h"
92 #endif
93
94 #define REG_COMP_C
95 #include "regcomp.h"
96
97 #ifdef op
98 #undef op
99 #endif /* op */
100
101 #ifdef MSDOS
102 # if defined(BUGGY_MSC6)
103  /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
104  # pragma optimize("a",off)
105  /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
106  # pragma optimize("w",on )
107 # endif /* BUGGY_MSC6 */
108 #endif /* MSDOS */
109
110 #ifndef STATIC
111 #define STATIC  static
112 #endif
113
114 typedef struct RExC_state_t {
115     U16         flags16;                /* are we folding, multilining? */
116     char        *precomp;               /* uncompiled string. */
117     regexp      *rx;
118     char        *start;                 /* Start of input for compile */
119     char        *end;                   /* End of input for compile */
120     char        *parse;                 /* Input-scan pointer. */
121     I32         whilem_seen;            /* number of WHILEM in this expr */
122     regnode     *emit_start;            /* Start of emitted-code area */
123     regnode     *emit;                  /* Code-emit pointer; &regdummy = don't = compiling */
124     I32         naughty;                /* How bad is this pattern? */
125     I32         sawback;                /* Did we see \1, ...? */
126     U32         seen;
127     I32         size;                   /* Code size. */
128     I32         npar;                   /* () count. */
129     I32         extralen;
130     I32         seen_zerolen;
131     I32         seen_evals;
132     I32         utf8;
133 #if ADD_TO_REGEXEC
134     char        *starttry;              /* -Dr: where regtry was called. */
135 #define RExC_starttry   (pRExC_state->starttry)
136 #endif
137 } RExC_state_t;
138
139 #define RExC_flags16    (pRExC_state->flags16)
140 #define RExC_precomp    (pRExC_state->precomp)
141 #define RExC_rx         (pRExC_state->rx)
142 #define RExC_start      (pRExC_state->start)
143 #define RExC_end        (pRExC_state->end)
144 #define RExC_parse      (pRExC_state->parse)
145 #define RExC_whilem_seen        (pRExC_state->whilem_seen)
146 #define RExC_offsets    (pRExC_state->rx->offsets) /* I am not like the others */
147 #define RExC_emit       (pRExC_state->emit)
148 #define RExC_emit_start (pRExC_state->emit_start)
149 #define RExC_naughty    (pRExC_state->naughty)
150 #define RExC_sawback    (pRExC_state->sawback)
151 #define RExC_seen       (pRExC_state->seen)
152 #define RExC_size       (pRExC_state->size)
153 #define RExC_npar       (pRExC_state->npar)
154 #define RExC_extralen   (pRExC_state->extralen)
155 #define RExC_seen_zerolen       (pRExC_state->seen_zerolen)
156 #define RExC_seen_evals (pRExC_state->seen_evals)
157 #define RExC_utf8       (pRExC_state->utf8)
158
159 #define ISMULT1(c)      ((c) == '*' || (c) == '+' || (c) == '?')
160 #define ISMULT2(s)      ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
161         ((*s) == '{' && regcurly(s)))
162
163 #ifdef SPSTART
164 #undef SPSTART          /* dratted cpp namespace... */
165 #endif
166 /*
167  * Flags to be passed up and down.
168  */
169 #define WORST           0       /* Worst case. */
170 #define HASWIDTH        0x1     /* Known to match non-null strings. */
171 #define SIMPLE          0x2     /* Simple enough to be STAR/PLUS operand. */
172 #define SPSTART         0x4     /* Starts with * or +. */
173 #define TRYAGAIN        0x8     /* Weeded out a declaration. */
174
175 /* Length of a variant. */
176
177 typedef struct scan_data_t {
178     I32 len_min;
179     I32 len_delta;
180     I32 pos_min;
181     I32 pos_delta;
182     SV *last_found;
183     I32 last_end;                       /* min value, <0 unless valid. */
184     I32 last_start_min;
185     I32 last_start_max;
186     SV **longest;                       /* Either &l_fixed, or &l_float. */
187     SV *longest_fixed;
188     I32 offset_fixed;
189     SV *longest_float;
190     I32 offset_float_min;
191     I32 offset_float_max;
192     I32 flags;
193     I32 whilem_c;
194     I32 *last_closep;
195     struct regnode_charclass_class *start_class;
196 } scan_data_t;
197
198 /*
199  * Forward declarations for pregcomp()'s friends.
200  */
201
202 static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
203                                       0, 0, 0, 0, 0, 0};
204
205 #define SF_BEFORE_EOL           (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
206 #define SF_BEFORE_SEOL          0x1
207 #define SF_BEFORE_MEOL          0x2
208 #define SF_FIX_BEFORE_EOL       (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
209 #define SF_FL_BEFORE_EOL        (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
210
211 #ifdef NO_UNARY_PLUS
212 #  define SF_FIX_SHIFT_EOL      (0+2)
213 #  define SF_FL_SHIFT_EOL               (0+4)
214 #else
215 #  define SF_FIX_SHIFT_EOL      (+2)
216 #  define SF_FL_SHIFT_EOL               (+4)
217 #endif
218
219 #define SF_FIX_BEFORE_SEOL      (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
220 #define SF_FIX_BEFORE_MEOL      (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
221
222 #define SF_FL_BEFORE_SEOL       (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
223 #define SF_FL_BEFORE_MEOL       (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
224 #define SF_IS_INF               0x40
225 #define SF_HAS_PAR              0x80
226 #define SF_IN_PAR               0x100
227 #define SF_HAS_EVAL             0x200
228 #define SCF_DO_SUBSTR           0x400
229 #define SCF_DO_STCLASS_AND      0x0800
230 #define SCF_DO_STCLASS_OR       0x1000
231 #define SCF_DO_STCLASS          (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
232 #define SCF_WHILEM_VISITED_POS  0x2000
233
234 #define UTF RExC_utf8
235 #define LOC (RExC_flags16 & PMf_LOCALE)
236 #define FOLD (RExC_flags16 & PMf_FOLD)
237
238 #define OOB_UNICODE             12345678
239 #define OOB_NAMEDCLASS          -1
240
241 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
242 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
243
244
245 /* length of regex to show in messages that don't mark a position within */
246 #define RegexLengthToShowInErrorMessages 127
247
248 /*
249  * If MARKER[12] are adjusted, be sure to adjust the constants at the top
250  * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
251  * op/pragma/warn/regcomp.
252  */
253 #define MARKER1 "<-- HERE"    /* marker as it appears in the description */
254 #define MARKER2 " <-- HERE "  /* marker as it appears within the regex */
255
256 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
257
258 /*
259  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
260  * arg. Show regex, up to a maximum length. If it's too long, chop and add
261  * "...".
262  */
263 #define FAIL(msg)                                                             \
264     STMT_START {                                                             \
265         char *ellipses = "";                                                 \
266         unsigned len = strlen(RExC_precomp);                                \
267                                                                              \
268         if (!SIZE_ONLY)                                                      \
269             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
270                                                                              \
271         if (len > RegexLengthToShowInErrorMessages) {                        \
272             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
273             len = RegexLengthToShowInErrorMessages - 10;                     \
274             ellipses = "...";                                                \
275         }                                                                    \
276         Perl_croak(aTHX_ "%s in regex m/%.*s%s/",                            \
277                    msg, (int)len, RExC_precomp, ellipses);                  \
278     } STMT_END
279
280 /*
281  * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
282  * args. Show regex, up to a maximum length. If it's too long, chop and add
283  * "...".
284  */
285 #define FAIL2(pat,msg)                                                        \
286     STMT_START {                                                             \
287         char *ellipses = "";                                                 \
288         unsigned len = strlen(RExC_precomp);                                \
289                                                                              \
290         if (!SIZE_ONLY)                                                      \
291             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
292                                                                              \
293         if (len > RegexLengthToShowInErrorMessages) {                        \
294             /* chop 10 shorter than the max, to ensure meaning of "..." */   \
295             len = RegexLengthToShowInErrorMessages - 10;                     \
296             ellipses = "...";                                                \
297         }                                                                    \
298         S_re_croak2(aTHX_ pat, " in regex m/%.*s%s/",                        \
299                     msg, (int)len, RExC_precomp, ellipses);                \
300     } STMT_END
301
302
303 /*
304  * Simple_vFAIL -- like FAIL, but marks the current location in the scan
305  */
306 #define Simple_vFAIL(m)                                                      \
307     STMT_START {                                                             \
308       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
309                                                                              \
310       Perl_croak(aTHX_ "%s" REPORT_LOCATION,               \
311                  m, (int)offset, RExC_precomp, RExC_precomp + offset);     \
312     } STMT_END
313
314 /*
315  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
316  */
317 #define vFAIL(m)                                                             \
318     STMT_START {                                                             \
319       if (!SIZE_ONLY)                                                        \
320             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
321       Simple_vFAIL(m);                                                       \
322     } STMT_END
323
324 /*
325  * Like Simple_vFAIL(), but accepts two arguments.
326  */
327 #define Simple_vFAIL2(m,a1)                                                  \
328     STMT_START {                                                             \
329       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
330                                                                              \
331       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1,       \
332                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
333     } STMT_END
334
335 /*
336  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
337  */
338 #define vFAIL2(m,a1)                                                         \
339     STMT_START {                                                             \
340       if (!SIZE_ONLY)                                                        \
341             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
342       Simple_vFAIL2(m, a1);                                                  \
343     } STMT_END
344
345
346 /*
347  * Like Simple_vFAIL(), but accepts three arguments.
348  */
349 #define Simple_vFAIL3(m, a1, a2)                                             \
350     STMT_START {                                                             \
351       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
352                                                                              \
353       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2,   \
354                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
355     } STMT_END
356
357 /*
358  * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
359  */
360 #define vFAIL3(m,a1,a2)                                                      \
361     STMT_START {                                                             \
362       if (!SIZE_ONLY)                                                        \
363             SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx);                 \
364       Simple_vFAIL3(m, a1, a2);                                              \
365     } STMT_END
366
367 /*
368  * Like Simple_vFAIL(), but accepts four arguments.
369  */
370 #define Simple_vFAIL4(m, a1, a2, a3)                                         \
371     STMT_START {                                                             \
372       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
373                                                                              \
374       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3,\
375                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
376     } STMT_END
377
378 /*
379  * Like Simple_vFAIL(), but accepts five arguments.
380  */
381 #define Simple_vFAIL5(m, a1, a2, a3, a4)                                     \
382     STMT_START {                                                             \
383       unsigned offset = strlen(RExC_precomp)-(RExC_end-RExC_parse); \
384       S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, a4,\
385                   (int)offset, RExC_precomp, RExC_precomp + offset);       \
386     } STMT_END
387
388
389 #define vWARN(loc,m)                                                         \
390     STMT_START {                                                             \
391         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
392         Perl_warner(aTHX_ WARN_REGEXP, "%s" REPORT_LOCATION,\
393                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
394     } STMT_END                                                               \
395
396 #define vWARNdep(loc,m)                                                         \
397     STMT_START {                                                             \
398         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
399         int warn_cat = ckWARN(WARN_REGEXP) ? WARN_REGEXP : WARN_DEPRECATED;  \
400         Perl_warner(aTHX_ warn_cat, "%s" REPORT_LOCATION,\
401                  m, (int)offset, RExC_precomp, RExC_precomp + offset);          \
402     } STMT_END                                                               \
403
404
405 #define vWARN2(loc, m, a1)                                                   \
406     STMT_START {                                                             \
407         unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));          \
408         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
409                  a1,                                                         \
410                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
411     } STMT_END
412
413 #define vWARN3(loc, m, a1, a2)                                               \
414     STMT_START {                                                             \
415       unsigned offset = strlen(RExC_precomp) - (RExC_end - (loc));        \
416         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,                    \
417                  a1, a2,                                                     \
418                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
419     } STMT_END
420
421 #define vWARN4(loc, m, a1, a2, a3)                                           \
422     STMT_START {                                                             \
423       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));            \
424         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,\
425                  a1, a2, a3,                                                 \
426                  (int)offset, RExC_precomp, RExC_precomp + offset);        \
427     } STMT_END
428
429 /* used for the parse_flags section for (?c) -- japhy */
430 #define vWARN5(loc, m, a1, a2, a3, a4)                                       \
431   STMT_START {                                                   \
432       unsigned offset = strlen(RExC_precomp)-(RExC_end-(loc));   \
433         Perl_warner(aTHX_ WARN_REGEXP, m REPORT_LOCATION,      \
434                  a1, a2, a3, a4,                                 \
435                  (int)offset, RExC_precomp, RExC_precomp + offset);  \
436     } STMT_END
437
438
439 /* Allow for side effects in s */
440 #define REGC(c,s) STMT_START { if (!SIZE_ONLY) *(s) = (c); else (void)(s);} STMT_END
441
442 /* Macros for recording node offsets.   20001227 mjd@plover.com 
443  * Nodes are numbered 1, 2, 3, 4.  Node #n's position is recorded in
444  * element 2*n-1 of the array.  Element #2n holds the byte length node #n.
445  * Element 0 holds the number n.
446  */
447
448 #define MJD_OFFSET_DEBUG(x)
449 /* #define MJD_OFFSET_DEBUG(x) fprintf x */
450
451
452 #  define Set_Node_Offset_To_R(node,byte)                           \
453    STMT_START {                                        \
454      if (! SIZE_ONLY) {                                  \
455        if((node) < 0) {                   \
456          Perl_croak(aTHX_ "value of node is %d in Offset macro", node); \
457        } else {                                                        \
458          RExC_offsets[2*(node)-1] = (byte);                               \
459        }                                                               \
460      }                                                                 \
461    } STMT_END
462
463 #  define Set_Node_Offset(node,byte) Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
464 #  define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
465
466 #  define Set_Node_Length_To_R(node,len)                            \
467    STMT_START {                                        \
468      if (! SIZE_ONLY) {                                  \
469        MJD_OFFSET_DEBUG((stderr, "** (%d) size of node %d is %d.\n", __LINE__, (node), (len))); \
470        if((node) < 0) {                   \
471          Perl_croak(aTHX_ "value of node is %d in Length macro", node); \
472        } else {                                                        \
473          RExC_offsets[2*(node)] = (len);                               \
474        }                                                               \
475      }                                                                 \
476    } STMT_END
477
478 #  define Set_Node_Length(node,len) Set_Node_Length_To_R((node)-RExC_emit_start, len)
479 #  define Set_Cur_Node_Length(len)  Set_Node_Length(RExC_emit, len)
480 #  define Set_Node_Cur_Length(node)   Set_Node_Length(node, RExC_parse - parse_start)
481
482 /* Get offsets and lengths */
483 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
484 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
485
486 static void clear_re(pTHXo_ void *r);
487
488 /* Mark that we cannot extend a found fixed substring at this point.
489    Updata the longest found anchored substring and the longest found
490    floating substrings if needed. */
491
492 STATIC void
493 S_scan_commit(pTHX_ RExC_state_t *pRExC_state, scan_data_t *data)
494 {
495     STRLEN l = CHR_SVLEN(data->last_found);
496     STRLEN old_l = CHR_SVLEN(*data->longest);
497
498     if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
499         sv_setsv(*data->longest, data->last_found);
500         if (*data->longest == data->longest_fixed) {
501             data->offset_fixed = l ? data->last_start_min : data->pos_min;
502             if (data->flags & SF_BEFORE_EOL)
503                 data->flags
504                     |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
505             else
506                 data->flags &= ~SF_FIX_BEFORE_EOL;
507         }
508         else {
509             data->offset_float_min = l ? data->last_start_min : data->pos_min;
510             data->offset_float_max = (l
511                                       ? data->last_start_max
512                                       : data->pos_min + data->pos_delta);
513             if (data->flags & SF_BEFORE_EOL)
514                 data->flags
515                     |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
516             else
517                 data->flags &= ~SF_FL_BEFORE_EOL;
518         }
519     }
520     SvCUR_set(data->last_found, 0);
521     data->last_end = -1;
522     data->flags &= ~SF_BEFORE_EOL;
523 }
524
525 /* Can match anything (initialization) */
526 STATIC void
527 S_cl_anything(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
528 {
529     int value;
530
531     ANYOF_CLASS_ZERO(cl);
532     for (value = 0; value < 256; ++value)
533         ANYOF_BITMAP_SET(cl, value);
534     cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
535     if (LOC)
536         cl->flags |= ANYOF_LOCALE;
537 }
538
539 /* Can match anything (initialization) */
540 STATIC int
541 S_cl_is_anything(pTHX_ struct regnode_charclass_class *cl)
542 {
543     int value;
544
545     for (value = 0; value <= ANYOF_MAX; value += 2)
546         if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
547             return 1;
548     if (!(cl->flags & ANYOF_UNICODE_ALL))
549         return 0;
550     for (value = 0; value < 256; ++value)
551         if (!ANYOF_BITMAP_TEST(cl, value))
552             return 0;
553     return 1;
554 }
555
556 /* Can match anything (initialization) */
557 STATIC void
558 S_cl_init(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
559 {
560     Zero(cl, 1, struct regnode_charclass_class);
561     cl->type = ANYOF;
562     cl_anything(pRExC_state, cl);
563 }
564
565 STATIC void
566 S_cl_init_zero(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
567 {
568     Zero(cl, 1, struct regnode_charclass_class);
569     cl->type = ANYOF;
570     cl_anything(pRExC_state, cl);
571     if (LOC)
572         cl->flags |= ANYOF_LOCALE;
573 }
574
575 /* 'And' a given class with another one.  Can create false positives */
576 /* We assume that cl is not inverted */
577 STATIC void
578 S_cl_and(pTHX_ struct regnode_charclass_class *cl,
579          struct regnode_charclass_class *and_with)
580 {
581     if (!(and_with->flags & ANYOF_CLASS)
582         && !(cl->flags & ANYOF_CLASS)
583         && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
584         && !(and_with->flags & ANYOF_FOLD)
585         && !(cl->flags & ANYOF_FOLD)) {
586         int i;
587
588         if (and_with->flags & ANYOF_INVERT)
589             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
590                 cl->bitmap[i] &= ~and_with->bitmap[i];
591         else
592             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
593                 cl->bitmap[i] &= and_with->bitmap[i];
594     } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
595     if (!(and_with->flags & ANYOF_EOS))
596         cl->flags &= ~ANYOF_EOS;
597
598     if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE) {
599         cl->flags &= ~ANYOF_UNICODE_ALL;
600         cl->flags |= ANYOF_UNICODE;
601         ARG_SET(cl, ARG(and_with));
602     }
603     if (!(and_with->flags & ANYOF_UNICODE_ALL))
604         cl->flags &= ~ANYOF_UNICODE_ALL;
605     if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)))
606         cl->flags &= ~ANYOF_UNICODE;
607 }
608
609 /* 'OR' a given class with another one.  Can create false positives */
610 /* We assume that cl is not inverted */
611 STATIC void
612 S_cl_or(pTHX_ RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
613 {
614     if (or_with->flags & ANYOF_INVERT) {
615         /* We do not use
616          * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
617          *   <= (B1 | !B2) | (CL1 | !CL2)
618          * which is wasteful if CL2 is small, but we ignore CL2:
619          *   (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
620          * XXXX Can we handle case-fold?  Unclear:
621          *   (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
622          *   (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
623          */
624         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
625              && !(or_with->flags & ANYOF_FOLD)
626              && !(cl->flags & ANYOF_FOLD) ) {
627             int i;
628
629             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
630                 cl->bitmap[i] |= ~or_with->bitmap[i];
631         } /* XXXX: logic is complicated otherwise */
632         else {
633             cl_anything(pRExC_state, cl);
634         }
635     } else {
636         /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
637         if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
638              && (!(or_with->flags & ANYOF_FOLD)
639                  || (cl->flags & ANYOF_FOLD)) ) {
640             int i;
641
642             /* OR char bitmap and class bitmap separately */
643             for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
644                 cl->bitmap[i] |= or_with->bitmap[i];
645             if (or_with->flags & ANYOF_CLASS) {
646                 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
647                     cl->classflags[i] |= or_with->classflags[i];
648                 cl->flags |= ANYOF_CLASS;
649             }
650         }
651         else { /* XXXX: logic is complicated, leave it along for a moment. */
652             cl_anything(pRExC_state, cl);
653         }
654     }
655     if (or_with->flags & ANYOF_EOS)
656         cl->flags |= ANYOF_EOS;
657
658     if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
659         ARG(cl) != ARG(or_with)) {
660         cl->flags |= ANYOF_UNICODE_ALL;
661         cl->flags &= ~ANYOF_UNICODE;
662     }
663     if (or_with->flags & ANYOF_UNICODE_ALL) {
664         cl->flags |= ANYOF_UNICODE_ALL;
665         cl->flags &= ~ANYOF_UNICODE;
666     }
667 }
668
669 /* REx optimizer.  Converts nodes into quickier variants "in place".
670    Finds fixed substrings.  */
671
672 /* Stops at toplevel WHILEM as well as at `last'. At end *scanp is set
673    to the position after last scanned or to NULL. */
674
675 STATIC I32
676 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
677                         /* scanp: Start here (read-write). */
678                         /* deltap: Write maxlen-minlen here. */
679                         /* last: Stop before this one. */
680 {
681     I32 min = 0, pars = 0, code;
682     regnode *scan = *scanp, *next;
683     I32 delta = 0;
684     int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
685     int is_inf_internal = 0;            /* The studied chunk is infinite */
686     I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
687     scan_data_t data_fake;
688     struct regnode_charclass_class and_with; /* Valid if flags & SCF_DO_STCLASS_OR */
689
690     while (scan && OP(scan) != END && scan < last) {
691         /* Peephole optimizer: */
692
693         if (PL_regkind[(U8)OP(scan)] == EXACT) {
694             /* Merge several consecutive EXACTish nodes into one. */
695             regnode *n = regnext(scan);
696             U32 stringok = 1;
697 #ifdef DEBUGGING
698             regnode *stop = scan;
699 #endif
700
701             next = scan + NODE_SZ_STR(scan);
702             /* Skip NOTHING, merge EXACT*. */
703             while (n &&
704                    ( PL_regkind[(U8)OP(n)] == NOTHING ||
705                      (stringok && (OP(n) == OP(scan))))
706                    && NEXT_OFF(n)
707                    && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
708                 if (OP(n) == TAIL || n > next)
709                     stringok = 0;
710                 if (PL_regkind[(U8)OP(n)] == NOTHING) {
711                     NEXT_OFF(scan) += NEXT_OFF(n);
712                     next = n + NODE_STEP_REGNODE;
713 #ifdef DEBUGGING
714                     if (stringok)
715                         stop = n;
716 #endif
717                     n = regnext(n);
718                 }
719                 else if (stringok) {
720                     int oldl = STR_LEN(scan);
721                     regnode *nnext = regnext(n);
722
723                     if (oldl + STR_LEN(n) > U8_MAX)
724                         break;
725                     NEXT_OFF(scan) += NEXT_OFF(n);
726                     STR_LEN(scan) += STR_LEN(n);
727                     next = n + NODE_SZ_STR(n);
728                     /* Now we can overwrite *n : */
729                     Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
730 #ifdef DEBUGGING
731                     stop = next - 1;
732 #endif
733                     n = nnext;
734                 }
735             }
736 #ifdef DEBUGGING
737             /* Allow dumping */
738             n = scan + NODE_SZ_STR(scan);
739             while (n <= stop) {
740                 if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
741                     OP(n) = OPTIMIZED;
742                     NEXT_OFF(n) = 0;
743                 }
744                 n++;
745             }
746 #endif
747         }
748         /* Follow the next-chain of the current node and optimize
749            away all the NOTHINGs from it.  */
750         if (OP(scan) != CURLYX) {
751             int max = (reg_off_by_arg[OP(scan)]
752                        ? I32_MAX
753                        /* I32 may be smaller than U16 on CRAYs! */
754                        : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
755             int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
756             int noff;
757             regnode *n = scan;
758         
759             /* Skip NOTHING and LONGJMP. */
760             while ((n = regnext(n))
761                    && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
762                        || ((OP(n) == LONGJMP) && (noff = ARG(n))))
763                    && off + noff < max)
764                 off += noff;
765             if (reg_off_by_arg[OP(scan)])
766                 ARG(scan) = off;
767             else
768                 NEXT_OFF(scan) = off;
769         }
770         /* The principal pseudo-switch.  Cannot be a switch, since we
771            look into several different things.  */
772         if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
773                    || OP(scan) == IFTHEN || OP(scan) == SUSPEND) {
774             next = regnext(scan);
775             code = OP(scan);
776         
777             if (OP(next) == code || code == IFTHEN || code == SUSPEND) {
778                 I32 max1 = 0, min1 = I32_MAX, num = 0;
779                 struct regnode_charclass_class accum;
780                 
781                 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
782                     scan_commit(pRExC_state, data); /* Cannot merge strings after this. */
783                 if (flags & SCF_DO_STCLASS)
784                     cl_init_zero(pRExC_state, &accum);
785                 while (OP(scan) == code) {
786                     I32 deltanext, minnext, f = 0, fake;
787                     struct regnode_charclass_class this_class;
788
789                     num++;
790                     data_fake.flags = 0;
791                     if (data) {         
792                         data_fake.whilem_c = data->whilem_c;
793                         data_fake.last_closep = data->last_closep;
794                     }
795                     else
796                         data_fake.last_closep = &fake;
797                     next = regnext(scan);
798                     scan = NEXTOPER(scan);
799                     if (code != BRANCH)
800                         scan = NEXTOPER(scan);
801                     if (flags & SCF_DO_STCLASS) {
802                         cl_init(pRExC_state, &this_class);
803                         data_fake.start_class = &this_class;
804                         f = SCF_DO_STCLASS_AND;
805                     }           
806                     if (flags & SCF_WHILEM_VISITED_POS)
807                         f |= SCF_WHILEM_VISITED_POS;
808                     /* we suppose the run is continuous, last=next...*/
809                     minnext = study_chunk(pRExC_state, &scan, &deltanext,
810                                           next, &data_fake, f);
811                     if (min1 > minnext)
812                         min1 = minnext;
813                     if (max1 < minnext + deltanext)
814                         max1 = minnext + deltanext;
815                     if (deltanext == I32_MAX)
816                         is_inf = is_inf_internal = 1;
817                     scan = next;
818                     if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
819                         pars++;
820                     if (data && (data_fake.flags & SF_HAS_EVAL))
821                         data->flags |= SF_HAS_EVAL;
822                     if (data)
823                         data->whilem_c = data_fake.whilem_c;
824                     if (flags & SCF_DO_STCLASS)
825                         cl_or(pRExC_state, &accum, &this_class);
826                     if (code == SUSPEND)
827                         break;
828                 }
829                 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
830                     min1 = 0;
831                 if (flags & SCF_DO_SUBSTR) {
832                     data->pos_min += min1;
833                     data->pos_delta += max1 - min1;
834                     if (max1 != min1 || is_inf)
835                         data->longest = &(data->longest_float);
836                 }
837                 min += min1;
838                 delta += max1 - min1;
839                 if (flags & SCF_DO_STCLASS_OR) {
840                     cl_or(pRExC_state, data->start_class, &accum);
841                     if (min1) {
842                         cl_and(data->start_class, &and_with);
843                         flags &= ~SCF_DO_STCLASS;
844                     }
845                 }
846                 else if (flags & SCF_DO_STCLASS_AND) {
847                     if (min1) {
848                         cl_and(data->start_class, &accum);
849                         flags &= ~SCF_DO_STCLASS;
850                     }
851                     else {
852                         /* Switch to OR mode: cache the old value of
853                          * data->start_class */
854                         StructCopy(data->start_class, &and_with,
855                                    struct regnode_charclass_class);
856                         flags &= ~SCF_DO_STCLASS_AND;
857                         StructCopy(&accum, data->start_class,
858                                    struct regnode_charclass_class);
859                         flags |= SCF_DO_STCLASS_OR;
860                         data->start_class->flags |= ANYOF_EOS;
861                     }
862                 }
863             }
864             else if (code == BRANCHJ)   /* single branch is optimized. */
865                 scan = NEXTOPER(NEXTOPER(scan));
866             else                        /* single branch is optimized. */
867                 scan = NEXTOPER(scan);
868             continue;
869         }
870         else if (OP(scan) == EXACT) {
871             I32 l = STR_LEN(scan);
872             UV uc = *((U8*)STRING(scan));
873             if (UTF) {
874                 U8 *s = (U8*)STRING(scan);
875                 l = utf8_length(s, s + l);
876                 uc = utf8_to_uvchr(s, NULL);
877             }
878             min += l;
879             if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
880                 /* The code below prefers earlier match for fixed
881                    offset, later match for variable offset.  */
882                 if (data->last_end == -1) { /* Update the start info. */
883                     data->last_start_min = data->pos_min;
884                     data->last_start_max = is_inf
885                         ? I32_MAX : data->pos_min + data->pos_delta;
886                 }
887                 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
888                 data->last_end = data->pos_min + l;
889                 data->pos_min += l; /* As in the first entry. */
890                 data->flags &= ~SF_BEFORE_EOL;
891             }
892             if (flags & SCF_DO_STCLASS_AND) {
893                 /* Check whether it is compatible with what we know already! */
894                 int compat = 1;
895
896                 if (uc >= 0x100 ||
897                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
898                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
899                     && (!(data->start_class->flags & ANYOF_FOLD)
900                         || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
901                     )
902                     compat = 0;
903                 ANYOF_CLASS_ZERO(data->start_class);
904                 ANYOF_BITMAP_ZERO(data->start_class);
905                 if (compat)
906                     ANYOF_BITMAP_SET(data->start_class, uc);
907                 data->start_class->flags &= ~ANYOF_EOS;
908                 if (uc < 0x100)
909                   data->start_class->flags &= ~ANYOF_UNICODE_ALL;
910             }
911             else if (flags & SCF_DO_STCLASS_OR) {
912                 /* false positive possible if the class is case-folded */
913                 if (uc < 0x100)
914                     ANYOF_BITMAP_SET(data->start_class, uc);
915                 else
916                     data->start_class->flags |= ANYOF_UNICODE_ALL;
917                 data->start_class->flags &= ~ANYOF_EOS;
918                 cl_and(data->start_class, &and_with);
919             }
920             flags &= ~SCF_DO_STCLASS;
921         }
922         else if (PL_regkind[(U8)OP(scan)] == EXACT) { /* But OP != EXACT! */
923             I32 l = STR_LEN(scan);
924             UV uc = *((U8*)STRING(scan));
925
926             /* Search for fixed substrings supports EXACT only. */
927             if (flags & SCF_DO_SUBSTR)
928                 scan_commit(pRExC_state, data);
929             if (UTF) {
930                 U8 *s = (U8 *)STRING(scan);
931                 l = utf8_length(s, s + l);
932                 uc = utf8_to_uvchr(s, NULL);
933             }
934             min += l;
935             if (data && (flags & SCF_DO_SUBSTR))
936                 data->pos_min += l;
937             if (flags & SCF_DO_STCLASS_AND) {
938                 /* Check whether it is compatible with what we know already! */
939                 int compat = 1;
940
941                 if (uc >= 0x100 ||
942                     (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
943                     && !ANYOF_BITMAP_TEST(data->start_class, uc)
944                      && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
945                     compat = 0;
946                 ANYOF_CLASS_ZERO(data->start_class);
947                 ANYOF_BITMAP_ZERO(data->start_class);
948                 if (compat) {
949                     ANYOF_BITMAP_SET(data->start_class, uc);
950                     data->start_class->flags &= ~ANYOF_EOS;
951                     data->start_class->flags |= ANYOF_FOLD;
952                     if (OP(scan) == EXACTFL)
953                         data->start_class->flags |= ANYOF_LOCALE;
954                 }
955             }
956             else if (flags & SCF_DO_STCLASS_OR) {
957                 if (data->start_class->flags & ANYOF_FOLD) {
958                     /* false positive possible if the class is case-folded.
959                        Assume that the locale settings are the same... */
960                     if (uc < 0x100)
961                         ANYOF_BITMAP_SET(data->start_class, uc);
962                     data->start_class->flags &= ~ANYOF_EOS;
963                 }
964                 cl_and(data->start_class, &and_with);
965             }
966             flags &= ~SCF_DO_STCLASS;
967         }
968         else if (strchr((char*)PL_varies,OP(scan))) {
969             I32 mincount, maxcount, minnext, deltanext, fl = 0;
970             I32 f = flags, pos_before = 0;
971             regnode *oscan = scan;
972             struct regnode_charclass_class this_class;
973             struct regnode_charclass_class *oclass = NULL;
974
975             switch (PL_regkind[(U8)OP(scan)]) {
976             case WHILEM:                /* End of (?:...)* . */
977                 scan = NEXTOPER(scan);
978                 goto finish;
979             case PLUS:
980                 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
981                     next = NEXTOPER(scan);
982                     if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
983                         mincount = 1;
984                         maxcount = REG_INFTY;
985                         next = regnext(scan);
986                         scan = NEXTOPER(scan);
987                         goto do_curly;
988                     }
989                 }
990                 if (flags & SCF_DO_SUBSTR)
991                     data->pos_min++;
992                 min++;
993                 /* Fall through. */
994             case STAR:
995                 if (flags & SCF_DO_STCLASS) {
996                     mincount = 0;
997                     maxcount = REG_INFTY;
998                     next = regnext(scan);
999                     scan = NEXTOPER(scan);
1000                     goto do_curly;
1001                 }
1002                 is_inf = is_inf_internal = 1;
1003                 scan = regnext(scan);
1004                 if (flags & SCF_DO_SUBSTR) {
1005                     scan_commit(pRExC_state, data); /* Cannot extend fixed substrings */
1006                     data->longest = &(data->longest_float);
1007                 }
1008                 goto optimize_curly_tail;
1009             case CURLY:
1010                 mincount = ARG1(scan);
1011                 maxcount = ARG2(scan);
1012                 next = regnext(scan);
1013                 if (OP(scan) == CURLYX) {
1014                     I32 lp = (data ? *(data->last_closep) : 0);
1015
1016                     scan->flags = ((lp <= U8_MAX) ? lp : U8_MAX);
1017                 }
1018                 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
1019               do_curly:
1020                 if (flags & SCF_DO_SUBSTR) {
1021                     if (mincount == 0) scan_commit(pRExC_state,data); /* Cannot extend fixed substrings */
1022                     pos_before = data->pos_min;
1023                 }
1024                 if (data) {
1025                     fl = data->flags;
1026                     data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
1027                     if (is_inf)
1028                         data->flags |= SF_IS_INF;
1029                 }
1030                 if (flags & SCF_DO_STCLASS) {
1031                     cl_init(pRExC_state, &this_class);
1032                     oclass = data->start_class;
1033                     data->start_class = &this_class;
1034                     f |= SCF_DO_STCLASS_AND;
1035                     f &= ~SCF_DO_STCLASS_OR;
1036                 }
1037                 /* These are the cases when once a subexpression
1038                    fails at a particular position, it cannot succeed
1039                    even after backtracking at the enclosing scope.
1040                 
1041                    XXXX what if minimal match and we are at the
1042                         initial run of {n,m}? */
1043                 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
1044                     f &= ~SCF_WHILEM_VISITED_POS;
1045
1046                 /* This will finish on WHILEM, setting scan, or on NULL: */
1047                 minnext = study_chunk(pRExC_state, &scan, &deltanext, last, data,
1048                                       mincount == 0
1049                                         ? (f & ~SCF_DO_SUBSTR) : f);
1050
1051                 if (flags & SCF_DO_STCLASS)
1052                     data->start_class = oclass;
1053                 if (mincount == 0 || minnext == 0) {
1054                     if (flags & SCF_DO_STCLASS_OR) {
1055                         cl_or(pRExC_state, data->start_class, &this_class);
1056                     }
1057                     else if (flags & SCF_DO_STCLASS_AND) {
1058                         /* Switch to OR mode: cache the old value of
1059                          * data->start_class */
1060                         StructCopy(data->start_class, &and_with,
1061                                    struct regnode_charclass_class);
1062                         flags &= ~SCF_DO_STCLASS_AND;
1063                         StructCopy(&this_class, data->start_class,
1064                                    struct regnode_charclass_class);
1065                         flags |= SCF_DO_STCLASS_OR;
1066                         data->start_class->flags |= ANYOF_EOS;
1067                     }
1068                 } else {                /* Non-zero len */
1069                     if (flags & SCF_DO_STCLASS_OR) {
1070                         cl_or(pRExC_state, data->start_class, &this_class);
1071                         cl_and(data->start_class, &and_with);
1072                     }
1073                     else if (flags & SCF_DO_STCLASS_AND)
1074                         cl_and(data->start_class, &this_class);
1075                     flags &= ~SCF_DO_STCLASS;
1076                 }
1077                 if (!scan)              /* It was not CURLYX, but CURLY. */
1078                     scan = next;
1079                 if (ckWARN(WARN_REGEXP)
1080                     && (minnext == 0) && (deltanext == 0)
1081                     && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
1082                     && maxcount <= REG_INFTY/3) /* Complement check for big count */
1083                 {
1084                     vWARN(RExC_parse,
1085                           "Quantifier unexpected on zero-length expression");
1086                 }
1087
1088                 min += minnext * mincount;
1089                 is_inf_internal |= ((maxcount == REG_INFTY
1090                                      && (minnext + deltanext) > 0)
1091                                     || deltanext == I32_MAX);
1092                 is_inf |= is_inf_internal;
1093                 delta += (minnext + deltanext) * maxcount - minnext * mincount;
1094
1095                 /* Try powerful optimization CURLYX => CURLYN. */
1096                 if (  OP(oscan) == CURLYX && data
1097                       && data->flags & SF_IN_PAR
1098                       && !(data->flags & SF_HAS_EVAL)
1099                       && !deltanext && minnext == 1 ) {
1100                     /* Try to optimize to CURLYN.  */
1101                     regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
1102                     regnode *nxt1 = nxt;
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;
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     Newz(1002, r->startp, RExC_npar, I32);
2020     Newz(1002, r->endp, RExC_npar, I32);
2021     PL_regdata = r->data; /* for regprop() */
2022     DEBUG_r(regdump(r));
2023     return(r);
2024 }
2025
2026 /*
2027  - reg - regular expression, i.e. main body or parenthesized thing
2028  *
2029  * Caller must absorb opening parenthesis.
2030  *
2031  * Combining parenthesis handling with the base level of regular expression
2032  * is a trifle forced, but the need to tie the tails of the branches to what
2033  * follows makes it hard to avoid.
2034  */
2035 STATIC regnode *
2036 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp)
2037     /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
2038 {
2039     register regnode *ret;              /* Will be the head of the group. */
2040     register regnode *br;
2041     register regnode *lastbr;
2042     register regnode *ender = 0;
2043     register I32 parno = 0;
2044     I32 flags, oregflags = RExC_flags16, have_branch = 0, open = 0;
2045
2046     /* for (?g), (?gc), and (?o) warnings; warning
2047        about (?c) will warn about (?g) -- japhy    */
2048
2049     I32 wastedflags = 0x00,
2050         wasted_o    = 0x01,
2051         wasted_g    = 0x02,
2052         wasted_gc   = 0x02 | 0x04,
2053         wasted_c    = 0x04;
2054
2055     char * parse_start = RExC_parse; /* MJD */
2056     char *oregcomp_parse = RExC_parse;
2057     char c;
2058
2059     *flagp = 0;                         /* Tentatively. */
2060
2061
2062     /* Make an OPEN node, if parenthesized. */
2063     if (paren) {
2064         if (*RExC_parse == '?') { /* (?...) */
2065             U16 posflags = 0, negflags = 0;
2066             U16 *flagsp = &posflags;
2067             int logical = 0;
2068             char *seqstart = RExC_parse;
2069
2070             RExC_parse++;
2071             paren = *RExC_parse++;
2072             ret = NULL;                 /* For look-ahead/behind. */
2073             switch (paren) {
2074             case '<':           /* (?<...) */
2075                 RExC_seen |= REG_SEEN_LOOKBEHIND;
2076                 if (*RExC_parse == '!')
2077                     paren = ',';
2078                 if (*RExC_parse != '=' && *RExC_parse != '!')
2079                     goto unknown;
2080                 RExC_parse++;
2081             case '=':           /* (?=...) */
2082             case '!':           /* (?!...) */
2083                 RExC_seen_zerolen++;
2084             case ':':           /* (?:...) */
2085             case '>':           /* (?>...) */
2086                 break;
2087             case '$':           /* (?$...) */
2088             case '@':           /* (?@...) */
2089                 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
2090                 break;
2091             case '#':           /* (?#...) */
2092                 while (*RExC_parse && *RExC_parse != ')')
2093                     RExC_parse++;
2094                 if (*RExC_parse != ')')
2095                     FAIL("Sequence (?#... not terminated");
2096                 nextchar(pRExC_state);
2097                 *flagp = TRYAGAIN;
2098                 return NULL;
2099             case 'p':           /* (?p...) */
2100                 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
2101                     vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
2102                 /* FALL THROUGH*/
2103             case '?':           /* (??...) */
2104                 logical = 1;
2105                 paren = *RExC_parse++;
2106                 /* FALL THROUGH */
2107             case '{':           /* (?{...}) */
2108             {
2109                 I32 count = 1, n = 0;
2110                 char c;
2111                 char *s = RExC_parse;
2112                 SV *sv;
2113                 OP_4tree *sop, *rop;
2114
2115                 RExC_seen_zerolen++;
2116                 RExC_seen |= REG_SEEN_EVAL;
2117                 while (count && (c = *RExC_parse)) {
2118                     if (c == '\\' && RExC_parse[1])
2119                         RExC_parse++;
2120                     else if (c == '{')
2121                         count++;
2122                     else if (c == '}')
2123                         count--;
2124                     RExC_parse++;
2125                 }
2126                 if (*RExC_parse != ')')
2127                 {
2128                     RExC_parse = s;             
2129                     vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
2130                 }
2131                 if (!SIZE_ONLY) {
2132                     AV *av;
2133                 
2134                     if (RExC_parse - 1 - s)
2135                         sv = newSVpvn(s, RExC_parse - 1 - s);
2136                     else
2137                         sv = newSVpvn("", 0);
2138
2139                     ENTER;
2140                     Perl_save_re_context(aTHX);
2141                     rop = sv_compile_2op(sv, &sop, "re", &av);
2142                     sop->op_private |= OPpREFCOUNTED;
2143                     /* re_dup will OpREFCNT_inc */
2144                     OpREFCNT_set(sop, 1);
2145                     LEAVE;
2146
2147                     n = add_data(pRExC_state, 3, "nop");
2148                     RExC_rx->data->data[n] = (void*)rop;
2149                     RExC_rx->data->data[n+1] = (void*)sop;
2150                     RExC_rx->data->data[n+2] = (void*)av;
2151                     SvREFCNT_dec(sv);
2152                 }
2153                 else {                                          /* First pass */
2154                     if (PL_reginterp_cnt < ++RExC_seen_evals
2155                         && PL_curcop != &PL_compiling)
2156                         /* No compiled RE interpolated, has runtime
2157                            components ===> unsafe.  */
2158                         FAIL("Eval-group not allowed at runtime, use re 'eval'");
2159                     if (PL_tainting && PL_tainted)
2160                         FAIL("Eval-group in insecure regular expression");
2161                 }
2162                 
2163                 nextchar(pRExC_state);
2164                 if (logical) {
2165                     ret = reg_node(pRExC_state, LOGICAL);
2166                     if (!SIZE_ONLY)
2167                         ret->flags = 2;
2168                     regtail(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
2169                     /* deal with the length of this later - MJD */
2170                     return ret;
2171                 }
2172                 return reganode(pRExC_state, EVAL, n);
2173             }
2174             case '(':           /* (?(?{...})...) and (?(?=...)...) */
2175             {
2176                 if (RExC_parse[0] == '?') {        /* (?(?...)) */
2177                     if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
2178                         || RExC_parse[1] == '<'
2179                         || RExC_parse[1] == '{') { /* Lookahead or eval. */
2180                         I32 flag;
2181                         
2182                         ret = reg_node(pRExC_state, LOGICAL);
2183                         if (!SIZE_ONLY)
2184                             ret->flags = 1;
2185                         regtail(pRExC_state, ret, reg(pRExC_state, 1, &flag));
2186                         goto insert_if;
2187                     }
2188                 }
2189                 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
2190                     /* (?(1)...) */
2191                     parno = atoi(RExC_parse++);
2192
2193                     while (isDIGIT(*RExC_parse))
2194                         RExC_parse++;
2195                     ret = reganode(pRExC_state, GROUPP, parno);
2196                     
2197                     if ((c = *nextchar(pRExC_state)) != ')')
2198                         vFAIL("Switch condition not recognized");
2199                   insert_if:
2200                     regtail(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
2201                     br = regbranch(pRExC_state, &flags, 1);
2202                     if (br == NULL)
2203                         br = reganode(pRExC_state, LONGJMP, 0);
2204                     else
2205                         regtail(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
2206                     c = *nextchar(pRExC_state);
2207                     if (flags&HASWIDTH)
2208                         *flagp |= HASWIDTH;
2209                     if (c == '|') {
2210                         lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
2211                         regbranch(pRExC_state, &flags, 1);
2212                         regtail(pRExC_state, ret, lastbr);
2213                         if (flags&HASWIDTH)
2214                             *flagp |= HASWIDTH;
2215                         c = *nextchar(pRExC_state);
2216                     }
2217                     else
2218                         lastbr = NULL;
2219                     if (c != ')')
2220                         vFAIL("Switch (?(condition)... contains too many branches");
2221                     ender = reg_node(pRExC_state, TAIL);
2222                     regtail(pRExC_state, br, ender);
2223                     if (lastbr) {
2224                         regtail(pRExC_state, lastbr, ender);
2225                         regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
2226                     }
2227                     else
2228                         regtail(pRExC_state, ret, ender);
2229                     return ret;
2230                 }
2231                 else {
2232                     vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
2233                 }
2234             }
2235             case 0:
2236                 RExC_parse--; /* for vFAIL to print correctly */
2237                 vFAIL("Sequence (? incomplete");
2238                 break;
2239             default:
2240                 --RExC_parse;
2241               parse_flags:      /* (?i) */
2242                 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
2243                     /* (?g), (?gc) and (?o) are useless here
2244                        and must be globally applied -- japhy */
2245
2246                     if (*RExC_parse == 'o' || *RExC_parse == 'g') {
2247                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2248                             I32 wflagbit = *RExC_parse == 'o' ? wasted_o : wasted_g;
2249                             if (! (wastedflags & wflagbit) ) {
2250                                 wastedflags |= wflagbit;
2251                                 vWARN5(
2252                                     RExC_parse + 1,
2253                                     "Useless (%s%c) - %suse /%c modifier",
2254                                     flagsp == &negflags ? "?-" : "?",
2255                                     *RExC_parse,
2256                                     flagsp == &negflags ? "don't " : "",
2257                                     *RExC_parse
2258                                 );
2259                             }
2260                         }
2261                     }
2262                     else if (*RExC_parse == 'c') {
2263                         if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
2264                             if (! (wastedflags & wasted_c) ) {
2265                                 wastedflags |= wasted_gc;
2266                                 vWARN3(
2267                                     RExC_parse + 1,
2268                                     "Useless (%sc) - %suse /gc modifier",
2269                                     flagsp == &negflags ? "?-" : "?",
2270                                     flagsp == &negflags ? "don't " : ""
2271                                 );
2272                             }
2273                         }
2274                     }
2275                     else { pmflag(flagsp, *RExC_parse); }
2276
2277                     ++RExC_parse;
2278                 }
2279                 if (*RExC_parse == '-') {
2280                     flagsp = &negflags;
2281                     wastedflags = 0;  /* reset so (?g-c) warns twice */
2282                     ++RExC_parse;
2283                     goto parse_flags;
2284                 }
2285                 RExC_flags16 |= posflags;
2286                 RExC_flags16 &= ~negflags;
2287                 if (*RExC_parse == ':') {
2288                     RExC_parse++;
2289                     paren = ':';
2290                     break;
2291                 }               
2292               unknown:
2293                 if (*RExC_parse != ')') {
2294                     RExC_parse++;
2295                     vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
2296                 }
2297                 nextchar(pRExC_state);
2298                 *flagp = TRYAGAIN;
2299                 return NULL;
2300             }
2301         }
2302         else {                  /* (...) */
2303             parno = RExC_npar;
2304             RExC_npar++;
2305             ret = reganode(pRExC_state, OPEN, parno);
2306             Set_Node_Length(ret, 1); /* MJD */
2307             Set_Node_Offset(ret, RExC_parse); /* MJD */
2308             open = 1;
2309         }
2310     }
2311     else                        /* ! paren */
2312         ret = NULL;
2313
2314     /* Pick up the branches, linking them together. */
2315     parse_start = RExC_parse;   /* MJD */
2316     br = regbranch(pRExC_state, &flags, 1);
2317     /*     branch_len = (paren != 0); */
2318     
2319     if (br == NULL)
2320         return(NULL);
2321     if (*RExC_parse == '|') {
2322         if (!SIZE_ONLY && RExC_extralen) {
2323             reginsert(pRExC_state, BRANCHJ, br);
2324         }
2325         else {                  /* MJD */
2326             reginsert(pRExC_state, BRANCH, br);
2327             Set_Node_Length(br, paren != 0);
2328             Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
2329         }
2330         have_branch = 1;
2331         if (SIZE_ONLY)
2332             RExC_extralen += 1;         /* For BRANCHJ-BRANCH. */
2333     }
2334     else if (paren == ':') {
2335         *flagp |= flags&SIMPLE;
2336     }
2337     if (open) {                         /* Starts with OPEN. */
2338         regtail(pRExC_state, ret, br);          /* OPEN -> first. */
2339     }
2340     else if (paren != '?')              /* Not Conditional */
2341         ret = br;
2342     if (flags&HASWIDTH)
2343         *flagp |= HASWIDTH;
2344     *flagp |= flags&SPSTART;
2345     lastbr = br;
2346     while (*RExC_parse == '|') {
2347         if (!SIZE_ONLY && RExC_extralen) {
2348             ender = reganode(pRExC_state, LONGJMP,0);
2349             regtail(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
2350         }
2351         if (SIZE_ONLY)
2352             RExC_extralen += 2;         /* Account for LONGJMP. */
2353         nextchar(pRExC_state);
2354         br = regbranch(pRExC_state, &flags, 0);
2355         
2356         if (br == NULL)
2357             return(NULL);
2358         regtail(pRExC_state, lastbr, br);               /* BRANCH -> BRANCH. */
2359         lastbr = br;
2360         if (flags&HASWIDTH)
2361             *flagp |= HASWIDTH;
2362         *flagp |= flags&SPSTART;
2363     }
2364
2365     if (have_branch || paren != ':') {
2366         /* Make a closing node, and hook it on the end. */
2367         switch (paren) {
2368         case ':':
2369             ender = reg_node(pRExC_state, TAIL);
2370             break;
2371         case 1:
2372             ender = reganode(pRExC_state, CLOSE, parno);
2373             Set_Node_Offset(ender,RExC_parse+1); /* MJD */
2374             Set_Node_Length(ender,1); /* MJD */
2375             break;
2376         case '<':
2377         case ',':
2378         case '=':
2379         case '!':
2380             *flagp &= ~HASWIDTH;
2381             /* FALL THROUGH */
2382         case '>':
2383             ender = reg_node(pRExC_state, SUCCEED);
2384             break;
2385         case 0:
2386             ender = reg_node(pRExC_state, END);
2387             break;
2388         }
2389         regtail(pRExC_state, lastbr, ender);
2390
2391         if (have_branch) {
2392             /* Hook the tails of the branches to the closing node. */
2393             for (br = ret; br != NULL; br = regnext(br)) {
2394                 regoptail(pRExC_state, br, ender);
2395             }
2396         }
2397     }
2398
2399     {
2400         char *p;
2401         static char parens[] = "=!<,>";
2402
2403         if (paren && (p = strchr(parens, paren))) {
2404             int node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
2405             int flag = (p - parens) > 1;
2406
2407             if (paren == '>')
2408                 node = SUSPEND, flag = 0;
2409             reginsert(pRExC_state, node,ret);
2410             ret->flags = flag;
2411             regtail(pRExC_state, ret, reg_node(pRExC_state, TAIL));
2412         }
2413     }
2414
2415     /* Check for proper termination. */
2416     if (paren) {
2417         RExC_flags16 = oregflags;
2418         if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
2419             RExC_parse = oregcomp_parse;
2420             vFAIL("Unmatched (");
2421         }
2422     }
2423     else if (!paren && RExC_parse < RExC_end) {
2424         if (*RExC_parse == ')') {
2425             RExC_parse++;
2426             vFAIL("Unmatched )");
2427         }
2428         else
2429             FAIL("Junk on end of regexp");      /* "Can't happen". */
2430         /* NOTREACHED */
2431     }
2432
2433     return(ret);
2434 }
2435
2436 /*
2437  - regbranch - one alternative of an | operator
2438  *
2439  * Implements the concatenation operator.
2440  */
2441 STATIC regnode *
2442 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first)
2443 {
2444     register regnode *ret;
2445     register regnode *chain = NULL;
2446     register regnode *latest;
2447     I32 flags = 0, c = 0;
2448
2449     if (first)
2450         ret = NULL;
2451     else {
2452         if (!SIZE_ONLY && RExC_extralen)
2453             ret = reganode(pRExC_state, BRANCHJ,0);
2454         else {
2455             ret = reg_node(pRExC_state, BRANCH);
2456             Set_Node_Length(ret, 1);
2457         }
2458     }
2459         
2460     if (!first && SIZE_ONLY)
2461         RExC_extralen += 1;                     /* BRANCHJ */
2462
2463     *flagp = WORST;                     /* Tentatively. */
2464
2465     RExC_parse--;
2466     nextchar(pRExC_state);
2467     while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
2468         flags &= ~TRYAGAIN;
2469         latest = regpiece(pRExC_state, &flags);
2470         if (latest == NULL) {
2471             if (flags & TRYAGAIN)
2472                 continue;
2473             return(NULL);
2474         }
2475         else if (ret == NULL)
2476             ret = latest;
2477         *flagp |= flags&HASWIDTH;
2478         if (chain == NULL)      /* First piece. */
2479             *flagp |= flags&SPSTART;
2480         else {
2481             RExC_naughty++;
2482             regtail(pRExC_state, chain, latest);
2483         }
2484         chain = latest;
2485         c++;
2486     }
2487     if (chain == NULL) {        /* Loop ran zero times. */
2488         chain = reg_node(pRExC_state, NOTHING);
2489         if (ret == NULL)
2490             ret = chain;
2491     }
2492     if (c == 1) {
2493         *flagp |= flags&SIMPLE;
2494     }
2495
2496     return(ret);
2497 }
2498
2499 /*
2500  - regpiece - something followed by possible [*+?]
2501  *
2502  * Note that the branching code sequences used for ? and the general cases
2503  * of * and + are somewhat optimized:  they use the same NOTHING node as
2504  * both the endmarker for their branch list and the body of the last branch.
2505  * It might seem that this node could be dispensed with entirely, but the
2506  * endmarker role is not redundant.
2507  */
2508 STATIC regnode *
2509 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2510 {
2511     register regnode *ret;
2512     register char op;
2513     register char *next;
2514     I32 flags;
2515     char *origparse = RExC_parse;
2516     char *maxpos;
2517     I32 min;
2518     I32 max = REG_INFTY;
2519     char *parse_start;
2520
2521     ret = regatom(pRExC_state, &flags);
2522     if (ret == NULL) {
2523         if (flags & TRYAGAIN)
2524             *flagp |= TRYAGAIN;
2525         return(NULL);
2526     }
2527
2528     op = *RExC_parse;
2529
2530     if (op == '{' && regcurly(RExC_parse)) {
2531         parse_start = RExC_parse; /* MJD */
2532         next = RExC_parse + 1;
2533         maxpos = Nullch;
2534         while (isDIGIT(*next) || *next == ',') {
2535             if (*next == ',') {
2536                 if (maxpos)
2537                     break;
2538                 else
2539                     maxpos = next;
2540             }
2541             next++;
2542         }
2543         if (*next == '}') {             /* got one */
2544             if (!maxpos)
2545                 maxpos = next;
2546             RExC_parse++;
2547             min = atoi(RExC_parse);
2548             if (*maxpos == ',')
2549                 maxpos++;
2550             else
2551                 maxpos = RExC_parse;
2552             max = atoi(maxpos);
2553             if (!max && *maxpos != '0')
2554                 max = REG_INFTY;                /* meaning "infinity" */
2555             else if (max >= REG_INFTY)
2556                 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
2557             RExC_parse = next;
2558             nextchar(pRExC_state);
2559
2560         do_curly:
2561             if ((flags&SIMPLE)) {
2562                 RExC_naughty += 2 + RExC_naughty / 2;
2563                 reginsert(pRExC_state, CURLY, ret);
2564                 Set_Node_Offset(ret, parse_start+1); /* MJD */
2565                 Set_Node_Cur_Length(ret);
2566             }
2567             else {
2568                 regnode *w = reg_node(pRExC_state, WHILEM);
2569
2570                 w->flags = 0;
2571                 regtail(pRExC_state, ret, w);
2572                 if (!SIZE_ONLY && RExC_extralen) {
2573                     reginsert(pRExC_state, LONGJMP,ret);
2574                     reginsert(pRExC_state, NOTHING,ret);
2575                     NEXT_OFF(ret) = 3;  /* Go over LONGJMP. */
2576                 }
2577                 reginsert(pRExC_state, CURLYX,ret);
2578                                 /* MJD hk */
2579                 Set_Node_Offset(ret, parse_start+1);
2580                 Set_Node_Length(ret, 
2581                                 op == '{' ? (RExC_parse - parse_start) : 1);
2582                 
2583                 if (!SIZE_ONLY && RExC_extralen)
2584                     NEXT_OFF(ret) = 3;  /* Go over NOTHING to LONGJMP. */
2585                 regtail(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
2586                 if (SIZE_ONLY)
2587                     RExC_whilem_seen++, RExC_extralen += 3;
2588                 RExC_naughty += 4 + RExC_naughty;       /* compound interest */
2589             }
2590             ret->flags = 0;
2591
2592             if (min > 0)
2593                 *flagp = WORST;
2594             if (max > 0)
2595                 *flagp |= HASWIDTH;
2596             if (max && max < min)
2597                 vFAIL("Can't do {n,m} with n > m");
2598             if (!SIZE_ONLY) {
2599                 ARG1_SET(ret, min);
2600                 ARG2_SET(ret, max);
2601             }
2602
2603             goto nest_check;
2604         }
2605     }
2606
2607     if (!ISMULT1(op)) {
2608         *flagp = flags;
2609         return(ret);
2610     }
2611
2612 #if 0                           /* Now runtime fix should be reliable. */
2613
2614     /* if this is reinstated, don't forget to put this back into perldiag:
2615
2616             =item Regexp *+ operand could be empty at {#} in regex m/%s/
2617
2618            (F) The part of the regexp subject to either the * or + quantifier
2619            could match an empty string. The {#} shows in the regular
2620            expression about where the problem was discovered.
2621
2622     */
2623
2624     if (!(flags&HASWIDTH) && op != '?')
2625       vFAIL("Regexp *+ operand could be empty");
2626 #endif
2627
2628     parse_start = RExC_parse;
2629     nextchar(pRExC_state);
2630
2631     *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
2632
2633     if (op == '*' && (flags&SIMPLE)) {
2634         reginsert(pRExC_state, STAR, ret);
2635         ret->flags = 0;
2636         RExC_naughty += 4;
2637     }
2638     else if (op == '*') {
2639         min = 0;
2640         goto do_curly;
2641     }
2642     else if (op == '+' && (flags&SIMPLE)) {
2643         reginsert(pRExC_state, PLUS, ret);
2644         ret->flags = 0;
2645         RExC_naughty += 3;
2646     }
2647     else if (op == '+') {
2648         min = 1;
2649         goto do_curly;
2650     }
2651     else if (op == '?') {
2652         min = 0; max = 1;
2653         goto do_curly;
2654     }
2655   nest_check:
2656     if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
2657         vWARN3(RExC_parse,
2658                "%.*s matches null string many times",
2659                RExC_parse - origparse,
2660                origparse);
2661     }
2662
2663     if (*RExC_parse == '?') {
2664         nextchar(pRExC_state);
2665         reginsert(pRExC_state, MINMOD, ret);
2666         regtail(pRExC_state, ret, ret + NODE_STEP_REGNODE);
2667     }
2668     if (ISMULT2(RExC_parse)) {
2669         RExC_parse++;
2670         vFAIL("Nested quantifiers");
2671     }
2672
2673     return(ret);
2674 }
2675
2676 /*
2677  - regatom - the lowest level
2678  *
2679  * Optimization:  gobbles an entire sequence of ordinary characters so that
2680  * it can turn them into a single node, which is smaller to store and
2681  * faster to run.  Backslashed characters are exceptions, each becoming a
2682  * separate node; the code is simpler that way and it's not worth fixing.
2683  *
2684  * [Yes, it is worth fixing, some scripts can run twice the speed.] */
2685 STATIC regnode *
2686 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp)
2687 {
2688     register regnode *ret = 0;
2689     I32 flags;
2690     char *parse_start = 0;
2691
2692     *flagp = WORST;             /* Tentatively. */
2693
2694 tryagain:
2695     switch (*RExC_parse) {
2696     case '^':
2697         RExC_seen_zerolen++;
2698         nextchar(pRExC_state);
2699         if (RExC_flags16 & PMf_MULTILINE)
2700             ret = reg_node(pRExC_state, MBOL);
2701         else if (RExC_flags16 & PMf_SINGLELINE)
2702             ret = reg_node(pRExC_state, SBOL);
2703         else
2704             ret = reg_node(pRExC_state, BOL);
2705         Set_Node_Length(ret, 1); /* MJD */
2706         break;
2707     case '$':
2708         nextchar(pRExC_state);
2709         if (*RExC_parse)
2710             RExC_seen_zerolen++;
2711         if (RExC_flags16 & PMf_MULTILINE)
2712             ret = reg_node(pRExC_state, MEOL);
2713         else if (RExC_flags16 & PMf_SINGLELINE)
2714             ret = reg_node(pRExC_state, SEOL);
2715         else
2716             ret = reg_node(pRExC_state, EOL);
2717         Set_Node_Length(ret, 1); /* MJD */
2718         break;
2719     case '.':
2720         nextchar(pRExC_state);
2721         if (RExC_flags16 & PMf_SINGLELINE)
2722             ret = reg_node(pRExC_state, SANY);
2723         else
2724             ret = reg_node(pRExC_state, REG_ANY);
2725         *flagp |= HASWIDTH|SIMPLE;
2726         RExC_naughty++;
2727         Set_Node_Length(ret, 1); /* MJD */
2728         break;
2729     case '[':
2730     {
2731         char *oregcomp_parse = ++RExC_parse;
2732         ret = regclass(pRExC_state);
2733         if (*RExC_parse != ']') {
2734             RExC_parse = oregcomp_parse;
2735             vFAIL("Unmatched [");
2736         }
2737         nextchar(pRExC_state);
2738         *flagp |= HASWIDTH|SIMPLE;
2739         Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
2740         break;
2741     }
2742     case '(':
2743         nextchar(pRExC_state);
2744         ret = reg(pRExC_state, 1, &flags);
2745         if (ret == NULL) {
2746                 if (flags & TRYAGAIN) {
2747                     if (RExC_parse == RExC_end) {
2748                          /* Make parent create an empty node if needed. */
2749                         *flagp |= TRYAGAIN;
2750                         return(NULL);
2751                     }
2752                     goto tryagain;
2753                 }
2754                 return(NULL);
2755         }
2756         *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
2757         break;
2758     case '|':
2759     case ')':
2760         if (flags & TRYAGAIN) {
2761             *flagp |= TRYAGAIN;
2762             return NULL;
2763         }
2764         vFAIL("Internal urp");
2765                                 /* Supposed to be caught earlier. */
2766         break;
2767     case '{':
2768         if (!regcurly(RExC_parse)) {
2769             RExC_parse++;
2770             goto defchar;
2771         }
2772         /* FALL THROUGH */
2773     case '?':
2774     case '+':
2775     case '*':
2776         RExC_parse++;
2777         vFAIL("Quantifier follows nothing");
2778         break;
2779     case '\\':
2780         switch (*++RExC_parse) {
2781         case 'A':
2782             RExC_seen_zerolen++;
2783             ret = reg_node(pRExC_state, SBOL);
2784             *flagp |= SIMPLE;
2785             nextchar(pRExC_state);
2786             Set_Node_Length(ret, 2); /* MJD */
2787             break;
2788         case 'G':
2789             ret = reg_node(pRExC_state, GPOS);
2790             RExC_seen |= REG_SEEN_GPOS;
2791             *flagp |= SIMPLE;
2792             nextchar(pRExC_state);
2793             Set_Node_Length(ret, 2); /* MJD */
2794             break;
2795         case 'Z':
2796             ret = reg_node(pRExC_state, SEOL);
2797             *flagp |= SIMPLE;
2798             nextchar(pRExC_state);
2799             break;
2800         case 'z':
2801             ret = reg_node(pRExC_state, EOS);
2802             *flagp |= SIMPLE;
2803             RExC_seen_zerolen++;                /* Do not optimize RE away */
2804             nextchar(pRExC_state);
2805             Set_Node_Length(ret, 2); /* MJD */
2806             break;
2807         case 'C':
2808             ret = reg_node(pRExC_state, CANY);
2809             RExC_seen |= REG_SEEN_CANY;
2810             *flagp |= HASWIDTH|SIMPLE;
2811             nextchar(pRExC_state);
2812             Set_Node_Length(ret, 2); /* MJD */
2813             break;
2814         case 'X':
2815             ret = reg_node(pRExC_state, CLUMP);
2816             *flagp |= HASWIDTH;
2817             nextchar(pRExC_state);
2818             Set_Node_Length(ret, 2); /* MJD */
2819             break;
2820         case 'w':
2821             ret = reg_node(pRExC_state, LOC ? ALNUML     : ALNUM);
2822             *flagp |= HASWIDTH|SIMPLE;
2823             nextchar(pRExC_state);
2824             Set_Node_Length(ret, 2); /* MJD */
2825             break;
2826         case 'W':
2827             ret = reg_node(pRExC_state, LOC ? NALNUML     : NALNUM);
2828             *flagp |= HASWIDTH|SIMPLE;
2829             nextchar(pRExC_state);
2830             Set_Node_Length(ret, 2); /* MJD */
2831             break;
2832         case 'b':
2833             RExC_seen_zerolen++;
2834             RExC_seen |= REG_SEEN_LOOKBEHIND;
2835             ret = reg_node(pRExC_state, LOC ? BOUNDL     : BOUND);
2836             *flagp |= SIMPLE;
2837             nextchar(pRExC_state);
2838             Set_Node_Length(ret, 2); /* MJD */
2839             break;
2840         case 'B':
2841             RExC_seen_zerolen++;
2842             RExC_seen |= REG_SEEN_LOOKBEHIND;
2843             ret = reg_node(pRExC_state, LOC ? NBOUNDL     : NBOUND);
2844             *flagp |= SIMPLE;
2845             nextchar(pRExC_state);
2846             Set_Node_Length(ret, 2); /* MJD */
2847             break;
2848         case 's':
2849             ret = reg_node(pRExC_state, LOC ? SPACEL     : SPACE);
2850             *flagp |= HASWIDTH|SIMPLE;
2851             nextchar(pRExC_state);
2852             Set_Node_Length(ret, 2); /* MJD */
2853             break;
2854         case 'S':
2855             ret = reg_node(pRExC_state, LOC ? NSPACEL     : NSPACE);
2856             *flagp |= HASWIDTH|SIMPLE;
2857             nextchar(pRExC_state);
2858             Set_Node_Length(ret, 2); /* MJD */
2859             break;
2860         case 'd':
2861             ret = reg_node(pRExC_state, DIGIT);
2862             *flagp |= HASWIDTH|SIMPLE;
2863             nextchar(pRExC_state);
2864             Set_Node_Length(ret, 2); /* MJD */
2865             break;
2866         case 'D':
2867             ret = reg_node(pRExC_state, NDIGIT);
2868             *flagp |= HASWIDTH|SIMPLE;
2869             nextchar(pRExC_state);
2870             Set_Node_Length(ret, 2); /* MJD */
2871             break;
2872         case 'p':
2873         case 'P':
2874             {   
2875                 char* oldregxend = RExC_end;
2876                 char* parse_start = RExC_parse;
2877
2878                 if (RExC_parse[1] == '{') {
2879                   /* a lovely hack--pretend we saw [\pX] instead */
2880                     RExC_end = strchr(RExC_parse, '}');
2881                     if (!RExC_end) {
2882                         RExC_parse += 2;
2883                         RExC_end = oldregxend;
2884                         vFAIL("Missing right brace on \\p{}");
2885                     }
2886                     RExC_end++;
2887                 }
2888                 else
2889                     RExC_end = RExC_parse + 2;
2890                 RExC_parse--;
2891
2892                 ret = regclass(pRExC_state);
2893
2894                 RExC_end = oldregxend;
2895                 RExC_parse--;
2896                 Set_Node_Cur_Length(ret); /* MJD */
2897                 nextchar(pRExC_state);
2898                 *flagp |= HASWIDTH|SIMPLE;
2899             }
2900             break;
2901         case 'n':
2902         case 'r':
2903         case 't':
2904         case 'f':
2905         case 'e':
2906         case 'a':
2907         case 'x':
2908         case 'c':
2909         case '0':
2910             goto defchar;
2911         case '1': case '2': case '3': case '4':
2912         case '5': case '6': case '7': case '8': case '9':
2913             {
2914                 I32 num = atoi(RExC_parse);
2915
2916                 if (num > 9 && num >= RExC_npar)
2917                     goto defchar;
2918                 else {
2919                     char * parse_start = RExC_parse - 1; /* MJD */
2920                     while (isDIGIT(*RExC_parse))
2921                         RExC_parse++;
2922
2923                     if (!SIZE_ONLY && num > RExC_rx->nparens)
2924                         vFAIL("Reference to nonexistent group");
2925                     RExC_sawback = 1;
2926                     ret = reganode(pRExC_state, FOLD
2927                                    ? (LOC ? REFFL : REFF)
2928                                    : REF, num);
2929                     *flagp |= HASWIDTH;
2930                     
2931                     /* override incorrect value set in reganode MJD */
2932                     Set_Node_Offset(ret, parse_start+1); 
2933                     Set_Node_Cur_Length(ret); /* MJD */
2934                     RExC_parse--;
2935                     nextchar(pRExC_state);
2936                 }
2937             }
2938             break;
2939         case '\0':
2940             if (RExC_parse >= RExC_end)
2941                 FAIL("Trailing \\");
2942             /* FALL THROUGH */
2943         default:
2944             /* Do not generate `unrecognized' warnings here, we fall
2945                back into the quick-grab loop below */
2946             goto defchar;
2947         }
2948         break;
2949
2950     case '#':
2951         if (RExC_flags16 & PMf_EXTENDED) {
2952             while (RExC_parse < RExC_end && *RExC_parse != '\n') RExC_parse++;
2953             if (RExC_parse < RExC_end)
2954                 goto tryagain;
2955         }
2956         /* FALL THROUGH */
2957
2958     default: {
2959             register STRLEN len;
2960             register UV ender;
2961             register char *p;
2962             char *oldp, *s;
2963             STRLEN numlen;
2964
2965             parse_start = RExC_parse - 1;
2966
2967             RExC_parse++;
2968
2969         defchar:
2970             ret = reg_node(pRExC_state, FOLD
2971                           ? (LOC ? EXACTFL : EXACTF)
2972                           : EXACT);
2973             s = STRING(ret);
2974             for (len = 0, p = RExC_parse - 1;
2975               len < 127 && p < RExC_end;
2976               len++)
2977             {
2978                 oldp = p;
2979
2980                 if (RExC_flags16 & PMf_EXTENDED)
2981                     p = regwhite(p, RExC_end);
2982                 switch (*p) {
2983                 case '^':
2984                 case '$':
2985                 case '.':
2986                 case '[':
2987                 case '(':
2988                 case ')':
2989                 case '|':
2990                     goto loopdone;
2991                 case '\\':
2992                     switch (*++p) {
2993                     case 'A':
2994                     case 'G':
2995                     case 'Z':
2996                     case 'z':
2997                     case 'w':
2998                     case 'W':
2999                     case 'b':
3000                     case 'B':
3001                     case 's':
3002                     case 'S':
3003                     case 'd':
3004                     case 'D':
3005                     case 'p':
3006                     case 'P':
3007                         --p;
3008                         goto loopdone;
3009                     case 'n':
3010                         ender = '\n';
3011                         p++;
3012                         break;
3013                     case 'r':
3014                         ender = '\r';
3015                         p++;
3016                         break;
3017                     case 't':
3018                         ender = '\t';
3019                         p++;
3020                         break;
3021                     case 'f':
3022                         ender = '\f';
3023                         p++;
3024                         break;
3025                     case 'e':
3026                           ender = ASCII_TO_NATIVE('\033');
3027                         p++;
3028                         break;
3029                     case 'a':
3030                           ender = ASCII_TO_NATIVE('\007');
3031                         p++;
3032                         break;
3033                     case 'x':
3034                         if (*++p == '{') {
3035                             char* e = strchr(p, '}');
3036         
3037                             if (!e) {
3038                                 RExC_parse = p + 1;
3039                                 vFAIL("Missing right brace on \\x{}");
3040                             }
3041                             else {
3042                                 numlen = 1;     /* allow underscores */
3043                                 ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
3044                                 if (ender > 0xff)
3045                                     RExC_utf8 = 1;
3046                                 /* numlen is generous */
3047                                 if (numlen + len >= 127) {
3048                                     p--;
3049                                     goto loopdone;
3050                                 }
3051                                 p = e + 1;
3052                             }
3053                         }
3054                         else {
3055                             numlen = 0;         /* disallow underscores */
3056                             ender = (UV)scan_hex(p, 2, &numlen);
3057                             p += numlen;
3058                         }
3059                         break;
3060                     case 'c':
3061                         p++;
3062                         ender = UCHARAT(p++);
3063                         ender = toCTRL(ender);
3064                         break;
3065                     case '0': case '1': case '2': case '3':case '4':
3066                     case '5': case '6': case '7': case '8':case '9':
3067                         if (*p == '0' ||
3068                           (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
3069                             numlen = 0;         /* disallow underscores */
3070                             ender = (UV)scan_oct(p, 3, &numlen);
3071                             p += numlen;
3072                         }
3073                         else {
3074                             --p;
3075                             goto loopdone;
3076                         }
3077                         break;
3078                     case '\0':
3079                         if (p >= RExC_end)
3080                             FAIL("Trailing \\");
3081                         /* FALL THROUGH */
3082                     default:
3083                         if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
3084                             vWARN2(p +1, "Unrecognized escape \\%c passed through", *p);
3085                         goto normal_default;
3086                     }
3087                     break;
3088                 default:
3089                   normal_default:
3090                     if (UTF8_IS_START(*p) && UTF) {
3091                         ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
3092                                                &numlen, 0);
3093                         p += numlen;
3094                     }
3095                     else
3096                         ender = *p++;
3097                     break;
3098                 }
3099                 if (RExC_flags16 & PMf_EXTENDED)
3100                     p = regwhite(p, RExC_end);
3101                 if (UTF && FOLD) {
3102                     if (LOC)
3103                         ender = toLOWER_LC_uvchr(ender);
3104                     else
3105                         ender = toLOWER_uni(ender);
3106                 }
3107                 if (ISMULT2(p)) { /* Back off on ?+*. */
3108                     if (len)
3109                         p = oldp;
3110                     else if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3111                         reguni(pRExC_state, ender, s, &numlen);
3112                         s += numlen;
3113                         len += numlen;
3114                     }
3115                     else {
3116                         len++;
3117                         REGC(ender, s++);
3118                     }
3119                     break;
3120                 }
3121                 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(ender)) && UTF) {
3122                     reguni(pRExC_state, ender, s, &numlen);
3123                     s += numlen;
3124                     len += numlen - 1;
3125                 }
3126                 else
3127                     REGC(ender, s++);
3128             }
3129         loopdone:
3130             RExC_parse = p - 1;
3131             Set_Node_Cur_Length(ret); /* MJD */
3132             nextchar(pRExC_state);
3133             {
3134                 /* len is STRLEN which is unsigned, need to copy to signed */
3135                 IV iv = len;
3136                 if (iv < 0)
3137                     vFAIL("Internal disaster");
3138             }
3139             if (len > 0)
3140                 *flagp |= HASWIDTH;
3141             if (len == 1)
3142                 *flagp |= SIMPLE;
3143             if (!SIZE_ONLY)
3144                 STR_LEN(ret) = len;
3145             if (SIZE_ONLY)
3146                 RExC_size += STR_SZ(len);
3147             else
3148                 RExC_emit += STR_SZ(len);
3149         }
3150         break;
3151     }
3152
3153     return(ret);
3154 }
3155
3156 STATIC char *
3157 S_regwhite(pTHX_ char *p, char *e)
3158 {
3159     while (p < e) {
3160         if (isSPACE(*p))
3161             ++p;
3162         else if (*p == '#') {
3163             do {
3164                 p++;
3165             } while (p < e && *p != '\n');
3166         }
3167         else
3168             break;
3169     }
3170     return p;
3171 }
3172
3173 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
3174    Character classes ([:foo:]) can also be negated ([:^foo:]).
3175    Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
3176    Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
3177    but trigger warnings because they are currently unimplemented. */
3178 STATIC I32
3179 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
3180 {
3181     char *posixcc = 0;
3182     I32 namedclass = OOB_NAMEDCLASS;
3183
3184     if (value == '[' && RExC_parse + 1 < RExC_end &&
3185         /* I smell either [: or [= or [. -- POSIX has been here, right? */
3186         (*RExC_parse == ':' ||
3187          *RExC_parse == '=' ||
3188          *RExC_parse == '.')) {
3189         char  c = *RExC_parse;
3190         char* s = RExC_parse++;
3191         
3192         while (RExC_parse < RExC_end && *RExC_parse != c)
3193             RExC_parse++;
3194         if (RExC_parse == RExC_end)
3195             /* Grandfather lone [:, [=, [. */
3196             RExC_parse = s;
3197         else {
3198             char* t = RExC_parse++; /* skip over the c */
3199
3200             if (*RExC_parse == ']') {
3201                 RExC_parse++; /* skip over the ending ] */
3202                 posixcc = s + 1;
3203                 if (*s == ':') {
3204                     I32 complement = *posixcc == '^' ? *posixcc++ : 0;
3205                     I32 skip = 5; /* the most common skip */
3206
3207                     switch (*posixcc) {
3208                     case 'a':
3209                         if (strnEQ(posixcc, "alnum", 5))
3210                             namedclass =
3211                                 complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
3212                         else if (strnEQ(posixcc, "alpha", 5))
3213                             namedclass =
3214                                 complement ? ANYOF_NALPHA : ANYOF_ALPHA;
3215                         else if (strnEQ(posixcc, "ascii", 5))
3216                             namedclass =
3217                                 complement ? ANYOF_NASCII : ANYOF_ASCII;
3218                         break;
3219                     case 'b':
3220                         if (strnEQ(posixcc, "blank", 5))
3221                             namedclass =
3222                                 complement ? ANYOF_NBLANK : ANYOF_BLANK;
3223                         break;
3224                     case 'c':
3225                         if (strnEQ(posixcc, "cntrl", 5))
3226                             namedclass =
3227                                 complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
3228                         break;
3229                     case 'd':
3230                         if (strnEQ(posixcc, "digit", 5))
3231                             namedclass =
3232                                 complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
3233                         break;
3234                     case 'g':
3235                         if (strnEQ(posixcc, "graph", 5))
3236                             namedclass =
3237                                 complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
3238                         break;
3239                     case 'l':
3240                         if (strnEQ(posixcc, "lower", 5))
3241                             namedclass =
3242                                 complement ? ANYOF_NLOWER : ANYOF_LOWER;
3243                         break;
3244                     case 'p':
3245                         if (strnEQ(posixcc, "print", 5))
3246                             namedclass =
3247                                 complement ? ANYOF_NPRINT : ANYOF_PRINT;
3248                         else if (strnEQ(posixcc, "punct", 5))
3249                             namedclass =
3250                                 complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
3251                         break;
3252                     case 's':
3253                         if (strnEQ(posixcc, "space", 5))
3254                             namedclass =
3255                                 complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
3256                         break;
3257                     case 'u':
3258                         if (strnEQ(posixcc, "upper", 5))
3259                             namedclass =
3260                                 complement ? ANYOF_NUPPER : ANYOF_UPPER;
3261                         break;
3262                     case 'w': /* this is not POSIX, this is the Perl \w */
3263                         if (strnEQ(posixcc, "word", 4)) {
3264                             namedclass =
3265                                 complement ? ANYOF_NALNUM : ANYOF_ALNUM;
3266                             skip = 4;
3267                         }
3268                         break;
3269                     case 'x':
3270                         if (strnEQ(posixcc, "xdigit", 6)) {
3271                             namedclass =
3272                                 complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
3273                             skip = 6;
3274                         }
3275                         break;
3276                     }
3277                     if (namedclass == OOB_NAMEDCLASS ||
3278                         posixcc[skip] != ':' ||
3279                         posixcc[skip+1] != ']')
3280                     {
3281                         Simple_vFAIL3("POSIX class [:%.*s:] unknown",
3282                                       t - s - 1, s + 1);
3283                     }
3284                 } else if (!SIZE_ONLY) {
3285                     /* [[=foo=]] and [[.foo.]] are still future. */
3286
3287                     /* adjust RExC_parse so the warning shows after
3288                        the class closes */
3289                     while (*RExC_parse && *RExC_parse != ']')
3290                         RExC_parse++;
3291                     Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3292                 }
3293             } else {
3294                 /* Maternal grandfather:
3295                  * "[:" ending in ":" but not in ":]" */
3296                 RExC_parse = s;
3297             }
3298         }
3299     }
3300
3301     return namedclass;
3302 }
3303
3304 STATIC void
3305 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
3306 {
3307     if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
3308         (*RExC_parse == ':' ||
3309          *RExC_parse == '=' ||
3310          *RExC_parse == '.')) {
3311         char *s = RExC_parse;
3312         char  c = *s++;
3313
3314         while(*s && isALNUM(*s))
3315             s++;
3316         if (*s && c == *s && s[1] == ']') {
3317             vWARN3(s+2, "POSIX syntax [%c %c] belongs inside character classes", c, c);
3318
3319             /* [[=foo=]] and [[.foo.]] are still future. */
3320             if (c == '=' || c == '.')
3321             {
3322                 /* adjust RExC_parse so the error shows after
3323                    the class closes */
3324                 while (*RExC_parse && *RExC_parse++ != ']')
3325                     ;
3326                 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
3327             }
3328         }
3329     }
3330 }
3331
3332 STATIC regnode *
3333 S_regclass(pTHX_ RExC_state_t *pRExC_state)
3334 {
3335     register UV value;
3336     register IV prevvalue = OOB_UNICODE;
3337     register IV range = 0;
3338     register regnode *ret;
3339     STRLEN numlen;
3340     IV namedclass;
3341     char *rangebegin = 0;
3342     bool need_class = 0;
3343     SV *listsv = Nullsv;
3344     register char *e;
3345     UV n;
3346     bool optimize_invert = TRUE;
3347
3348     ret = reganode(pRExC_state, ANYOF, 0);
3349
3350     if (!SIZE_ONLY)
3351         ANYOF_FLAGS(ret) = 0;
3352
3353     if (*RExC_parse == '^') {   /* Complement of range. */
3354         RExC_naughty++;
3355         RExC_parse++;
3356         if (!SIZE_ONLY)
3357             ANYOF_FLAGS(ret) |= ANYOF_INVERT;
3358     }
3359
3360     if (SIZE_ONLY)
3361         RExC_size += ANYOF_SKIP;
3362     else {
3363         RExC_emit += ANYOF_SKIP;
3364         if (FOLD)
3365             ANYOF_FLAGS(ret) |= ANYOF_FOLD;
3366         if (LOC)
3367             ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
3368         ANYOF_BITMAP_ZERO(ret);
3369         listsv = newSVpvn("# comment\n", 10);
3370     }
3371
3372     if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
3373         checkposixcc(pRExC_state);
3374
3375     if (*RExC_parse == ']' || *RExC_parse == '-')
3376         goto charclassloop;             /* allow 1st char to be ] or - */
3377
3378     while (RExC_parse < RExC_end && *RExC_parse != ']') {
3379
3380     charclassloop:
3381
3382         namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
3383
3384         if (!range)
3385             rangebegin = RExC_parse;
3386         if (UTF) {
3387             value = utf8n_to_uvchr((U8*)RExC_parse,
3388                                    RExC_end - RExC_parse,
3389                                    &numlen, 0);
3390             RExC_parse += numlen;
3391         }
3392         else
3393             value = UCHARAT(RExC_parse++);
3394         if (value == '[')
3395             namedclass = regpposixcc(pRExC_state, value);
3396         else if (value == '\\') {
3397             if (UTF) {
3398                 value = utf8n_to_uvchr((U8*)RExC_parse,
3399                                    RExC_end - RExC_parse,
3400                                    &numlen, 0);
3401                 RExC_parse += numlen;
3402             }
3403             else
3404                 value = UCHARAT(RExC_parse++);
3405             /* Some compilers cannot handle switching on 64-bit integer
3406              * values, therefore value cannot be an UV.  Yes, this will
3407              * be a problem later if we want switch on Unicode.
3408              * A similar issue a little bit later when switching on
3409              * namedclass. --jhi */
3410             switch ((I32)value) {
3411             case 'w':   namedclass = ANYOF_ALNUM;       break;
3412             case 'W':   namedclass = ANYOF_NALNUM;      break;
3413             case 's':   namedclass = ANYOF_SPACE;       break;
3414             case 'S':   namedclass = ANYOF_NSPACE;      break;
3415             case 'd':   namedclass = ANYOF_DIGIT;       break;
3416             case 'D':   namedclass = ANYOF_NDIGIT;      break;
3417             case 'p':
3418             case 'P':
3419                 if (*RExC_parse == '{') {
3420                     e = strchr(RExC_parse++, '}');
3421                     if (!e)
3422                         vFAIL("Missing right brace on \\p{}");
3423                     n = e - RExC_parse;
3424                 }
3425                 else {
3426                     e = RExC_parse;
3427                     n = 1;
3428                 }
3429                 if (!SIZE_ONLY) {
3430                     if (value == 'p')
3431                         Perl_sv_catpvf(aTHX_ listsv,
3432                                        "+utf8::%.*s\n", (int)n, RExC_parse);
3433                     else
3434                         Perl_sv_catpvf(aTHX_ listsv,
3435                                        "!utf8::%.*s\n", (int)n, RExC_parse);
3436                 }
3437                 RExC_parse = e + 1;
3438                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3439                 continue;
3440             case 'n':   value = '\n';                   break;
3441             case 'r':   value = '\r';                   break;
3442             case 't':   value = '\t';                   break;
3443             case 'f':   value = '\f';                   break;
3444             case 'b':   value = '\b';                   break;
3445             case 'e':   value = ASCII_TO_NATIVE('\033');break;
3446             case 'a':   value = ASCII_TO_NATIVE('\007');break;
3447             case 'x':
3448                 if (*RExC_parse == '{') {
3449                     e = strchr(RExC_parse++, '}');
3450                     if (!e)
3451                         vFAIL("Missing right brace on \\x{}");
3452                     numlen = 1;         /* allow underscores */
3453                     value = (UV)scan_hex(RExC_parse,
3454                                          e - RExC_parse,
3455                                          &numlen);
3456                     RExC_parse = e + 1;
3457                 }
3458                 else {
3459                     numlen = 0;         /* disallow underscores */
3460                     value = (UV)scan_hex(RExC_parse, 2, &numlen);
3461                     RExC_parse += numlen;
3462                 }
3463                 break;
3464             case 'c':
3465                 value = UCHARAT(RExC_parse++);
3466                 value = toCTRL(value);
3467                 break;
3468             case '0': case '1': case '2': case '3': case '4':
3469             case '5': case '6': case '7': case '8': case '9':
3470                 numlen = 0;             /* disallow underscores */
3471                 value = (UV)scan_oct(--RExC_parse, 3, &numlen);
3472                 RExC_parse += numlen;
3473                 break;
3474             default:
3475                 if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
3476                     vWARN2(RExC_parse,
3477                            "Unrecognized escape \\%c in character class passed through",
3478                            (int)value);
3479                 break;
3480             }
3481         } /* end of \blah */
3482
3483         if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
3484
3485             if (!SIZE_ONLY && !need_class)
3486                 ANYOF_CLASS_ZERO(ret);
3487
3488             need_class = 1;
3489
3490             /* a bad range like a-\d, a-[:digit:] ? */
3491             if (range) {
3492                 if (!SIZE_ONLY) {
3493                     if (ckWARN(WARN_REGEXP))
3494                         vWARN4(RExC_parse,
3495                                "False [] range \"%*.*s\"",
3496                                RExC_parse - rangebegin,
3497                                RExC_parse - rangebegin,
3498                                rangebegin);
3499                     if (prevvalue < 256) {
3500                         ANYOF_BITMAP_SET(ret, prevvalue);
3501                         ANYOF_BITMAP_SET(ret, '-');
3502                     }
3503                     else {
3504                         ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3505                         Perl_sv_catpvf(aTHX_ listsv,
3506                                        "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
3507                     }
3508                 }
3509
3510                 range = 0; /* this was not a true range */
3511             }
3512
3513             if (!SIZE_ONLY) {
3514                 if (namedclass > OOB_NAMEDCLASS)
3515                     optimize_invert = FALSE;
3516                 /* Possible truncation here but in some 64-bit environments
3517                  * the compiler gets heartburn about switch on 64-bit values.
3518                  * A similar issue a little earlier when switching on value.
3519                  * --jhi */
3520                 switch ((I32)namedclass) {
3521                 case ANYOF_ALNUM:
3522                     if (LOC)
3523                         ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
3524                     else {
3525                         for (value = 0; value < 256; value++)
3526                             if (isALNUM(value))
3527                                 ANYOF_BITMAP_SET(ret, value);
3528                     }
3529                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsWord\n");    
3530                     break;
3531                 case ANYOF_NALNUM:
3532                     if (LOC)
3533                         ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
3534                     else {
3535                         for (value = 0; value < 256; value++)
3536                             if (!isALNUM(value))
3537                                 ANYOF_BITMAP_SET(ret, value);
3538                     }
3539                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsWord\n");
3540                     break;
3541                 case ANYOF_ALNUMC:
3542                     if (LOC)
3543                         ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
3544                     else {
3545                         for (value = 0; value < 256; value++)
3546                             if (isALNUMC(value))
3547                                 ANYOF_BITMAP_SET(ret, value);
3548                     }
3549                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
3550                     break;
3551                 case ANYOF_NALNUMC:
3552                     if (LOC)
3553                         ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
3554                     else {
3555                         for (value = 0; value < 256; value++)
3556                             if (!isALNUMC(value))
3557                                 ANYOF_BITMAP_SET(ret, value);
3558                     }
3559                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlnum\n");
3560                     break;
3561                 case ANYOF_ALPHA:
3562                     if (LOC)
3563                         ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
3564                     else {
3565                         for (value = 0; value < 256; value++)
3566                             if (isALPHA(value))
3567                                 ANYOF_BITMAP_SET(ret, value);
3568                     }
3569                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlpha\n");
3570                     break;
3571                 case ANYOF_NALPHA:
3572                     if (LOC)
3573                         ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
3574                     else {
3575                         for (value = 0; value < 256; value++)
3576                             if (!isALPHA(value))
3577                                 ANYOF_BITMAP_SET(ret, value);
3578                     }
3579                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsAlpha\n");
3580                     break;
3581                 case ANYOF_ASCII:
3582                     if (LOC)
3583                         ANYOF_CLASS_SET(ret, ANYOF_ASCII);
3584                     else {
3585 #ifndef EBCDIC
3586                         for (value = 0; value < 128; value++)
3587                             ANYOF_BITMAP_SET(ret, value);
3588 #else  /* EBCDIC */
3589                         for (value = 0; value < 256; value++) {
3590                             if (isASCII(value))
3591                                 ANYOF_BITMAP_SET(ret, value);
3592                         }
3593 #endif /* EBCDIC */
3594                     }
3595                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsASCII\n");
3596                     break;
3597                 case ANYOF_NASCII:
3598                     if (LOC)
3599                         ANYOF_CLASS_SET(ret, ANYOF_NASCII);
3600                     else {
3601 #ifndef EBCDIC
3602                         for (value = 128; value < 256; value++)
3603                             ANYOF_BITMAP_SET(ret, value);
3604 #else  /* EBCDIC */
3605                         for (value = 0; value < 256; value++) {
3606                             if (!isASCII(value))
3607                                 ANYOF_BITMAP_SET(ret, value);
3608                         }
3609 #endif /* EBCDIC */
3610                     }
3611                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsASCII\n");
3612                     break;
3613                 case ANYOF_BLANK:
3614                     if (LOC)
3615                         ANYOF_CLASS_SET(ret, ANYOF_BLANK);
3616                     else {
3617                         for (value = 0; value < 256; value++)
3618                             if (isBLANK(value))
3619                                 ANYOF_BITMAP_SET(ret, value);
3620                     }
3621                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsBlank\n");
3622                     break;
3623                 case ANYOF_NBLANK:
3624                     if (LOC)
3625                         ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
3626                     else {
3627                         for (value = 0; value < 256; value++)
3628                             if (!isBLANK(value))
3629                                 ANYOF_BITMAP_SET(ret, value);
3630                     }
3631                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsBlank\n");
3632                     break;
3633                 case ANYOF_CNTRL:
3634                     if (LOC)
3635                         ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
3636                     else {
3637                         for (value = 0; value < 256; value++)
3638                             if (isCNTRL(value))
3639                                 ANYOF_BITMAP_SET(ret, value);
3640                     }
3641                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsCntrl\n");
3642                     break;
3643                 case ANYOF_NCNTRL:
3644                     if (LOC)
3645                         ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
3646                     else {
3647                         for (value = 0; value < 256; value++)
3648                             if (!isCNTRL(value))
3649                                 ANYOF_BITMAP_SET(ret, value);
3650                     }
3651                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsCntrl\n");
3652                     break;
3653                 case ANYOF_DIGIT:
3654                     if (LOC)
3655                         ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
3656                     else {
3657                         /* consecutive digits assumed */
3658                         for (value = '0'; value <= '9'; value++)
3659                             ANYOF_BITMAP_SET(ret, value);
3660                     }
3661                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
3662                     break;
3663                 case ANYOF_NDIGIT:
3664                     if (LOC)
3665                         ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
3666                     else {
3667                         /* consecutive digits assumed */
3668                         for (value = 0; value < '0'; value++)
3669                             ANYOF_BITMAP_SET(ret, value);
3670                         for (value = '9' + 1; value < 256; value++)
3671                             ANYOF_BITMAP_SET(ret, value);
3672                     }
3673                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsDigit\n");
3674                     break;
3675                 case ANYOF_GRAPH:
3676                     if (LOC)
3677                         ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
3678                     else {
3679                         for (value = 0; value < 256; value++)
3680                             if (isGRAPH(value))
3681                                 ANYOF_BITMAP_SET(ret, value);
3682                     }
3683                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsGraph\n");
3684                     break;
3685                 case ANYOF_NGRAPH:
3686                     if (LOC)
3687                         ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
3688                     else {
3689                         for (value = 0; value < 256; value++)
3690                             if (!isGRAPH(value))
3691                                 ANYOF_BITMAP_SET(ret, value);
3692                     }
3693                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsGraph\n");
3694                     break;
3695                 case ANYOF_LOWER:
3696                     if (LOC)
3697                         ANYOF_CLASS_SET(ret, ANYOF_LOWER);
3698                     else {
3699                         for (value = 0; value < 256; value++)
3700                             if (isLOWER(value))
3701                                 ANYOF_BITMAP_SET(ret, value);
3702                     }
3703                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsLower\n");
3704                     break;
3705                 case ANYOF_NLOWER:
3706                     if (LOC)
3707                         ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
3708                     else {
3709                         for (value = 0; value < 256; value++)
3710                             if (!isLOWER(value))
3711                                 ANYOF_BITMAP_SET(ret, value);
3712                     }
3713                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsLower\n");
3714                     break;
3715                 case ANYOF_PRINT:
3716                     if (LOC)
3717                         ANYOF_CLASS_SET(ret, ANYOF_PRINT);
3718                     else {
3719                         for (value = 0; value < 256; value++)
3720                             if (isPRINT(value))
3721                                 ANYOF_BITMAP_SET(ret, value);
3722                     }
3723                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPrint\n");
3724                     break;
3725                 case ANYOF_NPRINT:
3726                     if (LOC)
3727                         ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
3728                     else {
3729                         for (value = 0; value < 256; value++)
3730                             if (!isPRINT(value))
3731                                 ANYOF_BITMAP_SET(ret, value);
3732                     }
3733                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPrint\n");
3734                     break;
3735                 case ANYOF_PSXSPC:
3736                     if (LOC)
3737                         ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
3738                     else {
3739                         for (value = 0; value < 256; value++)
3740                             if (isPSXSPC(value))
3741                                 ANYOF_BITMAP_SET(ret, value);
3742                     }
3743                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
3744                     break;
3745                 case ANYOF_NPSXSPC:
3746                     if (LOC)
3747                         ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
3748                     else {
3749                         for (value = 0; value < 256; value++)
3750                             if (!isPSXSPC(value))
3751                                 ANYOF_BITMAP_SET(ret, value);
3752                     }
3753                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpace\n");
3754                     break;
3755                 case ANYOF_PUNCT:
3756                     if (LOC)
3757                         ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
3758                     else {
3759                         for (value = 0; value < 256; value++)
3760                             if (isPUNCT(value))
3761                                 ANYOF_BITMAP_SET(ret, value);
3762                     }
3763                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsPunct\n");
3764                     break;
3765                 case ANYOF_NPUNCT:
3766                     if (LOC)
3767                         ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
3768                     else {
3769                         for (value = 0; value < 256; value++)
3770                             if (!isPUNCT(value))
3771                                 ANYOF_BITMAP_SET(ret, value);
3772                     }
3773                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsPunct\n");
3774                     break;
3775                 case ANYOF_SPACE:
3776                     if (LOC)
3777                         ANYOF_CLASS_SET(ret, ANYOF_SPACE);
3778                     else {
3779                         for (value = 0; value < 256; value++)
3780                             if (isSPACE(value))
3781                                 ANYOF_BITMAP_SET(ret, value);
3782                     }
3783                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpacePerl\n");
3784                     break;
3785                 case ANYOF_NSPACE:
3786                     if (LOC)
3787                         ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
3788                     else {
3789                         for (value = 0; value < 256; value++)
3790                             if (!isSPACE(value))
3791                                 ANYOF_BITMAP_SET(ret, value);
3792                     }
3793                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsSpacePerl\n");
3794                     break;
3795                 case ANYOF_UPPER:
3796                     if (LOC)
3797                         ANYOF_CLASS_SET(ret, ANYOF_UPPER);
3798                     else {
3799                         for (value = 0; value < 256; value++)
3800                             if (isUPPER(value))
3801                                 ANYOF_BITMAP_SET(ret, value);
3802                     }
3803                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsUpper\n");
3804                     break;
3805                 case ANYOF_NUPPER:
3806                     if (LOC)
3807                         ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
3808                     else {
3809                         for (value = 0; value < 256; value++)
3810                             if (!isUPPER(value))
3811                                 ANYOF_BITMAP_SET(ret, value);
3812                     }
3813                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsUpper\n");
3814                     break;
3815                 case ANYOF_XDIGIT:
3816                     if (LOC)
3817                         ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
3818                     else {
3819                         for (value = 0; value < 256; value++)
3820                             if (isXDIGIT(value))
3821                                 ANYOF_BITMAP_SET(ret, value);
3822                     }
3823                     Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsXDigit\n");
3824                     break;
3825                 case ANYOF_NXDIGIT:
3826                     if (LOC)
3827                         ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
3828                     else {
3829                         for (value = 0; value < 256; value++)
3830                             if (!isXDIGIT(value))
3831                                 ANYOF_BITMAP_SET(ret, value);
3832                     }
3833                     Perl_sv_catpvf(aTHX_ listsv, "!utf8::IsXDigit\n");
3834                     break;
3835                 default:
3836                     vFAIL("Invalid [::] class");
3837                     break;
3838                 }
3839                 if (LOC)
3840                     ANYOF_FLAGS(ret) |= ANYOF_CLASS;
3841                 continue;
3842             }
3843         } /* end of namedclass \blah */
3844
3845         if (range) {
3846             if (prevvalue > value) /* b-a */ {
3847                 Simple_vFAIL4("Invalid [] range \"%*.*s\"",
3848                               RExC_parse - rangebegin,
3849                               RExC_parse - rangebegin,
3850                               rangebegin);
3851                 range = 0; /* not a valid range */
3852             }
3853         }
3854         else {
3855             prevvalue = value; /* save the beginning of the range */
3856             if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
3857                 RExC_parse[1] != ']') {
3858                 RExC_parse++;
3859
3860                 /* a bad range like \w-, [:word:]- ? */
3861                 if (namedclass > OOB_NAMEDCLASS) {
3862                     if (ckWARN(WARN_REGEXP))
3863                         vWARN4(RExC_parse,
3864                                "False [] range \"%*.*s\"",
3865                                RExC_parse - rangebegin,
3866                                RExC_parse - rangebegin,
3867                                rangebegin);
3868                     if (!SIZE_ONLY)
3869                         ANYOF_BITMAP_SET(ret, '-');
3870                 } else
3871                     range = 1;  /* yeah, it's a range! */
3872                 continue;       /* but do it the next time */
3873             }
3874         }
3875
3876         /* now is the next time */
3877         if (!SIZE_ONLY) {
3878             IV i;
3879
3880             if (prevvalue < 256) {
3881                 IV ceilvalue = value < 256 ? value : 255;
3882
3883 #ifdef EBCDIC
3884                 if ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
3885                     (isUPPER(prevvalue) && isUPPER(ceilvalue)))
3886                 {
3887                     if (isLOWER(prevvalue)) {
3888                         for (i = prevvalue; i <= ceilvalue; i++)
3889                             if (isLOWER(i))
3890                                 ANYOF_BITMAP_SET(ret, i);
3891                     } else {
3892                         for (i = prevvalue; i <= ceilvalue; i++)
3893                             if (isUPPER(i))
3894                                 ANYOF_BITMAP_SET(ret, i);
3895                     }
3896                 }
3897                 else
3898 #endif
3899                     for (i = prevvalue; i <= ceilvalue; i++)
3900                         ANYOF_BITMAP_SET(ret, i);
3901           }
3902           if (value > 255) {
3903                 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
3904                 if (prevvalue < value)
3905                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
3906                                    (UV)prevvalue, (UV)value);
3907                 else if (prevvalue == value)
3908                     Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
3909                                    (UV)value);
3910             }
3911         }
3912
3913         range = 0; /* this range (if it was one) is done now */
3914     }
3915
3916     if (need_class) {
3917         ANYOF_FLAGS(ret) |= ANYOF_LARGE;
3918         if (SIZE_ONLY)
3919             RExC_size += ANYOF_CLASS_ADD_SKIP;
3920         else
3921             RExC_emit += ANYOF_CLASS_ADD_SKIP;
3922     }
3923
3924     /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
3925     if (!SIZE_ONLY &&
3926          /* If the only flag is folding (plus possibly inversion). */
3927         ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
3928        ) {
3929         for (value = 0; value < 256; ++value) {
3930             if (ANYOF_BITMAP_TEST(ret, value)) {
3931                 IV fold = PL_fold[value];
3932
3933                 if (fold != value)
3934                     ANYOF_BITMAP_SET(ret, fold);
3935             }
3936         }
3937         ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
3938     }
3939
3940     /* optimize inverted simple patterns (e.g. [^a-z]) */
3941     if (!SIZE_ONLY && optimize_invert &&
3942         /* If the only flag is inversion. */
3943         (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
3944         for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
3945             ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
3946         ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
3947     }
3948
3949     if (!SIZE_ONLY) {
3950         AV *av = newAV();
3951         SV *rv;
3952
3953         av_store(av, 0, listsv);
3954         av_store(av, 1, NULL);
3955         rv = newRV_noinc((SV*)av);
3956         n = add_data(pRExC_state, 1, "s");
3957         RExC_rx->data->data[n] = (void*)rv;
3958         ARG_SET(ret, n);
3959     }
3960
3961     return ret;
3962 }
3963
3964 STATIC char*
3965 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
3966 {
3967     char* retval = RExC_parse++;
3968
3969     for (;;) {
3970         if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
3971                 RExC_parse[2] == '#') {
3972             while (*RExC_parse && *RExC_parse != ')')
3973                 RExC_parse++;
3974             RExC_parse++;
3975             continue;
3976         }
3977         if (RExC_flags16 & PMf_EXTENDED) {
3978             if (isSPACE(*RExC_parse)) {
3979                 RExC_parse++;
3980                 continue;
3981             }
3982             else if (*RExC_parse == '#') {
3983                 while (*RExC_parse && *RExC_parse != '\n')
3984                     RExC_parse++;
3985                 RExC_parse++;
3986                 continue;
3987             }
3988         }
3989         return retval;
3990     }
3991 }
3992
3993 /*
3994 - reg_node - emit a node
3995 */
3996 STATIC regnode *                        /* Location. */
3997 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
3998 {
3999     register regnode *ret;
4000     register regnode *ptr;
4001
4002     ret = RExC_emit;
4003     if (SIZE_ONLY) {
4004         SIZE_ALIGN(RExC_size);
4005         RExC_size += 1;
4006         return(ret);
4007     }
4008
4009     NODE_ALIGN_FILL(ret);
4010     ptr = ret;
4011     FILL_ADVANCE_NODE(ptr, op);
4012     if (RExC_offsets) {         /* MJD */
4013       MJD_OFFSET_DEBUG((stderr, "%s:%u: (op %s) %s %u <- %u (len %u) (max %u).\n", 
4014               "reg_node", __LINE__, 
4015               reg_name[op],
4016               RExC_emit - RExC_emit_start > RExC_offsets[0] 
4017               ? "Overwriting end of array!\n" : "OK",
4018               RExC_emit - RExC_emit_start,
4019               RExC_parse - RExC_start,
4020               RExC_offsets[0])); 
4021       Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
4022     }
4023             
4024     RExC_emit = ptr;
4025
4026     return(ret);
4027 }
4028
4029 /*
4030 - reganode - emit a node with an argument
4031 */
4032 STATIC regnode *                        /* Location. */
4033 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
4034 {
4035     register regnode *ret;
4036     register regnode *ptr;
4037
4038     ret = RExC_emit;
4039     if (SIZE_ONLY) {
4040         SIZE_ALIGN(RExC_size);
4041         RExC_size += 2;
4042         return(ret);
4043     }
4044
4045     NODE_ALIGN_FILL(ret);
4046     ptr = ret;
4047     FILL_ADVANCE_NODE_ARG(ptr, op, arg);
4048     if (RExC_offsets) {         /* MJD */
4049       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4050               "reganode",
4051               RExC_emit - RExC_emit_start > RExC_offsets[0] ? 
4052               "Overwriting end of array!\n" : "OK",
4053               RExC_emit - RExC_emit_start,
4054               RExC_parse - RExC_start,
4055               RExC_offsets[0])); 
4056       Set_Cur_Node_Offset;
4057     }
4058             
4059     RExC_emit = ptr;
4060
4061     return(ret);
4062 }
4063
4064 /*
4065 - reguni - emit (if appropriate) a Unicode character
4066 */
4067 STATIC void
4068 S_reguni(pTHX_ RExC_state_t *pRExC_state, UV uv, char* s, STRLEN* lenp)
4069 {
4070     *lenp = SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
4071 }
4072
4073 /*
4074 - reginsert - insert an operator in front of already-emitted operand
4075 *
4076 * Means relocating the operand.
4077 */
4078 STATIC void
4079 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd)
4080 {
4081     register regnode *src;
4082     register regnode *dst;
4083     register regnode *place;
4084     register int offset = regarglen[(U8)op];
4085
4086 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
4087
4088     if (SIZE_ONLY) {
4089         RExC_size += NODE_STEP_REGNODE + offset;
4090         return;
4091     }
4092
4093     src = RExC_emit;
4094     RExC_emit += NODE_STEP_REGNODE + offset;
4095     dst = RExC_emit;
4096     while (src > opnd) {
4097         StructCopy(--src, --dst, regnode);
4098         if (RExC_offsets) {     /* MJD 20010112 */
4099           MJD_OFFSET_DEBUG((stderr, "%s: %s copy %u -> %u (max %u).\n", 
4100                   "reg_insert",
4101                   dst - RExC_emit_start > RExC_offsets[0] 
4102                   ? "Overwriting end of array!\n" : "OK",
4103                   src - RExC_emit_start,
4104                   dst - RExC_emit_start,
4105                   RExC_offsets[0])); 
4106           Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
4107           Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
4108         }
4109     }
4110     
4111
4112     place = opnd;               /* Op node, where operand used to be. */
4113     if (RExC_offsets) {         /* MJD */
4114       MJD_OFFSET_DEBUG((stderr, "%s: %s %u <- %u (max %u).\n", 
4115               "reginsert",
4116               place - RExC_emit_start > RExC_offsets[0] 
4117               ? "Overwriting end of array!\n" : "OK",
4118               place - RExC_emit_start,
4119               RExC_parse - RExC_start,
4120               RExC_offsets[0])); 
4121       Set_Node_Offset(place, RExC_parse);
4122     }
4123     src = NEXTOPER(place);
4124     FILL_ADVANCE_NODE(place, op);
4125     Zero(src, offset, regnode);
4126 }
4127
4128 /*
4129 - regtail - set the next-pointer at the end of a node chain of p to val.
4130 */
4131 STATIC void
4132 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4133 {
4134     register regnode *scan;
4135     register regnode *temp;
4136
4137     if (SIZE_ONLY)
4138         return;
4139
4140     /* Find last node. */
4141     scan = p;
4142     for (;;) {
4143         temp = regnext(scan);
4144         if (temp == NULL)
4145             break;
4146         scan = temp;
4147     }
4148
4149     if (reg_off_by_arg[OP(scan)]) {
4150         ARG_SET(scan, val - scan);
4151     }
4152     else {
4153         NEXT_OFF(scan) = val - scan;
4154     }
4155 }
4156
4157 /*
4158 - regoptail - regtail on operand of first argument; nop if operandless
4159 */
4160 STATIC void
4161 S_regoptail(pTHX_ RExC_state_t *pRExC_state, regnode *p, regnode *val)
4162 {
4163     /* "Operandless" and "op != BRANCH" are synonymous in practice. */
4164     if (p == NULL || SIZE_ONLY)
4165         return;
4166     if (PL_regkind[(U8)OP(p)] == BRANCH) {
4167         regtail(pRExC_state, NEXTOPER(p), val);
4168     }
4169     else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
4170         regtail(pRExC_state, NEXTOPER(NEXTOPER(p)), val);
4171     }
4172     else
4173         return;
4174 }
4175
4176 /*
4177  - regcurly - a little FSA that accepts {\d+,?\d*}
4178  */
4179 STATIC I32
4180 S_regcurly(pTHX_ register char *s)
4181 {
4182     if (*s++ != '{')
4183         return FALSE;
4184     if (!isDIGIT(*s))
4185         return FALSE;
4186     while (isDIGIT(*s))
4187         s++;
4188     if (*s == ',')
4189         s++;
4190     while (isDIGIT(*s))
4191         s++;
4192     if (*s != '}')
4193         return FALSE;
4194     return TRUE;
4195 }
4196
4197
4198 #ifdef DEBUGGING
4199
4200 STATIC regnode *
4201 S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
4202 {
4203     register U8 op = EXACT;     /* Arbitrary non-END op. */
4204     register regnode *next;
4205
4206     while (op != END && (!last || node < last)) {
4207         /* While that wasn't END last time... */
4208
4209         NODE_ALIGN(node);
4210         op = OP(node);
4211         if (op == CLOSE)
4212             l--;        
4213         next = regnext(node);
4214         /* Where, what. */
4215         if (OP(node) == OPTIMIZED)
4216             goto after_print;
4217         regprop(sv, node);
4218         PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
4219                       (int)(2*l + 1), "", SvPVX(sv));
4220         if (next == NULL)               /* Next ptr. */
4221             PerlIO_printf(Perl_debug_log, "(0)");
4222         else
4223             PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
4224         (void)PerlIO_putc(Perl_debug_log, '\n');
4225       after_print:
4226         if (PL_regkind[(U8)op] == BRANCHJ) {
4227             register regnode *nnode = (OP(next) == LONGJMP
4228                                        ? regnext(next)
4229                                        : next);
4230             if (last && nnode > last)
4231                 nnode = last;
4232             node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
4233         }
4234         else if (PL_regkind[(U8)op] == BRANCH) {
4235             node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
4236         }
4237         else if ( op == CURLY) {   /* `next' might be very big: optimizer */
4238             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4239                              NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
4240         }
4241         else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
4242             node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
4243                              next, sv, l + 1);
4244         }
4245         else if ( op == PLUS || op == STAR) {
4246             node = dumpuntil(start, NEXTOPER(node), NEXTOPER(node) + 1, sv, l + 1);
4247         }
4248         else if (op == ANYOF) {
4249             /* arglen 1 + class block */
4250             node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
4251                     ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
4252             node = NEXTOPER(node);
4253         }
4254         else if (PL_regkind[(U8)op] == EXACT) {
4255             /* Literal string, where present. */
4256             node += NODE_SZ_STR(node) - 1;
4257             node = NEXTOPER(node);
4258         }
4259         else {
4260             node = NEXTOPER(node);
4261             node += regarglen[(U8)op];
4262         }
4263         if (op == CURLYX || op == OPEN)
4264             l++;
4265         else if (op == WHILEM)
4266             l--;
4267     }
4268     return node;
4269 }
4270
4271 #endif  /* DEBUGGING */
4272
4273 /*
4274  - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
4275  */
4276 void
4277 Perl_regdump(pTHX_ regexp *r)
4278 {
4279 #ifdef DEBUGGING
4280     SV *sv = sv_newmortal();
4281
4282     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
4283
4284     /* Header fields of interest. */
4285     if (r->anchored_substr)
4286         PerlIO_printf(Perl_debug_log,
4287                       "anchored `%s%.*s%s'%s at %"IVdf" ",
4288                       PL_colors[0],
4289                       (int)(SvCUR(r->anchored_substr) - (SvTAIL(r->anchored_substr)!=0)),
4290                       SvPVX(r->anchored_substr),
4291                       PL_colors[1],
4292                       SvTAIL(r->anchored_substr) ? "$" : "",
4293                       (IV)r->anchored_offset);
4294     if (r->float_substr)
4295         PerlIO_printf(Perl_debug_log,
4296                       "floating `%s%.*s%s'%s at %"IVdf"..%"UVuf" ",
4297                       PL_colors[0],
4298                       (int)(SvCUR(r->float_substr) - (SvTAIL(r->float_substr)!=0)),
4299                       SvPVX(r->float_substr),
4300                       PL_colors[1],
4301                       SvTAIL(r->float_substr) ? "$" : "",
4302                       (IV)r->float_min_offset, (UV)r->float_max_offset);
4303     if (r->check_substr)
4304         PerlIO_printf(Perl_debug_log,
4305                       r->check_substr == r->float_substr
4306                       ? "(checking floating" : "(checking anchored");
4307     if (r->reganch & ROPT_NOSCAN)
4308         PerlIO_printf(Perl_debug_log, " noscan");
4309     if (r->reganch & ROPT_CHECK_ALL)
4310         PerlIO_printf(Perl_debug_log, " isall");
4311     if (r->check_substr)
4312         PerlIO_printf(Perl_debug_log, ") ");
4313
4314     if (r->regstclass) {
4315         regprop(sv, r->regstclass);
4316         PerlIO_printf(Perl_debug_log, "stclass `%s' ", SvPVX(sv));
4317     }
4318     if (r->reganch & ROPT_ANCH) {
4319         PerlIO_printf(Perl_debug_log, "anchored");
4320         if (r->reganch & ROPT_ANCH_BOL)
4321             PerlIO_printf(Perl_debug_log, "(BOL)");
4322         if (r->reganch & ROPT_ANCH_MBOL)
4323             PerlIO_printf(Perl_debug_log, "(MBOL)");
4324         if (r->reganch & ROPT_ANCH_SBOL)
4325             PerlIO_printf(Perl_debug_log, "(SBOL)");
4326         if (r->reganch & ROPT_ANCH_GPOS)
4327             PerlIO_printf(Perl_debug_log, "(GPOS)");
4328         PerlIO_putc(Perl_debug_log, ' ');
4329     }
4330     if (r->reganch & ROPT_GPOS_SEEN)
4331         PerlIO_printf(Perl_debug_log, "GPOS ");
4332     if (r->reganch & ROPT_SKIP)
4333         PerlIO_printf(Perl_debug_log, "plus ");
4334     if (r->reganch & ROPT_IMPLICIT)
4335         PerlIO_printf(Perl_debug_log, "implicit ");
4336     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
4337     if (r->reganch & ROPT_EVAL_SEEN)
4338         PerlIO_printf(Perl_debug_log, "with eval ");
4339     PerlIO_printf(Perl_debug_log, "\n");
4340     if (r->offsets) {
4341       U32 i;
4342       U32 len = r->offsets[0];
4343       PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)r->offsets[0]);
4344       for (i = 1; i <= len; i++)
4345         PerlIO_printf(Perl_debug_log, "%"UVuf"[%"UVuf"] ", 
4346                       (UV)r->offsets[i*2-1], 
4347                       (UV)r->offsets[i*2]);
4348       PerlIO_printf(Perl_debug_log, "\n");
4349     }
4350 #endif  /* DEBUGGING */
4351 }
4352
4353 #ifdef DEBUGGING
4354
4355 STATIC void
4356 S_put_byte(pTHX_ SV *sv, int c)
4357 {
4358     if (isCNTRL(c) || c == 255 || !isPRINT(c))
4359         Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
4360     else if (c == '-' || c == ']' || c == '\\' || c == '^')
4361         Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
4362     else
4363         Perl_sv_catpvf(aTHX_ sv, "%c", c);
4364 }
4365
4366 #endif  /* DEBUGGING */
4367
4368 /*
4369 - regprop - printable representation of opcode
4370 */
4371 void
4372 Perl_regprop(pTHX_ SV *sv, regnode *o)
4373 {
4374 #ifdef DEBUGGING
4375     register int k;
4376
4377     sv_setpvn(sv, "", 0);
4378     if (OP(o) >= reg_num)               /* regnode.type is unsigned */
4379         /* It would be nice to FAIL() here, but this may be called from
4380            regexec.c, and it would be hard to supply pRExC_state. */
4381         Perl_croak(aTHX_ "Corrupted regexp opcode");
4382     sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
4383
4384     k = PL_regkind[(U8)OP(o)];
4385
4386     if (k == EXACT)
4387         Perl_sv_catpvf(aTHX_ sv, " <%s%.*s%s>", PL_colors[0],
4388                        STR_LEN(o), STRING(o), PL_colors[1]);
4389     else if (k == CURLY) {
4390         if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
4391             Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
4392         Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
4393     }
4394     else if (k == WHILEM && o->flags)                   /* Ordinal/of */
4395         Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
4396     else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
4397         Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o));    /* Parenth number */
4398     else if (k == LOGICAL)
4399         Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);     /* 2: embedded, otherwise 1 */
4400     else if (k == ANYOF) {
4401         int i, rangestart = -1;
4402         U8 flags = ANYOF_FLAGS(o);
4403         const char * const anyofs[] = { /* Should be syncronized with
4404                                          * ANYOF_ #xdefines in regcomp.h */
4405             "\\w",
4406             "\\W",
4407             "\\s",
4408             "\\S",
4409             "\\d",
4410             "\\D",
4411             "[:alnum:]",
4412             "[:^alnum:]",
4413             "[:alpha:]",
4414             "[:^alpha:]",
4415             "[:ascii:]",
4416             "[:^ascii:]",
4417             "[:ctrl:]",
4418             "[:^ctrl:]",
4419             "[:graph:]",
4420             "[:^graph:]",
4421             "[:lower:]",
4422             "[:^lower:]",
4423             "[:print:]",
4424             "[:^print:]",
4425             "[:punct:]",
4426             "[:^punct:]",
4427             "[:upper:]",
4428             "[:^upper:]",
4429             "[:xdigit:]",
4430             "[:^xdigit:]",
4431             "[:space:]",
4432             "[:^space:]",
4433             "[:blank:]",
4434             "[:^blank:]"
4435         };
4436
4437         if (flags & ANYOF_LOCALE)
4438             sv_catpv(sv, "{loc}");
4439         if (flags & ANYOF_FOLD)
4440             sv_catpv(sv, "{i}");
4441         Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
4442         if (flags & ANYOF_INVERT)
4443             sv_catpv(sv, "^");
4444         for (i = 0; i <= 256; i++) {
4445             if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
4446                 if (rangestart == -1)
4447                     rangestart = i;
4448             } else if (rangestart != -1) {
4449                 if (i <= rangestart + 3)
4450                     for (; rangestart < i; rangestart++)
4451                         put_byte(sv, rangestart);
4452                 else {
4453                     put_byte(sv, rangestart);
4454                     sv_catpv(sv, "-");
4455                     put_byte(sv, i - 1);
4456                 }
4457                 rangestart = -1;
4458             }
4459         }
4460
4461         if (o->flags & ANYOF_CLASS)
4462             for (i = 0; i < sizeof(anyofs)/sizeof(char*); i++)
4463                 if (ANYOF_CLASS_TEST(o,i))
4464                     sv_catpv(sv, anyofs[i]);
4465
4466         if (flags & ANYOF_UNICODE)
4467             sv_catpv(sv, "{unicode}");
4468         else if (flags & ANYOF_UNICODE_ALL)
4469             sv_catpv(sv, "{all-unicode}");
4470
4471         {
4472             SV *lv;
4473             SV *sw = regclass_swash(o, FALSE, &lv);
4474         
4475             if (lv) {
4476                 if (sw) {
4477                     UV i;
4478                     U8 s[UTF8_MAXLEN+1];
4479                 
4480                     for (i = 0; i <= 256; i++) { /* just the first 256 */
4481                         U8 *e = uvchr_to_utf8(s, i);
4482                         
4483                         if (i < 256 && swash_fetch(sw, s, TRUE)) {
4484                             if (rangestart == -1)
4485                                 rangestart = i;
4486                         } else if (rangestart != -1) {
4487                             U8 *p;
4488                         
4489                             if (i <= rangestart + 3)
4490                                 for (; rangestart < i; rangestart++) {
4491                                     for(e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4492                                         put_byte(sv, *p);
4493                                 }
4494                             else {
4495                                 for (e = uvchr_to_utf8(s, rangestart), p = s; p < e; p++)
4496                                     put_byte(sv, *p);
4497                                 sv_catpv(sv, "-");
4498                                     for (e = uvchr_to_utf8(s, i - 1), p = s; p < e; p++)
4499                                         put_byte(sv, *p);
4500                                 }
4501                                 rangestart = -1;
4502                             }
4503                         }
4504                         
4505                     sv_catpv(sv, "..."); /* et cetera */
4506                 }
4507
4508                 {
4509                     char *s = savepv(SvPVX(lv));
4510                     char *origs = s;
4511                 
4512                     while(*s && *s != '\n') s++;
4513                 
4514                     if (*s == '\n') {
4515                         char *t = ++s;
4516                         
4517                         while (*s) {
4518                             if (*s == '\n')
4519                                 *s = ' ';
4520                             s++;
4521                         }
4522                         if (s[-1] == ' ')
4523                             s[-1] = 0;
4524                         
4525                         sv_catpv(sv, t);
4526                     }
4527                 
4528                     Safefree(origs);
4529                 }
4530             }
4531         }
4532
4533         Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
4534     }
4535     else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
4536         Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
4537 #endif  /* DEBUGGING */
4538 }
4539
4540 SV *
4541 Perl_re_intuit_string(pTHX_ regexp *prog)
4542 {                               /* Assume that RE_INTUIT is set */
4543     DEBUG_r(
4544         {   STRLEN n_a;
4545             char *s = SvPV(prog->check_substr,n_a);
4546
4547             if (!PL_colorset) reginitcolors();
4548             PerlIO_printf(Perl_debug_log,
4549                       "%sUsing REx substr:%s `%s%.60s%s%s'\n",
4550                       PL_colors[4],PL_colors[5],PL_colors[0],
4551                       s,
4552                       PL_colors[1],
4553                       (strlen(s) > 60 ? "..." : ""));
4554         } );
4555
4556     return prog->check_substr;
4557 }
4558
4559 void
4560 Perl_pregfree(pTHX_ struct regexp *r)
4561 {
4562     DEBUG_r(if (!PL_colorset) reginitcolors());
4563
4564     if (!r || (--r->refcnt > 0))
4565         return;
4566     DEBUG_r(PerlIO_printf(Perl_debug_log,
4567                       "%sFreeing REx:%s `%s%.60s%s%s'\n",
4568                       PL_colors[4],PL_colors[5],PL_colors[0],
4569                       r->precomp,
4570                       PL_colors[1],
4571                       (strlen(r->precomp) > 60 ? "..." : "")));
4572
4573     if (r->precomp)
4574         Safefree(r->precomp);
4575     if (r->offsets)             /* 20010421 MJD */
4576         Safefree(r->offsets);
4577     if (RX_MATCH_COPIED(r))
4578         Safefree(r->subbeg);
4579     if (r->substrs) {
4580         if (r->anchored_substr)
4581             SvREFCNT_dec(r->anchored_substr);
4582         if (r->float_substr)
4583             SvREFCNT_dec(r->float_substr);
4584         Safefree(r->substrs);
4585     }
4586     if (r->data) {
4587         int n = r->data->count;
4588         AV* new_comppad = NULL;
4589         AV* old_comppad;
4590         SV** old_curpad;
4591
4592         while (--n >= 0) {
4593           /* If you add a ->what type here, update the comment in regcomp.h */
4594             switch (r->data->what[n]) {
4595             case 's':
4596                 SvREFCNT_dec((SV*)r->data->data[n]);
4597                 break;
4598             case 'f':
4599                 Safefree(r->data->data[n]);
4600                 break;
4601             case 'p':
4602                 new_comppad = (AV*)r->data->data[n];
4603                 break;
4604             case 'o':
4605                 if (new_comppad == NULL)
4606                     Perl_croak(aTHX_ "panic: pregfree comppad");
4607                 old_comppad = PL_comppad;
4608                 old_curpad = PL_curpad;
4609                 /* Watch out for global destruction's random ordering. */
4610                 if (SvTYPE(new_comppad) == SVt_PVAV) {
4611                     PL_comppad = new_comppad;
4612                     PL_curpad = AvARRAY(new_comppad);
4613                 }
4614                 else
4615                     PL_curpad = NULL;
4616
4617                 if (!OpREFCNT_dec((OP_4tree*)r->data->data[n])) {
4618                     op_free((OP_4tree*)r->data->data[n]);
4619                 }
4620
4621                 PL_comppad = old_comppad;
4622                 PL_curpad = old_curpad;
4623                 SvREFCNT_dec((SV*)new_comppad);
4624                 new_comppad = NULL;
4625                 break;
4626             case 'n':
4627                 break;
4628             default:
4629                 Perl_croak(aTHX_ "panic: regfree data code '%c'", r->data->what[n]);
4630             }
4631         }
4632         Safefree(r->data->what);
4633         Safefree(r->data);
4634     }
4635     Safefree(r->startp);
4636     Safefree(r->endp);
4637     Safefree(r);
4638 }
4639
4640 /*
4641  - regnext - dig the "next" pointer out of a node
4642  *
4643  * [Note, when REGALIGN is defined there are two places in regmatch()
4644  * that bypass this code for speed.]
4645  */
4646 regnode *
4647 Perl_regnext(pTHX_ register regnode *p)
4648 {
4649     register I32 offset;
4650
4651     if (p == &PL_regdummy)
4652         return(NULL);
4653
4654     offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
4655     if (offset == 0)
4656         return(NULL);
4657
4658     return(p+offset);
4659 }
4660
4661 STATIC void     
4662 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
4663 {
4664     va_list args;
4665     STRLEN l1 = strlen(pat1);
4666     STRLEN l2 = strlen(pat2);
4667     char buf[512];
4668     SV *msv;
4669     char *message;
4670
4671     if (l1 > 510)
4672         l1 = 510;
4673     if (l1 + l2 > 510)
4674         l2 = 510 - l1;
4675     Copy(pat1, buf, l1 , char);
4676     Copy(pat2, buf + l1, l2 , char);
4677     buf[l1 + l2] = '\n';
4678     buf[l1 + l2 + 1] = '\0';
4679 #ifdef I_STDARG
4680     /* ANSI variant takes additional second argument */
4681     va_start(args, pat2);
4682 #else
4683     va_start(args);
4684 #endif
4685     msv = vmess(buf, &args);
4686     va_end(args);
4687     message = SvPV(msv,l1);
4688     if (l1 > 512)
4689         l1 = 512;
4690     Copy(message, buf, l1 , char);
4691     buf[l1] = '\0';                     /* Overwrite \n */
4692     Perl_croak(aTHX_ "%s", buf);
4693 }
4694
4695 /* XXX Here's a total kludge.  But we need to re-enter for swash routines. */
4696
4697 void
4698 Perl_save_re_context(pTHX)
4699 {
4700 #if 0
4701     SAVEPPTR(RExC_precomp);             /* uncompiled string. */
4702     SAVEI32(RExC_npar);         /* () count. */
4703     SAVEI32(RExC_size);         /* Code size. */
4704     SAVEI16(RExC_flags16);              /* are we folding, multilining? */
4705     SAVEVPTR(RExC_rx);          /* from regcomp.c */
4706     SAVEI32(RExC_seen);         /* from regcomp.c */
4707     SAVEI32(RExC_sawback);              /* Did we see \1, ...? */
4708     SAVEI32(RExC_naughty);              /* How bad is this pattern? */
4709     SAVEVPTR(RExC_emit);                /* Code-emit pointer; &regdummy = don't */
4710     SAVEPPTR(RExC_end);         /* End of input for compile */
4711     SAVEPPTR(RExC_parse);               /* Input-scan pointer. */
4712 #endif
4713
4714     SAVEI32(PL_reg_flags);              /* from regexec.c */
4715     SAVEPPTR(PL_bostr);
4716     SAVEPPTR(PL_reginput);              /* String-input pointer. */
4717     SAVEPPTR(PL_regbol);                /* Beginning of input, for ^ check. */
4718     SAVEPPTR(PL_regeol);                /* End of input, for $ check. */
4719     SAVEVPTR(PL_regstartp);             /* Pointer to startp array. */
4720     SAVEVPTR(PL_regendp);               /* Ditto for endp. */
4721     SAVEVPTR(PL_reglastparen);          /* Similarly for lastparen. */
4722     SAVEPPTR(PL_regtill);               /* How far we are required to go. */
4723     SAVEGENERICPV(PL_reg_start_tmp);            /* from regexec.c */
4724     PL_reg_start_tmp = 0;
4725     SAVEI32(PL_reg_start_tmpl);         /* from regexec.c */
4726     PL_reg_start_tmpl = 0;
4727     SAVEVPTR(PL_regdata);
4728     SAVEI32(PL_reg_eval_set);           /* from regexec.c */
4729     SAVEI32(PL_regnarrate);             /* from regexec.c */
4730     SAVEVPTR(PL_regprogram);            /* from regexec.c */
4731     SAVEINT(PL_regindent);              /* from regexec.c */
4732     SAVEVPTR(PL_regcc);                 /* from regexec.c */
4733     SAVEVPTR(PL_curcop);
4734     SAVEVPTR(PL_reg_call_cc);           /* from regexec.c */
4735     SAVEVPTR(PL_reg_re);                /* from regexec.c */
4736     SAVEPPTR(PL_reg_ganch);             /* from regexec.c */
4737     SAVESPTR(PL_reg_sv);                /* from regexec.c */
4738     SAVEVPTR(PL_reg_magic);             /* from regexec.c */
4739     SAVEI32(PL_reg_oldpos);                     /* from regexec.c */
4740     SAVEVPTR(PL_reg_oldcurpm);          /* from regexec.c */
4741     SAVEVPTR(PL_reg_curpm);             /* from regexec.c */
4742     SAVEI32(PL_regnpar);                /* () count. */
4743     SAVEI32(PL_regsize);                /* from regexec.c */
4744 #ifdef DEBUGGING
4745     SAVEPPTR(PL_reg_starttry);          /* from regexec.c */
4746 #endif
4747 }
4748
4749 #ifdef PERL_OBJECT
4750 #include "XSUB.h"
4751 #undef this
4752 #define this pPerl
4753 #endif
4754
4755 static void
4756 clear_re(pTHXo_ void *r)
4757 {
4758     ReREFCNT_dec((regexp *)r);
4759 }
4760