9 |
|
|
10 |
Written by: Philip Hazel <ph10@cam.ac.uk> |
Written by: Philip Hazel <ph10@cam.ac.uk> |
11 |
|
|
12 |
Copyright (c) 1998 University of Cambridge |
Copyright (c) 1997-2000 University of Cambridge |
13 |
|
|
14 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
15 |
Permission is granted to anyone to use this software for any purpose on any |
Permission is granted to anyone to use this software for any purpose on any |
25 |
|
|
26 |
3. Altered versions must be plainly marked as such, and must not be |
3. Altered versions must be plainly marked as such, and must not be |
27 |
misrepresented as being the original software. |
misrepresented as being the original software. |
28 |
|
|
29 |
|
4. If PCRE is embedded in any software that is released under the GNU |
30 |
|
General Purpose Licence (GPL), then the terms of that licence shall |
31 |
|
supersede any condition above with which it is incompatible. |
32 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
33 |
*/ |
*/ |
34 |
|
|
41 |
|
|
42 |
|
|
43 |
/************************************************* |
/************************************************* |
44 |
|
* Set a bit and maybe its alternate case * |
45 |
|
*************************************************/ |
46 |
|
|
47 |
|
/* Given a character, set its bit in the table, and also the bit for the other |
48 |
|
version of a letter if we are caseless. |
49 |
|
|
50 |
|
Arguments: |
51 |
|
start_bits points to the bit map |
52 |
|
c is the character |
53 |
|
caseless the caseless flag |
54 |
|
cd the block with char table pointers |
55 |
|
|
56 |
|
Returns: nothing |
57 |
|
*/ |
58 |
|
|
59 |
|
static void |
60 |
|
set_bit(uschar *start_bits, int c, BOOL caseless, compile_data *cd) |
61 |
|
{ |
62 |
|
start_bits[c/8] |= (1 << (c&7)); |
63 |
|
if (caseless && (cd->ctypes[c] & ctype_letter) != 0) |
64 |
|
start_bits[cd->fcc[c]/8] |= (1 << (cd->fcc[c]&7)); |
65 |
|
} |
66 |
|
|
67 |
|
|
68 |
|
|
69 |
|
/************************************************* |
70 |
* Create bitmap of starting chars * |
* Create bitmap of starting chars * |
71 |
*************************************************/ |
*************************************************/ |
72 |
|
|
77 |
Arguments: |
Arguments: |
78 |
code points to an expression |
code points to an expression |
79 |
start_bits points to a 32-byte table, initialized to 0 |
start_bits points to a 32-byte table, initialized to 0 |
80 |
|
caseless the current state of the caseless flag |
81 |
|
cd the block with char table pointers |
82 |
|
|
83 |
Returns: TRUE if table built, FALSE otherwise |
Returns: TRUE if table built, FALSE otherwise |
84 |
*/ |
*/ |
85 |
|
|
86 |
static BOOL |
static BOOL |
87 |
set_start_bits(const uschar *code, uschar *start_bits) |
set_start_bits(const uschar *code, uschar *start_bits, BOOL caseless, |
88 |
|
compile_data *cd) |
89 |
{ |
{ |
90 |
register int c; |
register int c; |
91 |
|
|
92 |
|
/* This next statement and the later reference to dummy are here in order to |
93 |
|
trick the optimizer of the IBM C compiler for OS/2 into generating correct |
94 |
|
code. Apparently IBM isn't going to fix the problem, and we would rather not |
95 |
|
disable optimization (in this module it actually makes a big difference, and |
96 |
|
the pcre module can use all the optimization it can get). */ |
97 |
|
|
98 |
|
volatile int dummy; |
99 |
|
|
100 |
do |
do |
101 |
{ |
{ |
102 |
const uschar *tcode = code + 3; |
const uschar *tcode = code + 3; |
106 |
{ |
{ |
107 |
try_next = FALSE; |
try_next = FALSE; |
108 |
|
|
109 |
|
/* If a branch starts with a bracket or a positive lookahead assertion, |
110 |
|
recurse to set bits from within them. That's all for this branch. */ |
111 |
|
|
112 |
if ((int)*tcode >= OP_BRA || *tcode == OP_ASSERT) |
if ((int)*tcode >= OP_BRA || *tcode == OP_ASSERT) |
113 |
{ |
{ |
114 |
if (!set_start_bits(tcode, start_bits)) return FALSE; |
if (!set_start_bits(tcode, start_bits, caseless, cd)) |
115 |
|
return FALSE; |
116 |
} |
} |
117 |
|
|
118 |
else switch(*tcode) |
else switch(*tcode) |
120 |
default: |
default: |
121 |
return FALSE; |
return FALSE; |
122 |
|
|
123 |
|
/* Skip over lookbehind and negative lookahead assertions */ |
124 |
|
|
125 |
|
case OP_ASSERT_NOT: |
126 |
|
case OP_ASSERTBACK: |
127 |
|
case OP_ASSERTBACK_NOT: |
128 |
|
try_next = TRUE; |
129 |
|
do tcode += (tcode[1] << 8) + tcode[2]; while (*tcode == OP_ALT); |
130 |
|
tcode += 3; |
131 |
|
break; |
132 |
|
|
133 |
|
/* Skip over an option setting, changing the caseless flag */ |
134 |
|
|
135 |
|
case OP_OPT: |
136 |
|
caseless = (tcode[1] & PCRE_CASELESS) != 0; |
137 |
|
tcode += 2; |
138 |
|
try_next = TRUE; |
139 |
|
break; |
140 |
|
|
141 |
/* BRAZERO does the bracket, but carries on. */ |
/* BRAZERO does the bracket, but carries on. */ |
142 |
|
|
143 |
case OP_BRAZERO: |
case OP_BRAZERO: |
144 |
case OP_BRAMINZERO: |
case OP_BRAMINZERO: |
145 |
if (!set_start_bits(++tcode, start_bits)) return FALSE; |
if (!set_start_bits(++tcode, start_bits, caseless, cd)) |
146 |
|
return FALSE; |
147 |
|
dummy = 1; |
148 |
do tcode += (tcode[1] << 8) + tcode[2]; while (*tcode == OP_ALT); |
do tcode += (tcode[1] << 8) + tcode[2]; while (*tcode == OP_ALT); |
149 |
tcode += 3; |
tcode += 3; |
150 |
try_next = TRUE; |
try_next = TRUE; |
156 |
case OP_MINSTAR: |
case OP_MINSTAR: |
157 |
case OP_QUERY: |
case OP_QUERY: |
158 |
case OP_MINQUERY: |
case OP_MINQUERY: |
159 |
start_bits[tcode[1]/8] |= (1 << (tcode[1]&7)); |
set_bit(start_bits, tcode[1], caseless, cd); |
160 |
tcode += 2; |
tcode += 2; |
161 |
try_next = TRUE; |
try_next = TRUE; |
162 |
break; |
break; |
165 |
|
|
166 |
case OP_UPTO: |
case OP_UPTO: |
167 |
case OP_MINUPTO: |
case OP_MINUPTO: |
168 |
start_bits[tcode[3]/8] |= (1 << (tcode[3]&7)); |
set_bit(start_bits, tcode[3], caseless, cd); |
169 |
tcode += 4; |
tcode += 4; |
170 |
try_next = TRUE; |
try_next = TRUE; |
171 |
break; |
break; |
180 |
|
|
181 |
case OP_PLUS: |
case OP_PLUS: |
182 |
case OP_MINPLUS: |
case OP_MINPLUS: |
183 |
start_bits[tcode[1]/8] |= (1 << (tcode[1]&7)); |
set_bit(start_bits, tcode[1], caseless, cd); |
184 |
break; |
break; |
185 |
|
|
186 |
/* Single character type sets the bits and stops */ |
/* Single character type sets the bits and stops */ |
187 |
|
|
188 |
case OP_NOT_DIGIT: |
case OP_NOT_DIGIT: |
189 |
for (c = 0; c < 32; c++) start_bits[c] |= ~pcre_cbits[c+cbit_digit]; |
for (c = 0; c < 32; c++) |
190 |
|
start_bits[c] |= ~cd->cbits[c+cbit_digit]; |
191 |
break; |
break; |
192 |
|
|
193 |
case OP_DIGIT: |
case OP_DIGIT: |
194 |
for (c = 0; c < 32; c++) start_bits[c] |= pcre_cbits[c+cbit_digit]; |
for (c = 0; c < 32; c++) |
195 |
|
start_bits[c] |= cd->cbits[c+cbit_digit]; |
196 |
break; |
break; |
197 |
|
|
198 |
case OP_NOT_WHITESPACE: |
case OP_NOT_WHITESPACE: |
199 |
for (c = 0; c < 32; c++) start_bits[c] |= ~pcre_cbits[c+cbit_space]; |
for (c = 0; c < 32; c++) |
200 |
|
start_bits[c] |= ~cd->cbits[c+cbit_space]; |
201 |
break; |
break; |
202 |
|
|
203 |
case OP_WHITESPACE: |
case OP_WHITESPACE: |
204 |
for (c = 0; c < 32; c++) start_bits[c] |= pcre_cbits[c+cbit_space]; |
for (c = 0; c < 32; c++) |
205 |
|
start_bits[c] |= cd->cbits[c+cbit_space]; |
206 |
break; |
break; |
207 |
|
|
208 |
case OP_NOT_WORDCHAR: |
case OP_NOT_WORDCHAR: |
209 |
for (c = 0; c < 32; c++) |
for (c = 0; c < 32; c++) |
210 |
start_bits[c] |= ~(pcre_cbits[c] | pcre_cbits[c+cbit_word]); |
start_bits[c] |= ~cd->cbits[c+cbit_word]; |
211 |
break; |
break; |
212 |
|
|
213 |
case OP_WORDCHAR: |
case OP_WORDCHAR: |
214 |
for (c = 0; c < 32; c++) |
for (c = 0; c < 32; c++) |
215 |
start_bits[c] |= (pcre_cbits[c] | pcre_cbits[c+cbit_word]); |
start_bits[c] |= cd->cbits[c+cbit_word]; |
216 |
break; |
break; |
217 |
|
|
218 |
/* One or more character type fudges the pointer and restarts, knowing |
/* One or more character type fudges the pointer and restarts, knowing |
243 |
switch(tcode[1]) |
switch(tcode[1]) |
244 |
{ |
{ |
245 |
case OP_NOT_DIGIT: |
case OP_NOT_DIGIT: |
246 |
for (c = 0; c < 32; c++) start_bits[c] |= ~pcre_cbits[c+cbit_digit]; |
for (c = 0; c < 32; c++) |
247 |
|
start_bits[c] |= ~cd->cbits[c+cbit_digit]; |
248 |
break; |
break; |
249 |
|
|
250 |
case OP_DIGIT: |
case OP_DIGIT: |
251 |
for (c = 0; c < 32; c++) start_bits[c] |= pcre_cbits[c+cbit_digit]; |
for (c = 0; c < 32; c++) |
252 |
|
start_bits[c] |= cd->cbits[c+cbit_digit]; |
253 |
break; |
break; |
254 |
|
|
255 |
case OP_NOT_WHITESPACE: |
case OP_NOT_WHITESPACE: |
256 |
for (c = 0; c < 32; c++) start_bits[c] |= ~pcre_cbits[c+cbit_space]; |
for (c = 0; c < 32; c++) |
257 |
|
start_bits[c] |= ~cd->cbits[c+cbit_space]; |
258 |
break; |
break; |
259 |
|
|
260 |
case OP_WHITESPACE: |
case OP_WHITESPACE: |
261 |
for (c = 0; c < 32; c++) start_bits[c] |= pcre_cbits[c+cbit_space]; |
for (c = 0; c < 32; c++) |
262 |
|
start_bits[c] |= cd->cbits[c+cbit_space]; |
263 |
break; |
break; |
264 |
|
|
265 |
case OP_NOT_WORDCHAR: |
case OP_NOT_WORDCHAR: |
266 |
for (c = 0; c < 32; c++) |
for (c = 0; c < 32; c++) |
267 |
start_bits[c] |= ~(pcre_cbits[c] | pcre_cbits[c+cbit_word]); |
start_bits[c] |= ~cd->cbits[c+cbit_word]; |
268 |
break; |
break; |
269 |
|
|
270 |
case OP_WORDCHAR: |
case OP_WORDCHAR: |
271 |
for (c = 0; c < 32; c++) |
for (c = 0; c < 32; c++) |
272 |
start_bits[c] |= (pcre_cbits[c] | pcre_cbits[c+cbit_word]); |
start_bits[c] |= cd->cbits[c+cbit_word]; |
273 |
break; |
break; |
274 |
} |
} |
275 |
|
|
281 |
according to the repeat count. */ |
according to the repeat count. */ |
282 |
|
|
283 |
case OP_CLASS: |
case OP_CLASS: |
|
case OP_NEGCLASS: |
|
284 |
{ |
{ |
285 |
tcode++; |
tcode++; |
286 |
for (c = 0; c < 32; c++) start_bits[c] |= tcode[c]; |
for (c = 0; c < 32; c++) start_bits[c] |= tcode[c]; |
339 |
pcre_extra * |
pcre_extra * |
340 |
pcre_study(const pcre *external_re, int options, const char **errorptr) |
pcre_study(const pcre *external_re, int options, const char **errorptr) |
341 |
{ |
{ |
|
BOOL caseless; |
|
342 |
uschar start_bits[32]; |
uschar start_bits[32]; |
343 |
real_pcre_extra *extra; |
real_pcre_extra *extra; |
344 |
const real_pcre *re = (const real_pcre *)external_re; |
const real_pcre *re = (const real_pcre *)external_re; |
345 |
|
compile_data compile_block; |
346 |
|
|
347 |
*errorptr = NULL; |
*errorptr = NULL; |
348 |
|
|
358 |
return NULL; |
return NULL; |
359 |
} |
} |
360 |
|
|
|
/* Caseless can either be from the compiled regex or from options. */ |
|
|
|
|
|
caseless = ((re->options | options) & PCRE_CASELESS) != 0; |
|
|
|
|
361 |
/* For an anchored pattern, or an unchored pattern that has a first char, or a |
/* For an anchored pattern, or an unchored pattern that has a first char, or a |
362 |
multiline pattern that matches only at "line starts", no further processing at |
multiline pattern that matches only at "line starts", no further processing at |
363 |
present. */ |
present. */ |
365 |
if ((re->options & (PCRE_ANCHORED|PCRE_FIRSTSET|PCRE_STARTLINE)) != 0) |
if ((re->options & (PCRE_ANCHORED|PCRE_FIRSTSET|PCRE_STARTLINE)) != 0) |
366 |
return NULL; |
return NULL; |
367 |
|
|
368 |
/* See if we can find a fixed set of initial characters for the pattern. */ |
/* Set the character tables in the block which is passed around */ |
369 |
|
|
370 |
memset(start_bits, 0, 32 * sizeof(uschar)); |
compile_block.lcc = re->tables + lcc_offset; |
371 |
if (!set_start_bits(re->code, start_bits)) return NULL; |
compile_block.fcc = re->tables + fcc_offset; |
372 |
|
compile_block.cbits = re->tables + cbits_offset; |
373 |
|
compile_block.ctypes = re->tables + ctypes_offset; |
374 |
|
|
375 |
/* If this studying is caseless, scan the created bit map and duplicate the |
/* See if we can find a fixed set of initial characters for the pattern. */ |
|
bits for any letters. */ |
|
376 |
|
|
377 |
if (caseless) |
memset(start_bits, 0, 32 * sizeof(uschar)); |
378 |
{ |
if (!set_start_bits(re->code, start_bits, (re->options & PCRE_CASELESS) != 0, |
379 |
register int c; |
&compile_block)) return NULL; |
|
for (c = 0; c < 256; c++) |
|
|
{ |
|
|
if ((start_bits[c/8] & (1 << (c&7))) != 0 && |
|
|
(pcre_ctypes[c] & ctype_letter) != 0) |
|
|
{ |
|
|
int d = pcre_fcc[c]; |
|
|
start_bits[d/8] |= (1 << (d&7)); |
|
|
} |
|
|
} |
|
|
} |
|
380 |
|
|
381 |
/* Get an "extra" block and put the information therein. */ |
/* Get an "extra" block and put the information therein. */ |
382 |
|
|
388 |
return NULL; |
return NULL; |
389 |
} |
} |
390 |
|
|
391 |
extra->options = PCRE_STUDY_MAPPED | (caseless? PCRE_STUDY_CASELESS : 0); |
extra->options = PCRE_STUDY_MAPPED; |
392 |
memcpy(extra->start_bits, start_bits, sizeof(start_bits)); |
memcpy(extra->start_bits, start_bits, sizeof(start_bits)); |
393 |
|
|
394 |
return (pcre_extra *)extra; |
return (pcre_extra *)extra; |