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

Contents of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log


Revision 234 - (hide annotations) (download)
Tue Sep 11 11:36:27 2007 UTC (6 years, 10 months ago) by ph10
File size: 29182 byte(s)
Macros for snprintf, strtoll, strtoull when compiling the C++ things under 
Windows.

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 199 #include <config.h>
34 ph10 97 #endif
35    
36 ph10 234 #ifdef _WIN32
37     #define HAVE_STRTOQ 1
38     #define strtoll _strtoui64
39     #define strtoull _strtoi64
40     #endif
41    
42 nigel 77 #include <stdlib.h>
43     #include <stdio.h>
44     #include <ctype.h>
45     #include <limits.h> /* for SHRT_MIN, USHRT_MAX, etc */
46     #include <assert.h>
47     #include <errno.h>
48     #include <string>
49 nigel 81 #include <algorithm>
50 ph10 199
51     #include "pcrecpp_internal.h"
52 ph10 137 #include <pcre.h>
53 nigel 77 #include "pcrecpp.h"
54 ph10 199 #include "pcre_stringpiece.h"
55 nigel 77
56    
57     namespace pcrecpp {
58    
59     // Maximum number of args we can set
60     static const int kMaxArgs = 16;
61     static const int kVecSize = (1 + kMaxArgs) * 3; // results + PCRE workspace
62    
63     // Special object that stands-in for no argument
64 ph10 200 PCRECPP_EXP_DEFN Arg no_arg((void*)NULL);
65 nigel 77
66     // If a regular expression has no error, its error_ field points here
67     static const string empty_string;
68    
69     // If the user doesn't ask for any options, we just use this one
70     static RE_Options default_options;
71    
72 nigel 93 void RE::Init(const string& pat, const RE_Options* options) {
73 nigel 77 pattern_ = pat;
74     if (options == NULL) {
75     options_ = default_options;
76     } else {
77     options_ = *options;
78     }
79     error_ = &empty_string;
80     re_full_ = NULL;
81     re_partial_ = NULL;
82    
83     re_partial_ = Compile(UNANCHORED);
84     if (re_partial_ != NULL) {
85 ph10 179 re_full_ = Compile(ANCHOR_BOTH);
86 nigel 77 }
87     }
88    
89 nigel 93 void RE::Cleanup() {
90 ph10 179 if (re_full_ != NULL) (*pcre_free)(re_full_);
91     if (re_partial_ != NULL) (*pcre_free)(re_partial_);
92     if (error_ != &empty_string) delete error_;
93 nigel 77 }
94    
95 nigel 93
96     RE::~RE() {
97     Cleanup();
98     }
99    
100    
101 nigel 77 pcre* RE::Compile(Anchor anchor) {
102     // First, convert RE_Options into pcre options
103     int pcre_options = 0;
104 nigel 81 pcre_options = options_.all_options();
105 nigel 77
106     // Special treatment for anchoring. This is needed because at
107     // runtime pcre only provides an option for anchoring at the
108     // beginning of a string (unless you use offset).
109     //
110     // There are three types of anchoring we want:
111     // UNANCHORED Compile the original pattern, and use
112     // a pcre unanchored match.
113     // ANCHOR_START Compile the original pattern, and use
114     // a pcre anchored match.
115     // ANCHOR_BOTH Tack a "\z" to the end of the original pattern
116     // and use a pcre anchored match.
117    
118     const char* compile_error;
119     int eoffset;
120     pcre* re;
121     if (anchor != ANCHOR_BOTH) {
122     re = pcre_compile(pattern_.c_str(), pcre_options,
123     &compile_error, &eoffset, NULL);
124     } else {
125     // Tack a '\z' at the end of RE. Parenthesize it first so that
126     // the '\z' applies to all top-level alternatives in the regexp.
127     string wrapped = "(?:"; // A non-counting grouping operator
128     wrapped += pattern_;
129     wrapped += ")\\z";
130     re = pcre_compile(wrapped.c_str(), pcre_options,
131     &compile_error, &eoffset, NULL);
132     }
133     if (re == NULL) {
134     if (error_ == &empty_string) error_ = new string(compile_error);
135     }
136     return re;
137     }
138    
139     /***** Matching interfaces *****/
140    
141     bool RE::FullMatch(const StringPiece& text,
142     const Arg& ptr1,
143     const Arg& ptr2,
144     const Arg& ptr3,
145     const Arg& ptr4,
146     const Arg& ptr5,
147     const Arg& ptr6,
148     const Arg& ptr7,
149     const Arg& ptr8,
150     const Arg& ptr9,
151     const Arg& ptr10,
152     const Arg& ptr11,
153     const Arg& ptr12,
154     const Arg& ptr13,
155     const Arg& ptr14,
156     const Arg& ptr15,
157     const Arg& ptr16) const {
158     const Arg* args[kMaxArgs];
159     int n = 0;
160     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
161     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
162     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
163     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
164     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
165     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
166     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
167     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
168     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
169     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
170     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
171     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
172     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
173     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
174     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
175     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
176     done:
177    
178     int consumed;
179     int vec[kVecSize];
180     return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
181     }
182    
183     bool RE::PartialMatch(const StringPiece& text,
184     const Arg& ptr1,
185     const Arg& ptr2,
186     const Arg& ptr3,
187     const Arg& ptr4,
188     const Arg& ptr5,
189     const Arg& ptr6,
190     const Arg& ptr7,
191     const Arg& ptr8,
192     const Arg& ptr9,
193     const Arg& ptr10,
194     const Arg& ptr11,
195     const Arg& ptr12,
196     const Arg& ptr13,
197     const Arg& ptr14,
198     const Arg& ptr15,
199     const Arg& ptr16) const {
200     const Arg* args[kMaxArgs];
201     int n = 0;
202     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
203     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
204     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
205     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
206     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
207     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
208     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
209     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
210     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
211     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
212     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
213     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
214     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
215     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
216     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
217     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
218     done:
219    
220     int consumed;
221     int vec[kVecSize];
222     return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
223     }
224    
225     bool RE::Consume(StringPiece* input,
226     const Arg& ptr1,
227     const Arg& ptr2,
228     const Arg& ptr3,
229     const Arg& ptr4,
230     const Arg& ptr5,
231     const Arg& ptr6,
232     const Arg& ptr7,
233     const Arg& ptr8,
234     const Arg& ptr9,
235     const Arg& ptr10,
236     const Arg& ptr11,
237     const Arg& ptr12,
238     const Arg& ptr13,
239     const Arg& ptr14,
240     const Arg& ptr15,
241     const Arg& ptr16) const {
242     const Arg* args[kMaxArgs];
243     int n = 0;
244     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
245     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
246     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
247     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
248     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
249     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
250     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
251     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
252     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
253     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
254     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
255     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
256     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
257     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
258     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
259     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
260     done:
261    
262     int consumed;
263     int vec[kVecSize];
264     if (DoMatchImpl(*input, ANCHOR_START, &consumed,
265     args, n, vec, kVecSize)) {
266     input->remove_prefix(consumed);
267     return true;
268     } else {
269     return false;
270     }
271     }
272    
273     bool RE::FindAndConsume(StringPiece* input,
274     const Arg& ptr1,
275     const Arg& ptr2,
276     const Arg& ptr3,
277     const Arg& ptr4,
278     const Arg& ptr5,
279     const Arg& ptr6,
280     const Arg& ptr7,
281     const Arg& ptr8,
282     const Arg& ptr9,
283     const Arg& ptr10,
284     const Arg& ptr11,
285     const Arg& ptr12,
286     const Arg& ptr13,
287     const Arg& ptr14,
288     const Arg& ptr15,
289     const Arg& ptr16) const {
290     const Arg* args[kMaxArgs];
291     int n = 0;
292     if (&ptr1 == &no_arg) goto done; args[n++] = &ptr1;
293     if (&ptr2 == &no_arg) goto done; args[n++] = &ptr2;
294     if (&ptr3 == &no_arg) goto done; args[n++] = &ptr3;
295     if (&ptr4 == &no_arg) goto done; args[n++] = &ptr4;
296     if (&ptr5 == &no_arg) goto done; args[n++] = &ptr5;
297     if (&ptr6 == &no_arg) goto done; args[n++] = &ptr6;
298     if (&ptr7 == &no_arg) goto done; args[n++] = &ptr7;
299     if (&ptr8 == &no_arg) goto done; args[n++] = &ptr8;
300     if (&ptr9 == &no_arg) goto done; args[n++] = &ptr9;
301     if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
302     if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
303     if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
304     if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
305     if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
306     if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
307     if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
308     done:
309    
310     int consumed;
311     int vec[kVecSize];
312     if (DoMatchImpl(*input, UNANCHORED, &consumed,
313     args, n, vec, kVecSize)) {
314     input->remove_prefix(consumed);
315     return true;
316     } else {
317     return false;
318     }
319     }
320    
321     bool RE::Replace(const StringPiece& rewrite,
322     string *str) const {
323     int vec[kVecSize];
324     int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
325     if (matches == 0)
326     return false;
327    
328     string s;
329     if (!Rewrite(&s, rewrite, *str, vec, matches))
330     return false;
331    
332     assert(vec[0] >= 0);
333     assert(vec[1] >= 0);
334     str->replace(vec[0], vec[1] - vec[0], s);
335     return true;
336     }
337    
338 nigel 91 // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
339     // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
340     static int NewlineMode(int pcre_options) {
341     // TODO: if we can make it threadsafe, cache this var
342     int newline_mode = 0;
343     /* if (newline_mode) return newline_mode; */ // do this once it's cached
344     if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF)) {
345     newline_mode = (pcre_options &
346     (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF));
347     } else {
348     int newline;
349     pcre_config(PCRE_CONFIG_NEWLINE, &newline);
350     if (newline == 10)
351     newline_mode = PCRE_NEWLINE_LF;
352     else if (newline == 13)
353     newline_mode = PCRE_NEWLINE_CR;
354     else if (newline == 3338)
355     newline_mode = PCRE_NEWLINE_CRLF;
356     else
357     assert("" == "Unexpected return value from pcre_config(NEWLINE)");
358     }
359     return newline_mode;
360     }
361    
362 nigel 77 int RE::GlobalReplace(const StringPiece& rewrite,
363     string *str) const {
364     int count = 0;
365     int vec[kVecSize];
366     string out;
367     int start = 0;
368     int lastend = -1;
369    
370     for (; start <= static_cast<int>(str->length()); count++) {
371     int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
372     if (matches <= 0)
373     break;
374     int matchstart = vec[0], matchend = vec[1];
375     assert(matchstart >= start);
376     assert(matchend >= matchstart);
377     if (matchstart == matchend && matchstart == lastend) {
378     // advance one character if we matched an empty string at the same
379     // place as the last match occurred
380 nigel 91 matchend = start + 1;
381     // If the current char is CR and we're in CRLF mode, skip LF too.
382     // Note it's better to call pcre_fullinfo() than to examine
383     // all_options(), since options_ could have changed bewteen
384     // compile-time and now, but this is simpler and safe enough.
385     if (start+1 < static_cast<int>(str->length()) &&
386     (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
387     NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF) {
388     matchend++;
389     }
390     // We also need to advance more than one char if we're in utf8 mode.
391     #ifdef SUPPORT_UTF8
392     if (options_.utf8()) {
393     while (matchend < static_cast<int>(str->length()) &&
394     ((*str)[matchend] & 0xc0) == 0x80)
395     matchend++;
396     }
397     #endif
398     if (matchend <= static_cast<int>(str->length()))
399     out.append(*str, start, matchend - start);
400     start = matchend;
401 nigel 77 } else {
402     out.append(*str, start, matchstart - start);
403     Rewrite(&out, rewrite, *str, vec, matches);
404     start = matchend;
405     lastend = matchend;
406     count++;
407     }
408     }
409    
410     if (count == 0)
411     return 0;
412    
413     if (start < static_cast<int>(str->length()))
414     out.append(*str, start, str->length() - start);
415     swap(out, *str);
416     return count;
417     }
418    
419     bool RE::Extract(const StringPiece& rewrite,
420     const StringPiece& text,
421     string *out) const {
422     int vec[kVecSize];
423     int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
424     if (matches == 0)
425     return false;
426 nigel 81 out->erase();
427 nigel 77 return Rewrite(out, rewrite, text, vec, matches);
428     }
429    
430 nigel 93 /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
431     string result;
432    
433     // Escape any ascii character not in [A-Za-z_0-9].
434     //
435     // Note that it's legal to escape a character even if it has no
436     // special meaning in a regular expression -- so this function does
437     // that. (This also makes it identical to the perl function of the
438     // same name; see `perldoc -f quotemeta`.)
439     for (int ii = 0; ii < unquoted.size(); ++ii) {
440     // Note that using 'isalnum' here raises the benchmark time from
441     // 32ns to 58ns:
442     if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
443     (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
444     (unquoted[ii] < '0' || unquoted[ii] > '9') &&
445     unquoted[ii] != '_' &&
446     // If this is the part of a UTF8 or Latin1 character, we need
447     // to copy this byte without escaping. Experimentally this is
448     // what works correctly with the regexp library.
449     !(unquoted[ii] & 128)) {
450     result += '\\';
451     }
452     result += unquoted[ii];
453     }
454    
455     return result;
456     }
457    
458 nigel 77 /***** Actual matching and rewriting code *****/
459    
460     int RE::TryMatch(const StringPiece& text,
461     int startpos,
462     Anchor anchor,
463     int *vec,
464     int vecsize) const {
465     pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
466     if (re == NULL) {
467     //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
468     return 0;
469     }
470    
471 ph10 199 pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
472 nigel 77 if (options_.match_limit() > 0) {
473 nigel 87 extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
474 nigel 77 extra.match_limit = options_.match_limit();
475     }
476 nigel 87 if (options_.match_limit_recursion() > 0) {
477     extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
478     extra.match_limit_recursion = options_.match_limit_recursion();
479     }
480 nigel 77 int rc = pcre_exec(re, // The regular expression object
481     &extra,
482 nigel 87 (text.data() == NULL) ? "" : text.data(),
483 nigel 77 text.size(),
484     startpos,
485     (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
486     vec,
487     vecsize);
488    
489     // Handle errors
490     if (rc == PCRE_ERROR_NOMATCH) {
491     return 0;
492     } else if (rc < 0) {
493     //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
494     // re, pattern_.c_str());
495     return 0;
496     } else if (rc == 0) {
497     // pcre_exec() returns 0 as a special case when the number of
498     // capturing subpatterns exceeds the size of the vector.
499     // When this happens, there is a match and the output vector
500     // is filled, but we miss out on the positions of the extra subpatterns.
501     rc = vecsize / 2;
502     }
503    
504     return rc;
505     }
506    
507     bool RE::DoMatchImpl(const StringPiece& text,
508     Anchor anchor,
509     int* consumed,
510     const Arg* const* args,
511     int n,
512     int* vec,
513     int vecsize) const {
514     assert((1 + n) * 3 <= vecsize); // results + PCRE workspace
515     int matches = TryMatch(text, 0, anchor, vec, vecsize);
516     assert(matches >= 0); // TryMatch never returns negatives
517     if (matches == 0)
518     return false;
519    
520     *consumed = vec[1];
521    
522 nigel 87 if (n == 0 || args == NULL) {
523 nigel 77 // We are not interested in results
524     return true;
525     }
526    
527 nigel 87 if (NumberOfCapturingGroups() < n) {
528     // RE has fewer capturing groups than number of arg pointers passed in
529     return false;
530     }
531    
532 nigel 77 // If we got here, we must have matched the whole pattern.
533     // We do not need (can not do) any more checks on the value of 'matches' here
534     // -- see the comment for TryMatch.
535     for (int i = 0; i < n; i++) {
536     const int start = vec[2*(i+1)];
537     const int limit = vec[2*(i+1)+1];
538     if (!args[i]->Parse(text.data() + start, limit-start)) {
539     // TODO: Should we indicate what the error was?
540     return false;
541     }
542     }
543    
544     return true;
545     }
546    
547     bool RE::DoMatch(const StringPiece& text,
548     Anchor anchor,
549     int* consumed,
550     const Arg* const args[],
551     int n) const {
552     assert(n >= 0);
553     size_t const vecsize = (1 + n) * 3; // results + PCRE workspace
554     // (as for kVecSize)
555     int space[21]; // use stack allocation for small vecsize (common case)
556     int* vec = vecsize <= 21 ? space : new int[vecsize];
557     bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
558     if (vec != space) delete [] vec;
559     return retval;
560     }
561    
562     bool RE::Rewrite(string *out, const StringPiece &rewrite,
563     const StringPiece &text, int *vec, int veclen) const {
564     for (const char *s = rewrite.data(), *end = s + rewrite.size();
565     s < end; s++) {
566     int c = *s;
567     if (c == '\\') {
568     c = *++s;
569     if (isdigit(c)) {
570     int n = (c - '0');
571     if (n >= veclen) {
572     //fprintf(stderr, requested group %d in regexp %.*s\n",
573     // n, rewrite.size(), rewrite.data());
574     return false;
575     }
576     int start = vec[2 * n];
577     if (start >= 0)
578     out->append(text.data() + start, vec[2 * n + 1] - start);
579     } else if (c == '\\') {
580     out->push_back('\\');
581     } else {
582     //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
583     // rewrite.size(), rewrite.data());
584     return false;
585     }
586     } else {
587     out->push_back(c);
588     }
589     }
590     return true;
591     }
592    
593     // Return the number of capturing subpatterns, or -1 if the
594     // regexp wasn't valid on construction.
595 nigel 87 int RE::NumberOfCapturingGroups() const {
596 nigel 77 if (re_partial_ == NULL) return -1;
597    
598     int result;
599     int pcre_retval = pcre_fullinfo(re_partial_, // The regular expression object
600     NULL, // We did not study the pattern
601     PCRE_INFO_CAPTURECOUNT,
602     &result);
603     assert(pcre_retval == 0);
604     return result;
605     }
606    
607     /***** Parsers for various types *****/
608    
609     bool Arg::parse_null(const char* str, int n, void* dest) {
610     // We fail if somebody asked us to store into a non-NULL void* pointer
611     return (dest == NULL);
612     }
613    
614     bool Arg::parse_string(const char* str, int n, void* dest) {
615     reinterpret_cast<string*>(dest)->assign(str, n);
616     return true;
617     }
618    
619     bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
620     reinterpret_cast<StringPiece*>(dest)->set(str, n);
621     return true;
622     }
623    
624     bool Arg::parse_char(const char* str, int n, void* dest) {
625     if (n != 1) return false;
626     *(reinterpret_cast<char*>(dest)) = str[0];
627     return true;
628     }
629    
630     bool Arg::parse_uchar(const char* str, int n, void* dest) {
631     if (n != 1) return false;
632     *(reinterpret_cast<unsigned char*>(dest)) = str[0];
633     return true;
634     }
635    
636     // Largest number spec that we are willing to parse
637     static const int kMaxNumberLength = 32;
638    
639     // REQUIRES "buf" must have length at least kMaxNumberLength+1
640     // REQUIRES "n > 0"
641     // Copies "str" into "buf" and null-terminates if necessary.
642     // Returns one of:
643     // a. "str" if no termination is needed
644     // b. "buf" if the string was copied and null-terminated
645     // c. "" if the input was invalid and has no hope of being parsed
646     static const char* TerminateNumber(char* buf, const char* str, int n) {
647     if ((n > 0) && isspace(*str)) {
648     // We are less forgiving than the strtoxxx() routines and do not
649     // allow leading spaces.
650     return "";
651     }
652    
653     // See if the character right after the input text may potentially
654     // look like a digit.
655     if (isdigit(str[n]) ||
656     ((str[n] >= 'a') && (str[n] <= 'f')) ||
657     ((str[n] >= 'A') && (str[n] <= 'F'))) {
658     if (n > kMaxNumberLength) return ""; // Input too big to be a valid number
659     memcpy(buf, str, n);
660     buf[n] = '\0';
661     return buf;
662     } else {
663     // We can parse right out of the supplied string, so return it.
664     return str;
665     }
666     }
667    
668     bool Arg::parse_long_radix(const char* str,
669     int n,
670     void* dest,
671     int radix) {
672     if (n == 0) return false;
673     char buf[kMaxNumberLength+1];
674     str = TerminateNumber(buf, str, n);
675     char* end;
676     errno = 0;
677     long r = strtol(str, &end, radix);
678     if (end != str + n) return false; // Leftover junk
679     if (errno) return false;
680     *(reinterpret_cast<long*>(dest)) = r;
681     return true;
682     }
683    
684     bool Arg::parse_ulong_radix(const char* str,
685     int n,
686     void* dest,
687     int radix) {
688     if (n == 0) return false;
689     char buf[kMaxNumberLength+1];
690     str = TerminateNumber(buf, str, n);
691 nigel 87 if (str[0] == '-') return false; // strtoul() on a negative number?!
692 nigel 77 char* end;
693     errno = 0;
694     unsigned long r = strtoul(str, &end, radix);
695     if (end != str + n) return false; // Leftover junk
696     if (errno) return false;
697     *(reinterpret_cast<unsigned long*>(dest)) = r;
698     return true;
699     }
700    
701     bool Arg::parse_short_radix(const char* str,
702     int n,
703     void* dest,
704     int radix) {
705     long r;
706     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
707     if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range
708     *(reinterpret_cast<short*>(dest)) = r;
709     return true;
710     }
711    
712     bool Arg::parse_ushort_radix(const char* str,
713     int n,
714     void* dest,
715     int radix) {
716     unsigned long r;
717     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
718     if (r > USHRT_MAX) return false; // Out of range
719     *(reinterpret_cast<unsigned short*>(dest)) = r;
720     return true;
721     }
722    
723     bool Arg::parse_int_radix(const char* str,
724     int n,
725     void* dest,
726     int radix) {
727     long r;
728     if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
729     if (r < INT_MIN || r > INT_MAX) return false; // Out of range
730     *(reinterpret_cast<int*>(dest)) = r;
731     return true;
732     }
733    
734     bool Arg::parse_uint_radix(const char* str,
735     int n,
736     void* dest,
737     int radix) {
738     unsigned long r;
739     if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
740     if (r > UINT_MAX) return false; // Out of range
741     *(reinterpret_cast<unsigned int*>(dest)) = r;
742     return true;
743     }
744    
745     bool Arg::parse_longlong_radix(const char* str,
746     int n,
747     void* dest,
748     int radix) {
749     #ifndef HAVE_LONG_LONG
750     return false;
751     #else
752     if (n == 0) return false;
753     char buf[kMaxNumberLength+1];
754     str = TerminateNumber(buf, str, n);
755     char* end;
756     errno = 0;
757     #if defined HAVE_STRTOQ
758     long long r = strtoq(str, &end, radix);
759     #elif defined HAVE_STRTOLL
760     long long r = strtoll(str, &end, radix);
761     #else
762     #error parse_longlong_radix: cannot convert input to a long-long
763     #endif
764     if (end != str + n) return false; // Leftover junk
765     if (errno) return false;
766     *(reinterpret_cast<long long*>(dest)) = r;
767     return true;
768     #endif /* HAVE_LONG_LONG */
769     }
770    
771     bool Arg::parse_ulonglong_radix(const char* str,
772     int n,
773     void* dest,
774     int radix) {
775     #ifndef HAVE_UNSIGNED_LONG_LONG
776     return false;
777     #else
778     if (n == 0) return false;
779     char buf[kMaxNumberLength+1];
780     str = TerminateNumber(buf, str, n);
781 nigel 87 if (str[0] == '-') return false; // strtoull() on a negative number?!
782 nigel 77 char* end;
783     errno = 0;
784     #if defined HAVE_STRTOQ
785     unsigned long long r = strtouq(str, &end, radix);
786     #elif defined HAVE_STRTOLL
787     unsigned long long r = strtoull(str, &end, radix);
788     #else
789     #error parse_ulonglong_radix: cannot convert input to a long-long
790     #endif
791     if (end != str + n) return false; // Leftover junk
792     if (errno) return false;
793     *(reinterpret_cast<unsigned long long*>(dest)) = r;
794     return true;
795     #endif /* HAVE_UNSIGNED_LONG_LONG */
796     }
797    
798     bool Arg::parse_double(const char* str, int n, void* dest) {
799     if (n == 0) return false;
800     static const int kMaxLength = 200;
801     char buf[kMaxLength];
802     if (n >= kMaxLength) return false;
803     memcpy(buf, str, n);
804     buf[n] = '\0';
805     errno = 0;
806     char* end;
807     double r = strtod(buf, &end);
808     if (end != buf + n) return false; // Leftover junk
809     if (errno) return false;
810     *(reinterpret_cast<double*>(dest)) = r;
811     return true;
812     }
813    
814     bool Arg::parse_float(const char* str, int n, void* dest) {
815     double r;
816     if (!parse_double(str, n, &r)) return false;
817     *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
818     return true;
819     }
820    
821    
822     #define DEFINE_INTEGER_PARSERS(name) \
823     bool Arg::parse_##name(const char* str, int n, void* dest) { \
824     return parse_##name##_radix(str, n, dest, 10); \
825     } \
826     bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \
827     return parse_##name##_radix(str, n, dest, 16); \
828     } \
829     bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \
830     return parse_##name##_radix(str, n, dest, 8); \
831     } \
832     bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
833     return parse_##name##_radix(str, n, dest, 0); \
834     }
835    
836 nigel 93 DEFINE_INTEGER_PARSERS(short) /* */
837     DEFINE_INTEGER_PARSERS(ushort) /* */
838     DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */
839     DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */
840     DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */
841     DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */
842     DEFINE_INTEGER_PARSERS(longlong) /* */
843     DEFINE_INTEGER_PARSERS(ulonglong) /* */
844 nigel 77
845     #undef DEFINE_INTEGER_PARSERS
846    
847     } // 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