Gitiles
Code Review Sign In
review.jami.net / jami-client-android / 046164639702e8c26c5c1691c2ee0ac4b5f1e7c6 / . / jni / libpcre / pcre_dfa_exec.c
blob: 1ffbe3be842076e742a463ecb770a7acfafa128b [file] [log] [blame]
Tristan Matthews04616462013-11-14 16:09:34 -0500[diff] [blame^]1/*************************************************
2* Perl-Compatible Regular Expressions *
3*************************************************/
4
5/* PCRE is a library of functions to support regular expressions whose syntax
6and semantics are as close as possible to those of the Perl 5 language (but see
7below for why this module is different).
8
9 Written by Philip Hazel
10 Copyright (c) 1997-2011 University of Cambridge
11
12-----------------------------------------------------------------------------
13Redistribution and use in source and binary forms, with or without
14modification, are permitted provided that the following conditions are met:
15
16 * Redistributions of source code must retain the above copyright notice,
17 this list of conditions and the following disclaimer.
18
19 * Redistributions in binary form must reproduce the above copyright
20 notice, this list of conditions and the following disclaimer in the
21 documentation and/or other materials provided with the distribution.
22
23 * Neither the name of the University of Cambridge nor the names of its
24 contributors may be used to endorse or promote products derived from
25 this software without specific prior written permission.
26
27THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
28AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
31LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37POSSIBILITY OF SUCH DAMAGE.
38-----------------------------------------------------------------------------
39*/
40
41
42/* This module contains the external function pcre_dfa_exec(), which is an
43alternative matching function that uses a sort of DFA algorithm (not a true
44FSM). This is NOT Perl- compatible, but it has advantages in certain
45applications. */
46
47
48/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
49the performance of his patterns greatly. I could not use it as it stood, as it
50was not thread safe, and made assumptions about pattern sizes. Also, it caused
51test 7 to loop, and test 9 to crash with a segfault.
52
53The issue is the check for duplicate states, which is done by a simple linear
54search up the state list. (Grep for "duplicate" below to find the code.) For
55many patterns, there will never be many states active at one time, so a simple
56linear search is fine. In patterns that have many active states, it might be a
57bottleneck. The suggested code used an indexing scheme to remember which states
58had previously been used for each character, and avoided the linear search when
59it knew there was no chance of a duplicate. This was implemented when adding
60states to the state lists.
61
62I wrote some thread-safe, not-limited code to try something similar at the time
63of checking for duplicates (instead of when adding states), using index vectors
64on the stack. It did give a 13% improvement with one specially constructed
65pattern for certain subject strings, but on other strings and on many of the
66simpler patterns in the test suite it did worse. The major problem, I think,
67was the extra time to initialize the index. This had to be done for each call
68of internal_dfa_exec(). (The supplied patch used a static vector, initialized
69only once - I suspect this was the cause of the problems with the tests.)
70
71Overall, I concluded that the gains in some cases did not outweigh the losses
72in others, so I abandoned this code. */
73
74
75
76#ifdef HAVE_CONFIG_H
77#include "config.h"
78#endif
79
80#define NLBLOCK md /* Block containing newline information */
81#define PSSTART start_subject /* Field containing processed string start */
82#define PSEND end_subject /* Field containing processed string end */
83
84#include "pcre_internal.h"
85
86
87/* For use to indent debugging output */
88
89#define SP " "
90
91
92/*************************************************
93* Code parameters and static tables *
94*************************************************/
95
96/* These are offsets that are used to turn the OP_TYPESTAR and friends opcodes
97into others, under special conditions. A gap of 20 between the blocks should be
98enough. The resulting opcodes don't have to be less than 256 because they are
99never stored, so we push them well clear of the normal opcodes. */
100
101#define OP_PROP_EXTRA 300
102#define OP_EXTUNI_EXTRA 320
103#define OP_ANYNL_EXTRA 340
104#define OP_HSPACE_EXTRA 360
105#define OP_VSPACE_EXTRA 380
106
107
108/* This table identifies those opcodes that are followed immediately by a
109character that is to be tested in some way. This makes it possible to
110centralize the loading of these characters. In the case of Type * etc, the
111"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
112small value. Non-zero values in the table are the offsets from the opcode where
113the character is to be found. ***NOTE*** If the start of this table is
114modified, the three tables that follow must also be modified. */
115
116static const uschar coptable[] = {
117 0, /* End */
118 0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */
119 0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */
120 0, 0, 0, /* Any, AllAny, Anybyte */
121 0, 0, /* \P, \p */
122 0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */
123 0, /* \X */
124 0, 0, 0, 0, 0, 0, /* \Z, \z, ^, ^M, $, $M */
125 1, /* Char */
126 1, /* Chari */
127 1, /* not */
128 1, /* noti */
129 /* Positive single-char repeats */
130 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
131 3, 3, 3, /* upto, minupto, exact */
132 1, 1, 1, 3, /* *+, ++, ?+, upto+ */
133 1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
134 3, 3, 3, /* upto I, minupto I, exact I */
135 1, 1, 1, 3, /* *+I, ++I, ?+I, upto+I */
136 /* Negative single-char repeats - only for chars < 256 */
137 1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
138 3, 3, 3, /* NOT upto, minupto, exact */
139 1, 1, 1, 3, /* NOT *+, ++, ?+, upto+ */
140 1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
141 3, 3, 3, /* NOT upto I, minupto I, exact I */
142 1, 1, 1, 3, /* NOT *+I, ++I, ?+I, upto+I */
143 /* Positive type repeats */
144 1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
145 3, 3, 3, /* Type upto, minupto, exact */
146 1, 1, 1, 3, /* Type *+, ++, ?+, upto+ */
147 /* Character class & ref repeats */
148 0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */
149 0, 0, /* CRRANGE, CRMINRANGE */
150 0, /* CLASS */
151 0, /* NCLASS */
152 0, /* XCLASS - variable length */
153 0, /* REF */
154 0, /* REFI */
155 0, /* RECURSE */
156 0, /* CALLOUT */
157 0, /* Alt */
158 0, /* Ket */
159 0, /* KetRmax */
160 0, /* KetRmin */
161 0, /* KetRpos */
162 0, /* Reverse */
163 0, /* Assert */
164 0, /* Assert not */
165 0, /* Assert behind */
166 0, /* Assert behind not */
167 0, 0, /* ONCE, ONCE_NC */
168 0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
169 0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
170 0, 0, /* CREF, NCREF */
171 0, 0, /* RREF, NRREF */
172 0, /* DEF */
173 0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
174 0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
175 0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
176 0, 0, 0, 0, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */
177 0, 0 /* CLOSE, SKIPZERO */
178};
179
180/* This table identifies those opcodes that inspect a character. It is used to
181remember the fact that a character could have been inspected when the end of
182the subject is reached. ***NOTE*** If the start of this table is modified, the
183two tables that follow must also be modified. */
184
185static const uschar poptable[] = {
186 0, /* End */
187 0, 0, 0, 1, 1, /* \A, \G, \K, \B, \b */
188 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */
189 1, 1, 1, /* Any, AllAny, Anybyte */
190 1, 1, /* \P, \p */
191 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */
192 1, /* \X */
193 0, 0, 0, 0, 0, 0, /* \Z, \z, ^, ^M, $, $M */
194 1, /* Char */
195 1, /* Chari */
196 1, /* not */
197 1, /* noti */
198 /* Positive single-char repeats */
199 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
200 1, 1, 1, /* upto, minupto, exact */
201 1, 1, 1, 1, /* *+, ++, ?+, upto+ */
202 1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
203 1, 1, 1, /* upto I, minupto I, exact I */
204 1, 1, 1, 1, /* *+I, ++I, ?+I, upto+I */
205 /* Negative single-char repeats - only for chars < 256 */
206 1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
207 1, 1, 1, /* NOT upto, minupto, exact */
208 1, 1, 1, 1, /* NOT *+, ++, ?+, upto+ */
209 1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
210 1, 1, 1, /* NOT upto I, minupto I, exact I */
211 1, 1, 1, 1, /* NOT *+I, ++I, ?+I, upto+I */
212 /* Positive type repeats */
213 1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
214 1, 1, 1, /* Type upto, minupto, exact */
215 1, 1, 1, 1, /* Type *+, ++, ?+, upto+ */
216 /* Character class & ref repeats */
217 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
218 1, 1, /* CRRANGE, CRMINRANGE */
219 1, /* CLASS */
220 1, /* NCLASS */
221 1, /* XCLASS - variable length */
222 0, /* REF */
223 0, /* REFI */
224 0, /* RECURSE */
225 0, /* CALLOUT */
226 0, /* Alt */
227 0, /* Ket */
228 0, /* KetRmax */
229 0, /* KetRmin */
230 0, /* KetRpos */
231 0, /* Reverse */
232 0, /* Assert */
233 0, /* Assert not */
234 0, /* Assert behind */
235 0, /* Assert behind not */
236 0, 0, /* ONCE, ONCE_NC */
237 0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
238 0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
239 0, 0, /* CREF, NCREF */
240 0, 0, /* RREF, NRREF */
241 0, /* DEF */
242 0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
243 0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
244 0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
245 0, 0, 0, 0, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */
246 0, 0 /* CLOSE, SKIPZERO */
247};
248
249/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
250and \w */
251
252static const uschar toptable1[] = {
253 0, 0, 0, 0, 0, 0,
254 ctype_digit, ctype_digit,
255 ctype_space, ctype_space,
256 ctype_word, ctype_word,
257 0, 0 /* OP_ANY, OP_ALLANY */
258};
259
260static const uschar toptable2[] = {
261 0, 0, 0, 0, 0, 0,
262 ctype_digit, 0,
263 ctype_space, 0,
264 ctype_word, 0,
265 1, 1 /* OP_ANY, OP_ALLANY */
266};
267
268
269/* Structure for holding data about a particular state, which is in effect the
270current data for an active path through the match tree. It must consist
271entirely of ints because the working vector we are passed, and which we put
272these structures in, is a vector of ints. */
273
274typedef struct stateblock {
275 int offset; /* Offset to opcode */
276 int count; /* Count for repeats */
277 int data; /* Some use extra data */
278} stateblock;
279
280#define INTS_PER_STATEBLOCK (sizeof(stateblock)/sizeof(int))
281
282
283#ifdef PCRE_DEBUG
284/*************************************************
285* Print character string *
286*************************************************/
287
288/* Character string printing function for debugging.
289
290Arguments:
291 p points to string
292 length number of bytes
293 f where to print
294
295Returns: nothing
296*/
297
298static void
299pchars(unsigned char *p, int length, FILE *f)
300{
301int c;
302while (length-- > 0)
303 {
304 if (isprint(c = *(p++)))
305 fprintf(f, "%c", c);
306 else
307 fprintf(f, "\\x%02x", c);
308 }
309}
310#endif
311
312
313
314/*************************************************
315* Execute a Regular Expression - DFA engine *
316*************************************************/
317
318/* This internal function applies a compiled pattern to a subject string,
319starting at a given point, using a DFA engine. This function is called from the
320external one, possibly multiple times if the pattern is not anchored. The
321function calls itself recursively for some kinds of subpattern.
322
323Arguments:
324 md the match_data block with fixed information
325 this_start_code the opening bracket of this subexpression's code
326 current_subject where we currently are in the subject string
327 start_offset start offset in the subject string
328 offsets vector to contain the matching string offsets
329 offsetcount size of same
330 workspace vector of workspace
331 wscount size of same
332 rlevel function call recursion level
333
334Returns: > 0 => number of match offset pairs placed in offsets
335 = 0 => offsets overflowed; longest matches are present
336 -1 => failed to match
337 < -1 => some kind of unexpected problem
338
339The following macros are used for adding states to the two state vectors (one
340for the current character, one for the following character). */
341
342#define ADD_ACTIVE(x,y) \
343 if (active_count++ < wscount) \
344 { \
345 next_active_state->offset = (x); \
346 next_active_state->count = (y); \
347 next_active_state++; \
348 DPRINTF(("%.*sADD_ACTIVE(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \
349 } \
350 else return PCRE_ERROR_DFA_WSSIZE
351
352#define ADD_ACTIVE_DATA(x,y,z) \
353 if (active_count++ < wscount) \
354 { \
355 next_active_state->offset = (x); \
356 next_active_state->count = (y); \
357 next_active_state->data = (z); \
358 next_active_state++; \
359 DPRINTF(("%.*sADD_ACTIVE_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \
360 } \
361 else return PCRE_ERROR_DFA_WSSIZE
362
363#define ADD_NEW(x,y) \
364 if (new_count++ < wscount) \
365 { \
366 next_new_state->offset = (x); \
367 next_new_state->count = (y); \
368 next_new_state++; \
369 DPRINTF(("%.*sADD_NEW(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \
370 } \
371 else return PCRE_ERROR_DFA_WSSIZE
372
373#define ADD_NEW_DATA(x,y,z) \
374 if (new_count++ < wscount) \
375 { \
376 next_new_state->offset = (x); \
377 next_new_state->count = (y); \
378 next_new_state->data = (z); \
379 next_new_state++; \
380 DPRINTF(("%.*sADD_NEW_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \
381 } \
382 else return PCRE_ERROR_DFA_WSSIZE
383
384/* And now, here is the code */
385
386static int
387internal_dfa_exec(
388 dfa_match_data *md,
389 const uschar *this_start_code,
390 const uschar *current_subject,
391 int start_offset,
392 int *offsets,
393 int offsetcount,
394 int *workspace,
395 int wscount,
396 int rlevel)
397{
398stateblock *active_states, *new_states, *temp_states;
399stateblock *next_active_state, *next_new_state;
400
401const uschar *ctypes, *lcc, *fcc;
402const uschar *ptr;
403const uschar *end_code, *first_op;
404
405dfa_recursion_info new_recursive;
406
407int active_count, new_count, match_count;
408
409/* Some fields in the md block are frequently referenced, so we load them into
410independent variables in the hope that this will perform better. */
411
412const uschar *start_subject = md->start_subject;
413const uschar *end_subject = md->end_subject;
414const uschar *start_code = md->start_code;
415
416#ifdef SUPPORT_UTF8
417BOOL utf8 = (md->poptions & PCRE_UTF8) != 0;
418#else
419BOOL utf8 = FALSE;
420#endif
421
422rlevel++;
423offsetcount &= (-2);
424
425wscount -= 2;
426wscount = (wscount - (wscount % (INTS_PER_STATEBLOCK * 2))) /
427 (2 * INTS_PER_STATEBLOCK);
428
429DPRINTF(("\n%.*s---------------------\n"
430 "%.*sCall to internal_dfa_exec f=%d\n",
431 rlevel*2-2, SP, rlevel*2-2, SP, rlevel));
432
433ctypes = md->tables + ctypes_offset;
434lcc = md->tables + lcc_offset;
435fcc = md->tables + fcc_offset;
436
437match_count = PCRE_ERROR_NOMATCH; /* A negative number */
438
439active_states = (stateblock *)(workspace + 2);
440next_new_state = new_states = active_states + wscount;
441new_count = 0;
442
443first_op = this_start_code + 1 + LINK_SIZE +
444 ((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
445 *this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)? 2:0);
446
447/* The first thing in any (sub) pattern is a bracket of some sort. Push all
448the alternative states onto the list, and find out where the end is. This
449makes is possible to use this function recursively, when we want to stop at a
450matching internal ket rather than at the end.
451
452If the first opcode in the first alternative is OP_REVERSE, we are dealing with
453a backward assertion. In that case, we have to find out the maximum amount to
454move back, and set up each alternative appropriately. */
455
456if (*first_op == OP_REVERSE)
457 {
458 int max_back = 0;
459 int gone_back;
460
461 end_code = this_start_code;
462 do
463 {
464 int back = GET(end_code, 2+LINK_SIZE);
465 if (back > max_back) max_back = back;
466 end_code += GET(end_code, 1);
467 }
468 while (*end_code == OP_ALT);
469
470 /* If we can't go back the amount required for the longest lookbehind
471 pattern, go back as far as we can; some alternatives may still be viable. */
472
473#ifdef SUPPORT_UTF8
474 /* In character mode we have to step back character by character */
475
476 if (utf8)
477 {
478 for (gone_back = 0; gone_back < max_back; gone_back++)
479 {
480 if (current_subject <= start_subject) break;
481 current_subject--;
482 while (current_subject > start_subject &&
483 (*current_subject & 0xc0) == 0x80)
484 current_subject--;
485 }
486 }
487 else
488#endif
489
490 /* In byte-mode we can do this quickly. */
491
492 {
493 gone_back = (current_subject - max_back < start_subject)?
494 (int)(current_subject - start_subject) : max_back;
495 current_subject -= gone_back;
496 }
497
498 /* Save the earliest consulted character */
499
500 if (current_subject < md->start_used_ptr)
501 md->start_used_ptr = current_subject;
502
503 /* Now we can process the individual branches. */
504
505 end_code = this_start_code;
506 do
507 {
508 int back = GET(end_code, 2+LINK_SIZE);
509 if (back <= gone_back)
510 {
511 int bstate = (int)(end_code - start_code + 2 + 2*LINK_SIZE);
512 ADD_NEW_DATA(-bstate, 0, gone_back - back);
513 }
514 end_code += GET(end_code, 1);
515 }
516 while (*end_code == OP_ALT);
517 }
518
519/* This is the code for a "normal" subpattern (not a backward assertion). The
520start of a whole pattern is always one of these. If we are at the top level,
521we may be asked to restart matching from the same point that we reached for a
522previous partial match. We still have to scan through the top-level branches to
523find the end state. */
524
525else
526 {
527 end_code = this_start_code;
528
529 /* Restarting */
530
531 if (rlevel == 1 && (md->moptions & PCRE_DFA_RESTART) != 0)
532 {
533 do { end_code += GET(end_code, 1); } while (*end_code == OP_ALT);
534 new_count = workspace[1];
535 if (!workspace[0])
536 memcpy(new_states, active_states, new_count * sizeof(stateblock));
537 }
538
539 /* Not restarting */
540
541 else
542 {
543 int length = 1 + LINK_SIZE +
544 ((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
545 *this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)?
546 2:0);
547 do
548 {
549 ADD_NEW((int)(end_code - start_code + length), 0);
550 end_code += GET(end_code, 1);
551 length = 1 + LINK_SIZE;
552 }
553 while (*end_code == OP_ALT);
554 }
555 }
556
557workspace[0] = 0; /* Bit indicating which vector is current */
558
559DPRINTF(("%.*sEnd state = %d\n", rlevel*2-2, SP, end_code - start_code));
560
561/* Loop for scanning the subject */
562
563ptr = current_subject;
564for (;;)
565 {
566 int i, j;
567 int clen, dlen;
568 unsigned int c, d;
569 int forced_fail = 0;
570 BOOL could_continue = FALSE;
571
572 /* Make the new state list into the active state list and empty the
573 new state list. */
574
575 temp_states = active_states;
576 active_states = new_states;
577 new_states = temp_states;
578 active_count = new_count;
579 new_count = 0;
580
581 workspace[0] ^= 1; /* Remember for the restarting feature */
582 workspace[1] = active_count;
583
584#ifdef PCRE_DEBUG
585 printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP);
586 pchars((uschar *)ptr, strlen((char *)ptr), stdout);
587 printf("\"\n");
588
589 printf("%.*sActive states: ", rlevel*2-2, SP);
590 for (i = 0; i < active_count; i++)
591 printf("%d/%d ", active_states[i].offset, active_states[i].count);
592 printf("\n");
593#endif
594
595 /* Set the pointers for adding new states */
596
597 next_active_state = active_states + active_count;
598 next_new_state = new_states;
599
600 /* Load the current character from the subject outside the loop, as many
601 different states may want to look at it, and we assume that at least one
602 will. */
603
604 if (ptr < end_subject)
605 {
606 clen = 1; /* Number of bytes in the character */
607#ifdef SUPPORT_UTF8
608 if (utf8) { GETCHARLEN(c, ptr, clen); } else
609#endif /* SUPPORT_UTF8 */
610 c = *ptr;
611 }
612 else
613 {
614 clen = 0; /* This indicates the end of the subject */
615 c = NOTACHAR; /* This value should never actually be used */
616 }
617
618 /* Scan up the active states and act on each one. The result of an action
619 may be to add more states to the currently active list (e.g. on hitting a
620 parenthesis) or it may be to put states on the new list, for considering
621 when we move the character pointer on. */
622
623 for (i = 0; i < active_count; i++)
624 {
625 stateblock *current_state = active_states + i;
626 BOOL caseless = FALSE;
627 const uschar *code;
628 int state_offset = current_state->offset;
629 int count, codevalue, rrc;
630
631#ifdef PCRE_DEBUG
632 printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset);
633 if (clen == 0) printf("EOL\n");
634 else if (c > 32 && c < 127) printf("'%c'\n", c);
635 else printf("0x%02x\n", c);
636#endif
637
638 /* A negative offset is a special case meaning "hold off going to this
639 (negated) state until the number of characters in the data field have
640 been skipped". */
641
642 if (state_offset < 0)
643 {
644 if (current_state->data > 0)
645 {
646 DPRINTF(("%.*sSkipping this character\n", rlevel*2-2, SP));
647 ADD_NEW_DATA(state_offset, current_state->count,
648 current_state->data - 1);
649 continue;
650 }
651 else
652 {
653 current_state->offset = state_offset = -state_offset;
654 }
655 }
656
657 /* Check for a duplicate state with the same count, and skip if found.
658 See the note at the head of this module about the possibility of improving
659 performance here. */
660
661 for (j = 0; j < i; j++)
662 {
663 if (active_states[j].offset == state_offset &&
664 active_states[j].count == current_state->count)
665 {
666 DPRINTF(("%.*sDuplicate state: skipped\n", rlevel*2-2, SP));
667 goto NEXT_ACTIVE_STATE;
668 }
669 }
670
671 /* The state offset is the offset to the opcode */
672
673 code = start_code + state_offset;
674 codevalue = *code;
675
676 /* If this opcode inspects a character, but we are at the end of the
677 subject, remember the fact for use when testing for a partial match. */
678
679 if (clen == 0 && poptable[codevalue] != 0)
680 could_continue = TRUE;
681
682 /* If this opcode is followed by an inline character, load it. It is
683 tempting to test for the presence of a subject character here, but that
684 is wrong, because sometimes zero repetitions of the subject are
685 permitted.
686
687 We also use this mechanism for opcodes such as OP_TYPEPLUS that take an
688 argument that is not a data character - but is always one byte long. We
689 have to take special action to deal with \P, \p, \H, \h, \V, \v and \X in
690 this case. To keep the other cases fast, convert these ones to new opcodes.
691 */
692
693 if (coptable[codevalue] > 0)
694 {
695 dlen = 1;
696#ifdef SUPPORT_UTF8
697 if (utf8) { GETCHARLEN(d, (code + coptable[codevalue]), dlen); } else
698#endif /* SUPPORT_UTF8 */
699 d = code[coptable[codevalue]];
700 if (codevalue >= OP_TYPESTAR)
701 {
702 switch(d)
703 {
704 case OP_ANYBYTE: return PCRE_ERROR_DFA_UITEM;
705 case OP_NOTPROP:
706 case OP_PROP: codevalue += OP_PROP_EXTRA; break;
707 case OP_ANYNL: codevalue += OP_ANYNL_EXTRA; break;
708 case OP_EXTUNI: codevalue += OP_EXTUNI_EXTRA; break;
709 case OP_NOT_HSPACE:
710 case OP_HSPACE: codevalue += OP_HSPACE_EXTRA; break;
711 case OP_NOT_VSPACE:
712 case OP_VSPACE: codevalue += OP_VSPACE_EXTRA; break;
713 default: break;
714 }
715 }
716 }
717 else
718 {
719 dlen = 0; /* Not strictly necessary, but compilers moan */
720 d = NOTACHAR; /* if these variables are not set. */
721 }
722
723
724 /* Now process the individual opcodes */
725
726 switch (codevalue)
727 {
728/* ========================================================================== */
729 /* These cases are never obeyed. This is a fudge that causes a compile-
730 time error if the vectors coptable or poptable, which are indexed by
731 opcode, are not the correct length. It seems to be the only way to do
732 such a check at compile time, as the sizeof() operator does not work
733 in the C preprocessor. */
734
735 case OP_TABLE_LENGTH:
736 case OP_TABLE_LENGTH +
737 ((sizeof(coptable) == OP_TABLE_LENGTH) &&
738 (sizeof(poptable) == OP_TABLE_LENGTH)):
739 break;
740
741/* ========================================================================== */
742 /* Reached a closing bracket. If not at the end of the pattern, carry
743 on with the next opcode. For repeating opcodes, also add the repeat
744 state. Note that KETRPOS will always be encountered at the end of the
745 subpattern, because the possessive subpattern repeats are always handled
746 using recursive calls. Thus, it never adds any new states.
747
748 At the end of the (sub)pattern, unless we have an empty string and
749 PCRE_NOTEMPTY is set, or PCRE_NOTEMPTY_ATSTART is set and we are at the
750 start of the subject, save the match data, shifting up all previous
751 matches so we always have the longest first. */
752
753 case OP_KET:
754 case OP_KETRMIN:
755 case OP_KETRMAX:
756 case OP_KETRPOS:
757 if (code != end_code)
758 {
759 ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0);
760 if (codevalue != OP_KET)
761 {
762 ADD_ACTIVE(state_offset - GET(code, 1), 0);
763 }
764 }
765 else
766 {
767 if (ptr > current_subject ||
768 ((md->moptions & PCRE_NOTEMPTY) == 0 &&
769 ((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 ||
770 current_subject > start_subject + md->start_offset)))
771 {
772 if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
773 else if (match_count > 0 && ++match_count * 2 > offsetcount)
774 match_count = 0;
775 count = ((match_count == 0)? offsetcount : match_count * 2) - 2;
776 if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int));
777 if (offsetcount >= 2)
778 {
779 offsets[0] = (int)(current_subject - start_subject);
780 offsets[1] = (int)(ptr - start_subject);
781 DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP,
782 offsets[1] - offsets[0], current_subject));
783 }
784 if ((md->moptions & PCRE_DFA_SHORTEST) != 0)
785 {
786 DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
787 "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel,
788 match_count, rlevel*2-2, SP));
789 return match_count;
790 }
791 }
792 }
793 break;
794
795/* ========================================================================== */
796 /* These opcodes add to the current list of states without looking
797 at the current character. */
798
799 /*-----------------------------------------------------------------*/
800 case OP_ALT:
801 do { code += GET(code, 1); } while (*code == OP_ALT);
802 ADD_ACTIVE((int)(code - start_code), 0);
803 break;
804
805 /*-----------------------------------------------------------------*/
806 case OP_BRA:
807 case OP_SBRA:
808 do
809 {
810 ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
811 code += GET(code, 1);
812 }
813 while (*code == OP_ALT);
814 break;
815
816 /*-----------------------------------------------------------------*/
817 case OP_CBRA:
818 case OP_SCBRA:
819 ADD_ACTIVE((int)(code - start_code + 3 + LINK_SIZE), 0);
820 code += GET(code, 1);
821 while (*code == OP_ALT)
822 {
823 ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
824 code += GET(code, 1);
825 }
826 break;
827
828 /*-----------------------------------------------------------------*/
829 case OP_BRAZERO:
830 case OP_BRAMINZERO:
831 ADD_ACTIVE(state_offset + 1, 0);
832 code += 1 + GET(code, 2);
833 while (*code == OP_ALT) code += GET(code, 1);
834 ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
835 break;
836
837 /*-----------------------------------------------------------------*/
838 case OP_SKIPZERO:
839 code += 1 + GET(code, 2);
840 while (*code == OP_ALT) code += GET(code, 1);
841 ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
842 break;
843
844 /*-----------------------------------------------------------------*/
845 case OP_CIRC:
846 if (ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0)
847 { ADD_ACTIVE(state_offset + 1, 0); }
848 break;
849
850 /*-----------------------------------------------------------------*/
851 case OP_CIRCM:
852 if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) ||
853 (ptr != end_subject && WAS_NEWLINE(ptr)))
854 { ADD_ACTIVE(state_offset + 1, 0); }
855 break;
856
857 /*-----------------------------------------------------------------*/
858 case OP_EOD:
859 if (ptr >= end_subject)
860 {
861 if ((md->moptions & PCRE_PARTIAL_HARD) != 0)
862 could_continue = TRUE;
863 else { ADD_ACTIVE(state_offset + 1, 0); }
864 }
865 break;
866
867 /*-----------------------------------------------------------------*/
868 case OP_SOD:
869 if (ptr == start_subject) { ADD_ACTIVE(state_offset + 1, 0); }
870 break;
871
872 /*-----------------------------------------------------------------*/
873 case OP_SOM:
874 if (ptr == start_subject + start_offset) { ADD_ACTIVE(state_offset + 1, 0); }
875 break;
876
877
878/* ========================================================================== */
879 /* These opcodes inspect the next subject character, and sometimes
880 the previous one as well, but do not have an argument. The variable
881 clen contains the length of the current character and is zero if we are
882 at the end of the subject. */
883
884 /*-----------------------------------------------------------------*/
885 case OP_ANY:
886 if (clen > 0 && !IS_NEWLINE(ptr))
887 { ADD_NEW(state_offset + 1, 0); }
888 break;
889
890 /*-----------------------------------------------------------------*/
891 case OP_ALLANY:
892 if (clen > 0)
893 { ADD_NEW(state_offset + 1, 0); }
894 break;
895
896 /*-----------------------------------------------------------------*/
897 case OP_EODN:
898 if (clen == 0 && (md->moptions & PCRE_PARTIAL_HARD) != 0)
899 could_continue = TRUE;
900 else if (clen == 0 || (IS_NEWLINE(ptr) && ptr == end_subject - md->nllen))
901 { ADD_ACTIVE(state_offset + 1, 0); }
902 break;
903
904 /*-----------------------------------------------------------------*/
905 case OP_DOLL:
906 if ((md->moptions & PCRE_NOTEOL) == 0)
907 {
908 if (clen == 0 && (md->moptions & PCRE_PARTIAL_HARD) != 0)
909 could_continue = TRUE;
910 else if (clen == 0 ||
911 ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
912 (ptr == end_subject - md->nllen)
913 ))
914 { ADD_ACTIVE(state_offset + 1, 0); }
915 }
916 break;
917
918 /*-----------------------------------------------------------------*/
919 case OP_DOLLM:
920 if ((md->moptions & PCRE_NOTEOL) == 0)
921 {
922 if (clen == 0 && (md->moptions & PCRE_PARTIAL_HARD) != 0)
923 could_continue = TRUE;
924 else if (clen == 0 ||
925 ((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr)))
926 { ADD_ACTIVE(state_offset + 1, 0); }
927 }
928 else if (IS_NEWLINE(ptr))
929 { ADD_ACTIVE(state_offset + 1, 0); }
930 break;
931
932 /*-----------------------------------------------------------------*/
933
934 case OP_DIGIT:
935 case OP_WHITESPACE:
936 case OP_WORDCHAR:
937 if (clen > 0 && c < 256 &&
938 ((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0)
939 { ADD_NEW(state_offset + 1, 0); }
940 break;
941
942 /*-----------------------------------------------------------------*/
943 case OP_NOT_DIGIT:
944 case OP_NOT_WHITESPACE:
945 case OP_NOT_WORDCHAR:
946 if (clen > 0 && (c >= 256 ||
947 ((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0))
948 { ADD_NEW(state_offset + 1, 0); }
949 break;
950
951 /*-----------------------------------------------------------------*/
952 case OP_WORD_BOUNDARY:
953 case OP_NOT_WORD_BOUNDARY:
954 {
955 int left_word, right_word;
956
957 if (ptr > start_subject)
958 {
959 const uschar *temp = ptr - 1;
960 if (temp < md->start_used_ptr) md->start_used_ptr = temp;
961#ifdef SUPPORT_UTF8
962 if (utf8) BACKCHAR(temp);
963#endif
964 GETCHARTEST(d, temp);
965#ifdef SUPPORT_UCP
966 if ((md->poptions & PCRE_UCP) != 0)
967 {
968 if (d == '_') left_word = TRUE; else
969 {
970 int cat = UCD_CATEGORY(d);
971 left_word = (cat == ucp_L || cat == ucp_N);
972 }
973 }
974 else
975#endif
976 left_word = d < 256 && (ctypes[d] & ctype_word) != 0;
977 }
978 else left_word = FALSE;
979
980 if (clen > 0)
981 {
982#ifdef SUPPORT_UCP
983 if ((md->poptions & PCRE_UCP) != 0)
984 {
985 if (c == '_') right_word = TRUE; else
986 {
987 int cat = UCD_CATEGORY(c);
988 right_word = (cat == ucp_L || cat == ucp_N);
989 }
990 }
991 else
992#endif
993 right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
994 }
995 else right_word = FALSE;
996
997 if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY))
998 { ADD_ACTIVE(state_offset + 1, 0); }
999 }
1000 break;
1001
1002
1003 /*-----------------------------------------------------------------*/
1004 /* Check the next character by Unicode property. We will get here only
1005 if the support is in the binary; otherwise a compile-time error occurs.
1006 */
1007
1008#ifdef SUPPORT_UCP
1009 case OP_PROP:
1010 case OP_NOTPROP:
1011 if (clen > 0)
1012 {
1013 BOOL OK;
1014 const ucd_record * prop = GET_UCD(c);
1015 switch(code[1])
1016 {
1017 case PT_ANY:
1018 OK = TRUE;
1019 break;
1020
1021 case PT_LAMP:
1022 OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1023 prop->chartype == ucp_Lt;
1024 break;
1025
1026 case PT_GC:
1027 OK = _pcre_ucp_gentype[prop->chartype] == code[2];
1028 break;
1029
1030 case PT_PC:
1031 OK = prop->chartype == code[2];
1032 break;
1033
1034 case PT_SC:
1035 OK = prop->script == code[2];
1036 break;
1037
1038 /* These are specials for combination cases. */
1039
1040 case PT_ALNUM:
1041 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1042 _pcre_ucp_gentype[prop->chartype] == ucp_N;
1043 break;
1044
1045 case PT_SPACE: /* Perl space */
1046 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1047 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1048 break;
1049
1050 case PT_PXSPACE: /* POSIX space */
1051 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1052 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1053 c == CHAR_FF || c == CHAR_CR;
1054 break;
1055
1056 case PT_WORD:
1057 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1058 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1059 c == CHAR_UNDERSCORE;
1060 break;
1061
1062 /* Should never occur, but keep compilers from grumbling. */
1063
1064 default:
1065 OK = codevalue != OP_PROP;
1066 break;
1067 }
1068
1069 if (OK == (codevalue == OP_PROP)) { ADD_NEW(state_offset + 3, 0); }
1070 }
1071 break;
1072#endif
1073
1074
1075
1076/* ========================================================================== */
1077 /* These opcodes likewise inspect the subject character, but have an
1078 argument that is not a data character. It is one of these opcodes:
1079 OP_ANY, OP_ALLANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE,
1080 OP_WORDCHAR, OP_NOT_WORDCHAR. The value is loaded into d. */
1081
1082 case OP_TYPEPLUS:
1083 case OP_TYPEMINPLUS:
1084 case OP_TYPEPOSPLUS:
1085 count = current_state->count; /* Already matched */
1086 if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1087 if (clen > 0)
1088 {
1089 if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1090 (c < 256 &&
1091 (d != OP_ANY || !IS_NEWLINE(ptr)) &&
1092 ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1093 {
1094 if (count > 0 && codevalue == OP_TYPEPOSPLUS)
1095 {
1096 active_count--; /* Remove non-match possibility */
1097 next_active_state--;
1098 }
1099 count++;
1100 ADD_NEW(state_offset, count);
1101 }
1102 }
1103 break;
1104
1105 /*-----------------------------------------------------------------*/
1106 case OP_TYPEQUERY:
1107 case OP_TYPEMINQUERY:
1108 case OP_TYPEPOSQUERY:
1109 ADD_ACTIVE(state_offset + 2, 0);
1110 if (clen > 0)
1111 {
1112 if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1113 (c < 256 &&
1114 (d != OP_ANY || !IS_NEWLINE(ptr)) &&
1115 ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1116 {
1117 if (codevalue == OP_TYPEPOSQUERY)
1118 {
1119 active_count--; /* Remove non-match possibility */
1120 next_active_state--;
1121 }
1122 ADD_NEW(state_offset + 2, 0);
1123 }
1124 }
1125 break;
1126
1127 /*-----------------------------------------------------------------*/
1128 case OP_TYPESTAR:
1129 case OP_TYPEMINSTAR:
1130 case OP_TYPEPOSSTAR:
1131 ADD_ACTIVE(state_offset + 2, 0);
1132 if (clen > 0)
1133 {
1134 if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1135 (c < 256 &&
1136 (d != OP_ANY || !IS_NEWLINE(ptr)) &&
1137 ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1138 {
1139 if (codevalue == OP_TYPEPOSSTAR)
1140 {
1141 active_count--; /* Remove non-match possibility */
1142 next_active_state--;
1143 }
1144 ADD_NEW(state_offset, 0);
1145 }
1146 }
1147 break;
1148
1149 /*-----------------------------------------------------------------*/
1150 case OP_TYPEEXACT:
1151 count = current_state->count; /* Number already matched */
1152 if (clen > 0)
1153 {
1154 if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1155 (c < 256 &&
1156 (d != OP_ANY || !IS_NEWLINE(ptr)) &&
1157 ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1158 {
1159 if (++count >= GET2(code, 1))
1160 { ADD_NEW(state_offset + 4, 0); }
1161 else
1162 { ADD_NEW(state_offset, count); }
1163 }
1164 }
1165 break;
1166
1167 /*-----------------------------------------------------------------*/
1168 case OP_TYPEUPTO:
1169 case OP_TYPEMINUPTO:
1170 case OP_TYPEPOSUPTO:
1171 ADD_ACTIVE(state_offset + 4, 0);
1172 count = current_state->count; /* Number already matched */
1173 if (clen > 0)
1174 {
1175 if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
1176 (c < 256 &&
1177 (d != OP_ANY || !IS_NEWLINE(ptr)) &&
1178 ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
1179 {
1180 if (codevalue == OP_TYPEPOSUPTO)
1181 {
1182 active_count--; /* Remove non-match possibility */
1183 next_active_state--;
1184 }
1185 if (++count >= GET2(code, 1))
1186 { ADD_NEW(state_offset + 4, 0); }
1187 else
1188 { ADD_NEW(state_offset, count); }
1189 }
1190 }
1191 break;
1192
1193/* ========================================================================== */
1194 /* These are virtual opcodes that are used when something like
1195 OP_TYPEPLUS has OP_PROP, OP_NOTPROP, OP_ANYNL, or OP_EXTUNI as its
1196 argument. It keeps the code above fast for the other cases. The argument
1197 is in the d variable. */
1198
1199#ifdef SUPPORT_UCP
1200 case OP_PROP_EXTRA + OP_TYPEPLUS:
1201 case OP_PROP_EXTRA + OP_TYPEMINPLUS:
1202 case OP_PROP_EXTRA + OP_TYPEPOSPLUS:
1203 count = current_state->count; /* Already matched */
1204 if (count > 0) { ADD_ACTIVE(state_offset + 4, 0); }
1205 if (clen > 0)
1206 {
1207 BOOL OK;
1208 const ucd_record * prop = GET_UCD(c);
1209 switch(code[2])
1210 {
1211 case PT_ANY:
1212 OK = TRUE;
1213 break;
1214
1215 case PT_LAMP:
1216 OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1217 prop->chartype == ucp_Lt;
1218 break;
1219
1220 case PT_GC:
1221 OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1222 break;
1223
1224 case PT_PC:
1225 OK = prop->chartype == code[3];
1226 break;
1227
1228 case PT_SC:
1229 OK = prop->script == code[3];
1230 break;
1231
1232 /* These are specials for combination cases. */
1233
1234 case PT_ALNUM:
1235 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1236 _pcre_ucp_gentype[prop->chartype] == ucp_N;
1237 break;
1238
1239 case PT_SPACE: /* Perl space */
1240 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1241 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1242 break;
1243
1244 case PT_PXSPACE: /* POSIX space */
1245 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1246 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1247 c == CHAR_FF || c == CHAR_CR;
1248 break;
1249
1250 case PT_WORD:
1251 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1252 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1253 c == CHAR_UNDERSCORE;
1254 break;
1255
1256 /* Should never occur, but keep compilers from grumbling. */
1257
1258 default:
1259 OK = codevalue != OP_PROP;
1260 break;
1261 }
1262
1263 if (OK == (d == OP_PROP))
1264 {
1265 if (count > 0 && codevalue == OP_PROP_EXTRA + OP_TYPEPOSPLUS)
1266 {
1267 active_count--; /* Remove non-match possibility */
1268 next_active_state--;
1269 }
1270 count++;
1271 ADD_NEW(state_offset, count);
1272 }
1273 }
1274 break;
1275
1276 /*-----------------------------------------------------------------*/
1277 case OP_EXTUNI_EXTRA + OP_TYPEPLUS:
1278 case OP_EXTUNI_EXTRA + OP_TYPEMINPLUS:
1279 case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
1280 count = current_state->count; /* Already matched */
1281 if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1282 if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1283 {
1284 const uschar *nptr = ptr + clen;
1285 int ncount = 0;
1286 if (count > 0 && codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS)
1287 {
1288 active_count--; /* Remove non-match possibility */
1289 next_active_state--;
1290 }
1291 while (nptr < end_subject)
1292 {
1293 int nd;
1294 int ndlen = 1;
1295 GETCHARLEN(nd, nptr, ndlen);
1296 if (UCD_CATEGORY(nd) != ucp_M) break;
1297 ncount++;
1298 nptr += ndlen;
1299 }
1300 count++;
1301 ADD_NEW_DATA(-state_offset, count, ncount);
1302 }
1303 break;
1304#endif
1305
1306 /*-----------------------------------------------------------------*/
1307 case OP_ANYNL_EXTRA + OP_TYPEPLUS:
1308 case OP_ANYNL_EXTRA + OP_TYPEMINPLUS:
1309 case OP_ANYNL_EXTRA + OP_TYPEPOSPLUS:
1310 count = current_state->count; /* Already matched */
1311 if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1312 if (clen > 0)
1313 {
1314 int ncount = 0;
1315 switch (c)
1316 {
1317 case 0x000b:
1318 case 0x000c:
1319 case 0x0085:
1320 case 0x2028:
1321 case 0x2029:
1322 if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1323 goto ANYNL01;
1324
1325 case 0x000d:
1326 if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1327 /* Fall through */
1328
1329 ANYNL01:
1330 case 0x000a:
1331 if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS)
1332 {
1333 active_count--; /* Remove non-match possibility */
1334 next_active_state--;
1335 }
1336 count++;
1337 ADD_NEW_DATA(-state_offset, count, ncount);
1338 break;
1339
1340 default:
1341 break;
1342 }
1343 }
1344 break;
1345
1346 /*-----------------------------------------------------------------*/
1347 case OP_VSPACE_EXTRA + OP_TYPEPLUS:
1348 case OP_VSPACE_EXTRA + OP_TYPEMINPLUS:
1349 case OP_VSPACE_EXTRA + OP_TYPEPOSPLUS:
1350 count = current_state->count; /* Already matched */
1351 if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1352 if (clen > 0)
1353 {
1354 BOOL OK;
1355 switch (c)
1356 {
1357 case 0x000a:
1358 case 0x000b:
1359 case 0x000c:
1360 case 0x000d:
1361 case 0x0085:
1362 case 0x2028:
1363 case 0x2029:
1364 OK = TRUE;
1365 break;
1366
1367 default:
1368 OK = FALSE;
1369 break;
1370 }
1371
1372 if (OK == (d == OP_VSPACE))
1373 {
1374 if (count > 0 && codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSPLUS)
1375 {
1376 active_count--; /* Remove non-match possibility */
1377 next_active_state--;
1378 }
1379 count++;
1380 ADD_NEW_DATA(-state_offset, count, 0);
1381 }
1382 }
1383 break;
1384
1385 /*-----------------------------------------------------------------*/
1386 case OP_HSPACE_EXTRA + OP_TYPEPLUS:
1387 case OP_HSPACE_EXTRA + OP_TYPEMINPLUS:
1388 case OP_HSPACE_EXTRA + OP_TYPEPOSPLUS:
1389 count = current_state->count; /* Already matched */
1390 if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
1391 if (clen > 0)
1392 {
1393 BOOL OK;
1394 switch (c)
1395 {
1396 case 0x09: /* HT */
1397 case 0x20: /* SPACE */
1398 case 0xa0: /* NBSP */
1399 case 0x1680: /* OGHAM SPACE MARK */
1400 case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */
1401 case 0x2000: /* EN QUAD */
1402 case 0x2001: /* EM QUAD */
1403 case 0x2002: /* EN SPACE */
1404 case 0x2003: /* EM SPACE */
1405 case 0x2004: /* THREE-PER-EM SPACE */
1406 case 0x2005: /* FOUR-PER-EM SPACE */
1407 case 0x2006: /* SIX-PER-EM SPACE */
1408 case 0x2007: /* FIGURE SPACE */
1409 case 0x2008: /* PUNCTUATION SPACE */
1410 case 0x2009: /* THIN SPACE */
1411 case 0x200A: /* HAIR SPACE */
1412 case 0x202f: /* NARROW NO-BREAK SPACE */
1413 case 0x205f: /* MEDIUM MATHEMATICAL SPACE */
1414 case 0x3000: /* IDEOGRAPHIC SPACE */
1415 OK = TRUE;
1416 break;
1417
1418 default:
1419 OK = FALSE;
1420 break;
1421 }
1422
1423 if (OK == (d == OP_HSPACE))
1424 {
1425 if (count > 0 && codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSPLUS)
1426 {
1427 active_count--; /* Remove non-match possibility */
1428 next_active_state--;
1429 }
1430 count++;
1431 ADD_NEW_DATA(-state_offset, count, 0);
1432 }
1433 }
1434 break;
1435
1436 /*-----------------------------------------------------------------*/
1437#ifdef SUPPORT_UCP
1438 case OP_PROP_EXTRA + OP_TYPEQUERY:
1439 case OP_PROP_EXTRA + OP_TYPEMINQUERY:
1440 case OP_PROP_EXTRA + OP_TYPEPOSQUERY:
1441 count = 4;
1442 goto QS1;
1443
1444 case OP_PROP_EXTRA + OP_TYPESTAR:
1445 case OP_PROP_EXTRA + OP_TYPEMINSTAR:
1446 case OP_PROP_EXTRA + OP_TYPEPOSSTAR:
1447 count = 0;
1448
1449 QS1:
1450
1451 ADD_ACTIVE(state_offset + 4, 0);
1452 if (clen > 0)
1453 {
1454 BOOL OK;
1455 const ucd_record * prop = GET_UCD(c);
1456 switch(code[2])
1457 {
1458 case PT_ANY:
1459 OK = TRUE;
1460 break;
1461
1462 case PT_LAMP:
1463 OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1464 prop->chartype == ucp_Lt;
1465 break;
1466
1467 case PT_GC:
1468 OK = _pcre_ucp_gentype[prop->chartype] == code[3];
1469 break;
1470
1471 case PT_PC:
1472 OK = prop->chartype == code[3];
1473 break;
1474
1475 case PT_SC:
1476 OK = prop->script == code[3];
1477 break;
1478
1479 /* These are specials for combination cases. */
1480
1481 case PT_ALNUM:
1482 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1483 _pcre_ucp_gentype[prop->chartype] == ucp_N;
1484 break;
1485
1486 case PT_SPACE: /* Perl space */
1487 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1488 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1489 break;
1490
1491 case PT_PXSPACE: /* POSIX space */
1492 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1493 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1494 c == CHAR_FF || c == CHAR_CR;
1495 break;
1496
1497 case PT_WORD:
1498 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1499 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1500 c == CHAR_UNDERSCORE;
1501 break;
1502
1503 /* Should never occur, but keep compilers from grumbling. */
1504
1505 default:
1506 OK = codevalue != OP_PROP;
1507 break;
1508 }
1509
1510 if (OK == (d == OP_PROP))
1511 {
1512 if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSSTAR ||
1513 codevalue == OP_PROP_EXTRA + OP_TYPEPOSQUERY)
1514 {
1515 active_count--; /* Remove non-match possibility */
1516 next_active_state--;
1517 }
1518 ADD_NEW(state_offset + count, 0);
1519 }
1520 }
1521 break;
1522
1523 /*-----------------------------------------------------------------*/
1524 case OP_EXTUNI_EXTRA + OP_TYPEQUERY:
1525 case OP_EXTUNI_EXTRA + OP_TYPEMINQUERY:
1526 case OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY:
1527 count = 2;
1528 goto QS2;
1529
1530 case OP_EXTUNI_EXTRA + OP_TYPESTAR:
1531 case OP_EXTUNI_EXTRA + OP_TYPEMINSTAR:
1532 case OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR:
1533 count = 0;
1534
1535 QS2:
1536
1537 ADD_ACTIVE(state_offset + 2, 0);
1538 if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1539 {
1540 const uschar *nptr = ptr + clen;
1541 int ncount = 0;
1542 if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR ||
1543 codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY)
1544 {
1545 active_count--; /* Remove non-match possibility */
1546 next_active_state--;
1547 }
1548 while (nptr < end_subject)
1549 {
1550 int nd;
1551 int ndlen = 1;
1552 GETCHARLEN(nd, nptr, ndlen);
1553 if (UCD_CATEGORY(nd) != ucp_M) break;
1554 ncount++;
1555 nptr += ndlen;
1556 }
1557 ADD_NEW_DATA(-(state_offset + count), 0, ncount);
1558 }
1559 break;
1560#endif
1561
1562 /*-----------------------------------------------------------------*/
1563 case OP_ANYNL_EXTRA + OP_TYPEQUERY:
1564 case OP_ANYNL_EXTRA + OP_TYPEMINQUERY:
1565 case OP_ANYNL_EXTRA + OP_TYPEPOSQUERY:
1566 count = 2;
1567 goto QS3;
1568
1569 case OP_ANYNL_EXTRA + OP_TYPESTAR:
1570 case OP_ANYNL_EXTRA + OP_TYPEMINSTAR:
1571 case OP_ANYNL_EXTRA + OP_TYPEPOSSTAR:
1572 count = 0;
1573
1574 QS3:
1575 ADD_ACTIVE(state_offset + 2, 0);
1576 if (clen > 0)
1577 {
1578 int ncount = 0;
1579 switch (c)
1580 {
1581 case 0x000b:
1582 case 0x000c:
1583 case 0x0085:
1584 case 0x2028:
1585 case 0x2029:
1586 if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1587 goto ANYNL02;
1588
1589 case 0x000d:
1590 if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1591 /* Fall through */
1592
1593 ANYNL02:
1594 case 0x000a:
1595 if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR ||
1596 codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY)
1597 {
1598 active_count--; /* Remove non-match possibility */
1599 next_active_state--;
1600 }
1601 ADD_NEW_DATA(-(state_offset + count), 0, ncount);
1602 break;
1603
1604 default:
1605 break;
1606 }
1607 }
1608 break;
1609
1610 /*-----------------------------------------------------------------*/
1611 case OP_VSPACE_EXTRA + OP_TYPEQUERY:
1612 case OP_VSPACE_EXTRA + OP_TYPEMINQUERY:
1613 case OP_VSPACE_EXTRA + OP_TYPEPOSQUERY:
1614 count = 2;
1615 goto QS4;
1616
1617 case OP_VSPACE_EXTRA + OP_TYPESTAR:
1618 case OP_VSPACE_EXTRA + OP_TYPEMINSTAR:
1619 case OP_VSPACE_EXTRA + OP_TYPEPOSSTAR:
1620 count = 0;
1621
1622 QS4:
1623 ADD_ACTIVE(state_offset + 2, 0);
1624 if (clen > 0)
1625 {
1626 BOOL OK;
1627 switch (c)
1628 {
1629 case 0x000a:
1630 case 0x000b:
1631 case 0x000c:
1632 case 0x000d:
1633 case 0x0085:
1634 case 0x2028:
1635 case 0x2029:
1636 OK = TRUE;
1637 break;
1638
1639 default:
1640 OK = FALSE;
1641 break;
1642 }
1643 if (OK == (d == OP_VSPACE))
1644 {
1645 if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSSTAR ||
1646 codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSQUERY)
1647 {
1648 active_count--; /* Remove non-match possibility */
1649 next_active_state--;
1650 }
1651 ADD_NEW_DATA(-(state_offset + count), 0, 0);
1652 }
1653 }
1654 break;
1655
1656 /*-----------------------------------------------------------------*/
1657 case OP_HSPACE_EXTRA + OP_TYPEQUERY:
1658 case OP_HSPACE_EXTRA + OP_TYPEMINQUERY:
1659 case OP_HSPACE_EXTRA + OP_TYPEPOSQUERY:
1660 count = 2;
1661 goto QS5;
1662
1663 case OP_HSPACE_EXTRA + OP_TYPESTAR:
1664 case OP_HSPACE_EXTRA + OP_TYPEMINSTAR:
1665 case OP_HSPACE_EXTRA + OP_TYPEPOSSTAR:
1666 count = 0;
1667
1668 QS5:
1669 ADD_ACTIVE(state_offset + 2, 0);
1670 if (clen > 0)
1671 {
1672 BOOL OK;
1673 switch (c)
1674 {
1675 case 0x09: /* HT */
1676 case 0x20: /* SPACE */
1677 case 0xa0: /* NBSP */
1678 case 0x1680: /* OGHAM SPACE MARK */
1679 case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */
1680 case 0x2000: /* EN QUAD */
1681 case 0x2001: /* EM QUAD */
1682 case 0x2002: /* EN SPACE */
1683 case 0x2003: /* EM SPACE */
1684 case 0x2004: /* THREE-PER-EM SPACE */
1685 case 0x2005: /* FOUR-PER-EM SPACE */
1686 case 0x2006: /* SIX-PER-EM SPACE */
1687 case 0x2007: /* FIGURE SPACE */
1688 case 0x2008: /* PUNCTUATION SPACE */
1689 case 0x2009: /* THIN SPACE */
1690 case 0x200A: /* HAIR SPACE */
1691 case 0x202f: /* NARROW NO-BREAK SPACE */
1692 case 0x205f: /* MEDIUM MATHEMATICAL SPACE */
1693 case 0x3000: /* IDEOGRAPHIC SPACE */
1694 OK = TRUE;
1695 break;
1696
1697 default:
1698 OK = FALSE;
1699 break;
1700 }
1701
1702 if (OK == (d == OP_HSPACE))
1703 {
1704 if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSSTAR ||
1705 codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSQUERY)
1706 {
1707 active_count--; /* Remove non-match possibility */
1708 next_active_state--;
1709 }
1710 ADD_NEW_DATA(-(state_offset + count), 0, 0);
1711 }
1712 }
1713 break;
1714
1715 /*-----------------------------------------------------------------*/
1716#ifdef SUPPORT_UCP
1717 case OP_PROP_EXTRA + OP_TYPEEXACT:
1718 case OP_PROP_EXTRA + OP_TYPEUPTO:
1719 case OP_PROP_EXTRA + OP_TYPEMINUPTO:
1720 case OP_PROP_EXTRA + OP_TYPEPOSUPTO:
1721 if (codevalue != OP_PROP_EXTRA + OP_TYPEEXACT)
1722 { ADD_ACTIVE(state_offset + 6, 0); }
1723 count = current_state->count; /* Number already matched */
1724 if (clen > 0)
1725 {
1726 BOOL OK;
1727 const ucd_record * prop = GET_UCD(c);
1728 switch(code[4])
1729 {
1730 case PT_ANY:
1731 OK = TRUE;
1732 break;
1733
1734 case PT_LAMP:
1735 OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll ||
1736 prop->chartype == ucp_Lt;
1737 break;
1738
1739 case PT_GC:
1740 OK = _pcre_ucp_gentype[prop->chartype] == code[5];
1741 break;
1742
1743 case PT_PC:
1744 OK = prop->chartype == code[5];
1745 break;
1746
1747 case PT_SC:
1748 OK = prop->script == code[5];
1749 break;
1750
1751 /* These are specials for combination cases. */
1752
1753 case PT_ALNUM:
1754 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1755 _pcre_ucp_gentype[prop->chartype] == ucp_N;
1756 break;
1757
1758 case PT_SPACE: /* Perl space */
1759 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1760 c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR;
1761 break;
1762
1763 case PT_PXSPACE: /* POSIX space */
1764 OK = _pcre_ucp_gentype[prop->chartype] == ucp_Z ||
1765 c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
1766 c == CHAR_FF || c == CHAR_CR;
1767 break;
1768
1769 case PT_WORD:
1770 OK = _pcre_ucp_gentype[prop->chartype] == ucp_L ||
1771 _pcre_ucp_gentype[prop->chartype] == ucp_N ||
1772 c == CHAR_UNDERSCORE;
1773 break;
1774
1775 /* Should never occur, but keep compilers from grumbling. */
1776
1777 default:
1778 OK = codevalue != OP_PROP;
1779 break;
1780 }
1781
1782 if (OK == (d == OP_PROP))
1783 {
1784 if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSUPTO)
1785 {
1786 active_count--; /* Remove non-match possibility */
1787 next_active_state--;
1788 }
1789 if (++count >= GET2(code, 1))
1790 { ADD_NEW(state_offset + 6, 0); }
1791 else
1792 { ADD_NEW(state_offset, count); }
1793 }
1794 }
1795 break;
1796
1797 /*-----------------------------------------------------------------*/
1798 case OP_EXTUNI_EXTRA + OP_TYPEEXACT:
1799 case OP_EXTUNI_EXTRA + OP_TYPEUPTO:
1800 case OP_EXTUNI_EXTRA + OP_TYPEMINUPTO:
1801 case OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO:
1802 if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
1803 { ADD_ACTIVE(state_offset + 4, 0); }
1804 count = current_state->count; /* Number already matched */
1805 if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
1806 {
1807 const uschar *nptr = ptr + clen;
1808 int ncount = 0;
1809 if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO)
1810 {
1811 active_count--; /* Remove non-match possibility */
1812 next_active_state--;
1813 }
1814 while (nptr < end_subject)
1815 {
1816 int nd;
1817 int ndlen = 1;
1818 GETCHARLEN(nd, nptr, ndlen);
1819 if (UCD_CATEGORY(nd) != ucp_M) break;
1820 ncount++;
1821 nptr += ndlen;
1822 }
1823 if (++count >= GET2(code, 1))
1824 { ADD_NEW_DATA(-(state_offset + 4), 0, ncount); }
1825 else
1826 { ADD_NEW_DATA(-state_offset, count, ncount); }
1827 }
1828 break;
1829#endif
1830
1831 /*-----------------------------------------------------------------*/
1832 case OP_ANYNL_EXTRA + OP_TYPEEXACT:
1833 case OP_ANYNL_EXTRA + OP_TYPEUPTO:
1834 case OP_ANYNL_EXTRA + OP_TYPEMINUPTO:
1835 case OP_ANYNL_EXTRA + OP_TYPEPOSUPTO:
1836 if (codevalue != OP_ANYNL_EXTRA + OP_TYPEEXACT)
1837 { ADD_ACTIVE(state_offset + 4, 0); }
1838 count = current_state->count; /* Number already matched */
1839 if (clen > 0)
1840 {
1841 int ncount = 0;
1842 switch (c)
1843 {
1844 case 0x000b:
1845 case 0x000c:
1846 case 0x0085:
1847 case 0x2028:
1848 case 0x2029:
1849 if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
1850 goto ANYNL03;
1851
1852 case 0x000d:
1853 if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1;
1854 /* Fall through */
1855
1856 ANYNL03:
1857 case 0x000a:
1858 if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO)
1859 {
1860 active_count--; /* Remove non-match possibility */
1861 next_active_state--;
1862 }
1863 if (++count >= GET2(code, 1))
1864 { ADD_NEW_DATA(-(state_offset + 4), 0, ncount); }
1865 else
1866 { ADD_NEW_DATA(-state_offset, count, ncount); }
1867 break;
1868
1869 default:
1870 break;
1871 }
1872 }
1873 break;
1874
1875 /*-----------------------------------------------------------------*/
1876 case OP_VSPACE_EXTRA + OP_TYPEEXACT:
1877 case OP_VSPACE_EXTRA + OP_TYPEUPTO:
1878 case OP_VSPACE_EXTRA + OP_TYPEMINUPTO:
1879 case OP_VSPACE_EXTRA + OP_TYPEPOSUPTO:
1880 if (codevalue != OP_VSPACE_EXTRA + OP_TYPEEXACT)
1881 { ADD_ACTIVE(state_offset + 4, 0); }
1882 count = current_state->count; /* Number already matched */
1883 if (clen > 0)
1884 {
1885 BOOL OK;
1886 switch (c)
1887 {
1888 case 0x000a:
1889 case 0x000b:
1890 case 0x000c:
1891 case 0x000d:
1892 case 0x0085:
1893 case 0x2028:
1894 case 0x2029:
1895 OK = TRUE;
1896 break;
1897
1898 default:
1899 OK = FALSE;
1900 }
1901
1902 if (OK == (d == OP_VSPACE))
1903 {
1904 if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSUPTO)
1905 {
1906 active_count--; /* Remove non-match possibility */
1907 next_active_state--;
1908 }
1909 if (++count >= GET2(code, 1))
1910 { ADD_NEW_DATA(-(state_offset + 4), 0, 0); }
1911 else
1912 { ADD_NEW_DATA(-state_offset, count, 0); }
1913 }
1914 }
1915 break;
1916
1917 /*-----------------------------------------------------------------*/
1918 case OP_HSPACE_EXTRA + OP_TYPEEXACT:
1919 case OP_HSPACE_EXTRA + OP_TYPEUPTO:
1920 case OP_HSPACE_EXTRA + OP_TYPEMINUPTO:
1921 case OP_HSPACE_EXTRA + OP_TYPEPOSUPTO:
1922 if (codevalue != OP_HSPACE_EXTRA + OP_TYPEEXACT)
1923 { ADD_ACTIVE(state_offset + 4, 0); }
1924 count = current_state->count; /* Number already matched */
1925 if (clen > 0)
1926 {
1927 BOOL OK;
1928 switch (c)
1929 {
1930 case 0x09: /* HT */
1931 case 0x20: /* SPACE */
1932 case 0xa0: /* NBSP */
1933 case 0x1680: /* OGHAM SPACE MARK */
1934 case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */
1935 case 0x2000: /* EN QUAD */
1936 case 0x2001: /* EM QUAD */
1937 case 0x2002: /* EN SPACE */
1938 case 0x2003: /* EM SPACE */
1939 case 0x2004: /* THREE-PER-EM SPACE */
1940 case 0x2005: /* FOUR-PER-EM SPACE */
1941 case 0x2006: /* SIX-PER-EM SPACE */
1942 case 0x2007: /* FIGURE SPACE */
1943 case 0x2008: /* PUNCTUATION SPACE */
1944 case 0x2009: /* THIN SPACE */
1945 case 0x200A: /* HAIR SPACE */
1946 case 0x202f: /* NARROW NO-BREAK SPACE */
1947 case 0x205f: /* MEDIUM MATHEMATICAL SPACE */
1948 case 0x3000: /* IDEOGRAPHIC SPACE */
1949 OK = TRUE;
1950 break;
1951
1952 default:
1953 OK = FALSE;
1954 break;
1955 }
1956
1957 if (OK == (d == OP_HSPACE))
1958 {
1959 if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSUPTO)
1960 {
1961 active_count--; /* Remove non-match possibility */
1962 next_active_state--;
1963 }
1964 if (++count >= GET2(code, 1))
1965 { ADD_NEW_DATA(-(state_offset + 4), 0, 0); }
1966 else
1967 { ADD_NEW_DATA(-state_offset, count, 0); }
1968 }
1969 }
1970 break;
1971
1972/* ========================================================================== */
1973 /* These opcodes are followed by a character that is usually compared
1974 to the current subject character; it is loaded into d. We still get
1975 here even if there is no subject character, because in some cases zero
1976 repetitions are permitted. */
1977
1978 /*-----------------------------------------------------------------*/
1979 case OP_CHAR:
1980 if (clen > 0 && c == d) { ADD_NEW(state_offset + dlen + 1, 0); }
1981 break;
1982
1983 /*-----------------------------------------------------------------*/
1984 case OP_CHARI:
1985 if (clen == 0) break;
1986
1987#ifdef SUPPORT_UTF8
1988 if (utf8)
1989 {
1990 if (c == d) { ADD_NEW(state_offset + dlen + 1, 0); } else
1991 {
1992 unsigned int othercase;
1993 if (c < 128) othercase = fcc[c]; else
1994
1995 /* If we have Unicode property support, we can use it to test the
1996 other case of the character. */
1997
1998#ifdef SUPPORT_UCP
1999 othercase = UCD_OTHERCASE(c);
2000#else
2001 othercase = NOTACHAR;
2002#endif
2003
2004 if (d == othercase) { ADD_NEW(state_offset + dlen + 1, 0); }
2005 }
2006 }
2007 else
2008#endif /* SUPPORT_UTF8 */
2009
2010 /* Non-UTF-8 mode */
2011 {
2012 if (lcc[c] == lcc[d]) { ADD_NEW(state_offset + 2, 0); }
2013 }
2014 break;
2015
2016
2017#ifdef SUPPORT_UCP
2018 /*-----------------------------------------------------------------*/
2019 /* This is a tricky one because it can match more than one character.
2020 Find out how many characters to skip, and then set up a negative state
2021 to wait for them to pass before continuing. */
2022
2023 case OP_EXTUNI:
2024 if (clen > 0 && UCD_CATEGORY(c) != ucp_M)
2025 {
2026 const uschar *nptr = ptr + clen;
2027 int ncount = 0;
2028 while (nptr < end_subject)
2029 {
2030 int nclen = 1;
2031 GETCHARLEN(c, nptr, nclen);
2032 if (UCD_CATEGORY(c) != ucp_M) break;
2033 ncount++;
2034 nptr += nclen;
2035 }
2036 ADD_NEW_DATA(-(state_offset + 1), 0, ncount);
2037 }
2038 break;
2039#endif
2040
2041 /*-----------------------------------------------------------------*/
2042 /* This is a tricky like EXTUNI because it too can match more than one
2043 character (when CR is followed by LF). In this case, set up a negative
2044 state to wait for one character to pass before continuing. */
2045
2046 case OP_ANYNL:
2047 if (clen > 0) switch(c)
2048 {
2049 case 0x000b:
2050 case 0x000c:
2051 case 0x0085:
2052 case 0x2028:
2053 case 0x2029:
2054 if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break;
2055
2056 case 0x000a:
2057 ADD_NEW(state_offset + 1, 0);
2058 break;
2059
2060 case 0x000d:
2061 if (ptr + 1 < end_subject && ptr[1] == 0x0a)
2062 {
2063 ADD_NEW_DATA(-(state_offset + 1), 0, 1);
2064 }
2065 else
2066 {
2067 ADD_NEW(state_offset + 1, 0);
2068 }
2069 break;
2070 }
2071 break;
2072
2073 /*-----------------------------------------------------------------*/
2074 case OP_NOT_VSPACE:
2075 if (clen > 0) switch(c)
2076 {
2077 case 0x000a:
2078 case 0x000b:
2079 case 0x000c:
2080 case 0x000d:
2081 case 0x0085:
2082 case 0x2028:
2083 case 0x2029:
2084 break;
2085
2086 default:
2087 ADD_NEW(state_offset + 1, 0);
2088 break;
2089 }
2090 break;
2091
2092 /*-----------------------------------------------------------------*/
2093 case OP_VSPACE:
2094 if (clen > 0) switch(c)
2095 {
2096 case 0x000a:
2097 case 0x000b:
2098 case 0x000c:
2099 case 0x000d:
2100 case 0x0085:
2101 case 0x2028:
2102 case 0x2029:
2103 ADD_NEW(state_offset + 1, 0);
2104 break;
2105
2106 default: break;
2107 }
2108 break;
2109
2110 /*-----------------------------------------------------------------*/
2111 case OP_NOT_HSPACE:
2112 if (clen > 0) switch(c)
2113 {
2114 case 0x09: /* HT */
2115 case 0x20: /* SPACE */
2116 case 0xa0: /* NBSP */
2117 case 0x1680: /* OGHAM SPACE MARK */
2118 case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */
2119 case 0x2000: /* EN QUAD */
2120 case 0x2001: /* EM QUAD */
2121 case 0x2002: /* EN SPACE */
2122 case 0x2003: /* EM SPACE */
2123 case 0x2004: /* THREE-PER-EM SPACE */
2124 case 0x2005: /* FOUR-PER-EM SPACE */
2125 case 0x2006: /* SIX-PER-EM SPACE */
2126 case 0x2007: /* FIGURE SPACE */
2127 case 0x2008: /* PUNCTUATION SPACE */
2128 case 0x2009: /* THIN SPACE */
2129 case 0x200A: /* HAIR SPACE */
2130 case 0x202f: /* NARROW NO-BREAK SPACE */
2131 case 0x205f: /* MEDIUM MATHEMATICAL SPACE */
2132 case 0x3000: /* IDEOGRAPHIC SPACE */
2133 break;
2134
2135 default:
2136 ADD_NEW(state_offset + 1, 0);
2137 break;
2138 }
2139 break;
2140
2141 /*-----------------------------------------------------------------*/
2142 case OP_HSPACE:
2143 if (clen > 0) switch(c)
2144 {
2145 case 0x09: /* HT */
2146 case 0x20: /* SPACE */
2147 case 0xa0: /* NBSP */
2148 case 0x1680: /* OGHAM SPACE MARK */
2149 case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */
2150 case 0x2000: /* EN QUAD */
2151 case 0x2001: /* EM QUAD */
2152 case 0x2002: /* EN SPACE */
2153 case 0x2003: /* EM SPACE */
2154 case 0x2004: /* THREE-PER-EM SPACE */
2155 case 0x2005: /* FOUR-PER-EM SPACE */
2156 case 0x2006: /* SIX-PER-EM SPACE */
2157 case 0x2007: /* FIGURE SPACE */
2158 case 0x2008: /* PUNCTUATION SPACE */
2159 case 0x2009: /* THIN SPACE */
2160 case 0x200A: /* HAIR SPACE */
2161 case 0x202f: /* NARROW NO-BREAK SPACE */
2162 case 0x205f: /* MEDIUM MATHEMATICAL SPACE */
2163 case 0x3000: /* IDEOGRAPHIC SPACE */
2164 ADD_NEW(state_offset + 1, 0);
2165 break;
2166 }
2167 break;
2168
2169 /*-----------------------------------------------------------------*/
2170 /* Match a negated single character casefully. This is only used for
2171 one-byte characters, that is, we know that d < 256. The character we are
2172 checking (c) can be multibyte. */
2173
2174 case OP_NOT:
2175 if (clen > 0 && c != d) { ADD_NEW(state_offset + dlen + 1, 0); }
2176 break;
2177
2178 /*-----------------------------------------------------------------*/
2179 /* Match a negated single character caselessly. This is only used for
2180 one-byte characters, that is, we know that d < 256. The character we are
2181 checking (c) can be multibyte. */
2182
2183 case OP_NOTI:
2184 if (clen > 0 && c != d && c != fcc[d])
2185 { ADD_NEW(state_offset + dlen + 1, 0); }
2186 break;
2187
2188 /*-----------------------------------------------------------------*/
2189 case OP_PLUSI:
2190 case OP_MINPLUSI:
2191 case OP_POSPLUSI:
2192 case OP_NOTPLUSI:
2193 case OP_NOTMINPLUSI:
2194 case OP_NOTPOSPLUSI:
2195 caseless = TRUE;
2196 codevalue -= OP_STARI - OP_STAR;
2197
2198 /* Fall through */
2199 case OP_PLUS:
2200 case OP_MINPLUS:
2201 case OP_POSPLUS:
2202 case OP_NOTPLUS:
2203 case OP_NOTMINPLUS:
2204 case OP_NOTPOSPLUS:
2205 count = current_state->count; /* Already matched */
2206 if (count > 0) { ADD_ACTIVE(state_offset + dlen + 1, 0); }
2207 if (clen > 0)
2208 {
2209 unsigned int otherd = NOTACHAR;
2210 if (caseless)
2211 {
2212#ifdef SUPPORT_UTF8
2213 if (utf8 && d >= 128)
2214 {
2215#ifdef SUPPORT_UCP
2216 otherd = UCD_OTHERCASE(d);
2217#endif /* SUPPORT_UCP */
2218 }
2219 else
2220#endif /* SUPPORT_UTF8 */
2221 otherd = fcc[d];
2222 }
2223 if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
2224 {
2225 if (count > 0 &&
2226 (codevalue == OP_POSPLUS || codevalue == OP_NOTPOSPLUS))
2227 {
2228 active_count--; /* Remove non-match possibility */
2229 next_active_state--;
2230 }
2231 count++;
2232 ADD_NEW(state_offset, count);
2233 }
2234 }
2235 break;
2236
2237 /*-----------------------------------------------------------------*/
2238 case OP_QUERYI:
2239 case OP_MINQUERYI:
2240 case OP_POSQUERYI:
2241 case OP_NOTQUERYI:
2242 case OP_NOTMINQUERYI:
2243 case OP_NOTPOSQUERYI:
2244 caseless = TRUE;
2245 codevalue -= OP_STARI - OP_STAR;
2246 /* Fall through */
2247 case OP_QUERY:
2248 case OP_MINQUERY:
2249 case OP_POSQUERY:
2250 case OP_NOTQUERY:
2251 case OP_NOTMINQUERY:
2252 case OP_NOTPOSQUERY:
2253 ADD_ACTIVE(state_offset + dlen + 1, 0);
2254 if (clen > 0)
2255 {
2256 unsigned int otherd = NOTACHAR;
2257 if (caseless)
2258 {
2259#ifdef SUPPORT_UTF8
2260 if (utf8 && d >= 128)
2261 {
2262#ifdef SUPPORT_UCP
2263 otherd = UCD_OTHERCASE(d);
2264#endif /* SUPPORT_UCP */
2265 }
2266 else
2267#endif /* SUPPORT_UTF8 */
2268 otherd = fcc[d];
2269 }
2270 if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
2271 {
2272 if (codevalue == OP_POSQUERY || codevalue == OP_NOTPOSQUERY)
2273 {
2274 active_count--; /* Remove non-match possibility */
2275 next_active_state--;
2276 }
2277 ADD_NEW(state_offset + dlen + 1, 0);
2278 }
2279 }
2280 break;
2281
2282 /*-----------------------------------------------------------------*/
2283 case OP_STARI:
2284 case OP_MINSTARI:
2285 case OP_POSSTARI:
2286 case OP_NOTSTARI:
2287 case OP_NOTMINSTARI:
2288 case OP_NOTPOSSTARI:
2289 caseless = TRUE;
2290 codevalue -= OP_STARI - OP_STAR;
2291 /* Fall through */
2292 case OP_STAR:
2293 case OP_MINSTAR:
2294 case OP_POSSTAR:
2295 case OP_NOTSTAR:
2296 case OP_NOTMINSTAR:
2297 case OP_NOTPOSSTAR:
2298 ADD_ACTIVE(state_offset + dlen + 1, 0);
2299 if (clen > 0)
2300 {
2301 unsigned int otherd = NOTACHAR;
2302 if (caseless)
2303 {
2304#ifdef SUPPORT_UTF8
2305 if (utf8 && d >= 128)
2306 {
2307#ifdef SUPPORT_UCP
2308 otherd = UCD_OTHERCASE(d);
2309#endif /* SUPPORT_UCP */
2310 }
2311 else
2312#endif /* SUPPORT_UTF8 */
2313 otherd = fcc[d];
2314 }
2315 if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
2316 {
2317 if (codevalue == OP_POSSTAR || codevalue == OP_NOTPOSSTAR)
2318 {
2319 active_count--; /* Remove non-match possibility */
2320 next_active_state--;
2321 }
2322 ADD_NEW(state_offset, 0);
2323 }
2324 }
2325 break;
2326
2327 /*-----------------------------------------------------------------*/
2328 case OP_EXACTI:
2329 case OP_NOTEXACTI:
2330 caseless = TRUE;
2331 codevalue -= OP_STARI - OP_STAR;
2332 /* Fall through */
2333 case OP_EXACT:
2334 case OP_NOTEXACT:
2335 count = current_state->count; /* Number already matched */
2336 if (clen > 0)
2337 {
2338 unsigned int otherd = NOTACHAR;
2339 if (caseless)
2340 {
2341#ifdef SUPPORT_UTF8
2342 if (utf8 && d >= 128)
2343 {
2344#ifdef SUPPORT_UCP
2345 otherd = UCD_OTHERCASE(d);
2346#endif /* SUPPORT_UCP */
2347 }
2348 else
2349#endif /* SUPPORT_UTF8 */
2350 otherd = fcc[d];
2351 }
2352 if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
2353 {
2354 if (++count >= GET2(code, 1))
2355 { ADD_NEW(state_offset + dlen + 3, 0); }
2356 else
2357 { ADD_NEW(state_offset, count); }
2358 }
2359 }
2360 break;
2361
2362 /*-----------------------------------------------------------------*/
2363 case OP_UPTOI:
2364 case OP_MINUPTOI:
2365 case OP_POSUPTOI:
2366 case OP_NOTUPTOI:
2367 case OP_NOTMINUPTOI:
2368 case OP_NOTPOSUPTOI:
2369 caseless = TRUE;
2370 codevalue -= OP_STARI - OP_STAR;
2371 /* Fall through */
2372 case OP_UPTO:
2373 case OP_MINUPTO:
2374 case OP_POSUPTO:
2375 case OP_NOTUPTO:
2376 case OP_NOTMINUPTO:
2377 case OP_NOTPOSUPTO:
2378 ADD_ACTIVE(state_offset + dlen + 3, 0);
2379 count = current_state->count; /* Number already matched */
2380 if (clen > 0)
2381 {
2382 unsigned int otherd = NOTACHAR;
2383 if (caseless)
2384 {
2385#ifdef SUPPORT_UTF8
2386 if (utf8 && d >= 128)
2387 {
2388#ifdef SUPPORT_UCP
2389 otherd = UCD_OTHERCASE(d);
2390#endif /* SUPPORT_UCP */
2391 }
2392 else
2393#endif /* SUPPORT_UTF8 */
2394 otherd = fcc[d];
2395 }
2396 if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
2397 {
2398 if (codevalue == OP_POSUPTO || codevalue == OP_NOTPOSUPTO)
2399 {
2400 active_count--; /* Remove non-match possibility */
2401 next_active_state--;
2402 }
2403 if (++count >= GET2(code, 1))
2404 { ADD_NEW(state_offset + dlen + 3, 0); }
2405 else
2406 { ADD_NEW(state_offset, count); }
2407 }
2408 }
2409 break;
2410
2411
2412/* ========================================================================== */
2413 /* These are the class-handling opcodes */
2414
2415 case OP_CLASS:
2416 case OP_NCLASS:
2417 case OP_XCLASS:
2418 {
2419 BOOL isinclass = FALSE;
2420 int next_state_offset;
2421 const uschar *ecode;
2422
2423 /* For a simple class, there is always just a 32-byte table, and we
2424 can set isinclass from it. */
2425
2426 if (codevalue != OP_XCLASS)
2427 {
2428 ecode = code + 33;
2429 if (clen > 0)
2430 {
2431 isinclass = (c > 255)? (codevalue == OP_NCLASS) :
2432 ((code[1 + c/8] & (1 << (c&7))) != 0);
2433 }
2434 }
2435
2436 /* An extended class may have a table or a list of single characters,
2437 ranges, or both, and it may be positive or negative. There's a
2438 function that sorts all this out. */
2439
2440 else
2441 {
2442 ecode = code + GET(code, 1);
2443 if (clen > 0) isinclass = _pcre_xclass(c, code + 1 + LINK_SIZE);
2444 }
2445
2446 /* At this point, isinclass is set for all kinds of class, and ecode
2447 points to the byte after the end of the class. If there is a
2448 quantifier, this is where it will be. */
2449
2450 next_state_offset = (int)(ecode - start_code);
2451
2452 switch (*ecode)
2453 {
2454 case OP_CRSTAR:
2455 case OP_CRMINSTAR:
2456 ADD_ACTIVE(next_state_offset + 1, 0);
2457 if (isinclass) { ADD_NEW(state_offset, 0); }
2458 break;
2459
2460 case OP_CRPLUS:
2461 case OP_CRMINPLUS:
2462 count = current_state->count; /* Already matched */
2463 if (count > 0) { ADD_ACTIVE(next_state_offset + 1, 0); }
2464 if (isinclass) { count++; ADD_NEW(state_offset, count); }
2465 break;
2466
2467 case OP_CRQUERY:
2468 case OP_CRMINQUERY:
2469 ADD_ACTIVE(next_state_offset + 1, 0);
2470 if (isinclass) { ADD_NEW(next_state_offset + 1, 0); }
2471 break;
2472
2473 case OP_CRRANGE:
2474 case OP_CRMINRANGE:
2475 count = current_state->count; /* Already matched */
2476 if (count >= GET2(ecode, 1))
2477 { ADD_ACTIVE(next_state_offset + 5, 0); }
2478 if (isinclass)
2479 {
2480 int max = GET2(ecode, 3);
2481 if (++count >= max && max != 0) /* Max 0 => no limit */
2482 { ADD_NEW(next_state_offset + 5, 0); }
2483 else
2484 { ADD_NEW(state_offset, count); }
2485 }
2486 break;
2487
2488 default:
2489 if (isinclass) { ADD_NEW(next_state_offset, 0); }
2490 break;
2491 }
2492 }
2493 break;
2494
2495/* ========================================================================== */
2496 /* These are the opcodes for fancy brackets of various kinds. We have
2497 to use recursion in order to handle them. The "always failing" assertion
2498 (?!) is optimised to OP_FAIL when compiling, so we have to support that,
2499 though the other "backtracking verbs" are not supported. */
2500
2501 case OP_FAIL:
2502 forced_fail++; /* Count FAILs for multiple states */
2503 break;
2504
2505 case OP_ASSERT:
2506 case OP_ASSERT_NOT:
2507 case OP_ASSERTBACK:
2508 case OP_ASSERTBACK_NOT:
2509 {
2510 int rc;
2511 int local_offsets[2];
2512 int local_workspace[1000];
2513 const uschar *endasscode = code + GET(code, 1);
2514
2515 while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1);
2516
2517 rc = internal_dfa_exec(
2518 md, /* static match data */
2519 code, /* this subexpression's code */
2520 ptr, /* where we currently are */
2521 (int)(ptr - start_subject), /* start offset */
2522 local_offsets, /* offset vector */
2523 sizeof(local_offsets)/sizeof(int), /* size of same */
2524 local_workspace, /* workspace vector */
2525 sizeof(local_workspace)/sizeof(int), /* size of same */
2526 rlevel); /* function recursion level */
2527
2528 if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2529 if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
2530 { ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
2531 }
2532 break;
2533
2534 /*-----------------------------------------------------------------*/
2535 case OP_COND:
2536 case OP_SCOND:
2537 {
2538 int local_offsets[1000];
2539 int local_workspace[1000];
2540 int codelink = GET(code, 1);
2541 int condcode;
2542
2543 /* Because of the way auto-callout works during compile, a callout item
2544 is inserted between OP_COND and an assertion condition. This does not
2545 happen for the other conditions. */
2546
2547 if (code[LINK_SIZE+1] == OP_CALLOUT)
2548 {
2549 rrc = 0;
2550 if (pcre_callout != NULL)
2551 {
2552 pcre_callout_block cb;
2553 cb.version = 1; /* Version 1 of the callout block */
2554 cb.callout_number = code[LINK_SIZE+2];
2555 cb.offset_vector = offsets;
2556 cb.subject = (PCRE_SPTR)start_subject;
2557 cb.subject_length = (int)(end_subject - start_subject);
2558 cb.start_match = (int)(current_subject - start_subject);
2559 cb.current_position = (int)(ptr - start_subject);
2560 cb.pattern_position = GET(code, LINK_SIZE + 3);
2561 cb.next_item_length = GET(code, 3 + 2*LINK_SIZE);
2562 cb.capture_top = 1;
2563 cb.capture_last = -1;
2564 cb.callout_data = md->callout_data;
2565 cb.mark = NULL; /* No (*MARK) support */
2566 if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */
2567 }
2568 if (rrc > 0) break; /* Fail this thread */
2569 code += _pcre_OP_lengths[OP_CALLOUT]; /* Skip callout data */
2570 }
2571
2572 condcode = code[LINK_SIZE+1];
2573
2574 /* Back reference conditions are not supported */
2575
2576 if (condcode == OP_CREF || condcode == OP_NCREF)
2577 return PCRE_ERROR_DFA_UCOND;
2578
2579 /* The DEFINE condition is always false */
2580
2581 if (condcode == OP_DEF)
2582 { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2583
2584 /* The only supported version of OP_RREF is for the value RREF_ANY,
2585 which means "test if in any recursion". We can't test for specifically
2586 recursed groups. */
2587
2588 else if (condcode == OP_RREF || condcode == OP_NRREF)
2589 {
2590 int value = GET2(code, LINK_SIZE+2);
2591 if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND;
2592 if (md->recursive != NULL)
2593 { ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); }
2594 else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2595 }
2596
2597 /* Otherwise, the condition is an assertion */
2598
2599 else
2600 {
2601 int rc;
2602 const uschar *asscode = code + LINK_SIZE + 1;
2603 const uschar *endasscode = asscode + GET(asscode, 1);
2604
2605 while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1);
2606
2607 rc = internal_dfa_exec(
2608 md, /* fixed match data */
2609 asscode, /* this subexpression's code */
2610 ptr, /* where we currently are */
2611 (int)(ptr - start_subject), /* start offset */
2612 local_offsets, /* offset vector */
2613 sizeof(local_offsets)/sizeof(int), /* size of same */
2614 local_workspace, /* workspace vector */
2615 sizeof(local_workspace)/sizeof(int), /* size of same */
2616 rlevel); /* function recursion level */
2617
2618 if (rc == PCRE_ERROR_DFA_UITEM) return rc;
2619 if ((rc >= 0) ==
2620 (condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
2621 { ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
2622 else
2623 { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
2624 }
2625 }
2626 break;
2627
2628 /*-----------------------------------------------------------------*/
2629 case OP_RECURSE:
2630 {
2631 dfa_recursion_info *ri;
2632 int local_offsets[1000];
2633 int local_workspace[1000];
2634 const uschar *callpat = start_code + GET(code, 1);
2635 int recno = (callpat == md->start_code)? 0 :
2636 GET2(callpat, 1 + LINK_SIZE);
2637 int rc;
2638
2639 DPRINTF(("%.*sStarting regex recursion\n", rlevel*2-2, SP));
2640
2641 /* Check for repeating a recursion without advancing the subject
2642 pointer. This should catch convoluted mutual recursions. (Some simple
2643 cases are caught at compile time.) */
2644
2645 for (ri = md->recursive; ri != NULL; ri = ri->prevrec)
2646 if (recno == ri->group_num && ptr == ri->subject_position)
2647 return PCRE_ERROR_RECURSELOOP;
2648
2649 /* Remember this recursion and where we started it so as to
2650 catch infinite loops. */
2651
2652 new_recursive.group_num = recno;
2653 new_recursive.subject_position = ptr;
2654 new_recursive.prevrec = md->recursive;
2655 md->recursive = &new_recursive;
2656
2657 rc = internal_dfa_exec(
2658 md, /* fixed match data */
2659 callpat, /* this subexpression's code */
2660 ptr, /* where we currently are */
2661 (int)(ptr - start_subject), /* start offset */
2662 local_offsets, /* offset vector */
2663 sizeof(local_offsets)/sizeof(int), /* size of same */
2664 local_workspace, /* workspace vector */
2665 sizeof(local_workspace)/sizeof(int), /* size of same */
2666 rlevel); /* function recursion level */
2667
2668 md->recursive = new_recursive.prevrec; /* Done this recursion */
2669
2670 DPRINTF(("%.*sReturn from regex recursion: rc=%d\n", rlevel*2-2, SP,
2671 rc));
2672
2673 /* Ran out of internal offsets */
2674
2675 if (rc == 0) return PCRE_ERROR_DFA_RECURSE;
2676
2677 /* For each successful matched substring, set up the next state with a
2678 count of characters to skip before trying it. Note that the count is in
2679 characters, not bytes. */
2680
2681 if (rc > 0)
2682 {
2683 for (rc = rc*2 - 2; rc >= 0; rc -= 2)
2684 {
2685 const uschar *p = start_subject + local_offsets[rc];
2686 const uschar *pp = start_subject + local_offsets[rc+1];
2687 int charcount = local_offsets[rc+1] - local_offsets[rc];
2688 while (p < pp) if ((*p++ & 0xc0) == 0x80) charcount--;
2689 if (charcount > 0)
2690 {
2691 ADD_NEW_DATA(-(state_offset + LINK_SIZE + 1), 0, (charcount - 1));
2692 }
2693 else
2694 {
2695 ADD_ACTIVE(state_offset + LINK_SIZE + 1, 0);
2696 }
2697 }
2698 }
2699 else if (rc != PCRE_ERROR_NOMATCH) return rc;
2700 }
2701 break;
2702
2703 /*-----------------------------------------------------------------*/
2704 case OP_BRAPOS:
2705 case OP_SBRAPOS:
2706 case OP_CBRAPOS:
2707 case OP_SCBRAPOS:
2708 case OP_BRAPOSZERO:
2709 {
2710 int charcount, matched_count;
2711 const uschar *local_ptr = ptr;
2712 BOOL allow_zero;
2713
2714 if (codevalue == OP_BRAPOSZERO)
2715 {
2716 allow_zero = TRUE;
2717 codevalue = *(++code); /* Codevalue will be one of above BRAs */
2718 }
2719 else allow_zero = FALSE;
2720
2721 /* Loop to match the subpattern as many times as possible as if it were
2722 a complete pattern. */
2723
2724 for (matched_count = 0;; matched_count++)
2725 {
2726 int local_offsets[2];
2727 int local_workspace[1000];
2728
2729 int rc = internal_dfa_exec(
2730 md, /* fixed match data */
2731 code, /* this subexpression's code */
2732 local_ptr, /* where we currently are */
2733 (int)(ptr - start_subject), /* start offset */
2734 local_offsets, /* offset vector */
2735 sizeof(local_offsets)/sizeof(int), /* size of same */
2736 local_workspace, /* workspace vector */
2737 sizeof(local_workspace)/sizeof(int), /* size of same */
2738 rlevel); /* function recursion level */
2739
2740 /* Failed to match */
2741
2742 if (rc < 0)
2743 {
2744 if (rc != PCRE_ERROR_NOMATCH) return rc;
2745 break;
2746 }
2747
2748 /* Matched: break the loop if zero characters matched. */
2749
2750 charcount = local_offsets[1] - local_offsets[0];
2751 if (charcount == 0) break;
2752 local_ptr += charcount; /* Advance temporary position ptr */
2753 }
2754
2755 /* At this point we have matched the subpattern matched_count
2756 times, and local_ptr is pointing to the character after the end of the
2757 last match. */
2758
2759 if (matched_count > 0 || allow_zero)
2760 {
2761 const uschar *end_subpattern = code;
2762 int next_state_offset;
2763
2764 do { end_subpattern += GET(end_subpattern, 1); }
2765 while (*end_subpattern == OP_ALT);
2766 next_state_offset =
2767 (int)(end_subpattern - start_code + LINK_SIZE + 1);
2768
2769 /* Optimization: if there are no more active states, and there
2770 are no new states yet set up, then skip over the subject string
2771 right here, to save looping. Otherwise, set up the new state to swing
2772 into action when the end of the matched substring is reached. */
2773
2774 if (i + 1 >= active_count && new_count == 0)
2775 {
2776 ptr = local_ptr;
2777 clen = 0;
2778 ADD_NEW(next_state_offset, 0);
2779 }
2780 else
2781 {
2782 const uschar *p = ptr;
2783 const uschar *pp = local_ptr;
2784 charcount = (int)(pp - p);
2785 while (p < pp) if ((*p++ & 0xc0) == 0x80) charcount--;
2786 ADD_NEW_DATA(-next_state_offset, 0, (charcount - 1));
2787 }
2788 }
2789 }
2790 break;
2791
2792 /*-----------------------------------------------------------------*/
2793 case OP_ONCE:
2794 case OP_ONCE_NC:
2795 {
2796 int local_offsets[2];
2797 int local_workspace[1000];
2798
2799 int rc = internal_dfa_exec(
2800 md, /* fixed match data */
2801 code, /* this subexpression's code */
2802 ptr, /* where we currently are */
2803 (int)(ptr - start_subject), /* start offset */
2804 local_offsets, /* offset vector */
2805 sizeof(local_offsets)/sizeof(int), /* size of same */
2806 local_workspace, /* workspace vector */
2807 sizeof(local_workspace)/sizeof(int), /* size of same */
2808 rlevel); /* function recursion level */
2809
2810 if (rc >= 0)
2811 {
2812 const uschar *end_subpattern = code;
2813 int charcount = local_offsets[1] - local_offsets[0];
2814 int next_state_offset, repeat_state_offset;
2815
2816 do { end_subpattern += GET(end_subpattern, 1); }
2817 while (*end_subpattern == OP_ALT);
2818 next_state_offset =
2819 (int)(end_subpattern - start_code + LINK_SIZE + 1);
2820
2821 /* If the end of this subpattern is KETRMAX or KETRMIN, we must
2822 arrange for the repeat state also to be added to the relevant list.
2823 Calculate the offset, or set -1 for no repeat. */
2824
2825 repeat_state_offset = (*end_subpattern == OP_KETRMAX ||
2826 *end_subpattern == OP_KETRMIN)?
2827 (int)(end_subpattern - start_code - GET(end_subpattern, 1)) : -1;
2828
2829 /* If we have matched an empty string, add the next state at the
2830 current character pointer. This is important so that the duplicate
2831 checking kicks in, which is what breaks infinite loops that match an
2832 empty string. */
2833
2834 if (charcount == 0)
2835 {
2836 ADD_ACTIVE(next_state_offset, 0);
2837 }
2838
2839 /* Optimization: if there are no more active states, and there
2840 are no new states yet set up, then skip over the subject string
2841 right here, to save looping. Otherwise, set up the new state to swing
2842 into action when the end of the matched substring is reached. */
2843
2844 else if (i + 1 >= active_count && new_count == 0)
2845 {
2846 ptr += charcount;
2847 clen = 0;
2848 ADD_NEW(next_state_offset, 0);
2849
2850 /* If we are adding a repeat state at the new character position,
2851 we must fudge things so that it is the only current state.
2852 Otherwise, it might be a duplicate of one we processed before, and
2853 that would cause it to be skipped. */
2854
2855 if (repeat_state_offset >= 0)
2856 {
2857 next_active_state = active_states;
2858 active_count = 0;
2859 i = -1;
2860 ADD_ACTIVE(repeat_state_offset, 0);
2861 }
2862 }
2863 else
2864 {
2865 const uschar *p = start_subject + local_offsets[0];
2866 const uschar *pp = start_subject + local_offsets[1];
2867 while (p < pp) if ((*p++ & 0xc0) == 0x80) charcount--;
2868 ADD_NEW_DATA(-next_state_offset, 0, (charcount - 1));
2869 if (repeat_state_offset >= 0)
2870 { ADD_NEW_DATA(-repeat_state_offset, 0, (charcount - 1)); }
2871 }
2872 }
2873 else if (rc != PCRE_ERROR_NOMATCH) return rc;
2874 }
2875 break;
2876
2877
2878/* ========================================================================== */
2879 /* Handle callouts */
2880
2881 case OP_CALLOUT:
2882 rrc = 0;
2883 if (pcre_callout != NULL)
2884 {
2885 pcre_callout_block cb;
2886 cb.version = 1; /* Version 1 of the callout block */
2887 cb.callout_number = code[1];
2888 cb.offset_vector = offsets;
2889 cb.subject = (PCRE_SPTR)start_subject;
2890 cb.subject_length = (int)(end_subject - start_subject);
2891 cb.start_match = (int)(current_subject - start_subject);
2892 cb.current_position = (int)(ptr - start_subject);
2893 cb.pattern_position = GET(code, 2);
2894 cb.next_item_length = GET(code, 2 + LINK_SIZE);
2895 cb.capture_top = 1;
2896 cb.capture_last = -1;
2897 cb.callout_data = md->callout_data;
2898 cb.mark = NULL; /* No (*MARK) support */
2899 if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */
2900 }
2901 if (rrc == 0)
2902 { ADD_ACTIVE(state_offset + _pcre_OP_lengths[OP_CALLOUT], 0); }
2903 break;
2904
2905
2906/* ========================================================================== */
2907 default: /* Unsupported opcode */
2908 return PCRE_ERROR_DFA_UITEM;
2909 }
2910
2911 NEXT_ACTIVE_STATE: continue;
2912
2913 } /* End of loop scanning active states */
2914
2915 /* We have finished the processing at the current subject character. If no
2916 new states have been set for the next character, we have found all the
2917 matches that we are going to find. If we are at the top level and partial
2918 matching has been requested, check for appropriate conditions.
2919
2920 The "forced_ fail" variable counts the number of (*F) encountered for the
2921 character. If it is equal to the original active_count (saved in
2922 workspace[1]) it means that (*F) was found on every active state. In this
2923 case we don't want to give a partial match.
2924
2925 The "could_continue" variable is true if a state could have continued but
2926 for the fact that the end of the subject was reached. */
2927
2928 if (new_count <= 0)
2929 {
2930 if (rlevel == 1 && /* Top level, and */
2931 could_continue && /* Some could go on */
2932 forced_fail != workspace[1] && /* Not all forced fail & */
2933 ( /* either... */
2934 (md->moptions & PCRE_PARTIAL_HARD) != 0 /* Hard partial */
2935 || /* or... */
2936 ((md->moptions & PCRE_PARTIAL_SOFT) != 0 && /* Soft partial and */
2937 match_count < 0) /* no matches */
2938 ) && /* And... */
2939 ptr >= end_subject && /* Reached end of subject */
2940 ptr > md->start_used_ptr) /* Inspected non-empty string */
2941 {
2942 if (offsetcount >= 2)
2943 {
2944 offsets[0] = (int)(md->start_used_ptr - start_subject);
2945 offsets[1] = (int)(end_subject - start_subject);
2946 }
2947 match_count = PCRE_ERROR_PARTIAL;
2948 }
2949
2950 DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n"
2951 "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel, match_count,
2952 rlevel*2-2, SP));
2953 break; /* In effect, "return", but see the comment below */
2954 }
2955
2956 /* One or more states are active for the next character. */
2957
2958 ptr += clen; /* Advance to next subject character */
2959 } /* Loop to move along the subject string */
2960
2961/* Control gets here from "break" a few lines above. We do it this way because
2962if we use "return" above, we have compiler trouble. Some compilers warn if
2963there's nothing here because they think the function doesn't return a value. On
2964the other hand, if we put a dummy statement here, some more clever compilers
2965complain that it can't be reached. Sigh. */
2966
2967return match_count;
2968}
2969
2970
2971
2972
2973/*************************************************
2974* Execute a Regular Expression - DFA engine *
2975*************************************************/
2976
2977/* This external function applies a compiled re to a subject string using a DFA
2978engine. This function calls the internal function multiple times if the pattern
2979is not anchored.
2980
2981Arguments:
2982 argument_re points to the compiled expression
2983 extra_data points to extra data or is NULL
2984 subject points to the subject string
2985 length length of subject string (may contain binary zeros)
2986 start_offset where to start in the subject string
2987 options option bits
2988 offsets vector of match offsets
2989 offsetcount size of same
2990 workspace workspace vector
2991 wscount size of same
2992
2993Returns: > 0 => number of match offset pairs placed in offsets
2994 = 0 => offsets overflowed; longest matches are present
2995 -1 => failed to match
2996 < -1 => some kind of unexpected problem
2997*/
2998
2999PCRE_EXP_DEFN int PCRE_CALL_CONVENTION
3000pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data,
3001 const char *subject, int length, int start_offset, int options, int *offsets,
3002 int offsetcount, int *workspace, int wscount)
3003{
3004real_pcre *re = (real_pcre *)argument_re;
3005dfa_match_data match_block;
3006dfa_match_data *md = &match_block;
3007BOOL utf8, anchored, startline, firstline;
3008const uschar *current_subject, *end_subject, *lcc;
3009
3010pcre_study_data internal_study;
3011const pcre_study_data *study = NULL;
3012real_pcre internal_re;
3013
3014const uschar *req_byte_ptr;
3015const uschar *start_bits = NULL;
3016BOOL first_byte_caseless = FALSE;
3017BOOL req_byte_caseless = FALSE;
3018int first_byte = -1;
3019int req_byte = -1;
3020int req_byte2 = -1;
3021int newline;
3022
3023/* Plausibility checks */
3024
3025if ((options & ~PUBLIC_DFA_EXEC_OPTIONS) != 0) return PCRE_ERROR_BADOPTION;
3026if (re == NULL || subject == NULL || workspace == NULL ||
3027 (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
3028if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
3029if (wscount < 20) return PCRE_ERROR_DFA_WSSIZE;
3030if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
3031
3032/* We need to find the pointer to any study data before we test for byte
3033flipping, so we scan the extra_data block first. This may set two fields in the
3034match block, so we must initialize them beforehand. However, the other fields
3035in the match block must not be set until after the byte flipping. */
3036
3037md->tables = re->tables;
3038md->callout_data = NULL;
3039
3040if (extra_data != NULL)
3041 {
3042 unsigned int flags = extra_data->flags;
3043 if ((flags & PCRE_EXTRA_STUDY_DATA) != 0)
3044 study = (const pcre_study_data *)extra_data->study_data;
3045 if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0) return PCRE_ERROR_DFA_UMLIMIT;
3046 if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0)
3047 return PCRE_ERROR_DFA_UMLIMIT;
3048 if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0)
3049 md->callout_data = extra_data->callout_data;
3050 if ((flags & PCRE_EXTRA_TABLES) != 0)
3051 md->tables = extra_data->tables;
3052 }
3053
3054/* Check that the first field in the block is the magic number. If it is not,
3055test for a regex that was compiled on a host of opposite endianness. If this is
3056the case, flipped values are put in internal_re and internal_study if there was
3057study data too. */
3058
3059if (re->magic_number != MAGIC_NUMBER)
3060 {
3061 re = _pcre_try_flipped(re, &internal_re, study, &internal_study);
3062 if (re == NULL) return PCRE_ERROR_BADMAGIC;
3063 if (study != NULL) study = &internal_study;
3064 }
3065
3066/* Set some local values */
3067
3068current_subject = (const unsigned char *)subject + start_offset;
3069end_subject = (const unsigned char *)subject + length;
3070req_byte_ptr = current_subject - 1;
3071
3072#ifdef SUPPORT_UTF8
3073utf8 = (re->options & PCRE_UTF8) != 0;
3074#else
3075utf8 = FALSE;
3076#endif
3077
3078anchored = (options & (PCRE_ANCHORED|PCRE_DFA_RESTART)) != 0 ||
3079 (re->options & PCRE_ANCHORED) != 0;
3080
3081/* The remaining fixed data for passing around. */
3082
3083md->start_code = (const uschar *)argument_re +
3084 re->name_table_offset + re->name_count * re->name_entry_size;
3085md->start_subject = (const unsigned char *)subject;
3086md->end_subject = end_subject;
3087md->start_offset = start_offset;
3088md->moptions = options;
3089md->poptions = re->options;
3090
3091/* If the BSR option is not set at match time, copy what was set
3092at compile time. */
3093
3094if ((md->moptions & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == 0)
3095 {
3096 if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0)
3097 md->moptions |= re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE);
3098#ifdef BSR_ANYCRLF
3099 else md->moptions |= PCRE_BSR_ANYCRLF;
3100#endif
3101 }
3102
3103/* Handle different types of newline. The three bits give eight cases. If
3104nothing is set at run time, whatever was used at compile time applies. */
3105
3106switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : (pcre_uint32)options) &
3107 PCRE_NEWLINE_BITS)
3108 {
3109 case 0: newline = NEWLINE; break; /* Compile-time default */
3110 case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
3111 case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
3112 case PCRE_NEWLINE_CR+
3113 PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
3114 case PCRE_NEWLINE_ANY: newline = -1; break;
3115 case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
3116 default: return PCRE_ERROR_BADNEWLINE;
3117 }
3118
3119if (newline == -2)
3120 {
3121 md->nltype = NLTYPE_ANYCRLF;
3122 }
3123else if (newline < 0)
3124 {
3125 md->nltype = NLTYPE_ANY;
3126 }
3127else
3128 {
3129 md->nltype = NLTYPE_FIXED;
3130 if (newline > 255)
3131 {
3132 md->nllen = 2;
3133 md->nl[0] = (newline >> 8) & 255;
3134 md->nl[1] = newline & 255;
3135 }
3136 else
3137 {
3138 md->nllen = 1;
3139 md->nl[0] = newline;
3140 }
3141 }
3142
3143/* Check a UTF-8 string if required. Unfortunately there's no way of passing
3144back the character offset. */
3145
3146#ifdef SUPPORT_UTF8
3147if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
3148 {
3149 int erroroffset;
3150 int errorcode = _pcre_valid_utf8((uschar *)subject, length, &erroroffset);
3151 if (errorcode != 0)
3152 {
3153 if (offsetcount >= 2)
3154 {
3155 offsets[0] = erroroffset;
3156 offsets[1] = errorcode;
3157 }
3158 return (errorcode <= PCRE_UTF8_ERR5 && (options & PCRE_PARTIAL_HARD) != 0)?
3159 PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
3160 }
3161 if (start_offset > 0 && start_offset < length &&
3162 (((USPTR)subject)[start_offset] & 0xc0) == 0x80)
3163 return PCRE_ERROR_BADUTF8_OFFSET;
3164 }
3165#endif
3166
3167/* If the exec call supplied NULL for tables, use the inbuilt ones. This
3168is a feature that makes it possible to save compiled regex and re-use them
3169in other programs later. */
3170
3171if (md->tables == NULL) md->tables = _pcre_default_tables;
3172
3173/* The lower casing table and the "must be at the start of a line" flag are
3174used in a loop when finding where to start. */
3175
3176lcc = md->tables + lcc_offset;
3177startline = (re->flags & PCRE_STARTLINE) != 0;
3178firstline = (re->options & PCRE_FIRSTLINE) != 0;
3179
3180/* Set up the first character to match, if available. The first_byte value is
3181never set for an anchored regular expression, but the anchoring may be forced
3182at run time, so we have to test for anchoring. The first char may be unset for
3183an unanchored pattern, of course. If there's no first char and the pattern was
3184studied, there may be a bitmap of possible first characters. */
3185
3186if (!anchored)
3187 {
3188 if ((re->flags & PCRE_FIRSTSET) != 0)
3189 {
3190 first_byte = re->first_byte & 255;
3191 if ((first_byte_caseless = ((re->first_byte & REQ_CASELESS) != 0)) == TRUE)
3192 first_byte = lcc[first_byte];
3193 }
3194 else
3195 {
3196 if (!startline && study != NULL &&
3197 (study->flags & PCRE_STUDY_MAPPED) != 0)
3198 start_bits = study->start_bits;
3199 }
3200 }
3201
3202/* For anchored or unanchored matches, there may be a "last known required
3203character" set. */
3204
3205if ((re->flags & PCRE_REQCHSET) != 0)
3206 {
3207 req_byte = re->req_byte & 255;
3208 req_byte_caseless = (re->req_byte & REQ_CASELESS) != 0;
3209 req_byte2 = (md->tables + fcc_offset)[req_byte]; /* case flipped */
3210 }
3211
3212/* Call the main matching function, looping for a non-anchored regex after a
3213failed match. If not restarting, perform certain optimizations at the start of
3214a match. */
3215
3216for (;;)
3217 {
3218 int rc;
3219
3220 if ((options & PCRE_DFA_RESTART) == 0)
3221 {
3222 const uschar *save_end_subject = end_subject;
3223
3224 /* If firstline is TRUE, the start of the match is constrained to the first
3225 line of a multiline string. Implement this by temporarily adjusting
3226 end_subject so that we stop scanning at a newline. If the match fails at
3227 the newline, later code breaks this loop. */
3228
3229 if (firstline)
3230 {
3231 USPTR t = current_subject;
3232#ifdef SUPPORT_UTF8
3233 if (utf8)
3234 {
3235 while (t < md->end_subject && !IS_NEWLINE(t))
3236 {
3237 t++;
3238 while (t < end_subject && (*t & 0xc0) == 0x80) t++;
3239 }
3240 }
3241 else
3242#endif
3243 while (t < md->end_subject && !IS_NEWLINE(t)) t++;
3244 end_subject = t;
3245 }
3246
3247 /* There are some optimizations that avoid running the match if a known
3248 starting point is not found. However, there is an option that disables
3249 these, for testing and for ensuring that all callouts do actually occur.
3250 The option can be set in the regex by (*NO_START_OPT) or passed in
3251 match-time options. */
3252
3253 if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
3254 {
3255 /* Advance to a known first byte. */
3256
3257 if (first_byte >= 0)
3258 {
3259 if (first_byte_caseless)
3260 while (current_subject < end_subject &&
3261 lcc[*current_subject] != first_byte)
3262 current_subject++;
3263 else
3264 while (current_subject < end_subject &&
3265 *current_subject != first_byte)
3266 current_subject++;
3267 }
3268
3269 /* Or to just after a linebreak for a multiline match if possible */
3270
3271 else if (startline)
3272 {
3273 if (current_subject > md->start_subject + start_offset)
3274 {
3275#ifdef SUPPORT_UTF8
3276 if (utf8)
3277 {
3278 while (current_subject < end_subject &&
3279 !WAS_NEWLINE(current_subject))
3280 {
3281 current_subject++;
3282 while(current_subject < end_subject &&
3283 (*current_subject & 0xc0) == 0x80)
3284 current_subject++;
3285 }
3286 }
3287 else
3288#endif
3289 while (current_subject < end_subject && !WAS_NEWLINE(current_subject))
3290 current_subject++;
3291
3292 /* If we have just passed a CR and the newline option is ANY or
3293 ANYCRLF, and we are now at a LF, advance the match position by one
3294 more character. */
3295
3296 if (current_subject[-1] == CHAR_CR &&
3297 (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) &&
3298 current_subject < end_subject &&
3299 *current_subject == CHAR_NL)
3300 current_subject++;
3301 }
3302 }
3303
3304 /* Or to a non-unique first char after study */
3305
3306 else if (start_bits != NULL)
3307 {
3308 while (current_subject < end_subject)
3309 {
3310 register unsigned int c = *current_subject;
3311 if ((start_bits[c/8] & (1 << (c&7))) == 0)
3312 {
3313 current_subject++;
3314#ifdef SUPPORT_UTF8
3315 if (utf8)
3316 while(current_subject < end_subject &&
3317 (*current_subject & 0xc0) == 0x80) current_subject++;
3318#endif
3319 }
3320 else break;
3321 }
3322 }
3323 }
3324
3325 /* Restore fudged end_subject */
3326
3327 end_subject = save_end_subject;
3328
3329 /* The following two optimizations are disabled for partial matching or if
3330 disabling is explicitly requested (and of course, by the test above, this
3331 code is not obeyed when restarting after a partial match). */
3332
3333 if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 &&
3334 (options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0)
3335 {
3336 /* If the pattern was studied, a minimum subject length may be set. This
3337 is a lower bound; no actual string of that length may actually match the
3338 pattern. Although the value is, strictly, in characters, we treat it as
3339 bytes to avoid spending too much time in this optimization. */
3340
3341 if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 &&
3342 (pcre_uint32)(end_subject - current_subject) < study->minlength)
3343 return PCRE_ERROR_NOMATCH;
3344
3345 /* If req_byte is set, we know that that character must appear in the
3346 subject for the match to succeed. If the first character is set, req_byte
3347 must be later in the subject; otherwise the test starts at the match
3348 point. This optimization can save a huge amount of work in patterns with
3349 nested unlimited repeats that aren't going to match. Writing separate
3350 code for cased/caseless versions makes it go faster, as does using an
3351 autoincrement and backing off on a match.
3352
3353 HOWEVER: when the subject string is very, very long, searching to its end
3354 can take a long time, and give bad performance on quite ordinary
3355 patterns. This showed up when somebody was matching /^C/ on a 32-megabyte
3356 string... so we don't do this when the string is sufficiently long. */
3357
3358 if (req_byte >= 0 && end_subject - current_subject < REQ_BYTE_MAX)
3359 {
3360 register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0);
3361
3362 /* We don't need to repeat the search if we haven't yet reached the
3363 place we found it at last time. */
3364
3365 if (p > req_byte_ptr)
3366 {
3367 if (req_byte_caseless)
3368 {
3369 while (p < end_subject)
3370 {
3371 register int pp = *p++;
3372 if (pp == req_byte || pp == req_byte2) { p--; break; }
3373 }
3374 }
3375 else
3376 {
3377 while (p < end_subject)
3378 {
3379 if (*p++ == req_byte) { p--; break; }
3380 }
3381 }
3382
3383 /* If we can't find the required character, break the matching loop,
3384 which will cause a return or PCRE_ERROR_NOMATCH. */
3385
3386 if (p >= end_subject) break;
3387
3388 /* If we have found the required character, save the point where we
3389 found it, so that we don't search again next time round the loop if
3390 the start hasn't passed this character yet. */
3391
3392 req_byte_ptr = p;
3393 }
3394 }
3395 }
3396 } /* End of optimizations that are done when not restarting */
3397
3398 /* OK, now we can do the business */
3399
3400 md->start_used_ptr = current_subject;
3401 md->recursive = NULL;
3402
3403 rc = internal_dfa_exec(
3404 md, /* fixed match data */
3405 md->start_code, /* this subexpression's code */
3406 current_subject, /* where we currently are */
3407 start_offset, /* start offset in subject */
3408 offsets, /* offset vector */
3409 offsetcount, /* size of same */
3410 workspace, /* workspace vector */
3411 wscount, /* size of same */
3412 0); /* function recurse level */
3413
3414 /* Anything other than "no match" means we are done, always; otherwise, carry
3415 on only if not anchored. */
3416
3417 if (rc != PCRE_ERROR_NOMATCH || anchored) return rc;
3418
3419 /* Advance to the next subject character unless we are at the end of a line
3420 and firstline is set. */
3421
3422 if (firstline && IS_NEWLINE(current_subject)) break;
3423 current_subject++;
3424 if (utf8)
3425 {
3426 while (current_subject < end_subject && (*current_subject & 0xc0) == 0x80)
3427 current_subject++;
3428 }
3429 if (current_subject > end_subject) break;
3430
3431 /* If we have just passed a CR and we are now at a LF, and the pattern does
3432 not contain any explicit matches for \r or \n, and the newline option is CRLF
3433 or ANY or ANYCRLF, advance the match position by one more character. */
3434
3435 if (current_subject[-1] == CHAR_CR &&
3436 current_subject < end_subject &&
3437 *current_subject == CHAR_NL &&
3438 (re->flags & PCRE_HASCRORLF) == 0 &&
3439 (md->nltype == NLTYPE_ANY ||
3440 md->nltype == NLTYPE_ANYCRLF ||
3441 md->nllen == 2))
3442 current_subject++;
3443
3444 } /* "Bumpalong" loop */
3445
3446return PCRE_ERROR_NOMATCH;
3447}
3448
3449/* End of pcre_dfa_exec.c */