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

Contents of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 96 - (hide annotations) (download)
Fri Mar 2 13:10:43 2007 UTC (7 years, 8 months ago) by nigel
File size: 30088 byte(s)
 r6896@hex:  nm | 2007-03-02 13:09:14 +0000
 Added EOL and keywork properties throughout

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     #include <stdlib.h>
33     #include <stdio.h>
34     #include <ctype.h>
35     #include <limits.h> /* for SHRT_MIN, USHRT_MAX, etc */
36     #include <assert.h>
37     #include <errno.h>
38     #include <string>
39 nigel 81 #include <algorithm>
40 nigel 77 #include "config.h"
41     // We need this to compile the proper dll on windows/msys. This is copied
42     // from pcre_internal.h. It would probably be better just to include that.
43     #define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */
44     #include "pcre.h"
45     #include "pcre_stringpiece.h"
46     #include "pcrecpp.h"
47    
48    
49     namespace pcrecpp {
50    
51     // Maximum number of args we can set
52     static const int kMaxArgs = 16;
53     static const int kVecSize = (1 + kMaxArgs) * 3; // results + PCRE workspace
54    
55     // Special object that stands-in for no argument
56     Arg no_arg((void*)NULL);
57    
58     // If a regular expression has no error, its error_ field points here
59     static const string empty_string;
60    
61     // If the user doesn't ask for any options, we just use this one
62     static RE_Options default_options;
63    
64 nigel 93 void RE::Init(const string& pat, const RE_Options* options) {
65 nigel 77 pattern_ = pat;
66     if (options == NULL) {
67     options_ = default_options;
68     } else {
69     options_ = *options;
70     }
71     error_ = &empty_string;
72     re_full_ = NULL;
73     re_partial_ = NULL;
74    
75     re_partial_ = Compile(UNANCHORED);
76     if (re_partial_ != NULL) {
77     // Check for complicated patterns. The following change is
78     // conservative in that it may treat some "simple" patterns
79     // as "complex" (e.g., if the vertical bar is in a character
80     // class or is escaped). But it seems good enough.
81 nigel 93 if (strchr(pat.c_str(), '|') == NULL) {
82 nigel 77 // Simple pattern: we can use position-based checks to perform
83     // fully anchored matches
84     re_full_ = re_partial_;
85     } else {
86     // We need a special pattern for anchored matches
87     re_full_ = Compile(ANCHOR_BOTH);
88     }
89     }
90     }
91    
92 nigel 93 void RE::Cleanup() {
93 nigel 77 if (re_full_ != NULL && re_full_ != re_partial_) (*pcre_free)(re_full_);
94     if (re_partial_ != NULL) (*pcre_free)(re_partial_);
95     if (error_ != &empty_string) delete error_;
96     }
97    
98 nigel 93
99     RE::~RE() {
100     Cleanup();
101     }
102    
103    
104 nigel 77 pcre* RE::Compile(Anchor anchor) {
105     // First, convert RE_Options into pcre options
106     int pcre_options = 0;
107 nigel 81 pcre_options = options_.all_options();
108 nigel 77
109     // Special treatment for anchoring. This is needed because at
110     // runtime pcre only provides an option for anchoring at the
111     // beginning of a string (unless you use offset).
112     //
113     // There are three types of anchoring we want:
114     // UNANCHORED Compile the original pattern, and use
115     // a pcre unanchored match.
116     // ANCHOR_START Compile the original pattern, and use
117     // a pcre anchored match.
118     // ANCHOR_BOTH Tack a "\z" to the end of the original pattern
119     // and use a pcre anchored match.
120    
121     const char* compile_error;
122     int eoffset;
123     pcre* re;
124     if (anchor != ANCHOR_BOTH) {
125     re = pcre_compile(pattern_.c_str(), pcre_options,
126     &compile_error, &eoffset, NULL);
127     } else {
128     // Tack a '\z' at the end of RE. Parenthesize it first so that
129     // the '\z' applies to all top-level alternatives in the regexp.
130     string wrapped = "(?:"; // A non-counting grouping operator
131     wrapped += pattern_;
132     wrapped += ")\\z";
133     re = pcre_compile(wrapped.c_str(), pcre_options,
134     &compile_error, &eoffset, NULL);
135     }
136     if (re == NULL) {
137     if (error_ == &empty_string) error_ = new string(compile_error);
138     }
139     return re;
140     }
141    
142     /***** Matching interfaces *****/
143    
144     bool RE::FullMatch(const StringPiece& text,
145     const Arg& ptr1,
146     const Arg& ptr2,
147     const Arg& ptr3,
148     const Arg& ptr4,
149     const Arg& ptr5,
150     const Arg& ptr6,
151     const Arg& ptr7,
152     const Arg& ptr8,
153     const Arg& ptr9,
154     const Arg& ptr10,
155     const Arg& ptr11,
156     const Arg& ptr12,
157     const Arg& ptr13,
158     const Arg& ptr14,
159     const Arg& ptr15,
160     const Arg& ptr16) const {
161     const Arg* args[kMaxArgs];
162     int n = 0;
163     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
164     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
165     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
166     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
167     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
168     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
169     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
170     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
171     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
172     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
173     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
174     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
175     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
176     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
177     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
178     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
179     done:
180    
181     int consumed;
182     int vec[kVecSize];
183     return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
184     }
185    
186     bool RE::PartialMatch(const StringPiece& text,
187     const Arg& ptr1,
188     const Arg& ptr2,
189     const Arg& ptr3,
190     const Arg& ptr4,
191     const Arg& ptr5,
192     const Arg& ptr6,
193     const Arg& ptr7,
194     const Arg& ptr8,
195     const Arg& ptr9,
196     const Arg& ptr10,
197     const Arg& ptr11,
198     const Arg& ptr12,
199     const Arg& ptr13,
200     const Arg& ptr14,
201     const Arg& ptr15,
202     const Arg& ptr16) const {
203     const Arg* args[kMaxArgs];
204     int n = 0;
205     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
206     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
207     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
208     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
209     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
210     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
211     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
212     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
213     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
214     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
215     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
216     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
217     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
218     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
219     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
220     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
221     done:
222    
223     int consumed;
224     int vec[kVecSize];
225     return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
226     }
227    
228     bool RE::Consume(StringPiece* input,
229     const Arg& ptr1,
230     const Arg& ptr2,
231     const Arg& ptr3,
232     const Arg& ptr4,
233     const Arg& ptr5,
234     const Arg& ptr6,
235     const Arg& ptr7,
236     const Arg& ptr8,
237     const Arg& ptr9,
238     const Arg& ptr10,
239     const Arg& ptr11,
240     const Arg& ptr12,
241     const Arg& ptr13,
242     const Arg& ptr14,
243     const Arg& ptr15,
244     const Arg& ptr16) const {
245     const Arg* args[kMaxArgs];
246     int n = 0;
247     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
248     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
249     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
250     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
251     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
252     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
253     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
254     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
255     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
256     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
257     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
258     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
259     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
260     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
261     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
262     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
263     done:
264    
265     int consumed;
266     int vec[kVecSize];
267     if (DoMatchImpl(*input, ANCHOR_START, &consumed,
268     args, n, vec, kVecSize)) {
269     input->remove_prefix(consumed);
270     return true;
271     } else {
272     return false;
273     }
274     }
275    
276     bool RE::FindAndConsume(StringPiece* input,
277     const Arg& ptr1,
278     const Arg& ptr2,
279     const Arg& ptr3,
280     const Arg& ptr4,
281     const Arg& ptr5,
282     const Arg& ptr6,
283     const Arg& ptr7,
284     const Arg& ptr8,
285     const Arg& ptr9,
286     const Arg& ptr10,
287     const Arg& ptr11,
288     const Arg& ptr12,
289     const Arg& ptr13,
290     const Arg& ptr14,
291     const Arg& ptr15,
292     const Arg& ptr16) const {
293     const Arg* args[kMaxArgs];
294     int n = 0;
295     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
296     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
297     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
298     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
299     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
300     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
301     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
302     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
303     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
304     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
305     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
306     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
307     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
308     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
309     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
310     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
311     done:
312    
313     int consumed;
314     int vec[kVecSize];
315     if (DoMatchImpl(*input, UNANCHORED, &consumed,
316     args, n, vec, kVecSize)) {
317     input->remove_prefix(consumed);
318     return true;
319     } else {
320     return false;
321     }
322     }
323    
324     bool RE::Replace(const StringPiece& rewrite,
325     string *str) const {
326     int vec[kVecSize];
327     int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
328     if (matches == 0)
329     return false;
330    
331     string s;
332     if (!Rewrite(&s, rewrite, *str, vec, matches))
333     return false;
334    
335     assert(vec[0] >= 0);
336     assert(vec[1] >= 0);
337     str->replace(vec[0], vec[1] - vec[0], s);
338     return true;
339     }
340    
341 nigel 91 // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
342     // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
343     static int NewlineMode(int pcre_options) {
344     // TODO: if we can make it threadsafe, cache this var
345     int newline_mode = 0;
346     /* if (newline_mode) return newline_mode; */ // do this once it's cached
347     if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF)) {
348     newline_mode = (pcre_options &
349     (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF));
350     } else {
351     int newline;
352     pcre_config(PCRE_CONFIG_NEWLINE, &newline);
353     if (newline == 10)
354     newline_mode = PCRE_NEWLINE_LF;
355     else if (newline == 13)
356     newline_mode = PCRE_NEWLINE_CR;
357     else if (newline == 3338)
358     newline_mode = PCRE_NEWLINE_CRLF;
359     else
360     assert("" == "Unexpected return value from pcre_config(NEWLINE)");
361     }
362     return newline_mode;
363     }
364    
365 nigel 77 int RE::GlobalReplace(const StringPiece& rewrite,
366     string *str) const {
367     int count = 0;
368     int vec[kVecSize];
369     string out;
370     int start = 0;
371     int lastend = -1;
372    
373     for (; start <= static_cast<int>(str->length()); count++) {
374     int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
375     if (matches <= 0)
376     break;
377     int matchstart = vec[0], matchend = vec[1];
378     assert(matchstart >= start);
379     assert(matchend >= matchstart);
380     if (matchstart == matchend && matchstart == lastend) {
381     // advance one character if we matched an empty string at the same
382     // place as the last match occurred
383 nigel 91 matchend = start + 1;
384     // If the current char is CR and we're in CRLF mode, skip LF too.
385     // Note it's better to call pcre_fullinfo() than to examine
386     // all_options(), since options_ could have changed bewteen
387     // compile-time and now, but this is simpler and safe enough.
388     if (start+1 < static_cast<int>(str->length()) &&
389     (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
390     NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF) {
391     matchend++;
392     }
393     // We also need to advance more than one char if we're in utf8 mode.
394     #ifdef SUPPORT_UTF8
395     if (options_.utf8()) {
396     while (matchend < static_cast<int>(str->length()) &&
397     ((*str)[matchend] & 0xc0) == 0x80)
398     matchend++;
399     }
400     #endif
401     if (matchend <= static_cast<int>(str->length()))
402     out.append(*str, start, matchend - start);
403     start = matchend;
404 nigel 77 } else {
405     out.append(*str, start, matchstart - start);
406     Rewrite(&out, rewrite, *str, vec, matches);
407     start = matchend;
408     lastend = matchend;
409     count++;
410     }
411     }
412    
413     if (count == 0)
414     return 0;
415    
416     if (start < static_cast<int>(str->length()))
417     out.append(*str, start, str->length() - start);
418     swap(out, *str);
419     return count;
420     }
421    
422     bool RE::Extract(const StringPiece& rewrite,
423     const StringPiece& text,
424     string *out) const {
425     int vec[kVecSize];
426     int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
427     if (matches == 0)
428     return false;
429 nigel 81 out->erase();
430 nigel 77 return Rewrite(out, rewrite, text, vec, matches);
431     }
432    
433 nigel 93 /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
434     string result;
435    
436     // Escape any ascii character not in [A-Za-z_0-9].
437     //
438     // Note that it's legal to escape a character even if it has no
439     // special meaning in a regular expression -- so this function does
440     // that. (This also makes it identical to the perl function of the
441     // same name; see `perldoc -f quotemeta`.)
442     for (int ii = 0; ii < unquoted.size(); ++ii) {
443     // Note that using 'isalnum' here raises the benchmark time from
444     // 32ns to 58ns:
445     if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
446     (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
447     (unquoted[ii] < '0' || unquoted[ii] > '9') &&
448     unquoted[ii] != '_' &&
449     // If this is the part of a UTF8 or Latin1 character, we need
450     // to copy this byte without escaping. Experimentally this is
451     // what works correctly with the regexp library.
452     !(unquoted[ii] & 128)) {
453     result += '\\';
454     }
455     result += unquoted[ii];
456     }
457    
458     return result;
459     }
460    
461 nigel 77 /***** Actual matching and rewriting code *****/
462    
463     int RE::TryMatch(const StringPiece& text,
464     int startpos,
465     Anchor anchor,
466     int *vec,
467     int vecsize) const {
468     pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
469     if (re == NULL) {
470     //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
471     return 0;
472     }
473    
474     pcre_extra extra = { 0 };
475     if (options_.match_limit() > 0) {
476 nigel 87 extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
477 nigel 77 extra.match_limit = options_.match_limit();
478     }
479 nigel 87 if (options_.match_limit_recursion() > 0) {
480     extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
481     extra.match_limit_recursion = options_.match_limit_recursion();
482     }
483 nigel 77 int rc = pcre_exec(re, // The regular expression object
484     &extra,
485 nigel 87 (text.data() == NULL) ? "" : text.data(),
486 nigel 77 text.size(),
487     startpos,
488     (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
489     vec,
490     vecsize);
491    
492     // Handle errors
493     if (rc == PCRE_ERROR_NOMATCH) {
494     return 0;
495     } else if (rc < 0) {
496     //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
497     // re, pattern_.c_str());
498     return 0;
499     } else if (rc == 0) {
500     // pcre_exec() returns 0 as a special case when the number of
501     // capturing subpatterns exceeds the size of the vector.
502     // When this happens, there is a match and the output vector
503     // is filled, but we miss out on the positions of the extra subpatterns.
504     rc = vecsize / 2;
505     }
506    
507     if ((anchor == ANCHOR_BOTH) && (re_full_ == re_partial_)) {
508     // We need an extra check to make sure that the match extended
509     // to the end of the input string
510     assert(vec[0] == 0); // PCRE_ANCHORED forces starting match
511     if (vec[1] != text.size()) return 0; // Did not get ending match
512     }
513    
514     return rc;
515     }
516    
517     bool RE::DoMatchImpl(const StringPiece& text,
518     Anchor anchor,
519     int* consumed,
520     const Arg* const* args,
521     int n,
522     int* vec,
523     int vecsize) const {
524     assert((1 + n) * 3 <= vecsize); // results + PCRE workspace
525     int matches = TryMatch(text, 0, anchor, vec, vecsize);
526     assert(matches >= 0); // TryMatch never returns negatives
527     if (matches == 0)
528     return false;
529    
530     *consumed = vec[1];
531    
532 nigel 87 if (n == 0 || args == NULL) {
533 nigel 77 // We are not interested in results
534     return true;
535     }
536    
537 nigel 87 if (NumberOfCapturingGroups() < n) {
538     // RE has fewer capturing groups than number of arg pointers passed in
539     return false;
540     }
541    
542 nigel 77 // If we got here, we must have matched the whole pattern.
543     // We do not need (can not do) any more checks on the value of 'matches' here
544     // -- see the comment for TryMatch.
545     for (int i = 0; i < n; i++) {
546     const int start = vec[2*(i+1)];
547     const int limit = vec[2*(i+1)+1];
548     if (!args[i]->Parse(text.data() + start, limit-start)) {
549     // TODO: Should we indicate what the error was?
550     return false;
551     }
552     }
553    
554     return true;
555     }
556    
557     bool RE::DoMatch(const StringPiece& text,
558     Anchor anchor,
559     int* consumed,
560     const Arg* const args[],
561     int n) const {
562     assert(n >= 0);
563     size_t const vecsize = (1 + n) * 3; // results + PCRE workspace
564     // (as for kVecSize)
565     int space[21]; // use stack allocation for small vecsize (common case)
566     int* vec = vecsize <= 21 ? space : new int[vecsize];
567     bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
568     if (vec != space) delete [] vec;
569     return retval;
570     }
571    
572     bool RE::Rewrite(string *out, const StringPiece &rewrite,
573     const StringPiece &text, int *vec, int veclen) const {
574     for (const char *s = rewrite.data(), *end = s + rewrite.size();
575     s < end; s++) {
576     int c = *s;
577     if (c == '\\') {
578     c = *++s;
579     if (isdigit(c)) {
580     int n = (c - '0');
581     if (n >= veclen) {
582     //fprintf(stderr, requested group %d in regexp %.*s\n",
583     // n, rewrite.size(), rewrite.data());
584     return false;
585     }
586     int start = vec[2 * n];
587     if (start >= 0)
588     out->append(text.data() + start, vec[2 * n + 1] - start);
589     } else if (c == '\\') {
590     out->push_back('\\');
591     } else {
592     //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
593     // rewrite.size(), rewrite.data());
594     return false;
595     }
596     } else {
597     out->push_back(c);
598     }
599     }
600     return true;
601     }
602    
603     // Return the number of capturing subpatterns, or -1 if the
604     // regexp wasn't valid on construction.
605 nigel 87 int RE::NumberOfCapturingGroups() const {
606 nigel 77 if (re_partial_ == NULL) return -1;
607    
608     int result;
609     int pcre_retval = pcre_fullinfo(re_partial_, // The regular expression object
610     NULL, // We did not study the pattern
611     PCRE_INFO_CAPTURECOUNT,
612     &result);
613     assert(pcre_retval == 0);
614     return result;
615     }
616    
617     /***** Parsers for various types *****/
618    
619     bool Arg::parse_null(const char* str, int n, void* dest) {
620     // We fail if somebody asked us to store into a non-NULL void* pointer
621     return (dest == NULL);
622     }
623    
624     bool Arg::parse_string(const char* str, int n, void* dest) {
625     reinterpret_cast<string*>(dest)->assign(str, n);
626     return true;
627     }
628    
629     bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
630     reinterpret_cast<StringPiece*>(dest)->set(str, n);
631     return true;
632     }
633    
634     bool Arg::parse_char(const char* str, int n, void* dest) {
635     if (n != 1) return false;
636     *(reinterpret_cast<char*>(dest)) = str[0];
637     return true;
638     }
639    
640     bool Arg::parse_uchar(const char* str, int n, void* dest) {
641     if (n != 1) return false;
642     *(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     *(reinterpret_cast<long*>(dest)) = r;
691     return true;
692     }
693    
694     bool Arg::parse_ulong_radix(const char* str,
695     int n,
696     void* dest,
697     int radix) {
698     if (n == 0) return false;
699     char buf[kMaxNumberLength+1];
700     str = TerminateNumber(buf, str, n);
701 nigel 87 if (str[0] == '-') return false; // strtoul() on a negative number?!
702 nigel 77 char* end;
703     errno = 0;
704     unsigned long r = strtoul(str, &end, radix);
705     if (end != str + n) return false; // Leftover junk
706     if (errno) return false;
707     *(reinterpret_cast<unsigned long*>(dest)) = r;
708     return true;
709     }
710    
711     bool Arg::parse_short_radix(const char* str,
712     int n,
713     void* dest,
714     int radix) {
715     long r;
716     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
717     if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range
718     *(reinterpret_cast<short*>(dest)) = r;
719     return true;
720     }
721    
722     bool Arg::parse_ushort_radix(const char* str,
723     int n,
724     void* dest,
725     int radix) {
726     unsigned long r;
727     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
728     if (r > USHRT_MAX) return false; // Out of range
729     *(reinterpret_cast<unsigned short*>(dest)) = r;
730     return true;
731     }
732    
733     bool Arg::parse_int_radix(const char* str,
734     int n,
735     void* dest,
736     int radix) {
737     long r;
738     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
739     if (r < INT_MIN || r > INT_MAX) return false; // Out of range
740     *(reinterpret_cast<int*>(dest)) = r;
741     return true;
742     }
743    
744     bool Arg::parse_uint_radix(const char* str,
745     int n,
746     void* dest,
747     int radix) {
748     unsigned long r;
749     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
750     if (r > UINT_MAX) return false; // Out of range
751     *(reinterpret_cast<unsigned int*>(dest)) = r;
752     return true;
753     }
754    
755     bool Arg::parse_longlong_radix(const char* str,
756     int n,
757     void* dest,
758     int radix) {
759     #ifndef HAVE_LONG_LONG
760     return false;
761     #else
762     if (n == 0) return false;
763     char buf[kMaxNumberLength+1];
764     str = TerminateNumber(buf, str, n);
765     char* end;
766     errno = 0;
767     #if defined HAVE_STRTOQ
768     long long r = strtoq(str, &end, radix);
769     #elif defined HAVE_STRTOLL
770     long long r = strtoll(str, &end, radix);
771     #else
772     #error parse_longlong_radix: cannot convert input to a long-long
773     #endif
774     if (end != str + n) return false; // Leftover junk
775     if (errno) return false;
776     *(reinterpret_cast<long long*>(dest)) = r;
777     return true;
778     #endif /* HAVE_LONG_LONG */
779     }
780    
781     bool Arg::parse_ulonglong_radix(const char* str,
782     int n,
783     void* dest,
784     int radix) {
785     #ifndef HAVE_UNSIGNED_LONG_LONG
786     return false;
787     #else
788     if (n == 0) return false;
789     char buf[kMaxNumberLength+1];
790     str = TerminateNumber(buf, str, n);
791 nigel 87 if (str[0] == '-') return false; // strtoull() on a negative number?!
792 nigel 77 char* end;
793     errno = 0;
794     #if defined HAVE_STRTOQ
795     unsigned long long r = strtouq(str, &end, radix);
796     #elif defined HAVE_STRTOLL
797     unsigned long long r = strtoull(str, &end, radix);
798     #else
799     #error parse_ulonglong_radix: cannot convert input to a long-long
800     #endif
801     if (end != str + n) return false; // Leftover junk
802     if (errno) return false;
803     *(reinterpret_cast<unsigned long long*>(dest)) = r;
804     return true;
805     #endif /* HAVE_UNSIGNED_LONG_LONG */
806     }
807    
808     bool Arg::parse_double(const char* str, int n, void* dest) {
809     if (n == 0) return false;
810     static const int kMaxLength = 200;
811     char buf[kMaxLength];
812     if (n >= kMaxLength) return false;
813     memcpy(buf, str, n);
814     buf[n] = '\0';
815     errno = 0;
816     char* end;
817     double r = strtod(buf, &end);
818     if (end != buf + n) return false; // Leftover junk
819     if (errno) return false;
820     *(reinterpret_cast<double*>(dest)) = r;
821     return true;
822     }
823    
824     bool Arg::parse_float(const char* str, int n, void* dest) {
825     double r;
826     if (!parse_double(str, n, &r)) return false;
827     *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
828     return true;
829     }
830    
831    
832     #define DEFINE_INTEGER_PARSERS(name) \
833     bool Arg::parse_##name(const char* str, int n, void* dest) { \
834     return parse_##name##_radix(str, n, dest, 10); \
835     } \
836     bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \
837     return parse_##name##_radix(str, n, dest, 16); \
838     } \
839     bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \
840     return parse_##name##_radix(str, n, dest, 8); \
841     } \
842     bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
843     return parse_##name##_radix(str, n, dest, 0); \
844     }
845    
846 nigel 93 DEFINE_INTEGER_PARSERS(short) /* */
847     DEFINE_INTEGER_PARSERS(ushort) /* */
848     DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */
849     DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */
850     DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */
851     DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */
852     DEFINE_INTEGER_PARSERS(longlong) /* */
853     DEFINE_INTEGER_PARSERS(ulonglong) /* */
854 nigel 77
855     #undef DEFINE_INTEGER_PARSERS
856    
857     } // 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