/[pcre]/code/trunk/pcrecpp.cc
ViewVC logotype

Contents of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 356 - (show annotations) (download)
Tue Jul 8 14:14:34 2008 UTC (6 years, 3 months ago) by ph10
File size: 31384 byte(s)
Minor bugfix in pcrecpp.cc.

1 // Copyright (c) 2005, Google Inc.
2 // All rights reserved.
3 //
4 // Redistribution and use in source and binary forms, with or without
5 // modification, are permitted provided that the following conditions are
6 // met:
7 //
8 // * Redistributions of source code must retain the above copyright
9 // notice, this list of conditions and the following disclaimer.
10 // * Redistributions in binary form must reproduce the above
11 // copyright notice, this list of conditions and the following disclaimer
12 // in the documentation and/or other materials provided with the
13 // distribution.
14 // * Neither the name of Google Inc. nor the names of its
15 // contributors may be used to endorse or promote products derived from
16 // this software without specific prior written permission.
17 //
18 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29 //
30 // Author: Sanjay Ghemawat
31
32 #ifdef HAVE_CONFIG_H
33 #include "config.h"
34 #endif
35
36 #include <stdlib.h>
37 #include <stdio.h>
38 #include <ctype.h>
39 #include <limits.h> /* for SHRT_MIN, USHRT_MAX, etc */
40 #include <assert.h>
41 #include <errno.h>
42 #include <string>
43 #include <algorithm>
44
45 #include "pcrecpp_internal.h"
46 #include "pcre.h"
47 #include "pcrecpp.h"
48 #include "pcre_stringpiece.h"
49
50
51 namespace pcrecpp {
52
53 // Maximum number of args we can set
54 static const int kMaxArgs = 16;
55 static const int kVecSize = (1 + kMaxArgs) * 3; // results + PCRE workspace
56
57 // Special object that stands-in for no argument
58 Arg RE::no_arg((void*)NULL);
59
60 // This is for ABI compatibility with old versions of pcre (pre-7.6),
61 // which defined a global no_arg variable instead of putting it in the
62 // RE class. This works on GCC >= 3, at least. It definitely works
63 // for ELF, but may not for other object formats (Mach-O, for
64 // instance, does not support aliases.) We could probably have a more
65 // inclusive test if we ever needed it. (Note that not only the
66 // __attribute__ syntax, but also __USER_LABEL_PREFIX__, are
67 // gnu-specific.)
68 #if defined(__GNUC__) && __GNUC__ >= 3 && defined(__ELF__)
69 # define ULP_AS_STRING(x) ULP_AS_STRING_INTERNAL(x)
70 # define ULP_AS_STRING_INTERNAL(x) #x
71 # define USER_LABEL_PREFIX_STR ULP_AS_STRING(__USER_LABEL_PREFIX__)
72 extern Arg no_arg
73 __attribute__((alias(USER_LABEL_PREFIX_STR "_ZN7pcrecpp2RE6no_argE")));
74 #endif
75
76 // If a regular expression has no error, its error_ field points here
77 static const string empty_string;
78
79 // If the user doesn't ask for any options, we just use this one
80 static RE_Options default_options;
81
82 void RE::Init(const string& pat, const RE_Options* options) {
83 pattern_ = pat;
84 if (options == NULL) {
85 options_ = default_options;
86 } else {
87 options_ = *options;
88 }
89 error_ = &empty_string;
90 re_full_ = NULL;
91 re_partial_ = NULL;
92
93 re_partial_ = Compile(UNANCHORED);
94 if (re_partial_ != NULL) {
95 re_full_ = Compile(ANCHOR_BOTH);
96 }
97 }
98
99 void RE::Cleanup() {
100 if (re_full_ != NULL) (*pcre_free)(re_full_);
101 if (re_partial_ != NULL) (*pcre_free)(re_partial_);
102 if (error_ != &empty_string) delete error_;
103 }
104
105
106 RE::~RE() {
107 Cleanup();
108 }
109
110
111 pcre* RE::Compile(Anchor anchor) {
112 // First, convert RE_Options into pcre options
113 int pcre_options = 0;
114 pcre_options = options_.all_options();
115
116 // Special treatment for anchoring. This is needed because at
117 // runtime pcre only provides an option for anchoring at the
118 // beginning of a string (unless you use offset).
119 //
120 // There are three types of anchoring we want:
121 // UNANCHORED Compile the original pattern, and use
122 // a pcre unanchored match.
123 // ANCHOR_START Compile the original pattern, and use
124 // a pcre anchored match.
125 // ANCHOR_BOTH Tack a "\z" to the end of the original pattern
126 // and use a pcre anchored match.
127
128 const char* compile_error;
129 int eoffset;
130 pcre* re;
131 if (anchor != ANCHOR_BOTH) {
132 re = pcre_compile(pattern_.c_str(), pcre_options,
133 &compile_error, &eoffset, NULL);
134 } else {
135 // Tack a '\z' at the end of RE. Parenthesize it first so that
136 // the '\z' applies to all top-level alternatives in the regexp.
137 string wrapped = "(?:"; // A non-counting grouping operator
138 wrapped += pattern_;
139 wrapped += ")\\z";
140 re = pcre_compile(wrapped.c_str(), pcre_options,
141 &compile_error, &eoffset, NULL);
142 }
143 if (re == NULL) {
144 if (error_ == &empty_string) error_ = new string(compile_error);
145 }
146 return re;
147 }
148
149 /***** Matching interfaces *****/
150
151 bool RE::FullMatch(const StringPiece& text,
152 const Arg& ptr1,
153 const Arg& ptr2,
154 const Arg& ptr3,
155 const Arg& ptr4,
156 const Arg& ptr5,
157 const Arg& ptr6,
158 const Arg& ptr7,
159 const Arg& ptr8,
160 const Arg& ptr9,
161 const Arg& ptr10,
162 const Arg& ptr11,
163 const Arg& ptr12,
164 const Arg& ptr13,
165 const Arg& ptr14,
166 const Arg& ptr15,
167 const Arg& ptr16) const {
168 const Arg* args[kMaxArgs];
169 int n = 0;
170 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
171 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
172 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
173 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
174 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
175 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
176 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
177 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
178 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
179 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
180 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
181 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
182 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
183 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
184 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
185 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
186 done:
187
188 int consumed;
189 int vec[kVecSize];
190 return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
191 }
192
193 bool RE::PartialMatch(const StringPiece& text,
194 const Arg& ptr1,
195 const Arg& ptr2,
196 const Arg& ptr3,
197 const Arg& ptr4,
198 const Arg& ptr5,
199 const Arg& ptr6,
200 const Arg& ptr7,
201 const Arg& ptr8,
202 const Arg& ptr9,
203 const Arg& ptr10,
204 const Arg& ptr11,
205 const Arg& ptr12,
206 const Arg& ptr13,
207 const Arg& ptr14,
208 const Arg& ptr15,
209 const Arg& ptr16) const {
210 const Arg* args[kMaxArgs];
211 int n = 0;
212 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
213 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
214 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
215 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
216 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
217 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
218 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
219 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
220 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
221 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
222 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
223 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
224 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
225 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
226 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
227 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
228 done:
229
230 int consumed;
231 int vec[kVecSize];
232 return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
233 }
234
235 bool RE::Consume(StringPiece* input,
236 const Arg& ptr1,
237 const Arg& ptr2,
238 const Arg& ptr3,
239 const Arg& ptr4,
240 const Arg& ptr5,
241 const Arg& ptr6,
242 const Arg& ptr7,
243 const Arg& ptr8,
244 const Arg& ptr9,
245 const Arg& ptr10,
246 const Arg& ptr11,
247 const Arg& ptr12,
248 const Arg& ptr13,
249 const Arg& ptr14,
250 const Arg& ptr15,
251 const Arg& ptr16) const {
252 const Arg* args[kMaxArgs];
253 int n = 0;
254 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
255 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
256 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
257 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
258 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
259 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
260 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
261 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
262 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
263 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
264 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
265 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
266 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
267 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
268 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
269 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
270 done:
271
272 int consumed;
273 int vec[kVecSize];
274 if (DoMatchImpl(*input, ANCHOR_START, &consumed,
275 args, n, vec, kVecSize)) {
276 input->remove_prefix(consumed);
277 return true;
278 } else {
279 return false;
280 }
281 }
282
283 bool RE::FindAndConsume(StringPiece* input,
284 const Arg& ptr1,
285 const Arg& ptr2,
286 const Arg& ptr3,
287 const Arg& ptr4,
288 const Arg& ptr5,
289 const Arg& ptr6,
290 const Arg& ptr7,
291 const Arg& ptr8,
292 const Arg& ptr9,
293 const Arg& ptr10,
294 const Arg& ptr11,
295 const Arg& ptr12,
296 const Arg& ptr13,
297 const Arg& ptr14,
298 const Arg& ptr15,
299 const Arg& ptr16) const {
300 const Arg* args[kMaxArgs];
301 int n = 0;
302 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
303 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
304 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
305 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
306 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
307 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
308 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
309 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
310 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
311 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
312 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
313 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
314 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
315 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
316 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
317 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
318 done:
319
320 int consumed;
321 int vec[kVecSize];
322 if (DoMatchImpl(*input, UNANCHORED, &consumed,
323 args, n, vec, kVecSize)) {
324 input->remove_prefix(consumed);
325 return true;
326 } else {
327 return false;
328 }
329 }
330
331 bool RE::Replace(const StringPiece& rewrite,
332 string *str) const {
333 int vec[kVecSize];
334 int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
335 if (matches == 0)
336 return false;
337
338 string s;
339 if (!Rewrite(&s, rewrite, *str, vec, matches))
340 return false;
341
342 assert(vec[0] >= 0);
343 assert(vec[1] >= 0);
344 str->replace(vec[0], vec[1] - vec[0], s);
345 return true;
346 }
347
348 // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
349 // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
350 // Modified by PH to add PCRE_NEWLINE_ANY and PCRE_NEWLINE_ANYCRLF.
351
352 static int NewlineMode(int pcre_options) {
353 // TODO: if we can make it threadsafe, cache this var
354 int newline_mode = 0;
355 /* if (newline_mode) return newline_mode; */ // do this once it's cached
356 if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
357 PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF)) {
358 newline_mode = (pcre_options &
359 (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
360 PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF));
361 } else {
362 int newline;
363 pcre_config(PCRE_CONFIG_NEWLINE, &newline);
364 if (newline == 10)
365 newline_mode = PCRE_NEWLINE_LF;
366 else if (newline == 13)
367 newline_mode = PCRE_NEWLINE_CR;
368 else if (newline == 3338)
369 newline_mode = PCRE_NEWLINE_CRLF;
370 else if (newline == -1)
371 newline_mode = PCRE_NEWLINE_ANY;
372 else if (newline == -2)
373 newline_mode = PCRE_NEWLINE_ANYCRLF;
374 else
375 assert(NULL == "Unexpected return value from pcre_config(NEWLINE)");
376 }
377 return newline_mode;
378 }
379
380 int RE::GlobalReplace(const StringPiece& rewrite,
381 string *str) const {
382 int count = 0;
383 int vec[kVecSize];
384 string out;
385 int start = 0;
386 int lastend = -1;
387
388 while (start <= static_cast<int>(str->length())) {
389 int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
390 if (matches <= 0)
391 break;
392 int matchstart = vec[0], matchend = vec[1];
393 assert(matchstart >= start);
394 assert(matchend >= matchstart);
395 if (matchstart == matchend && matchstart == lastend) {
396 // advance one character if we matched an empty string at the same
397 // place as the last match occurred
398 matchend = start + 1;
399 // If the current char is CR and we're in CRLF mode, skip LF too.
400 // Note it's better to call pcre_fullinfo() than to examine
401 // all_options(), since options_ could have changed bewteen
402 // compile-time and now, but this is simpler and safe enough.
403 // Modified by PH to add ANY and ANYCRLF.
404 if (start+1 < static_cast<int>(str->length()) &&
405 (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
406 (NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF ||
407 NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANY ||
408 NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANYCRLF)
409 ) {
410 matchend++;
411 }
412 // We also need to advance more than one char if we're in utf8 mode.
413 #ifdef SUPPORT_UTF8
414 if (options_.utf8()) {
415 while (matchend < static_cast<int>(str->length()) &&
416 ((*str)[matchend] & 0xc0) == 0x80)
417 matchend++;
418 }
419 #endif
420 if (matchend <= static_cast<int>(str->length()))
421 out.append(*str, start, matchend - start);
422 start = matchend;
423 } else {
424 out.append(*str, start, matchstart - start);
425 Rewrite(&out, rewrite, *str, vec, matches);
426 start = matchend;
427 lastend = matchend;
428 count++;
429 }
430 }
431
432 if (count == 0)
433 return 0;
434
435 if (start < static_cast<int>(str->length()))
436 out.append(*str, start, str->length() - start);
437 swap(out, *str);
438 return count;
439 }
440
441 bool RE::Extract(const StringPiece& rewrite,
442 const StringPiece& text,
443 string *out) const {
444 int vec[kVecSize];
445 int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
446 if (matches == 0)
447 return false;
448 out->erase();
449 return Rewrite(out, rewrite, text, vec, matches);
450 }
451
452 /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
453 string result;
454
455 // Escape any ascii character not in [A-Za-z_0-9].
456 //
457 // Note that it's legal to escape a character even if it has no
458 // special meaning in a regular expression -- so this function does
459 // that. (This also makes it identical to the perl function of the
460 // same name; see `perldoc -f quotemeta`.) The one exception is
461 // escaping NUL: rather than doing backslash + NUL, like perl does,
462 // we do '\0', because pcre itself doesn't take embedded NUL chars.
463 for (int ii = 0; ii < unquoted.size(); ++ii) {
464 // Note that using 'isalnum' here raises the benchmark time from
465 // 32ns to 58ns:
466 if (unquoted[ii] == '\0') {
467 result += "\\0";
468 } else if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
469 (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
470 (unquoted[ii] < '0' || unquoted[ii] > '9') &&
471 unquoted[ii] != '_' &&
472 // If this is the part of a UTF8 or Latin1 character, we need
473 // to copy this byte without escaping. Experimentally this is
474 // what works correctly with the regexp library.
475 !(unquoted[ii] & 128)) {
476 result += '\\';
477 result += unquoted[ii];
478 } else {
479 result += unquoted[ii];
480 }
481 }
482
483 return result;
484 }
485
486 /***** Actual matching and rewriting code *****/
487
488 int RE::TryMatch(const StringPiece& text,
489 int startpos,
490 Anchor anchor,
491 int *vec,
492 int vecsize) const {
493 pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
494 if (re == NULL) {
495 //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
496 return 0;
497 }
498
499 pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
500 if (options_.match_limit() > 0) {
501 extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
502 extra.match_limit = options_.match_limit();
503 }
504 if (options_.match_limit_recursion() > 0) {
505 extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
506 extra.match_limit_recursion = options_.match_limit_recursion();
507 }
508 int rc = pcre_exec(re, // The regular expression object
509 &extra,
510 (text.data() == NULL) ? "" : text.data(),
511 text.size(),
512 startpos,
513 (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
514 vec,
515 vecsize);
516
517 // Handle errors
518 if (rc == PCRE_ERROR_NOMATCH) {
519 return 0;
520 } else if (rc < 0) {
521 //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
522 // re, pattern_.c_str());
523 return 0;
524 } else if (rc == 0) {
525 // pcre_exec() returns 0 as a special case when the number of
526 // capturing subpatterns exceeds the size of the vector.
527 // When this happens, there is a match and the output vector
528 // is filled, but we miss out on the positions of the extra subpatterns.
529 rc = vecsize / 2;
530 }
531
532 return rc;
533 }
534
535 bool RE::DoMatchImpl(const StringPiece& text,
536 Anchor anchor,
537 int* consumed,
538 const Arg* const* args,
539 int n,
540 int* vec,
541 int vecsize) const {
542 assert((1 + n) * 3 <= vecsize); // results + PCRE workspace
543 int matches = TryMatch(text, 0, anchor, vec, vecsize);
544 assert(matches >= 0); // TryMatch never returns negatives
545 if (matches == 0)
546 return false;
547
548 *consumed = vec[1];
549
550 if (n == 0 || args == NULL) {
551 // We are not interested in results
552 return true;
553 }
554
555 if (NumberOfCapturingGroups() < n) {
556 // RE has fewer capturing groups than number of arg pointers passed in
557 return false;
558 }
559
560 // If we got here, we must have matched the whole pattern.
561 // We do not need (can not do) any more checks on the value of 'matches' here
562 // -- see the comment for TryMatch.
563 for (int i = 0; i < n; i++) {
564 const int start = vec[2*(i+1)];
565 const int limit = vec[2*(i+1)+1];
566 if (!args[i]->Parse(text.data() + start, limit-start)) {
567 // TODO: Should we indicate what the error was?
568 return false;
569 }
570 }
571
572 return true;
573 }
574
575 bool RE::DoMatch(const StringPiece& text,
576 Anchor anchor,
577 int* consumed,
578 const Arg* const args[],
579 int n) const {
580 assert(n >= 0);
581 size_t const vecsize = (1 + n) * 3; // results + PCRE workspace
582 // (as for kVecSize)
583 int space[21]; // use stack allocation for small vecsize (common case)
584 int* vec = vecsize <= 21 ? space : new int[vecsize];
585 bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
586 if (vec != space) delete [] vec;
587 return retval;
588 }
589
590 bool RE::Rewrite(string *out, const StringPiece &rewrite,
591 const StringPiece &text, int *vec, int veclen) const {
592 for (const char *s = rewrite.data(), *end = s + rewrite.size();
593 s < end; s++) {
594 int c = *s;
595 if (c == '\\') {
596 c = *++s;
597 if (isdigit(c)) {
598 int n = (c - '0');
599 if (n >= veclen) {
600 //fprintf(stderr, requested group %d in regexp %.*s\n",
601 // n, rewrite.size(), rewrite.data());
602 return false;
603 }
604 int start = vec[2 * n];
605 if (start >= 0)
606 out->append(text.data() + start, vec[2 * n + 1] - start);
607 } else if (c == '\\') {
608 *out += '\\';
609 } else {
610 //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
611 // rewrite.size(), rewrite.data());
612 return false;
613 }
614 } else {
615 *out += c;
616 }
617 }
618 return true;
619 }
620
621 // Return the number of capturing subpatterns, or -1 if the
622 // regexp wasn't valid on construction.
623 int RE::NumberOfCapturingGroups() const {
624 if (re_partial_ == NULL) return -1;
625
626 int result;
627 int pcre_retval = pcre_fullinfo(re_partial_, // The regular expression object
628 NULL, // We did not study the pattern
629 PCRE_INFO_CAPTURECOUNT,
630 &result);
631 assert(pcre_retval == 0);
632 return result;
633 }
634
635 /***** Parsers for various types *****/
636
637 bool Arg::parse_null(const char* str, int n, void* dest) {
638 // We fail if somebody asked us to store into a non-NULL void* pointer
639 return (dest == NULL);
640 }
641
642 bool Arg::parse_string(const char* str, int n, void* dest) {
643 if (dest == NULL) return true;
644 reinterpret_cast<string*>(dest)->assign(str, n);
645 return true;
646 }
647
648 bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
649 if (dest == NULL) return true;
650 reinterpret_cast<StringPiece*>(dest)->set(str, n);
651 return true;
652 }
653
654 bool Arg::parse_char(const char* str, int n, void* dest) {
655 if (n != 1) return false;
656 if (dest == NULL) return true;
657 *(reinterpret_cast<char*>(dest)) = str[0];
658 return true;
659 }
660
661 bool Arg::parse_uchar(const char* str, int n, void* dest) {
662 if (n != 1) return false;
663 if (dest == NULL) return true;
664 *(reinterpret_cast<unsigned char*>(dest)) = str[0];
665 return true;
666 }
667
668 // Largest number spec that we are willing to parse
669 static const int kMaxNumberLength = 32;
670
671 // REQUIRES "buf" must have length at least kMaxNumberLength+1
672 // REQUIRES "n > 0"
673 // Copies "str" into "buf" and null-terminates if necessary.
674 // Returns one of:
675 // a. "str" if no termination is needed
676 // b. "buf" if the string was copied and null-terminated
677 // c. "" if the input was invalid and has no hope of being parsed
678 static const char* TerminateNumber(char* buf, const char* str, int n) {
679 if ((n > 0) && isspace(*str)) {
680 // We are less forgiving than the strtoxxx() routines and do not
681 // allow leading spaces.
682 return "";
683 }
684
685 // See if the character right after the input text may potentially
686 // look like a digit.
687 if (isdigit(str[n]) ||
688 ((str[n] >= 'a') && (str[n] <= 'f')) ||
689 ((str[n] >= 'A') && (str[n] <= 'F'))) {
690 if (n > kMaxNumberLength) return ""; // Input too big to be a valid number
691 memcpy(buf, str, n);
692 buf[n] = '\0';
693 return buf;
694 } else {
695 // We can parse right out of the supplied string, so return it.
696 return str;
697 }
698 }
699
700 bool Arg::parse_long_radix(const char* str,
701 int n,
702 void* dest,
703 int radix) {
704 if (n == 0) return false;
705 char buf[kMaxNumberLength+1];
706 str = TerminateNumber(buf, str, n);
707 char* end;
708 errno = 0;
709 long r = strtol(str, &end, radix);
710 if (end != str + n) return false; // Leftover junk
711 if (errno) return false;
712 if (dest == NULL) return true;
713 *(reinterpret_cast<long*>(dest)) = r;
714 return true;
715 }
716
717 bool Arg::parse_ulong_radix(const char* str,
718 int n,
719 void* dest,
720 int radix) {
721 if (n == 0) return false;
722 char buf[kMaxNumberLength+1];
723 str = TerminateNumber(buf, str, n);
724 if (str[0] == '-') return false; // strtoul() on a negative number?!
725 char* end;
726 errno = 0;
727 unsigned long r = strtoul(str, &end, radix);
728 if (end != str + n) return false; // Leftover junk
729 if (errno) return false;
730 if (dest == NULL) return true;
731 *(reinterpret_cast<unsigned long*>(dest)) = r;
732 return true;
733 }
734
735 bool Arg::parse_short_radix(const char* str,
736 int n,
737 void* dest,
738 int radix) {
739 long r;
740 if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
741 if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range
742 if (dest == NULL) return true;
743 *(reinterpret_cast<short*>(dest)) = static_cast<short>(r);
744 return true;
745 }
746
747 bool Arg::parse_ushort_radix(const char* str,
748 int n,
749 void* dest,
750 int radix) {
751 unsigned long r;
752 if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
753 if (r > USHRT_MAX) return false; // Out of range
754 if (dest == NULL) return true;
755 *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r);
756 return true;
757 }
758
759 bool Arg::parse_int_radix(const char* str,
760 int n,
761 void* dest,
762 int radix) {
763 long r;
764 if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
765 if (r < INT_MIN || r > INT_MAX) return false; // Out of range
766 if (dest == NULL) return true;
767 *(reinterpret_cast<int*>(dest)) = r;
768 return true;
769 }
770
771 bool Arg::parse_uint_radix(const char* str,
772 int n,
773 void* dest,
774 int radix) {
775 unsigned long r;
776 if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
777 if (r > UINT_MAX) return false; // Out of range
778 if (dest == NULL) return true;
779 *(reinterpret_cast<unsigned int*>(dest)) = r;
780 return true;
781 }
782
783 bool Arg::parse_longlong_radix(const char* str,
784 int n,
785 void* dest,
786 int radix) {
787 #ifndef HAVE_LONG_LONG
788 return false;
789 #else
790 if (n == 0) return false;
791 char buf[kMaxNumberLength+1];
792 str = TerminateNumber(buf, str, n);
793 char* end;
794 errno = 0;
795 #if defined HAVE_STRTOQ
796 long long r = strtoq(str, &end, radix);
797 #elif defined HAVE_STRTOLL
798 long long r = strtoll(str, &end, radix);
799 #elif defined HAVE__STRTOI64
800 long long r = _strtoi64(str, &end, radix);
801 #else
802 #error parse_longlong_radix: cannot convert input to a long-long
803 #endif
804 if (end != str + n) return false; // Leftover junk
805 if (errno) return false;
806 if (dest == NULL) return true;
807 *(reinterpret_cast<long long*>(dest)) = r;
808 return true;
809 #endif /* HAVE_LONG_LONG */
810 }
811
812 bool Arg::parse_ulonglong_radix(const char* str,
813 int n,
814 void* dest,
815 int radix) {
816 #ifndef HAVE_UNSIGNED_LONG_LONG
817 return false;
818 #else
819 if (n == 0) return false;
820 char buf[kMaxNumberLength+1];
821 str = TerminateNumber(buf, str, n);
822 if (str[0] == '-') return false; // strtoull() on a negative number?!
823 char* end;
824 errno = 0;
825 #if defined HAVE_STRTOQ
826 unsigned long long r = strtouq(str, &end, radix);
827 #elif defined HAVE_STRTOLL
828 unsigned long long r = strtoull(str, &end, radix);
829 #elif defined HAVE__STRTOI64
830 unsigned long long r = _strtoui64(str, &end, radix);
831 #else
832 #error parse_ulonglong_radix: cannot convert input to a long-long
833 #endif
834 if (end != str + n) return false; // Leftover junk
835 if (errno) return false;
836 if (dest == NULL) return true;
837 *(reinterpret_cast<unsigned long long*>(dest)) = r;
838 return true;
839 #endif /* HAVE_UNSIGNED_LONG_LONG */
840 }
841
842 bool Arg::parse_double(const char* str, int n, void* dest) {
843 if (n == 0) return false;
844 static const int kMaxLength = 200;
845 char buf[kMaxLength];
846 if (n >= kMaxLength) return false;
847 memcpy(buf, str, n);
848 buf[n] = '\0';
849 errno = 0;
850 char* end;
851 double r = strtod(buf, &end);
852 if (end != buf + n) return false; // Leftover junk
853 if (errno) return false;
854 if (dest == NULL) return true;
855 *(reinterpret_cast<double*>(dest)) = r;
856 return true;
857 }
858
859 bool Arg::parse_float(const char* str, int n, void* dest) {
860 double r;
861 if (!parse_double(str, n, &r)) return false;
862 if (dest == NULL) return true;
863 *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
864 return true;
865 }
866
867
868 #define DEFINE_INTEGER_PARSERS(name) \
869 bool Arg::parse_##name(const char* str, int n, void* dest) { \
870 return parse_##name##_radix(str, n, dest, 10); \
871 } \
872 bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \
873 return parse_##name##_radix(str, n, dest, 16); \
874 } \
875 bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \
876 return parse_##name##_radix(str, n, dest, 8); \
877 } \
878 bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
879 return parse_##name##_radix(str, n, dest, 0); \
880 }
881
882 DEFINE_INTEGER_PARSERS(short) /* */
883 DEFINE_INTEGER_PARSERS(ushort) /* */
884 DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */
885 DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */
886 DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */
887 DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */
888 DEFINE_INTEGER_PARSERS(longlong) /* */
889 DEFINE_INTEGER_PARSERS(ulonglong) /* */
890
891 #undef DEFINE_INTEGER_PARSERS
892
893 } // namespace pcrecpp

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12