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

Contents of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 308 - (hide annotations) (download)
Mon Jan 21 14:57:19 2008 UTC (6 years, 10 months ago) by ph10
File size: 30257 byte(s)
Apply Craig's patch to move no_arg into the RE class.

1 nigel 77 // 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 ph10 97 #ifdef HAVE_CONFIG_H
33 ph10 236 #include "config.h"
34 ph10 97 #endif
35    
36 nigel 77 #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 nigel 81 #include <algorithm>
44 ph10 199
45     #include "pcrecpp_internal.h"
46 ph10 236 #include "pcre.h"
47 nigel 77 #include "pcrecpp.h"
48 ph10 199 #include "pcre_stringpiece.h"
49 nigel 77
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 ph10 308 Arg RE::no_arg((void*)NULL);
59 nigel 77
60     // If a regular expression has no error, its error_ field points here
61     static const string empty_string;
62    
63     // If the user doesn't ask for any options, we just use this one
64     static RE_Options default_options;
65    
66 nigel 93 void RE::Init(const string& pat, const RE_Options* options) {
67 nigel 77 pattern_ = pat;
68     if (options == NULL) {
69     options_ = default_options;
70     } else {
71     options_ = *options;
72     }
73     error_ = &empty_string;
74     re_full_ = NULL;
75     re_partial_ = NULL;
76    
77     re_partial_ = Compile(UNANCHORED);
78     if (re_partial_ != NULL) {
79 ph10 179 re_full_ = Compile(ANCHOR_BOTH);
80 nigel 77 }
81     }
82    
83 nigel 93 void RE::Cleanup() {
84 ph10 179 if (re_full_ != NULL) (*pcre_free)(re_full_);
85     if (re_partial_ != NULL) (*pcre_free)(re_partial_);
86     if (error_ != &empty_string) delete error_;
87 nigel 77 }
88    
89 nigel 93
90     RE::~RE() {
91     Cleanup();
92     }
93    
94    
95 nigel 77 pcre* RE::Compile(Anchor anchor) {
96     // First, convert RE_Options into pcre options
97     int pcre_options = 0;
98 nigel 81 pcre_options = options_.all_options();
99 nigel 77
100     // Special treatment for anchoring. This is needed because at
101     // runtime pcre only provides an option for anchoring at the
102     // beginning of a string (unless you use offset).
103     //
104     // There are three types of anchoring we want:
105     // UNANCHORED Compile the original pattern, and use
106     // a pcre unanchored match.
107     // ANCHOR_START Compile the original pattern, and use
108     // a pcre anchored match.
109     // ANCHOR_BOTH Tack a "\z" to the end of the original pattern
110     // and use a pcre anchored match.
111    
112     const char* compile_error;
113     int eoffset;
114     pcre* re;
115     if (anchor != ANCHOR_BOTH) {
116     re = pcre_compile(pattern_.c_str(), pcre_options,
117     &compile_error, &eoffset, NULL);
118     } else {
119     // Tack a '\z' at the end of RE. Parenthesize it first so that
120     // the '\z' applies to all top-level alternatives in the regexp.
121     string wrapped = "(?:"; // A non-counting grouping operator
122     wrapped += pattern_;
123     wrapped += ")\\z";
124     re = pcre_compile(wrapped.c_str(), pcre_options,
125     &compile_error, &eoffset, NULL);
126     }
127     if (re == NULL) {
128     if (error_ == &empty_string) error_ = new string(compile_error);
129     }
130     return re;
131     }
132    
133     /***** Matching interfaces *****/
134    
135     bool RE::FullMatch(const StringPiece& text,
136     const Arg& ptr1,
137     const Arg& ptr2,
138     const Arg& ptr3,
139     const Arg& ptr4,
140     const Arg& ptr5,
141     const Arg& ptr6,
142     const Arg& ptr7,
143     const Arg& ptr8,
144     const Arg& ptr9,
145     const Arg& ptr10,
146     const Arg& ptr11,
147     const Arg& ptr12,
148     const Arg& ptr13,
149     const Arg& ptr14,
150     const Arg& ptr15,
151     const Arg& ptr16) const {
152     const Arg* args[kMaxArgs];
153     int n = 0;
154     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
155     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
156     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
157     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
158     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
159     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
160     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
161     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
162     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
163     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
164     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
165     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
166     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
167     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
168     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
169     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
170     done:
171    
172     int consumed;
173     int vec[kVecSize];
174     return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
175     }
176    
177     bool RE::PartialMatch(const StringPiece& text,
178     const Arg& ptr1,
179     const Arg& ptr2,
180     const Arg& ptr3,
181     const Arg& ptr4,
182     const Arg& ptr5,
183     const Arg& ptr6,
184     const Arg& ptr7,
185     const Arg& ptr8,
186     const Arg& ptr9,
187     const Arg& ptr10,
188     const Arg& ptr11,
189     const Arg& ptr12,
190     const Arg& ptr13,
191     const Arg& ptr14,
192     const Arg& ptr15,
193     const Arg& ptr16) const {
194     const Arg* args[kMaxArgs];
195     int n = 0;
196     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
197     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
198     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
199     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
200     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
201     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
202     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
203     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
204     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
205     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
206     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
207     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
208     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
209     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
210     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
211     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
212     done:
213    
214     int consumed;
215     int vec[kVecSize];
216     return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
217     }
218    
219     bool RE::Consume(StringPiece* input,
220     const Arg& ptr1,
221     const Arg& ptr2,
222     const Arg& ptr3,
223     const Arg& ptr4,
224     const Arg& ptr5,
225     const Arg& ptr6,
226     const Arg& ptr7,
227     const Arg& ptr8,
228     const Arg& ptr9,
229     const Arg& ptr10,
230     const Arg& ptr11,
231     const Arg& ptr12,
232     const Arg& ptr13,
233     const Arg& ptr14,
234     const Arg& ptr15,
235     const Arg& ptr16) const {
236     const Arg* args[kMaxArgs];
237     int n = 0;
238     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
239     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
240     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
241     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
242     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
243     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
244     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
245     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
246     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
247     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
248     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
249     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
250     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
251     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
252     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
253     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
254     done:
255    
256     int consumed;
257     int vec[kVecSize];
258     if (DoMatchImpl(*input, ANCHOR_START, &consumed,
259     args, n, vec, kVecSize)) {
260     input->remove_prefix(consumed);
261     return true;
262     } else {
263     return false;
264     }
265     }
266    
267     bool RE::FindAndConsume(StringPiece* input,
268     const Arg& ptr1,
269     const Arg& ptr2,
270     const Arg& ptr3,
271     const Arg& ptr4,
272     const Arg& ptr5,
273     const Arg& ptr6,
274     const Arg& ptr7,
275     const Arg& ptr8,
276     const Arg& ptr9,
277     const Arg& ptr10,
278     const Arg& ptr11,
279     const Arg& ptr12,
280     const Arg& ptr13,
281     const Arg& ptr14,
282     const Arg& ptr15,
283     const Arg& ptr16) const {
284     const Arg* args[kMaxArgs];
285     int n = 0;
286     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
287     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
288     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
289     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
290     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
291     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
292     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
293     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
294     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
295     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
296     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
297     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
298     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
299     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
300     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
301     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
302     done:
303    
304     int consumed;
305     int vec[kVecSize];
306     if (DoMatchImpl(*input, UNANCHORED, &consumed,
307     args, n, vec, kVecSize)) {
308     input->remove_prefix(consumed);
309     return true;
310     } else {
311     return false;
312     }
313     }
314    
315     bool RE::Replace(const StringPiece& rewrite,
316     string *str) const {
317     int vec[kVecSize];
318     int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
319     if (matches == 0)
320     return false;
321    
322     string s;
323     if (!Rewrite(&s, rewrite, *str, vec, matches))
324     return false;
325    
326     assert(vec[0] >= 0);
327     assert(vec[1] >= 0);
328     str->replace(vec[0], vec[1] - vec[0], s);
329     return true;
330     }
331    
332 nigel 91 // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
333     // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
334 ph10 253 // Modified by PH to add PCRE_NEWLINE_ANY and PCRE_NEWLINE_ANYCRLF.
335    
336 nigel 91 static int NewlineMode(int pcre_options) {
337     // TODO: if we can make it threadsafe, cache this var
338     int newline_mode = 0;
339     /* if (newline_mode) return newline_mode; */ // do this once it's cached
340 ph10 253 if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
341     PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF)) {
342 nigel 91 newline_mode = (pcre_options &
343 ph10 253 (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
344     PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF));
345 nigel 91 } else {
346     int newline;
347     pcre_config(PCRE_CONFIG_NEWLINE, &newline);
348     if (newline == 10)
349     newline_mode = PCRE_NEWLINE_LF;
350     else if (newline == 13)
351     newline_mode = PCRE_NEWLINE_CR;
352     else if (newline == 3338)
353     newline_mode = PCRE_NEWLINE_CRLF;
354 ph10 253 else if (newline == -1)
355     newline_mode = PCRE_NEWLINE_ANY;
356     else if (newline == -2)
357 ph10 259 newline_mode = PCRE_NEWLINE_ANYCRLF;
358 nigel 91 else
359     assert("" == "Unexpected return value from pcre_config(NEWLINE)");
360     }
361     return newline_mode;
362     }
363    
364 nigel 77 int RE::GlobalReplace(const StringPiece& rewrite,
365     string *str) const {
366     int count = 0;
367     int vec[kVecSize];
368     string out;
369     int start = 0;
370     int lastend = -1;
371    
372 ph10 297 while (start <= static_cast<int>(str->length())) {
373 nigel 77 int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
374     if (matches <= 0)
375     break;
376     int matchstart = vec[0], matchend = vec[1];
377     assert(matchstart >= start);
378     assert(matchend >= matchstart);
379     if (matchstart == matchend && matchstart == lastend) {
380     // advance one character if we matched an empty string at the same
381     // place as the last match occurred
382 nigel 91 matchend = start + 1;
383     // If the current char is CR and we're in CRLF mode, skip LF too.
384     // Note it's better to call pcre_fullinfo() than to examine
385     // all_options(), since options_ could have changed bewteen
386     // compile-time and now, but this is simpler and safe enough.
387 ph10 259 // Modified by PH to add ANY and ANYCRLF.
388 nigel 91 if (start+1 < static_cast<int>(str->length()) &&
389     (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
390 ph10 253 (NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF ||
391     NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANY ||
392     NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANYCRLF)
393     ) {
394 nigel 91 matchend++;
395     }
396     // We also need to advance more than one char if we're in utf8 mode.
397     #ifdef SUPPORT_UTF8
398     if (options_.utf8()) {
399     while (matchend < static_cast<int>(str->length()) &&
400     ((*str)[matchend] & 0xc0) == 0x80)
401     matchend++;
402     }
403     #endif
404     if (matchend <= static_cast<int>(str->length()))
405     out.append(*str, start, matchend - start);
406     start = matchend;
407 nigel 77 } else {
408     out.append(*str, start, matchstart - start);
409     Rewrite(&out, rewrite, *str, vec, matches);
410     start = matchend;
411     lastend = matchend;
412 ph10 297 count++;
413 nigel 77 }
414     }
415    
416     if (count == 0)
417     return 0;
418    
419     if (start < static_cast<int>(str->length()))
420     out.append(*str, start, str->length() - start);
421     swap(out, *str);
422     return count;
423     }
424    
425     bool RE::Extract(const StringPiece& rewrite,
426     const StringPiece& text,
427     string *out) const {
428     int vec[kVecSize];
429     int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
430     if (matches == 0)
431     return false;
432 nigel 81 out->erase();
433 nigel 77 return Rewrite(out, rewrite, text, vec, matches);
434     }
435    
436 nigel 93 /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
437     string result;
438    
439     // Escape any ascii character not in [A-Za-z_0-9].
440     //
441     // Note that it's legal to escape a character even if it has no
442     // special meaning in a regular expression -- so this function does
443     // that. (This also makes it identical to the perl function of the
444     // same name; see `perldoc -f quotemeta`.)
445     for (int ii = 0; ii < unquoted.size(); ++ii) {
446     // Note that using 'isalnum' here raises the benchmark time from
447     // 32ns to 58ns:
448     if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
449     (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
450     (unquoted[ii] < '0' || unquoted[ii] > '9') &&
451     unquoted[ii] != '_' &&
452     // If this is the part of a UTF8 or Latin1 character, we need
453     // to copy this byte without escaping. Experimentally this is
454     // what works correctly with the regexp library.
455     !(unquoted[ii] & 128)) {
456     result += '\\';
457     }
458     result += unquoted[ii];
459     }
460    
461     return result;
462     }
463    
464 nigel 77 /***** Actual matching and rewriting code *****/
465    
466     int RE::TryMatch(const StringPiece& text,
467     int startpos,
468     Anchor anchor,
469     int *vec,
470     int vecsize) const {
471     pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
472     if (re == NULL) {
473     //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
474     return 0;
475     }
476    
477 ph10 199 pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
478 nigel 77 if (options_.match_limit() > 0) {
479 nigel 87 extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
480 nigel 77 extra.match_limit = options_.match_limit();
481     }
482 nigel 87 if (options_.match_limit_recursion() > 0) {
483     extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
484     extra.match_limit_recursion = options_.match_limit_recursion();
485     }
486 nigel 77 int rc = pcre_exec(re, // The regular expression object
487     &extra,
488 nigel 87 (text.data() == NULL) ? "" : text.data(),
489 nigel 77 text.size(),
490     startpos,
491     (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
492     vec,
493     vecsize);
494    
495     // Handle errors
496     if (rc == PCRE_ERROR_NOMATCH) {
497     return 0;
498     } else if (rc < 0) {
499     //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
500     // re, pattern_.c_str());
501     return 0;
502     } else if (rc == 0) {
503     // pcre_exec() returns 0 as a special case when the number of
504     // capturing subpatterns exceeds the size of the vector.
505     // When this happens, there is a match and the output vector
506     // is filled, but we miss out on the positions of the extra subpatterns.
507     rc = vecsize / 2;
508     }
509    
510     return rc;
511     }
512    
513     bool RE::DoMatchImpl(const StringPiece& text,
514     Anchor anchor,
515     int* consumed,
516     const Arg* const* args,
517     int n,
518     int* vec,
519     int vecsize) const {
520     assert((1 + n) * 3 <= vecsize); // results + PCRE workspace
521     int matches = TryMatch(text, 0, anchor, vec, vecsize);
522     assert(matches >= 0); // TryMatch never returns negatives
523     if (matches == 0)
524     return false;
525    
526     *consumed = vec[1];
527    
528 nigel 87 if (n == 0 || args == NULL) {
529 nigel 77 // We are not interested in results
530     return true;
531     }
532    
533 nigel 87 if (NumberOfCapturingGroups() < n) {
534     // RE has fewer capturing groups than number of arg pointers passed in
535     return false;
536     }
537    
538 nigel 77 // If we got here, we must have matched the whole pattern.
539     // We do not need (can not do) any more checks on the value of 'matches' here
540     // -- see the comment for TryMatch.
541     for (int i = 0; i < n; i++) {
542     const int start = vec[2*(i+1)];
543     const int limit = vec[2*(i+1)+1];
544     if (!args[i]->Parse(text.data() + start, limit-start)) {
545     // TODO: Should we indicate what the error was?
546     return false;
547     }
548     }
549    
550     return true;
551     }
552    
553     bool RE::DoMatch(const StringPiece& text,
554     Anchor anchor,
555     int* consumed,
556     const Arg* const args[],
557     int n) const {
558     assert(n >= 0);
559     size_t const vecsize = (1 + n) * 3; // results + PCRE workspace
560     // (as for kVecSize)
561     int space[21]; // use stack allocation for small vecsize (common case)
562     int* vec = vecsize <= 21 ? space : new int[vecsize];
563     bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
564     if (vec != space) delete [] vec;
565     return retval;
566     }
567    
568     bool RE::Rewrite(string *out, const StringPiece &rewrite,
569     const StringPiece &text, int *vec, int veclen) const {
570     for (const char *s = rewrite.data(), *end = s + rewrite.size();
571     s < end; s++) {
572     int c = *s;
573     if (c == '\\') {
574     c = *++s;
575     if (isdigit(c)) {
576     int n = (c - '0');
577     if (n >= veclen) {
578     //fprintf(stderr, requested group %d in regexp %.*s\n",
579     // n, rewrite.size(), rewrite.data());
580     return false;
581     }
582     int start = vec[2 * n];
583     if (start >= 0)
584     out->append(text.data() + start, vec[2 * n + 1] - start);
585     } else if (c == '\\') {
586     out->push_back('\\');
587     } else {
588     //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
589     // rewrite.size(), rewrite.data());
590     return false;
591     }
592     } else {
593     out->push_back(c);
594     }
595     }
596     return true;
597     }
598    
599     // Return the number of capturing subpatterns, or -1 if the
600     // regexp wasn't valid on construction.
601 nigel 87 int RE::NumberOfCapturingGroups() const {
602 nigel 77 if (re_partial_ == NULL) return -1;
603    
604     int result;
605     int pcre_retval = pcre_fullinfo(re_partial_, // The regular expression object
606     NULL, // We did not study the pattern
607     PCRE_INFO_CAPTURECOUNT,
608     &result);
609     assert(pcre_retval == 0);
610     return result;
611     }
612    
613     /***** Parsers for various types *****/
614    
615     bool Arg::parse_null(const char* str, int n, void* dest) {
616     // We fail if somebody asked us to store into a non-NULL void* pointer
617     return (dest == NULL);
618     }
619    
620     bool Arg::parse_string(const char* str, int n, void* dest) {
621 ph10 263 if (dest == NULL) return true;
622 nigel 77 reinterpret_cast<string*>(dest)->assign(str, n);
623     return true;
624     }
625    
626     bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
627 ph10 263 if (dest == NULL) return true;
628 nigel 77 reinterpret_cast<StringPiece*>(dest)->set(str, n);
629     return true;
630     }
631    
632     bool Arg::parse_char(const char* str, int n, void* dest) {
633     if (n != 1) return false;
634 ph10 263 if (dest == NULL) return true;
635 nigel 77 *(reinterpret_cast<char*>(dest)) = str[0];
636     return true;
637     }
638    
639     bool Arg::parse_uchar(const char* str, int n, void* dest) {
640     if (n != 1) return false;
641 ph10 263 if (dest == NULL) return true;
642 nigel 77 *(reinterpret_cast<unsigned char*>(dest)) = str[0];
643     return true;
644     }
645    
646     // Largest number spec that we are willing to parse
647     static const int kMaxNumberLength = 32;
648    
649     // REQUIRES "buf" must have length at least kMaxNumberLength+1
650     // REQUIRES "n > 0"
651     // Copies "str" into "buf" and null-terminates if necessary.
652     // Returns one of:
653     // a. "str" if no termination is needed
654     // b. "buf" if the string was copied and null-terminated
655     // c. "" if the input was invalid and has no hope of being parsed
656     static const char* TerminateNumber(char* buf, const char* str, int n) {
657     if ((n > 0) && isspace(*str)) {
658     // We are less forgiving than the strtoxxx() routines and do not
659     // allow leading spaces.
660     return "";
661     }
662    
663     // See if the character right after the input text may potentially
664     // look like a digit.
665     if (isdigit(str[n]) ||
666     ((str[n] >= 'a') && (str[n] <= 'f')) ||
667     ((str[n] >= 'A') && (str[n] <= 'F'))) {
668     if (n > kMaxNumberLength) return ""; // Input too big to be a valid number
669     memcpy(buf, str, n);
670     buf[n] = '\0';
671     return buf;
672     } else {
673     // We can parse right out of the supplied string, so return it.
674     return str;
675     }
676     }
677    
678     bool Arg::parse_long_radix(const char* str,
679     int n,
680     void* dest,
681     int radix) {
682     if (n == 0) return false;
683     char buf[kMaxNumberLength+1];
684     str = TerminateNumber(buf, str, n);
685     char* end;
686     errno = 0;
687     long r = strtol(str, &end, radix);
688     if (end != str + n) return false; // Leftover junk
689     if (errno) return false;
690 ph10 263 if (dest == NULL) return true;
691 nigel 77 *(reinterpret_cast<long*>(dest)) = r;
692     return true;
693     }
694    
695     bool Arg::parse_ulong_radix(const char* str,
696     int n,
697     void* dest,
698     int radix) {
699     if (n == 0) return false;
700     char buf[kMaxNumberLength+1];
701     str = TerminateNumber(buf, str, n);
702 nigel 87 if (str[0] == '-') return false; // strtoul() on a negative number?!
703 nigel 77 char* end;
704     errno = 0;
705     unsigned long r = strtoul(str, &end, radix);
706     if (end != str + n) return false; // Leftover junk
707     if (errno) return false;
708 ph10 263 if (dest == NULL) return true;
709 nigel 77 *(reinterpret_cast<unsigned long*>(dest)) = r;
710     return true;
711     }
712    
713     bool Arg::parse_short_radix(const char* str,
714     int n,
715     void* dest,
716     int radix) {
717     long r;
718     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
719     if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range
720 ph10 263 if (dest == NULL) return true;
721 ph10 256 *(reinterpret_cast<short*>(dest)) = static_cast<short>(r);
722 nigel 77 return true;
723     }
724    
725     bool Arg::parse_ushort_radix(const char* str,
726     int n,
727     void* dest,
728     int radix) {
729     unsigned long r;
730     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
731     if (r > USHRT_MAX) return false; // Out of range
732 ph10 263 if (dest == NULL) return true;
733 ph10 256 *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r);
734 nigel 77 return true;
735     }
736    
737     bool Arg::parse_int_radix(const char* str,
738     int n,
739     void* dest,
740     int radix) {
741     long r;
742     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
743     if (r < INT_MIN || r > INT_MAX) return false; // Out of range
744 ph10 263 if (dest == NULL) return true;
745 nigel 77 *(reinterpret_cast<int*>(dest)) = r;
746     return true;
747     }
748    
749     bool Arg::parse_uint_radix(const char* str,
750     int n,
751     void* dest,
752     int radix) {
753     unsigned long r;
754     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
755     if (r > UINT_MAX) return false; // Out of range
756 ph10 263 if (dest == NULL) return true;
757 nigel 77 *(reinterpret_cast<unsigned int*>(dest)) = r;
758     return true;
759     }
760    
761     bool Arg::parse_longlong_radix(const char* str,
762     int n,
763     void* dest,
764     int radix) {
765     #ifndef HAVE_LONG_LONG
766     return false;
767     #else
768     if (n == 0) return false;
769     char buf[kMaxNumberLength+1];
770     str = TerminateNumber(buf, str, n);
771     char* end;
772     errno = 0;
773     #if defined HAVE_STRTOQ
774     long long r = strtoq(str, &end, radix);
775     #elif defined HAVE_STRTOLL
776     long long r = strtoll(str, &end, radix);
777 ph10 257 #elif defined HAVE__STRTOI64
778     long long r = _strtoi64(str, &end, radix);
779 nigel 77 #else
780     #error parse_longlong_radix: cannot convert input to a long-long
781     #endif
782     if (end != str + n) return false; // Leftover junk
783     if (errno) return false;
784 ph10 263 if (dest == NULL) return true;
785 nigel 77 *(reinterpret_cast<long long*>(dest)) = r;
786     return true;
787     #endif /* HAVE_LONG_LONG */
788     }
789    
790     bool Arg::parse_ulonglong_radix(const char* str,
791     int n,
792     void* dest,
793     int radix) {
794     #ifndef HAVE_UNSIGNED_LONG_LONG
795     return false;
796     #else
797     if (n == 0) return false;
798     char buf[kMaxNumberLength+1];
799     str = TerminateNumber(buf, str, n);
800 nigel 87 if (str[0] == '-') return false; // strtoull() on a negative number?!
801 nigel 77 char* end;
802     errno = 0;
803     #if defined HAVE_STRTOQ
804     unsigned long long r = strtouq(str, &end, radix);
805     #elif defined HAVE_STRTOLL
806     unsigned long long r = strtoull(str, &end, radix);
807 ph10 257 #elif defined HAVE__STRTOI64
808     unsigned long long r = _strtoui64(str, &end, radix);
809 nigel 77 #else
810     #error parse_ulonglong_radix: cannot convert input to a long-long
811     #endif
812     if (end != str + n) return false; // Leftover junk
813     if (errno) return false;
814 ph10 263 if (dest == NULL) return true;
815 nigel 77 *(reinterpret_cast<unsigned long long*>(dest)) = r;
816     return true;
817     #endif /* HAVE_UNSIGNED_LONG_LONG */
818     }
819    
820     bool Arg::parse_double(const char* str, int n, void* dest) {
821     if (n == 0) return false;
822     static const int kMaxLength = 200;
823     char buf[kMaxLength];
824     if (n >= kMaxLength) return false;
825     memcpy(buf, str, n);
826     buf[n] = '\0';
827     errno = 0;
828     char* end;
829     double r = strtod(buf, &end);
830     if (end != buf + n) return false; // Leftover junk
831     if (errno) return false;
832 ph10 263 if (dest == NULL) return true;
833 nigel 77 *(reinterpret_cast<double*>(dest)) = r;
834     return true;
835     }
836    
837     bool Arg::parse_float(const char* str, int n, void* dest) {
838     double r;
839     if (!parse_double(str, n, &r)) return false;
840 ph10 263 if (dest == NULL) return true;
841 nigel 77 *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
842     return true;
843     }
844    
845    
846     #define DEFINE_INTEGER_PARSERS(name) \
847     bool Arg::parse_##name(const char* str, int n, void* dest) { \
848     return parse_##name##_radix(str, n, dest, 10); \
849     } \
850     bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \
851     return parse_##name##_radix(str, n, dest, 16); \
852     } \
853     bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \
854     return parse_##name##_radix(str, n, dest, 8); \
855     } \
856     bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
857     return parse_##name##_radix(str, n, dest, 0); \
858     }
859    
860 nigel 93 DEFINE_INTEGER_PARSERS(short) /* */
861     DEFINE_INTEGER_PARSERS(ushort) /* */
862     DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */
863     DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */
864     DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */
865     DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */
866     DEFINE_INTEGER_PARSERS(longlong) /* */
867     DEFINE_INTEGER_PARSERS(ulonglong) /* */
868 nigel 77
869     #undef DEFINE_INTEGER_PARSERS
870    
871     } // 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