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

Contents of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 326 - (hide annotations) (download)
Sat Mar 8 17:24:02 2008 UTC (6 years, 4 months ago) by ph10
File size: 30950 byte(s)
Craig's patch to the QuoteMeta function in pcrecpp.cc so that it escapes the
NUL character as backslash + 0 rather than backslash + NUL, because PCRE
doesn't support NULs in patterns.

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