7 |
below for why this module is different). |
below for why this module is different). |
8 |
|
|
9 |
Written by Philip Hazel |
Written by Philip Hazel |
10 |
Copyright (c) 1997-2009 University of Cambridge |
Copyright (c) 1997-2011 University of Cambridge |
11 |
|
|
12 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
13 |
Redistribution and use in source and binary forms, with or without |
Redistribution and use in source and binary forms, with or without |
45 |
applications. */ |
applications. */ |
46 |
|
|
47 |
|
|
48 |
/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved |
/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved |
49 |
the performance of his patterns greatly. I could not use it as it stood, as it |
the performance of his patterns greatly. I could not use it as it stood, as it |
50 |
was not thread safe, and made assumptions about pattern sizes. Also, it caused |
was not thread safe, and made assumptions about pattern sizes. Also, it caused |
51 |
test 7 to loop, and test 9 to crash with a segfault. |
test 7 to loop, and test 9 to crash with a segfault. |
52 |
|
|
53 |
The issue is the check for duplicate states, which is done by a simple linear |
The issue is the check for duplicate states, which is done by a simple linear |
68 |
of internal_dfa_exec(). (The supplied patch used a static vector, initialized |
of internal_dfa_exec(). (The supplied patch used a static vector, initialized |
69 |
only once - I suspect this was the cause of the problems with the tests.) |
only once - I suspect this was the cause of the problems with the tests.) |
70 |
|
|
71 |
Overall, I concluded that the gains in some cases did not outweigh the losses |
Overall, I concluded that the gains in some cases did not outweigh the losses |
72 |
in others, so I abandoned this code. */ |
in others, so I abandoned this code. */ |
73 |
|
|
74 |
|
|
106 |
|
|
107 |
|
|
108 |
/* This table identifies those opcodes that are followed immediately by a |
/* This table identifies those opcodes that are followed immediately by a |
109 |
character that is to be tested in some way. This makes is possible to |
character that is to be tested in some way. This makes it possible to |
110 |
centralize the loading of these characters. In the case of Type * etc, the |
centralize 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 |
"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a |
112 |
small value. ***NOTE*** If the start of this table is modified, the two tables |
small value. Non-zero values in the table are the offsets from the opcode where |
113 |
that follow must also be modified. */ |
the character is to be found. ***NOTE*** If the start of this table is |
114 |
|
modified, the three tables that follow must also be modified. */ |
115 |
|
|
116 |
static const uschar coptable[] = { |
static const uschar coptable[] = { |
117 |
0, /* End */ |
0, /* End */ |
118 |
0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */ |
0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */ |
119 |
0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */ |
0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */ |
120 |
0, 0, 0, /* Any, AllAny, Anybyte */ |
0, 0, 0, /* Any, AllAny, Anybyte */ |
121 |
0, 0, 0, /* NOTPROP, PROP, EXTUNI */ |
0, 0, /* \P, \p */ |
122 |
0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */ |
0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */ |
123 |
0, 0, 0, 0, 0, /* \Z, \z, Opt, ^, $ */ |
0, /* \X */ |
124 |
|
0, 0, 0, 0, 0, 0, /* \Z, \z, ^, ^M, $, $M */ |
125 |
1, /* Char */ |
1, /* Char */ |
126 |
1, /* Charnc */ |
1, /* Chari */ |
127 |
1, /* not */ |
1, /* not */ |
128 |
|
1, /* noti */ |
129 |
/* Positive single-char repeats */ |
/* Positive single-char repeats */ |
130 |
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ |
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ |
131 |
3, 3, 3, /* upto, minupto, exact */ |
3, 3, 3, /* upto, minupto, exact */ |
132 |
1, 1, 1, 3, /* *+, ++, ?+, upto+ */ |
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 */ |
/* Negative single-char repeats - only for chars < 256 */ |
137 |
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */ |
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */ |
138 |
3, 3, 3, /* NOT upto, minupto, exact */ |
3, 3, 3, /* NOT upto, minupto, exact */ |
139 |
1, 1, 1, 3, /* NOT *+, ++, ?+, updo+ */ |
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 */ |
/* Positive type repeats */ |
144 |
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */ |
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */ |
145 |
3, 3, 3, /* Type upto, minupto, exact */ |
3, 3, 3, /* Type upto, minupto, exact */ |
151 |
0, /* NCLASS */ |
0, /* NCLASS */ |
152 |
0, /* XCLASS - variable length */ |
0, /* XCLASS - variable length */ |
153 |
0, /* REF */ |
0, /* REF */ |
154 |
|
0, /* REFI */ |
155 |
0, /* RECURSE */ |
0, /* RECURSE */ |
156 |
0, /* CALLOUT */ |
0, /* CALLOUT */ |
157 |
0, /* Alt */ |
0, /* Alt */ |
158 |
0, /* Ket */ |
0, /* Ket */ |
159 |
0, /* KetRmax */ |
0, /* KetRmax */ |
160 |
0, /* KetRmin */ |
0, /* KetRmin */ |
161 |
|
0, /* KetRpos */ |
162 |
|
0, /* Reverse */ |
163 |
0, /* Assert */ |
0, /* Assert */ |
164 |
0, /* Assert not */ |
0, /* Assert not */ |
165 |
0, /* Assert behind */ |
0, /* Assert behind */ |
166 |
0, /* Assert behind not */ |
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 |
181 |
|
remember the fact that a character could have been inspected when the end of |
182 |
|
the subject is reached. ***NOTE*** If the start of this table is modified, the |
183 |
|
two tables that follow must also be modified. */ |
184 |
|
|
185 |
|
static 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 */ |
0, /* Reverse */ |
232 |
0, 0, 0, 0, /* ONCE, BRA, CBRA, COND */ |
0, /* Assert */ |
233 |
0, 0, 0, /* SBRA, SCBRA, SCOND */ |
0, /* Assert not */ |
234 |
0, /* CREF */ |
0, /* Assert behind */ |
235 |
0, /* RREF */ |
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 */ |
0, /* DEF */ |
242 |
0, 0, /* BRAZERO, BRAMINZERO */ |
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ |
243 |
0, 0, 0, 0, /* PRUNE, SKIP, THEN, COMMIT */ |
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */ |
244 |
0, 0, 0 /* FAIL, ACCEPT, SKIPZERO */ |
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, |
/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W, |
274 |
typedef struct stateblock { |
typedef struct stateblock { |
275 |
int offset; /* Offset to opcode */ |
int offset; /* Offset to opcode */ |
276 |
int count; /* Count for repeats */ |
int count; /* Count for repeats */ |
|
int ims; /* ims flag bits */ |
|
277 |
int data; /* Some use extra data */ |
int data; /* Some use extra data */ |
278 |
} stateblock; |
} stateblock; |
279 |
|
|
280 |
#define INTS_PER_STATEBLOCK (sizeof(stateblock)/sizeof(int)) |
#define INTS_PER_STATEBLOCK (sizeof(stateblock)/sizeof(int)) |
281 |
|
|
282 |
|
|
283 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
284 |
/************************************************* |
/************************************************* |
285 |
* Print character string * |
* Print character string * |
286 |
*************************************************/ |
*************************************************/ |
329 |
offsetcount size of same |
offsetcount size of same |
330 |
workspace vector of workspace |
workspace vector of workspace |
331 |
wscount size of same |
wscount size of same |
|
ims the current ims flags |
|
332 |
rlevel function call recursion level |
rlevel function call recursion level |
|
recursing regex recursive call level |
|
333 |
|
|
334 |
Returns: > 0 => number of match offset pairs placed in offsets |
Returns: > 0 => number of match offset pairs placed in offsets |
335 |
= 0 => offsets overflowed; longest matches are present |
= 0 => offsets overflowed; longest matches are present |
344 |
{ \ |
{ \ |
345 |
next_active_state->offset = (x); \ |
next_active_state->offset = (x); \ |
346 |
next_active_state->count = (y); \ |
next_active_state->count = (y); \ |
|
next_active_state->ims = ims; \ |
|
347 |
next_active_state++; \ |
next_active_state++; \ |
348 |
DPRINTF(("%.*sADD_ACTIVE(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ |
DPRINTF(("%.*sADD_ACTIVE(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ |
349 |
} \ |
} \ |
354 |
{ \ |
{ \ |
355 |
next_active_state->offset = (x); \ |
next_active_state->offset = (x); \ |
356 |
next_active_state->count = (y); \ |
next_active_state->count = (y); \ |
|
next_active_state->ims = ims; \ |
|
357 |
next_active_state->data = (z); \ |
next_active_state->data = (z); \ |
358 |
next_active_state++; \ |
next_active_state++; \ |
359 |
DPRINTF(("%.*sADD_ACTIVE_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ |
DPRINTF(("%.*sADD_ACTIVE_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ |
365 |
{ \ |
{ \ |
366 |
next_new_state->offset = (x); \ |
next_new_state->offset = (x); \ |
367 |
next_new_state->count = (y); \ |
next_new_state->count = (y); \ |
|
next_new_state->ims = ims; \ |
|
368 |
next_new_state++; \ |
next_new_state++; \ |
369 |
DPRINTF(("%.*sADD_NEW(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ |
DPRINTF(("%.*sADD_NEW(%d,%d)\n", rlevel*2-2, SP, (x), (y))); \ |
370 |
} \ |
} \ |
375 |
{ \ |
{ \ |
376 |
next_new_state->offset = (x); \ |
next_new_state->offset = (x); \ |
377 |
next_new_state->count = (y); \ |
next_new_state->count = (y); \ |
|
next_new_state->ims = ims; \ |
|
378 |
next_new_state->data = (z); \ |
next_new_state->data = (z); \ |
379 |
next_new_state++; \ |
next_new_state++; \ |
380 |
DPRINTF(("%.*sADD_NEW_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ |
DPRINTF(("%.*sADD_NEW_DATA(%d,%d,%d)\n", rlevel*2-2, SP, (x), (y), (z))); \ |
393 |
int offsetcount, |
int offsetcount, |
394 |
int *workspace, |
int *workspace, |
395 |
int wscount, |
int wscount, |
396 |
int ims, |
int rlevel) |
|
int rlevel, |
|
|
int recursing) |
|
397 |
{ |
{ |
398 |
stateblock *active_states, *new_states, *temp_states; |
stateblock *active_states, *new_states, *temp_states; |
399 |
stateblock *next_active_state, *next_new_state; |
stateblock *next_active_state, *next_new_state; |
402 |
const uschar *ptr; |
const uschar *ptr; |
403 |
const uschar *end_code, *first_op; |
const uschar *end_code, *first_op; |
404 |
|
|
405 |
|
dfa_recursion_info new_recursive; |
406 |
|
|
407 |
int active_count, new_count, match_count; |
int active_count, new_count, match_count; |
408 |
|
|
409 |
/* Some fields in the md block are frequently referenced, so we load them into |
/* Some fields in the md block are frequently referenced, so we load them into |
427 |
(2 * INTS_PER_STATEBLOCK); |
(2 * INTS_PER_STATEBLOCK); |
428 |
|
|
429 |
DPRINTF(("\n%.*s---------------------\n" |
DPRINTF(("\n%.*s---------------------\n" |
430 |
"%.*sCall to internal_dfa_exec f=%d r=%d\n", |
"%.*sCall to internal_dfa_exec f=%d\n", |
431 |
rlevel*2-2, SP, rlevel*2-2, SP, rlevel, recursing)); |
rlevel*2-2, SP, rlevel*2-2, SP, rlevel)); |
432 |
|
|
433 |
ctypes = md->tables + ctypes_offset; |
ctypes = md->tables + ctypes_offset; |
434 |
lcc = md->tables + lcc_offset; |
lcc = md->tables + lcc_offset; |
441 |
new_count = 0; |
new_count = 0; |
442 |
|
|
443 |
first_op = this_start_code + 1 + LINK_SIZE + |
first_op = this_start_code + 1 + LINK_SIZE + |
444 |
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA)? 2:0); |
((*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 |
/* The first thing in any (sub) pattern is a bracket of some sort. Push all |
448 |
the alternative states onto the list, and find out where the end is. This |
the alternative states onto the list, and find out where the end is. This |
491 |
|
|
492 |
{ |
{ |
493 |
gone_back = (current_subject - max_back < start_subject)? |
gone_back = (current_subject - max_back < start_subject)? |
494 |
current_subject - start_subject : max_back; |
(int)(current_subject - start_subject) : max_back; |
495 |
current_subject -= gone_back; |
current_subject -= gone_back; |
496 |
} |
} |
497 |
|
|
498 |
/* Save the earliest consulted character */ |
/* Save the earliest consulted character */ |
499 |
|
|
500 |
if (current_subject < md->start_used_ptr) |
if (current_subject < md->start_used_ptr) |
501 |
md->start_used_ptr = current_subject; |
md->start_used_ptr = current_subject; |
502 |
|
|
503 |
/* Now we can process the individual branches. */ |
/* Now we can process the individual branches. */ |
504 |
|
|
508 |
int back = GET(end_code, 2+LINK_SIZE); |
int back = GET(end_code, 2+LINK_SIZE); |
509 |
if (back <= gone_back) |
if (back <= gone_back) |
510 |
{ |
{ |
511 |
int bstate = end_code - start_code + 2 + 2*LINK_SIZE; |
int bstate = (int)(end_code - start_code + 2 + 2*LINK_SIZE); |
512 |
ADD_NEW_DATA(-bstate, 0, gone_back - back); |
ADD_NEW_DATA(-bstate, 0, gone_back - back); |
513 |
} |
} |
514 |
end_code += GET(end_code, 1); |
end_code += GET(end_code, 1); |
541 |
else |
else |
542 |
{ |
{ |
543 |
int length = 1 + LINK_SIZE + |
int length = 1 + LINK_SIZE + |
544 |
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA)? 2:0); |
((*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 |
do |
548 |
{ |
{ |
549 |
ADD_NEW(end_code - start_code + length, 0); |
ADD_NEW((int)(end_code - start_code + length), 0); |
550 |
end_code += GET(end_code, 1); |
end_code += GET(end_code, 1); |
551 |
length = 1 + LINK_SIZE; |
length = 1 + LINK_SIZE; |
552 |
} |
} |
567 |
int clen, dlen; |
int clen, dlen; |
568 |
unsigned int c, d; |
unsigned int c, d; |
569 |
int forced_fail = 0; |
int forced_fail = 0; |
570 |
int reached_end = 0; |
BOOL could_continue = FALSE; |
571 |
|
|
572 |
/* Make the new state list into the active state list and empty the |
/* Make the new state list into the active state list and empty the |
573 |
new state list. */ |
new state list. */ |
581 |
workspace[0] ^= 1; /* Remember for the restarting feature */ |
workspace[0] ^= 1; /* Remember for the restarting feature */ |
582 |
workspace[1] = active_count; |
workspace[1] = active_count; |
583 |
|
|
584 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
585 |
printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP); |
printf("%.*sNext character: rest of subject = \"", rlevel*2-2, SP); |
586 |
pchars((uschar *)ptr, strlen((char *)ptr), stdout); |
pchars((uschar *)ptr, strlen((char *)ptr), stdout); |
587 |
printf("\"\n"); |
printf("\"\n"); |
623 |
for (i = 0; i < active_count; i++) |
for (i = 0; i < active_count; i++) |
624 |
{ |
{ |
625 |
stateblock *current_state = active_states + i; |
stateblock *current_state = active_states + i; |
626 |
|
BOOL caseless = FALSE; |
627 |
const uschar *code; |
const uschar *code; |
628 |
int state_offset = current_state->offset; |
int state_offset = current_state->offset; |
629 |
int count, codevalue, rrc; |
int count, codevalue, rrc; |
630 |
|
|
631 |
#ifdef DEBUG |
#ifdef PCRE_DEBUG |
632 |
printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset); |
printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset); |
633 |
if (clen == 0) printf("EOL\n"); |
if (clen == 0) printf("EOL\n"); |
634 |
else if (c > 32 && c < 127) printf("'%c'\n", c); |
else if (c > 32 && c < 127) printf("'%c'\n", c); |
635 |
else printf("0x%02x\n", c); |
else printf("0x%02x\n", c); |
636 |
#endif |
#endif |
637 |
|
|
|
/* This variable is referred to implicity in the ADD_xxx macros. */ |
|
|
|
|
|
ims = current_state->ims; |
|
|
|
|
638 |
/* A negative offset is a special case meaning "hold off going to this |
/* 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 |
(negated) state until the number of characters in the data field have |
640 |
been skipped". */ |
been skipped". */ |
654 |
} |
} |
655 |
} |
} |
656 |
|
|
657 |
/* Check for a duplicate state with the same count, and skip if found. |
/* 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 |
See the note at the head of this module about the possibility of improving |
659 |
performance here. */ |
performance here. */ |
660 |
|
|
673 |
code = start_code + state_offset; |
code = start_code + state_offset; |
674 |
codevalue = *code; |
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 |
/* 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 |
tempting to test for the presence of a subject character here, but that |
684 |
is wrong, because sometimes zero repetitions of the subject are |
is wrong, because sometimes zero repetitions of the subject are |
725 |
|
|
726 |
switch (codevalue) |
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 |
/* Reached a closing bracket. If not at the end of the pattern, carry |
743 |
on with the next opcode. Otherwise, unless we have an empty string and |
on with the next opcode. For repeating opcodes, also add the repeat |
744 |
PCRE_NOTEMPTY is set, or PCRE_NOTEMPTY_ATSTART is set and we are at the |
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 |
start of the subject, save the match data, shifting up all previous |
751 |
matches so we always have the longest first. */ |
matches so we always have the longest first. */ |
752 |
|
|
753 |
case OP_KET: |
case OP_KET: |
754 |
case OP_KETRMIN: |
case OP_KETRMIN: |
755 |
case OP_KETRMAX: |
case OP_KETRMAX: |
756 |
|
case OP_KETRPOS: |
757 |
if (code != end_code) |
if (code != end_code) |
758 |
{ |
{ |
759 |
ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0); |
ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0); |
762 |
ADD_ACTIVE(state_offset - GET(code, 1), 0); |
ADD_ACTIVE(state_offset - GET(code, 1), 0); |
763 |
} |
} |
764 |
} |
} |
765 |
else |
else |
766 |
{ |
{ |
767 |
reached_end++; /* Count branches that reach the end */ |
if (ptr > current_subject || |
|
if (ptr > current_subject || |
|
768 |
((md->moptions & PCRE_NOTEMPTY) == 0 && |
((md->moptions & PCRE_NOTEMPTY) == 0 && |
769 |
((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 || |
((md->moptions & PCRE_NOTEMPTY_ATSTART) == 0 || |
770 |
current_subject > start_subject + md->start_offset))) |
current_subject > start_subject + md->start_offset))) |
771 |
{ |
{ |
772 |
if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0; |
if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0; |
773 |
else if (match_count > 0 && ++match_count * 2 >= offsetcount) |
else if (match_count > 0 && ++match_count * 2 > offsetcount) |
774 |
match_count = 0; |
match_count = 0; |
775 |
count = ((match_count == 0)? offsetcount : match_count * 2) - 2; |
count = ((match_count == 0)? offsetcount : match_count * 2) - 2; |
776 |
if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int)); |
if (count > 0) memmove(offsets + 2, offsets, count * sizeof(int)); |
777 |
if (offsetcount >= 2) |
if (offsetcount >= 2) |
778 |
{ |
{ |
779 |
offsets[0] = current_subject - start_subject; |
offsets[0] = (int)(current_subject - start_subject); |
780 |
offsets[1] = ptr - start_subject; |
offsets[1] = (int)(ptr - start_subject); |
781 |
DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP, |
DPRINTF(("%.*sSet matched string = \"%.*s\"\n", rlevel*2-2, SP, |
782 |
offsets[1] - offsets[0], current_subject)); |
offsets[1] - offsets[0], current_subject)); |
783 |
} |
} |
788 |
match_count, rlevel*2-2, SP)); |
match_count, rlevel*2-2, SP)); |
789 |
return match_count; |
return match_count; |
790 |
} |
} |
791 |
} |
} |
792 |
} |
} |
793 |
break; |
break; |
794 |
|
|
799 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
800 |
case OP_ALT: |
case OP_ALT: |
801 |
do { code += GET(code, 1); } while (*code == OP_ALT); |
do { code += GET(code, 1); } while (*code == OP_ALT); |
802 |
ADD_ACTIVE(code - start_code, 0); |
ADD_ACTIVE((int)(code - start_code), 0); |
803 |
break; |
break; |
804 |
|
|
805 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
807 |
case OP_SBRA: |
case OP_SBRA: |
808 |
do |
do |
809 |
{ |
{ |
810 |
ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); |
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0); |
811 |
code += GET(code, 1); |
code += GET(code, 1); |
812 |
} |
} |
813 |
while (*code == OP_ALT); |
while (*code == OP_ALT); |
816 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
817 |
case OP_CBRA: |
case OP_CBRA: |
818 |
case OP_SCBRA: |
case OP_SCBRA: |
819 |
ADD_ACTIVE(code - start_code + 3 + LINK_SIZE, 0); |
ADD_ACTIVE((int)(code - start_code + 3 + LINK_SIZE), 0); |
820 |
code += GET(code, 1); |
code += GET(code, 1); |
821 |
while (*code == OP_ALT) |
while (*code == OP_ALT) |
822 |
{ |
{ |
823 |
ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); |
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0); |
824 |
code += GET(code, 1); |
code += GET(code, 1); |
825 |
} |
} |
826 |
break; |
break; |
831 |
ADD_ACTIVE(state_offset + 1, 0); |
ADD_ACTIVE(state_offset + 1, 0); |
832 |
code += 1 + GET(code, 2); |
code += 1 + GET(code, 2); |
833 |
while (*code == OP_ALT) code += GET(code, 1); |
while (*code == OP_ALT) code += GET(code, 1); |
834 |
ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); |
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0); |
835 |
break; |
break; |
836 |
|
|
837 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
838 |
case OP_SKIPZERO: |
case OP_SKIPZERO: |
839 |
code += 1 + GET(code, 2); |
code += 1 + GET(code, 2); |
840 |
while (*code == OP_ALT) code += GET(code, 1); |
while (*code == OP_ALT) code += GET(code, 1); |
841 |
ADD_ACTIVE(code - start_code + 1 + LINK_SIZE, 0); |
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0); |
842 |
break; |
break; |
843 |
|
|
844 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
845 |
case OP_CIRC: |
case OP_CIRC: |
846 |
if ((ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) || |
if (ptr == start_subject && (md->moptions & PCRE_NOTBOL) == 0) |
|
((ims & PCRE_MULTILINE) != 0 && |
|
|
ptr != end_subject && |
|
|
WAS_NEWLINE(ptr))) |
|
847 |
{ ADD_ACTIVE(state_offset + 1, 0); } |
{ ADD_ACTIVE(state_offset + 1, 0); } |
848 |
break; |
break; |
849 |
|
|
850 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
851 |
case OP_EOD: |
case OP_CIRCM: |
852 |
if (ptr >= end_subject) { ADD_ACTIVE(state_offset + 1, 0); } |
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; |
break; |
856 |
|
|
857 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
858 |
case OP_OPT: |
case OP_EOD: |
859 |
ims = code[1]; |
if (ptr >= end_subject) |
860 |
ADD_ACTIVE(state_offset + 2, 0); |
{ |
861 |
|
if ((md->moptions & PCRE_PARTIAL_HARD) != 0) |
862 |
|
could_continue = TRUE; |
863 |
|
else { ADD_ACTIVE(state_offset + 1, 0); } |
864 |
|
} |
865 |
break; |
break; |
866 |
|
|
867 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
895 |
|
|
896 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
897 |
case OP_EODN: |
case OP_EODN: |
898 |
if (clen == 0 || (IS_NEWLINE(ptr) && ptr == end_subject - md->nllen)) |
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); } |
{ ADD_ACTIVE(state_offset + 1, 0); } |
902 |
break; |
break; |
903 |
|
|
905 |
case OP_DOLL: |
case OP_DOLL: |
906 |
if ((md->moptions & PCRE_NOTEOL) == 0) |
if ((md->moptions & PCRE_NOTEOL) == 0) |
907 |
{ |
{ |
908 |
if (clen == 0 || |
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) && |
((md->poptions & PCRE_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) && |
912 |
((ims & PCRE_MULTILINE) != 0 || ptr == end_subject - md->nllen) |
(ptr == end_subject - md->nllen) |
913 |
)) |
)) |
914 |
{ ADD_ACTIVE(state_offset + 1, 0); } |
{ ADD_ACTIVE(state_offset + 1, 0); } |
915 |
} |
} |
916 |
else if ((ims & PCRE_MULTILINE) != 0 && IS_NEWLINE(ptr)) |
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); } |
{ ADD_ACTIVE(state_offset + 1, 0); } |
930 |
break; |
break; |
931 |
|
|
957 |
if (ptr > start_subject) |
if (ptr > start_subject) |
958 |
{ |
{ |
959 |
const uschar *temp = ptr - 1; |
const uschar *temp = ptr - 1; |
960 |
if (temp < md->start_used_ptr) md->start_used_ptr = temp; |
if (temp < md->start_used_ptr) md->start_used_ptr = temp; |
961 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
962 |
if (utf8) BACKCHAR(temp); |
if (utf8) BACKCHAR(temp); |
963 |
#endif |
#endif |
964 |
GETCHARTEST(d, temp); |
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; |
left_word = d < 256 && (ctypes[d] & ctype_word) != 0; |
977 |
} |
} |
978 |
else left_word = 0; |
else left_word = FALSE; |
979 |
|
|
980 |
if (clen > 0) |
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; |
right_word = c < 256 && (ctypes[c] & ctype_word) != 0; |
994 |
else /* This is a fudge to ensure that if this is the */ |
} |
995 |
{ /* last item in the pattern, we don't count it as */ |
else right_word = FALSE; |
|
reached_end--; /* reached, thus disabling a partial match. */ |
|
|
right_word = 0; |
|
|
} |
|
996 |
|
|
997 |
if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY)) |
if ((left_word == right_word) == (codevalue == OP_NOT_WORD_BOUNDARY)) |
998 |
{ ADD_ACTIVE(state_offset + 1, 0); } |
{ ADD_ACTIVE(state_offset + 1, 0); } |
1019 |
break; |
break; |
1020 |
|
|
1021 |
case PT_LAMP: |
case PT_LAMP: |
1022 |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt; |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || |
1023 |
|
prop->chartype == ucp_Lt; |
1024 |
break; |
break; |
1025 |
|
|
1026 |
case PT_GC: |
case PT_GC: |
1035 |
OK = prop->script == code[2]; |
OK = prop->script == code[2]; |
1036 |
break; |
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. */ |
/* Should never occur, but keep compilers from grumbling. */ |
1063 |
|
|
1064 |
default: |
default: |
1213 |
break; |
break; |
1214 |
|
|
1215 |
case PT_LAMP: |
case PT_LAMP: |
1216 |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt; |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || |
1217 |
|
prop->chartype == ucp_Lt; |
1218 |
break; |
break; |
1219 |
|
|
1220 |
case PT_GC: |
case PT_GC: |
1229 |
OK = prop->script == code[3]; |
OK = prop->script == code[3]; |
1230 |
break; |
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. */ |
/* Should never occur, but keep compilers from grumbling. */ |
1257 |
|
|
1258 |
default: |
default: |
1460 |
break; |
break; |
1461 |
|
|
1462 |
case PT_LAMP: |
case PT_LAMP: |
1463 |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt; |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || |
1464 |
|
prop->chartype == ucp_Lt; |
1465 |
break; |
break; |
1466 |
|
|
1467 |
case PT_GC: |
case PT_GC: |
1476 |
OK = prop->script == code[3]; |
OK = prop->script == code[3]; |
1477 |
break; |
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. */ |
/* Should never occur, but keep compilers from grumbling. */ |
1504 |
|
|
1505 |
default: |
default: |
1732 |
break; |
break; |
1733 |
|
|
1734 |
case PT_LAMP: |
case PT_LAMP: |
1735 |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || prop->chartype == ucp_Lt; |
OK = prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || |
1736 |
|
prop->chartype == ucp_Lt; |
1737 |
break; |
break; |
1738 |
|
|
1739 |
case PT_GC: |
case PT_GC: |
1748 |
OK = prop->script == code[5]; |
OK = prop->script == code[5]; |
1749 |
break; |
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. */ |
/* Should never occur, but keep compilers from grumbling. */ |
1776 |
|
|
1777 |
default: |
default: |
1981 |
break; |
break; |
1982 |
|
|
1983 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
1984 |
case OP_CHARNC: |
case OP_CHARI: |
1985 |
if (clen == 0) break; |
if (clen == 0) break; |
1986 |
|
|
1987 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2167 |
break; |
break; |
2168 |
|
|
2169 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
2170 |
/* Match a negated single character. This is only used for one-byte |
/* Match a negated single character casefully. This is only used for |
2171 |
characters, that is, we know that d < 256. The character we are |
one-byte characters, that is, we know that d < 256. The character we are |
2172 |
checking (c) can be multibyte. */ |
checking (c) can be multibyte. */ |
2173 |
|
|
2174 |
case OP_NOT: |
case OP_NOT: |
2175 |
if (clen > 0) |
if (clen > 0 && c != d) { ADD_NEW(state_offset + dlen + 1, 0); } |
2176 |
{ |
break; |
2177 |
unsigned int otherd = ((ims & PCRE_CASELESS) != 0)? fcc[d] : d; |
|
2178 |
if (c != d && c != otherd) { ADD_NEW(state_offset + dlen + 1, 0); } |
/*-----------------------------------------------------------------*/ |
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; |
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: |
case OP_PLUS: |
2200 |
case OP_MINPLUS: |
case OP_MINPLUS: |
2201 |
case OP_POSPLUS: |
case OP_POSPLUS: |
2207 |
if (clen > 0) |
if (clen > 0) |
2208 |
{ |
{ |
2209 |
unsigned int otherd = NOTACHAR; |
unsigned int otherd = NOTACHAR; |
2210 |
if ((ims & PCRE_CASELESS) != 0) |
if (caseless) |
2211 |
{ |
{ |
2212 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2213 |
if (utf8 && d >= 128) |
if (utf8 && d >= 128) |
2235 |
break; |
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: |
case OP_QUERY: |
2248 |
case OP_MINQUERY: |
case OP_MINQUERY: |
2249 |
case OP_POSQUERY: |
case OP_POSQUERY: |
2254 |
if (clen > 0) |
if (clen > 0) |
2255 |
{ |
{ |
2256 |
unsigned int otherd = NOTACHAR; |
unsigned int otherd = NOTACHAR; |
2257 |
if ((ims & PCRE_CASELESS) != 0) |
if (caseless) |
2258 |
{ |
{ |
2259 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2260 |
if (utf8 && d >= 128) |
if (utf8 && d >= 128) |
2280 |
break; |
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: |
case OP_STAR: |
2293 |
case OP_MINSTAR: |
case OP_MINSTAR: |
2294 |
case OP_POSSTAR: |
case OP_POSSTAR: |
2299 |
if (clen > 0) |
if (clen > 0) |
2300 |
{ |
{ |
2301 |
unsigned int otherd = NOTACHAR; |
unsigned int otherd = NOTACHAR; |
2302 |
if ((ims & PCRE_CASELESS) != 0) |
if (caseless) |
2303 |
{ |
{ |
2304 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2305 |
if (utf8 && d >= 128) |
if (utf8 && d >= 128) |
2325 |
break; |
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: |
case OP_EXACT: |
2334 |
case OP_NOTEXACT: |
case OP_NOTEXACT: |
2335 |
count = current_state->count; /* Number already matched */ |
count = current_state->count; /* Number already matched */ |
2336 |
if (clen > 0) |
if (clen > 0) |
2337 |
{ |
{ |
2338 |
unsigned int otherd = NOTACHAR; |
unsigned int otherd = NOTACHAR; |
2339 |
if ((ims & PCRE_CASELESS) != 0) |
if (caseless) |
2340 |
{ |
{ |
2341 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2342 |
if (utf8 && d >= 128) |
if (utf8 && d >= 128) |
2360 |
break; |
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: |
case OP_UPTO: |
2373 |
case OP_MINUPTO: |
case OP_MINUPTO: |
2374 |
case OP_POSUPTO: |
case OP_POSUPTO: |
2380 |
if (clen > 0) |
if (clen > 0) |
2381 |
{ |
{ |
2382 |
unsigned int otherd = NOTACHAR; |
unsigned int otherd = NOTACHAR; |
2383 |
if ((ims & PCRE_CASELESS) != 0) |
if (caseless) |
2384 |
{ |
{ |
2385 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
2386 |
if (utf8 && d >= 128) |
if (utf8 && d >= 128) |
2447 |
points to the byte after the end of the class. If there is a |
points to the byte after the end of the class. If there is a |
2448 |
quantifier, this is where it will be. */ |
quantifier, this is where it will be. */ |
2449 |
|
|
2450 |
next_state_offset = ecode - start_code; |
next_state_offset = (int)(ecode - start_code); |
2451 |
|
|
2452 |
switch (*ecode) |
switch (*ecode) |
2453 |
{ |
{ |
2518 |
md, /* static match data */ |
md, /* static match data */ |
2519 |
code, /* this subexpression's code */ |
code, /* this subexpression's code */ |
2520 |
ptr, /* where we currently are */ |
ptr, /* where we currently are */ |
2521 |
ptr - start_subject, /* start offset */ |
(int)(ptr - start_subject), /* start offset */ |
2522 |
local_offsets, /* offset vector */ |
local_offsets, /* offset vector */ |
2523 |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
2524 |
local_workspace, /* workspace vector */ |
local_workspace, /* workspace vector */ |
2525 |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
2526 |
ims, /* the current ims flags */ |
rlevel); /* function recursion level */ |
|
rlevel, /* function recursion level */ |
|
|
recursing); /* pass on regex recursion */ |
|
2527 |
|
|
2528 |
|
if (rc == PCRE_ERROR_DFA_UITEM) return rc; |
2529 |
if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK)) |
if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK)) |
2530 |
{ ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); } |
{ ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); } |
2531 |
} |
} |
2532 |
break; |
break; |
2533 |
|
|
2554 |
cb.callout_number = code[LINK_SIZE+2]; |
cb.callout_number = code[LINK_SIZE+2]; |
2555 |
cb.offset_vector = offsets; |
cb.offset_vector = offsets; |
2556 |
cb.subject = (PCRE_SPTR)start_subject; |
cb.subject = (PCRE_SPTR)start_subject; |
2557 |
cb.subject_length = end_subject - start_subject; |
cb.subject_length = (int)(end_subject - start_subject); |
2558 |
cb.start_match = current_subject - start_subject; |
cb.start_match = (int)(current_subject - start_subject); |
2559 |
cb.current_position = ptr - start_subject; |
cb.current_position = (int)(ptr - start_subject); |
2560 |
cb.pattern_position = GET(code, LINK_SIZE + 3); |
cb.pattern_position = GET(code, LINK_SIZE + 3); |
2561 |
cb.next_item_length = GET(code, 3 + 2*LINK_SIZE); |
cb.next_item_length = GET(code, 3 + 2*LINK_SIZE); |
2562 |
cb.capture_top = 1; |
cb.capture_top = 1; |
2563 |
cb.capture_last = -1; |
cb.capture_last = -1; |
2564 |
cb.callout_data = md->callout_data; |
cb.callout_data = md->callout_data; |
2565 |
|
cb.mark = NULL; /* No (*MARK) support */ |
2566 |
if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */ |
if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */ |
2567 |
} |
} |
2568 |
if (rrc > 0) break; /* Fail this thread */ |
if (rrc > 0) break; /* Fail this thread */ |
2573 |
|
|
2574 |
/* Back reference conditions are not supported */ |
/* Back reference conditions are not supported */ |
2575 |
|
|
2576 |
if (condcode == OP_CREF) return PCRE_ERROR_DFA_UCOND; |
if (condcode == OP_CREF || condcode == OP_NCREF) |
2577 |
|
return PCRE_ERROR_DFA_UCOND; |
2578 |
|
|
2579 |
/* The DEFINE condition is always false */ |
/* The DEFINE condition is always false */ |
2580 |
|
|
2585 |
which means "test if in any recursion". We can't test for specifically |
which means "test if in any recursion". We can't test for specifically |
2586 |
recursed groups. */ |
recursed groups. */ |
2587 |
|
|
2588 |
else if (condcode == OP_RREF) |
else if (condcode == OP_RREF || condcode == OP_NRREF) |
2589 |
{ |
{ |
2590 |
int value = GET2(code, LINK_SIZE+2); |
int value = GET2(code, LINK_SIZE+2); |
2591 |
if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND; |
if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND; |
2592 |
if (recursing > 0) |
if (md->recursive != NULL) |
2593 |
{ ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); } |
{ ADD_ACTIVE(state_offset + LINK_SIZE + 4, 0); } |
2594 |
else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); } |
else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); } |
2595 |
} |
} |
2608 |
md, /* fixed match data */ |
md, /* fixed match data */ |
2609 |
asscode, /* this subexpression's code */ |
asscode, /* this subexpression's code */ |
2610 |
ptr, /* where we currently are */ |
ptr, /* where we currently are */ |
2611 |
ptr - start_subject, /* start offset */ |
(int)(ptr - start_subject), /* start offset */ |
2612 |
local_offsets, /* offset vector */ |
local_offsets, /* offset vector */ |
2613 |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
2614 |
local_workspace, /* workspace vector */ |
local_workspace, /* workspace vector */ |
2615 |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
2616 |
ims, /* the current ims flags */ |
rlevel); /* function recursion level */ |
|
rlevel, /* function recursion level */ |
|
|
recursing); /* pass on regex recursion */ |
|
2617 |
|
|
2618 |
|
if (rc == PCRE_ERROR_DFA_UITEM) return rc; |
2619 |
if ((rc >= 0) == |
if ((rc >= 0) == |
2620 |
(condcode == OP_ASSERT || condcode == OP_ASSERTBACK)) |
(condcode == OP_ASSERT || condcode == OP_ASSERTBACK)) |
2621 |
{ ADD_ACTIVE(endasscode + LINK_SIZE + 1 - start_code, 0); } |
{ ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); } |
2622 |
else |
else |
2623 |
{ ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); } |
{ ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); } |
2624 |
} |
} |
2628 |
/*-----------------------------------------------------------------*/ |
/*-----------------------------------------------------------------*/ |
2629 |
case OP_RECURSE: |
case OP_RECURSE: |
2630 |
{ |
{ |
2631 |
|
dfa_recursion_info *ri; |
2632 |
int local_offsets[1000]; |
int local_offsets[1000]; |
2633 |
int local_workspace[1000]; |
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; |
int rc; |
2638 |
|
|
2639 |
DPRINTF(("%.*sStarting regex recursion %d\n", rlevel*2-2, SP, |
DPRINTF(("%.*sStarting regex recursion\n", rlevel*2-2, SP)); |
2640 |
recursing + 1)); |
|
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( |
rc = internal_dfa_exec( |
2658 |
md, /* fixed match data */ |
md, /* fixed match data */ |
2659 |
start_code + GET(code, 1), /* this subexpression's code */ |
callpat, /* this subexpression's code */ |
2660 |
ptr, /* where we currently are */ |
ptr, /* where we currently are */ |
2661 |
ptr - start_subject, /* start offset */ |
(int)(ptr - start_subject), /* start offset */ |
2662 |
local_offsets, /* offset vector */ |
local_offsets, /* offset vector */ |
2663 |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
2664 |
local_workspace, /* workspace vector */ |
local_workspace, /* workspace vector */ |
2665 |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
2666 |
ims, /* the current ims flags */ |
rlevel); /* function recursion level */ |
|
rlevel, /* function recursion level */ |
|
|
recursing + 1); /* regex recurse level */ |
|
2667 |
|
|
2668 |
DPRINTF(("%.*sReturn from regex recursion %d: rc=%d\n", rlevel*2-2, SP, |
md->recursive = new_recursive.prevrec; /* Done this recursion */ |
2669 |
recursing + 1, rc)); |
|
2670 |
|
DPRINTF(("%.*sReturn from regex recursion: rc=%d\n", rlevel*2-2, SP, |
2671 |
|
rc)); |
2672 |
|
|
2673 |
/* Ran out of internal offsets */ |
/* Ran out of internal offsets */ |
2674 |
|
|
2701 |
break; |
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 = 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: |
case OP_ONCE: |
2794 |
|
case OP_ONCE_NC: |
2795 |
{ |
{ |
2796 |
int local_offsets[2]; |
int local_offsets[2]; |
2797 |
int local_workspace[1000]; |
int local_workspace[1000]; |
2800 |
md, /* fixed match data */ |
md, /* fixed match data */ |
2801 |
code, /* this subexpression's code */ |
code, /* this subexpression's code */ |
2802 |
ptr, /* where we currently are */ |
ptr, /* where we currently are */ |
2803 |
ptr - start_subject, /* start offset */ |
(int)(ptr - start_subject), /* start offset */ |
2804 |
local_offsets, /* offset vector */ |
local_offsets, /* offset vector */ |
2805 |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
sizeof(local_offsets)/sizeof(int), /* size of same */ |
2806 |
local_workspace, /* workspace vector */ |
local_workspace, /* workspace vector */ |
2807 |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
sizeof(local_workspace)/sizeof(int), /* size of same */ |
2808 |
ims, /* the current ims flags */ |
rlevel); /* function recursion level */ |
|
rlevel, /* function recursion level */ |
|
|
recursing); /* pass on regex recursion */ |
|
2809 |
|
|
2810 |
if (rc >= 0) |
if (rc >= 0) |
2811 |
{ |
{ |
2815 |
|
|
2816 |
do { end_subpattern += GET(end_subpattern, 1); } |
do { end_subpattern += GET(end_subpattern, 1); } |
2817 |
while (*end_subpattern == OP_ALT); |
while (*end_subpattern == OP_ALT); |
2818 |
next_state_offset = end_subpattern - start_code + LINK_SIZE + 1; |
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 |
/* 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. |
arrange for the repeat state also to be added to the relevant list. |
2824 |
|
|
2825 |
repeat_state_offset = (*end_subpattern == OP_KETRMAX || |
repeat_state_offset = (*end_subpattern == OP_KETRMAX || |
2826 |
*end_subpattern == OP_KETRMIN)? |
*end_subpattern == OP_KETRMIN)? |
2827 |
end_subpattern - start_code - GET(end_subpattern, 1) : -1; |
(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 |
/* If we have matched an empty string, add the next state at the |
2830 |
current character pointer. This is important so that the duplicate |
current character pointer. This is important so that the duplicate |
2839 |
/* Optimization: if there are no more active states, and there |
/* Optimization: if there are no more active states, and there |
2840 |
are no new states yet set up, then skip over the subject string |
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 |
right here, to save looping. Otherwise, set up the new state to swing |
2842 |
into action when the end of the substring is reached. */ |
into action when the end of the matched substring is reached. */ |
2843 |
|
|
2844 |
else if (i + 1 >= active_count && new_count == 0) |
else if (i + 1 >= active_count && new_count == 0) |
2845 |
{ |
{ |
2869 |
if (repeat_state_offset >= 0) |
if (repeat_state_offset >= 0) |
2870 |
{ ADD_NEW_DATA(-repeat_state_offset, 0, (charcount - 1)); } |
{ ADD_NEW_DATA(-repeat_state_offset, 0, (charcount - 1)); } |
2871 |
} |
} |
|
|
|
2872 |
} |
} |
2873 |
else if (rc != PCRE_ERROR_NOMATCH) return rc; |
else if (rc != PCRE_ERROR_NOMATCH) return rc; |
2874 |
} |
} |
2887 |
cb.callout_number = code[1]; |
cb.callout_number = code[1]; |
2888 |
cb.offset_vector = offsets; |
cb.offset_vector = offsets; |
2889 |
cb.subject = (PCRE_SPTR)start_subject; |
cb.subject = (PCRE_SPTR)start_subject; |
2890 |
cb.subject_length = end_subject - start_subject; |
cb.subject_length = (int)(end_subject - start_subject); |
2891 |
cb.start_match = current_subject - start_subject; |
cb.start_match = (int)(current_subject - start_subject); |
2892 |
cb.current_position = ptr - start_subject; |
cb.current_position = (int)(ptr - start_subject); |
2893 |
cb.pattern_position = GET(code, 2); |
cb.pattern_position = GET(code, 2); |
2894 |
cb.next_item_length = GET(code, 2 + LINK_SIZE); |
cb.next_item_length = GET(code, 2 + LINK_SIZE); |
2895 |
cb.capture_top = 1; |
cb.capture_top = 1; |
2896 |
cb.capture_last = -1; |
cb.capture_last = -1; |
2897 |
cb.callout_data = md->callout_data; |
cb.callout_data = md->callout_data; |
2898 |
|
cb.mark = NULL; /* No (*MARK) support */ |
2899 |
if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */ |
if ((rrc = (*pcre_callout)(&cb)) < 0) return rrc; /* Abandon */ |
2900 |
} |
} |
2901 |
if (rrc == 0) |
if (rrc == 0) |
2915 |
/* We have finished the processing at the current subject character. If no |
/* 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 |
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 |
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. The "forced_ |
matching has been requested, check for appropriate conditions. |
2919 |
fail" variable counts the number of (*F) encountered for the character. If it |
|
2920 |
is equal to the original active_count (saved in workspace[1]) it means that |
The "forced_ fail" variable counts the number of (*F) encountered for the |
2921 |
(*F) was found on every active state. In this case we don't want to give a |
character. If it is equal to the original active_count (saved in |
2922 |
partial match. */ |
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) |
if (new_count <= 0) |
2929 |
{ |
{ |
2930 |
if (rlevel == 1 && /* Top level, and */ |
if (rlevel == 1 && /* Top level, and */ |
2931 |
reached_end != workspace[1] && /* Not all reached end */ |
could_continue && /* Some could go on */ |
2932 |
forced_fail != workspace[1] && /* Not all forced fail & */ |
forced_fail != workspace[1] && /* Not all forced fail & */ |
2933 |
( /* either... */ |
( /* either... */ |
2934 |
(md->moptions & PCRE_PARTIAL_HARD) != 0 /* Hard partial */ |
(md->moptions & PCRE_PARTIAL_HARD) != 0 /* Hard partial */ |
2936 |
((md->moptions & PCRE_PARTIAL_SOFT) != 0 && /* Soft partial and */ |
((md->moptions & PCRE_PARTIAL_SOFT) != 0 && /* Soft partial and */ |
2937 |
match_count < 0) /* no matches */ |
match_count < 0) /* no matches */ |
2938 |
) && /* And... */ |
) && /* And... */ |
2939 |
ptr >= end_subject && /* Reached end of subject */ |
ptr >= end_subject && /* Reached end of subject */ |
2940 |
ptr > current_subject) /* Matched non-empty string */ |
ptr > md->start_used_ptr) /* Inspected non-empty string */ |
2941 |
{ |
{ |
2942 |
if (offsetcount >= 2) |
if (offsetcount >= 2) |
2943 |
{ |
{ |
2944 |
offsets[0] = md->start_used_ptr - start_subject; |
offsets[0] = (int)(md->start_used_ptr - start_subject); |
2945 |
offsets[1] = end_subject - start_subject; |
offsets[1] = (int)(end_subject - start_subject); |
2946 |
} |
} |
2947 |
match_count = PCRE_ERROR_PARTIAL; |
match_count = PCRE_ERROR_PARTIAL; |
2948 |
} |
} |
3027 |
(offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; |
(offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; |
3028 |
if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; |
if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; |
3029 |
if (wscount < 20) return PCRE_ERROR_DFA_WSSIZE; |
if (wscount < 20) return PCRE_ERROR_DFA_WSSIZE; |
3030 |
|
if (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 |
/* We need to find the pointer to any study data before we test for byte |
3033 |
flipping, so we scan the extra_data block first. This may set two fields in the |
flipping, so we scan the extra_data block first. This may set two fields in the |
3146 |
#ifdef SUPPORT_UTF8 |
#ifdef SUPPORT_UTF8 |
3147 |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) |
if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) |
3148 |
{ |
{ |
3149 |
if (_pcre_valid_utf8((uschar *)subject, length) >= 0) |
int erroroffset; |
3150 |
return PCRE_ERROR_BADUTF8; |
int errorcode = _pcre_valid_utf8((uschar *)subject, length, &erroroffset); |
3151 |
if (start_offset > 0 && start_offset < length) |
if (errorcode != 0) |
3152 |
{ |
{ |
3153 |
int tb = ((uschar *)subject)[start_offset]; |
if (offsetcount >= 2) |
|
if (tb > 127) |
|
3154 |
{ |
{ |
3155 |
tb &= 0xc0; |
offsets[0] = erroroffset; |
3156 |
if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET; |
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 |
#endif |
3166 |
|
|
3193 |
} |
} |
3194 |
else |
else |
3195 |
{ |
{ |
3196 |
if (startline && study != NULL && |
if (!startline && study != NULL && |
3197 |
(study->options & PCRE_STUDY_MAPPED) != 0) |
(study->flags & PCRE_STUDY_MAPPED) != 0) |
3198 |
start_bits = study->start_bits; |
start_bits = study->start_bits; |
3199 |
} |
} |
3200 |
} |
} |
3245 |
} |
} |
3246 |
|
|
3247 |
/* There are some optimizations that avoid running the match if a known |
/* There are some optimizations that avoid running the match if a known |
3248 |
starting point is not found, or if a known later character is not present. |
starting point is not found. However, there is an option that disables |
3249 |
However, there is an option that disables these, for testing and for |
these, for testing and for ensuring that all callouts do actually occur. |
3250 |
ensuring that all callouts do actually occur. */ |
The option can be set in the regex by (*NO_START_OPT) or passed in |
3251 |
|
match-time options. */ |
3252 |
|
|
3253 |
if ((options & PCRE_NO_START_OPTIMIZE) == 0) |
if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0) |
3254 |
{ |
{ |
|
|
|
3255 |
/* Advance to a known first byte. */ |
/* Advance to a known first byte. */ |
3256 |
|
|
3257 |
if (first_byte >= 0) |
if (first_byte >= 0) |
3308 |
while (current_subject < end_subject) |
while (current_subject < end_subject) |
3309 |
{ |
{ |
3310 |
register unsigned int c = *current_subject; |
register unsigned int c = *current_subject; |
3311 |
if ((start_bits[c/8] & (1 << (c&7))) == 0) current_subject++; |
if ((start_bits[c/8] & (1 << (c&7))) == 0) |
3312 |
else break; |
{ |
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 |
} |
} |
3325 |
/* Restore fudged end_subject */ |
/* Restore fudged end_subject */ |
3326 |
|
|
3327 |
end_subject = save_end_subject; |
end_subject = save_end_subject; |
|
} |
|
3328 |
|
|
3329 |
/* If req_byte is set, we know that that character must appear in the subject |
/* The following two optimizations are disabled for partial matching or if |
3330 |
for the match to succeed. If the first character is set, req_byte must be |
disabling is explicitly requested (and of course, by the test above, this |
3331 |
later in the subject; otherwise the test starts at the match point. This |
code is not obeyed when restarting after a partial match). */ |
|
optimization can save a huge amount of work in patterns with nested unlimited |
|
|
repeats that aren't going to match. Writing separate code for cased/caseless |
|
|
versions makes it go faster, as does using an autoincrement and backing off |
|
|
on a match. |
|
|
|
|
|
HOWEVER: when the subject string is very, very long, searching to its end can |
|
|
take a long time, and give bad performance on quite ordinary patterns. This |
|
|
showed up when somebody was matching /^C/ on a 32-megabyte string... so we |
|
|
don't do this when the string is sufficiently long. |
|
|
|
|
|
ALSO: this processing is disabled when partial matching is requested, and can |
|
|
also be explicitly deactivated. Furthermore, we have to disable when |
|
|
restarting after a partial match, because the required character may have |
|
|
already been matched. */ |
|
|
|
|
|
if ((options & PCRE_NO_START_OPTIMIZE) == 0 && |
|
|
req_byte >= 0 && |
|
|
end_subject - current_subject < REQ_BYTE_MAX && |
|
|
(options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_RESTART)) == 0) |
|
|
{ |
|
|
register const uschar *p = current_subject + ((first_byte >= 0)? 1 : 0); |
|
|
|
|
|
/* We don't need to repeat the search if we haven't yet reached the |
|
|
place we found it at last time. */ |
|
3332 |
|
|
3333 |
if (p > req_byte_ptr) |
if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 && |
3334 |
|
(options & (PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT)) == 0) |
3335 |
{ |
{ |
3336 |
if (req_byte_caseless) |
/* 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 |
while (p < end_subject) |
pattern. Although the value is, strictly, in characters, we treat it as |
3339 |
{ |
bytes to avoid spending too much time in this optimization. */ |
3340 |
register int pp = *p++; |
|
3341 |
if (pp == req_byte || pp == req_byte2) { p--; break; } |
if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 && |
3342 |
} |
(pcre_uint32)(end_subject - current_subject) < study->minlength) |
3343 |
} |
return PCRE_ERROR_NOMATCH; |
3344 |
else |
|
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 |
while (p < end_subject) |
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 (*p++ == req_byte) { p--; break; } |
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, |
/* If we can't find the required character, break the matching loop, |
3384 |
which will cause a return or PCRE_ERROR_NOMATCH. */ |
which will cause a return or PCRE_ERROR_NOMATCH. */ |
3385 |
|
|
3386 |
if (p >= end_subject) break; |
if (p >= end_subject) break; |
3387 |
|
|
3388 |
/* If we have found the required character, save the point where we |
/* 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 |
found it, so that we don't search again next time round the loop if |
3390 |
the start hasn't passed this character yet. */ |
the start hasn't passed this character yet. */ |
3391 |
|
|
3392 |
req_byte_ptr = p; |
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 */ |
/* OK, now we can do the business */ |
3399 |
|
|
3400 |
md->start_used_ptr = current_subject; |
md->start_used_ptr = current_subject; |
3401 |
|
md->recursive = NULL; |
3402 |
|
|
3403 |
rc = internal_dfa_exec( |
rc = internal_dfa_exec( |
3404 |
md, /* fixed match data */ |
md, /* fixed match data */ |
3405 |
md->start_code, /* this subexpression's code */ |
md->start_code, /* this subexpression's code */ |
3409 |
offsetcount, /* size of same */ |
offsetcount, /* size of same */ |
3410 |
workspace, /* workspace vector */ |
workspace, /* workspace vector */ |
3411 |
wscount, /* size of same */ |
wscount, /* size of same */ |
3412 |
re->options & (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL), /* ims flags */ |
0); /* function recurse level */ |
|
0, /* function recurse level */ |
|
|
0); /* regex recurse level */ |
|
3413 |
|
|
3414 |
/* Anything other than "no match" means we are done, always; otherwise, carry |
/* Anything other than "no match" means we are done, always; otherwise, carry |
3415 |
on only if not anchored. */ |
on only if not anchored. */ |