/[pcre]/code/branches/oldtrunk8/pcrecpp.cc
ViewVC logotype

Contents of /code/branches/oldtrunk8/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 328 - (show annotations)
Wed Mar 26 17:39:06 2008 UTC (7 years, 5 months ago) by ph10
Original Path: code/trunk/pcrecpp.cc
File size: 31487 byte(s)
Error occurred while calculating annotation data.
Patch to pcrecpp.cc to fix an OS X linking problem.
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. We could probably
63 // have a more inclusive test if we ever needed it. (Note that not
64 // only the __attribute__ syntax, but also __USER_LABEL_PREFIX__, are
65 // gnu-specific.)
66 #if defined(__GNUC__) && __GNUC__ >= 3
67 #if defined(__ELF__)
68 extern Arg no_arg
69 __attribute__((alias(__USER_LABEL_PREFIX__ "_ZN7pcrecpp2RE6no_argE")));
70 #else
71 // While we know elf supports strong aliases, not all formats do (Mach
72 // doesn't, for instance). So make aliases weak by default. This is
73 // a smidge less safe in theory (conceivably, someone could override
74 // this symbol in their own binary), but perfectly ok in practice.
75 extern Arg no_arg
76 __attribute__((weak, alias(__USER_LABEL_PREFIX__ "_ZN7pcrecpp2RE6no_argE")));
77 #endif
78 #endif
79
80 // If a regular expression has no error, its error_ field points here
81 static const string empty_string;
82
83 // If the user doesn't ask for any options, we just use this one
84 static RE_Options default_options;
85
86 void RE::Init(const string& pat, const RE_Options* options) {
87 pattern_ = pat;
88 if (options == NULL) {
89 options_ = default_options;
90 } else {
91 options_ = *options;
92 }
93 error_ = &empty_string;
94 re_full_ = NULL;
95 re_partial_ = NULL;
96
97 re_partial_ = Compile(UNANCHORED);
98 if (re_partial_ != NULL) {
99 re_full_ = Compile(ANCHOR_BOTH);
100 }
101 }
102
103 void RE::Cleanup() {
104 if (re_full_ != NULL) (*pcre_free)(re_full_);
105 if (re_partial_ != NULL) (*pcre_free)(re_partial_);
106 if (error_ != &empty_string) delete error_;
107 }
108
109
110 RE::~RE() {
111 Cleanup();
112 }
113
114
115 pcre* RE::Compile(Anchor anchor) {
116 // First, convert RE_Options into pcre options
117 int pcre_options = 0;
118 pcre_options = options_.all_options();
119
120 // Special treatment for anchoring. This is needed because at
121 // runtime pcre only provides an option for anchoring at the
122 // beginning of a string (unless you use offset).
123 //
124 // There are three types of anchoring we want:
125 // UNANCHORED Compile the original pattern, and use
126 // a pcre unanchored match.
127 // ANCHOR_START Compile the original pattern, and use
128 // a pcre anchored match.
129 // ANCHOR_BOTH Tack a "\z" to the end of the original pattern
130 // and use a pcre anchored match.
131
132 const char* compile_error;
133 int eoffset;
134 pcre* re;
135 if (anchor != ANCHOR_BOTH) {
136 re = pcre_compile(pattern_.c_str(), pcre_options,
137 &compile_error, &eoffset, NULL);
138 } else {
139 // Tack a '\z' at the end of RE. Parenthesize it first so that
140 // the '\z' applies to all top-level alternatives in the regexp.
141 string wrapped = "(?:"; // A non-counting grouping operator
142 wrapped += pattern_;
143 wrapped += ")\\z";
144 re = pcre_compile(wrapped.c_str(), pcre_options,
145 &compile_error, &eoffset, NULL);
146 }
147 if (re == NULL) {
148 if (error_ == &empty_string) error_ = new string(compile_error);
149 }
150 return re;
151 }
152
153 /***** Matching interfaces *****/
154
155 bool RE::FullMatch(const StringPiece& text,
156 const Arg& ptr1,
157 const Arg& ptr2,
158 const Arg& ptr3,
159 const Arg& ptr4,
160 const Arg& ptr5,
161 const Arg& ptr6,
162 const Arg& ptr7,
163 const Arg& ptr8,
164 const Arg& ptr9,
165 const Arg& ptr10,
166 const Arg& ptr11,
167 const Arg& ptr12,
168 const Arg& ptr13,
169 const Arg& ptr14,
170 const Arg& ptr15,
171 const Arg& ptr16) const {
172 const Arg* args[kMaxArgs];
173 int n = 0;
174 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
175 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
176 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
177 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
178 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
179 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
180 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
181 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
182 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
183 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
184 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
185 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
186 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
187 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
188 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
189 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
190 done:
191
192 int consumed;
193 int vec[kVecSize];
194 return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
195 }
196
197 bool RE::PartialMatch(const StringPiece& text,
198 const Arg& ptr1,
199 const Arg& ptr2,
200 const Arg& ptr3,
201 const Arg& ptr4,
202 const Arg& ptr5,
203 const Arg& ptr6,
204 const Arg& ptr7,
205 const Arg& ptr8,
206 const Arg& ptr9,
207 const Arg& ptr10,
208 const Arg& ptr11,
209 const Arg& ptr12,
210 const Arg& ptr13,
211 const Arg& ptr14,
212 const Arg& ptr15,
213 const Arg& ptr16) const {
214 const Arg* args[kMaxArgs];
215 int n = 0;
216 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
217 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
218 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
219 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
220 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
221 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
222 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
223 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
224 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
225 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
226 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
227 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
228 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
229 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
230 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
231 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
232 done:
233
234 int consumed;
235 int vec[kVecSize];
236 return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
237 }
238
239 bool RE::Consume(StringPiece* input,
240 const Arg& ptr1,
241 const Arg& ptr2,
242 const Arg& ptr3,
243 const Arg& ptr4,
244 const Arg& ptr5,
245 const Arg& ptr6,
246 const Arg& ptr7,
247 const Arg& ptr8,
248 const Arg& ptr9,
249 const Arg& ptr10,
250 const Arg& ptr11,
251 const Arg& ptr12,
252 const Arg& ptr13,
253 const Arg& ptr14,
254 const Arg& ptr15,
255 const Arg& ptr16) const {
256 const Arg* args[kMaxArgs];
257 int n = 0;
258 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
259 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
260 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
261 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
262 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
263 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
264 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
265 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
266 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
267 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
268 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
269 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
270 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
271 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
272 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
273 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
274 done:
275
276 int consumed;
277 int vec[kVecSize];
278 if (DoMatchImpl(*input, ANCHOR_START, &consumed,
279 args, n, vec, kVecSize)) {
280 input->remove_prefix(consumed);
281 return true;
282 } else {
283 return false;
284 }
285 }
286
287 bool RE::FindAndConsume(StringPiece* input,
288 const Arg& ptr1,
289 const Arg& ptr2,
290 const Arg& ptr3,
291 const Arg& ptr4,
292 const Arg& ptr5,
293 const Arg& ptr6,
294 const Arg& ptr7,
295 const Arg& ptr8,
296 const Arg& ptr9,
297 const Arg& ptr10,
298 const Arg& ptr11,
299 const Arg& ptr12,
300 const Arg& ptr13,
301 const Arg& ptr14,
302 const Arg& ptr15,
303 const Arg& ptr16) const {
304 const Arg* args[kMaxArgs];
305 int n = 0;
306 if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
307 if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
308 if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
309 if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
310 if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
311 if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
312 if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
313 if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
314 if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
315 if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
316 if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
317 if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
318 if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
319 if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
320 if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
321 if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
322 done:
323
324 int consumed;
325 int vec[kVecSize];
326 if (DoMatchImpl(*input, UNANCHORED, &consumed,
327 args, n, vec, kVecSize)) {
328 input->remove_prefix(consumed);
329 return true;
330 } else {
331 return false;
332 }
333 }
334
335 bool RE::Replace(const StringPiece& rewrite,
336 string *str) const {
337 int vec[kVecSize];
338 int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
339 if (matches == 0)
340 return false;
341
342 string s;
343 if (!Rewrite(&s, rewrite, *str, vec, matches))
344 return false;
345
346 assert(vec[0] >= 0);
347 assert(vec[1] >= 0);
348 str->replace(vec[0], vec[1] - vec[0], s);
349 return true;
350 }
351
352 // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
353 // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
354 // Modified by PH to add PCRE_NEWLINE_ANY and PCRE_NEWLINE_ANYCRLF.
355
356 static int NewlineMode(int pcre_options) {
357 // TODO: if we can make it threadsafe, cache this var
358 int newline_mode = 0;
359 /* if (newline_mode) return newline_mode; */ // do this once it's cached
360 if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
361 PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF)) {
362 newline_mode = (pcre_options &
363 (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
364 PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF));
365 } else {
366 int newline;
367 pcre_config(PCRE_CONFIG_NEWLINE, &newline);
368 if (newline == 10)
369 newline_mode = PCRE_NEWLINE_LF;
370 else if (newline == 13)
371 newline_mode = PCRE_NEWLINE_CR;
372 else if (newline == 3338)
373 newline_mode = PCRE_NEWLINE_CRLF;
374 else if (newline == -1)
375 newline_mode = PCRE_NEWLINE_ANY;
376 else if (newline == -2)
377 newline_mode = PCRE_NEWLINE_ANYCRLF;
378 else
379 assert("" == "Unexpected return value from pcre_config(NEWLINE)");
380 }
381 return newline_mode;
382 }
383
384 int RE::GlobalReplace(const StringPiece& rewrite,
385 string *str) const {
386 int count = 0;
387 int vec[kVecSize];
388 string out;
389 int start = 0;
390 int lastend = -1;
391
392 while (start <= static_cast<int>(str->length())) {
393 int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
394 if (matches <= 0)
395 break;
396 int matchstart = vec[0], matchend = vec[1];
397 assert(matchstart >= start);
398 assert(matchend >= matchstart);
399 if (matchstart == matchend && matchstart == lastend) {
400 // advance one character if we matched an empty string at the same
401 // place as the last match occurred
402 matchend = start + 1;
403 // If the current char is CR and we're in CRLF mode, skip LF too.
404 // Note it's better to call pcre_fullinfo() than to examine
405 // all_options(), since options_ could have changed bewteen
406 // compile-time and now, but this is simpler and safe enough.
407 // Modified by PH to add ANY and ANYCRLF.
408 if (start+1 < static_cast<int>(str->length()) &&
409 (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
410 (NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF ||
411 NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANY ||
412 NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANYCRLF)
413 ) {
414 matchend++;
415 }
416 // We also need to advance more than one char if we're in utf8 mode.
417 #ifdef SUPPORT_UTF8
418 if (options_.utf8()) {
419 while (matchend < static_cast<int>(str->length()) &&
420 ((*str)[matchend] & 0xc0) == 0x80)
421 matchend++;
422 }
423 #endif
424 if (matchend <= static_cast<int>(str->length()))
425 out.append(*str, start, matchend - start);
426 start = matchend;
427 } else {
428 out.append(*str, start, matchstart - start);
429 Rewrite(&out, rewrite, *str, vec, matches);
430 start = matchend;
431 lastend = matchend;
432 count++;
433 }
434 }
435
436 if (count == 0)
437 return 0;
438
439 if (start < static_cast<int>(str->length()))
440 out.append(*str, start, str->length() - start);
441 swap(out, *str);
442 return count;
443 }
444
445 bool RE::Extract(const StringPiece& rewrite,
446 const StringPiece& text,
447 string *out) const {
448 int vec[kVecSize];
449 int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
450 if (matches == 0)
451 return false;
452 out->erase();
453 return Rewrite(out, rewrite, text, vec, matches);
454 }
455
456 /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
457 string result;
458
459 // Escape any ascii character not in [A-Za-z_0-9].
460 //
461 // Note that it's legal to escape a character even if it has no
462 // special meaning in a regular expression -- so this function does
463 // that. (This also makes it identical to the perl function of the
464 // same name; see `perldoc -f quotemeta`.) The one exception is
465 // escaping NUL: rather than doing backslash + NUL, like perl does,
466 // we do '\0', because pcre itself doesn't take embedded NUL chars.
467 for (int ii = 0; ii < unquoted.size(); ++ii) {
468 // Note that using 'isalnum' here raises the benchmark time from
469 // 32ns to 58ns:
470 if (unquoted[ii] == '\0') {
471 result += "\\0";
472 } else if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
473 (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
474 (unquoted[ii] < '0' || unquoted[ii] > '9') &&
475 unquoted[ii] != '_' &&
476 // If this is the part of a UTF8 or Latin1 character, we need
477 // to copy this byte without escaping. Experimentally this is
478 // what works correctly with the regexp library.
479 !(unquoted[ii] & 128)) {
480 result += '\\';
481 result += unquoted[ii];
482 } else {
483 result += unquoted[ii];
484 }
485 }
486
487 return result;
488 }
489
490 /***** Actual matching and rewriting code *****/
491
492 int RE::TryMatch(const StringPiece& text,
493 int startpos,
494 Anchor anchor,
495 int *vec,
496 int vecsize) const {
497 pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
498 if (re == NULL) {
499 //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
500 return 0;
501 }
502
503 pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
504 if (options_.match_limit() > 0) {
505 extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
506 extra.match_limit = options_.match_limit();
507 }
508 if (options_.match_limit_recursion() > 0) {
509 extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
510 extra.match_limit_recursion = options_.match_limit_recursion();
511 }
512 int rc = pcre_exec(re, // The regular expression object
513 &extra,
514 (text.data() == NULL) ? "" : text.data(),
515 text.size(),
516 startpos,
517 (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
518 vec,
519 vecsize);
520
521 // Handle errors
522 if (rc == PCRE_ERROR_NOMATCH) {
523 return 0;
524 } else if (rc < 0) {
525 //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
526 // re, pattern_.c_str());
527 return 0;
528 } else if (rc == 0) {
529 // pcre_exec() returns 0 as a special case when the number of
530 // capturing subpatterns exceeds the size of the vector.
531 // When this happens, there is a match and the output vector
532 // is filled, but we miss out on the positions of the extra subpatterns.
533 rc = vecsize / 2;
534 }
535
536 return rc;
537 }
538
539 bool RE::DoMatchImpl(const StringPiece& text,
540 Anchor anchor,
541 int* consumed,
542 const Arg* const* args,
543 int n,
544 int* vec,
545 int vecsize) const {
546 assert((1 + n) * 3 <= vecsize); // results + PCRE workspace
547 int matches = TryMatch(text, 0, anchor, vec, vecsize);
548 assert(matches >= 0); // TryMatch never returns negatives
549 if (matches == 0)
550 return false;
551
552 *consumed = vec[1];
553
554 if (n == 0 || args == NULL) {
555 // We are not interested in results
556 return true;
557 }
558
559 if (NumberOfCapturingGroups() < n) {
560 // RE has fewer capturing groups than number of arg pointers passed in
561 return false;
562 }
563
564 // If we got here, we must have matched the whole pattern.
565 // We do not need (can not do) any more checks on the value of 'matches' here
566 // -- see the comment for TryMatch.
567 for (int i = 0; i < n; i++) {
568 const int start = vec[2*(i+1)];
569 const int limit = vec[2*(i+1)+1];
570 if (!args[i]->Parse(text.data() + start, limit-start)) {
571 // TODO: Should we indicate what the error was?
572 return false;
573 }
574 }
575
576 return true;
577 }
578
579 bool RE::DoMatch(const StringPiece& text,
580 Anchor anchor,
581 int* consumed,
582 const Arg* const args[],
583 int n) const {
584 assert(n >= 0);
585 size_t const vecsize = (1 + n) * 3; // results + PCRE workspace
586 // (as for kVecSize)
587 int space[21]; // use stack allocation for small vecsize (common case)
588 int* vec = vecsize <= 21 ? space : new int[vecsize];
589 bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
590 if (vec != space) delete [] vec;
591 return retval;
592 }
593
594 bool RE::Rewrite(string *out, const StringPiece &rewrite,
595 const StringPiece &text, int *vec, int veclen) const {
596 for (const char *s = rewrite.data(), *end = s + rewrite.size();
597 s < end; s++) {
598 int c = *s;
599 if (c == '\\') {
600 c = *++s;
601 if (isdigit(c)) {
602 int n = (c - '0');
603 if (n >= veclen) {
604 //fprintf(stderr, requested group %d in regexp %.*s\n",
605 // n, rewrite.size(), rewrite.data());
606 return false;
607 }
608 int start = vec[2 * n];
609 if (start >= 0)
610 out->append(text.data() + start, vec[2 * n + 1] - start);
611 } else if (c == '\\') {
612 out->push_back('\\');
613 } else {
614 //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
615 // rewrite.size(), rewrite.data());
616 return false;
617 }
618 } else {
619 out->push_back(c);
620 }
621 }
622 return true;
623 }
624
625 // Return the number of capturing subpatterns, or -1 if the
626 // regexp wasn't valid on construction.
627 int RE::NumberOfCapturingGroups() const {
628 if (re_partial_ == NULL) return -1;
629
630 int result;
631 int pcre_retval = pcre_fullinfo(re_partial_, // The regular expression object
632 NULL, // We did not study the pattern
633 PCRE_INFO_CAPTURECOUNT,
634 &result);
635 assert(pcre_retval == 0);
636 return result;
637 }
638
639 /***** Parsers for various types *****/
640
641 bool Arg::parse_null(const char* str, int n, void* dest) {
642 // We fail if somebody asked us to store into a non-NULL void* pointer
643 return (dest == NULL);
644 }
645
646 bool Arg::parse_string(const char* str, int n, void* dest) {
647 if (dest == NULL) return true;
648 reinterpret_cast<string*>(dest)->assign(str, n);
649 return true;
650 }
651
652 bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
653 if (dest == NULL) return true;
654 reinterpret_cast<StringPiece*>(dest)->set(str, n);
655 return true;
656 }
657
658 bool Arg::parse_char(const char* str, int n, void* dest) {
659 if (n != 1) return false;
660 if (dest == NULL) return true;
661 *(reinterpret_cast<char*>(dest)) = str[0];
662 return true;
663 }
664
665 bool Arg::parse_uchar(const char* str, int n, void* dest) {
666 if (n != 1) return false;
667 if (dest == NULL) return true;
668 *(reinterpret_cast<unsigned char*>(dest)) = str[0];
669 return true;
670 }
671
672 // Largest number spec that we are willing to parse
673 static const int kMaxNumberLength = 32;
674
675 // REQUIRES "buf" must have length at least kMaxNumberLength+1
676 // REQUIRES "n > 0"
677 // Copies "str" into "buf" and null-terminates if necessary.
678 // Returns one of:
679 // a. "str" if no termination is needed
680 // b. "buf" if the string was copied and null-terminated
681 // c. "" if the input was invalid and has no hope of being parsed
682 static const char* TerminateNumber(char* buf, const char* str, int n) {
683 if ((n > 0) && isspace(*str)) {
684 // We are less forgiving than the strtoxxx() routines and do not
685 // allow leading spaces.
686 return "";
687 }
688
689 // See if the character right after the input text may potentially
690 // look like a digit.
691 if (isdigit(str[n]) ||
692 ((str[n] >= 'a') && (str[n] <= 'f')) ||
693 ((str[n] >= 'A') && (str[n] <= 'F'))) {
694 if (n > kMaxNumberLength) return ""; // Input too big to be a valid number
695 memcpy(buf, str, n);
696 buf[n] = '\0';
697 return buf;
698 } else {
699 // We can parse right out of the supplied string, so return it.
700 return str;
701 }
702 }
703
704 bool Arg::parse_long_radix(const char* str,
705 int n,
706 void* dest,
707 int radix) {
708 if (n == 0) return false;
709 char buf[kMaxNumberLength+1];
710 str = TerminateNumber(buf, str, n);
711 char* end;
712 errno = 0;
713 long r = strtol(str, &end, radix);
714 if (end != str + n) return false; // Leftover junk
715 if (errno) return false;
716 if (dest == NULL) return true;
717 *(reinterpret_cast<long*>(dest)) = r;
718 return true;
719 }
720
721 bool Arg::parse_ulong_radix(const char* str,
722 int n,
723 void* dest,
724 int radix) {
725 if (n == 0) return false;
726 char buf[kMaxNumberLength+1];
727 str = TerminateNumber(buf, str, n);
728 if (str[0] == '-') return false; // strtoul() on a negative number?!
729 char* end;
730 errno = 0;
731 unsigned long r = strtoul(str, &end, radix);
732 if (end != str + n) return false; // Leftover junk
733 if (errno) return false;
734 if (dest == NULL) return true;
735 *(reinterpret_cast<unsigned long*>(dest)) = r;
736 return true;
737 }
738
739 bool Arg::parse_short_radix(const char* str,
740 int n,
741 void* dest,
742 int radix) {
743 long r;
744 if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
745 if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range
746 if (dest == NULL) return true;
747 *(reinterpret_cast<short*>(dest)) = static_cast<short>(r);
748 return true;
749 }
750
751 bool Arg::parse_ushort_radix(const char* str,
752 int n,
753 void* dest,
754 int radix) {
755 unsigned long r;
756 if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
757 if (r > USHRT_MAX) return false; // Out of range
758 if (dest == NULL) return true;
759 *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r);
760 return true;
761 }
762
763 bool Arg::parse_int_radix(const char* str,
764 int n,
765 void* dest,
766 int radix) {
767 long r;
768 if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
769 if (r < INT_MIN || r > INT_MAX) return false; // Out of range
770 if (dest == NULL) return true;
771 *(reinterpret_cast<int*>(dest)) = r;
772 return true;
773 }
774
775 bool Arg::parse_uint_radix(const char* str,
776 int n,
777 void* dest,
778 int radix) {
779 unsigned long r;
780 if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
781 if (r > UINT_MAX) return false; // Out of range
782 if (dest == NULL) return true;
783 *(reinterpret_cast<unsigned int*>(dest)) = r;
784 return true;
785 }
786
787 bool Arg::parse_longlong_radix(const char* str,
788 int n,
789 void* dest,
790 int radix) {
791 #ifndef HAVE_LONG_LONG
792 return false;
793 #else
794 if (n == 0) return false;
795 char buf[kMaxNumberLength+1];
796 str = TerminateNumber(buf, str, n);
797 char* end;
798 errno = 0;
799 #if defined HAVE_STRTOQ
800 long long r = strtoq(str, &end, radix);
801 #elif defined HAVE_STRTOLL
802 long long r = strtoll(str, &end, radix);
803 #elif defined HAVE__STRTOI64
804 long long r = _strtoi64(str, &end, radix);
805 #else
806 #error parse_longlong_radix: cannot convert input to a long-long
807 #endif
808 if (end != str + n) return false; // Leftover junk
809 if (errno) return false;
810 if (dest == NULL) return true;
811 *(reinterpret_cast<long long*>(dest)) = r;
812 return true;
813 #endif /* HAVE_LONG_LONG */
814 }
815
816 bool Arg::parse_ulonglong_radix(const char* str,
817 int n,
818 void* dest,
819 int radix) {
820 #ifndef HAVE_UNSIGNED_LONG_LONG
821 return false;
822 #else
823 if (n == 0) return false;
824 char buf[kMaxNumberLength+1];
825 str = TerminateNumber(buf, str, n);
826 if (str[0] == '-') return false; // strtoull() on a negative number?!
827 char* end;
828 errno = 0;
829 #if defined HAVE_STRTOQ
830 unsigned long long r = strtouq(str, &end, radix);
831 #elif defined HAVE_STRTOLL
832 unsigned long long r = strtoull(str, &end, radix);
833 #elif defined HAVE__STRTOI64
834 unsigned long long r = _strtoui64(str, &end, radix);
835 #else
836 #error parse_ulonglong_radix: cannot convert input to a long-long
837 #endif
838 if (end != str + n) return false; // Leftover junk
839 if (errno) return false;
840 if (dest == NULL) return true;
841 *(reinterpret_cast<unsigned long long*>(dest)) = r;
842 return true;
843 #endif /* HAVE_UNSIGNED_LONG_LONG */
844 }
845
846 bool Arg::parse_double(const char* str, int n, void* dest) {
847 if (n == 0) return false;
848 static const int kMaxLength = 200;
849 char buf[kMaxLength];
850 if (n >= kMaxLength) return false;
851 memcpy(buf, str, n);
852 buf[n] = '\0';
853 errno = 0;
854 char* end;
855 double r = strtod(buf, &end);
856 if (end != buf + n) return false; // Leftover junk
857 if (errno) return false;
858 if (dest == NULL) return true;
859 *(reinterpret_cast<double*>(dest)) = r;
860 return true;
861 }
862
863 bool Arg::parse_float(const char* str, int n, void* dest) {
864 double r;
865 if (!parse_double(str, n, &r)) return false;
866 if (dest == NULL) return true;
867 *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
868 return true;
869 }
870
871
872 #define DEFINE_INTEGER_PARSERS(name) \
873 bool Arg::parse_##name(const char* str, int n, void* dest) { \
874 return parse_##name##_radix(str, n, dest, 10); \
875 } \
876 bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \
877 return parse_##name##_radix(str, n, dest, 16); \
878 } \
879 bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \
880 return parse_##name##_radix(str, n, dest, 8); \
881 } \
882 bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
883 return parse_##name##_radix(str, n, dest, 0); \
884 }
885
886 DEFINE_INTEGER_PARSERS(short) /* */
887 DEFINE_INTEGER_PARSERS(ushort) /* */
888 DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */
889 DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */
890 DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */
891 DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */
892 DEFINE_INTEGER_PARSERS(longlong) /* */
893 DEFINE_INTEGER_PARSERS(ulonglong) /* */
894
895 #undef DEFINE_INTEGER_PARSERS
896
897 } // namespace pcrecpp

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5