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

Diff of /code/trunk/pcrecpp.cc

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 77 by nigel, Sat Feb 24 21:40:45 2007 UTC revision 257 by ph10, Wed Sep 19 09:11:19 2007 UTC
# Line 29  Line 29 
29  //  //
30  // Author: Sanjay Ghemawat  // Author: Sanjay Ghemawat
31    
32    #ifdef HAVE_CONFIG_H
33    #include "config.h"
34    #endif
35    
36  #include <stdlib.h>  #include <stdlib.h>
37  #include <stdio.h>  #include <stdio.h>
38  #include <ctype.h>  #include <ctype.h>
# Line 36  Line 40 
40  #include <assert.h>  #include <assert.h>
41  #include <errno.h>  #include <errno.h>
42  #include <string>  #include <string>
43  #include "config.h"  #include <algorithm>
44  // We need this to compile the proper dll on windows/msys.  This is copied  
45  // from pcre_internal.h.  It would probably be better just to include that.  #include "pcrecpp_internal.h"
 #define PCRE_DEFINITION  /* Win32 __declspec(export) trigger for .dll */  
46  #include "pcre.h"  #include "pcre.h"
 #include "pcre_stringpiece.h"  
47  #include "pcrecpp.h"  #include "pcrecpp.h"
48    #include "pcre_stringpiece.h"
49    
50    
51  namespace pcrecpp {  namespace pcrecpp {
# Line 52  static const int kMaxArgs = 16; Line 55  static const int kMaxArgs = 16;
55  static const int kVecSize = (1 + kMaxArgs) * 3;  // results + PCRE workspace  static const int kVecSize = (1 + kMaxArgs) * 3;  // results + PCRE workspace
56    
57  // Special object that stands-in for no argument  // Special object that stands-in for no argument
58  Arg no_arg((void*)NULL);  PCRECPP_EXP_DEFN Arg no_arg((void*)NULL);
59    
60  // If a regular expression has no error, its error_ field points here  // If a regular expression has no error, its error_ field points here
61  static const string empty_string;  static const string empty_string;
# Line 60  static const string empty_string; Line 63  static const string empty_string;
63  // If the user doesn't ask for any options, we just use this one  // If the user doesn't ask for any options, we just use this one
64  static RE_Options default_options;  static RE_Options default_options;
65    
66  void RE::Init(const char* pat, const RE_Options* options) {  void RE::Init(const string& pat, const RE_Options* options) {
67    pattern_ = pat;    pattern_ = pat;
68    if (options == NULL) {    if (options == NULL) {
69      options_ = default_options;      options_ = default_options;
# Line 73  void RE::Init(const char* pat, const RE_ Line 76  void RE::Init(const char* pat, const RE_
76    
77    re_partial_ = Compile(UNANCHORED);    re_partial_ = Compile(UNANCHORED);
78    if (re_partial_ != NULL) {    if (re_partial_ != NULL) {
79      // Check for complicated patterns.  The following change is      re_full_ = Compile(ANCHOR_BOTH);
     // conservative in that it may treat some "simple" patterns  
     // as "complex" (e.g., if the vertical bar is in a character  
     // class or is escaped).  But it seems good enough.  
     if (strchr(pat, '|') == NULL) {  
       // Simple pattern: we can use position-based checks to perform  
       // fully anchored matches  
       re_full_ = re_partial_;  
     } else {  
       // We need a special pattern for anchored matches  
       re_full_ = Compile(ANCHOR_BOTH);  
     }  
80    }    }
81  }  }
82    
83    void RE::Cleanup() {
84      if (re_full_ != NULL)         (*pcre_free)(re_full_);
85      if (re_partial_ != NULL)      (*pcre_free)(re_partial_);
86      if (error_ != &empty_string)  delete error_;
87    }
88    
89    
90  RE::~RE() {  RE::~RE() {
91    if (re_full_ != NULL && re_full_ != re_partial_) (*pcre_free)(re_full_);    Cleanup();
   if (re_partial_ != NULL)                         (*pcre_free)(re_partial_);  
   if (error_ != &empty_string)                     delete error_;  
92  }  }
93    
94    
95  pcre* RE::Compile(Anchor anchor) {  pcre* RE::Compile(Anchor anchor) {
96    // First, convert RE_Options into pcre options    // First, convert RE_Options into pcre options
97    int pcre_options = 0;    int pcre_options = 0;
98    if (options_.utf8())    pcre_options = options_.all_options();
     pcre_options |= PCRE_UTF8;  
99    
100    // Special treatment for anchoring.  This is needed because at    // Special treatment for anchoring.  This is needed because at
101    // runtime pcre only provides an option for anchoring at the    // runtime pcre only provides an option for anchoring at the
# Line 332  bool RE::Replace(const StringPiece& rewr Line 329  bool RE::Replace(const StringPiece& rewr
329    return true;    return true;
330  }  }
331    
332    // Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
333    // Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
334    // Modified by PH to add PCRE_NEWLINE_ANY and PCRE_NEWLINE_ANYCRLF.
335    
336    static int NewlineMode(int pcre_options) {
337      // TODO: if we can make it threadsafe, cache this var
338      int newline_mode = 0;
339      /* if (newline_mode) return newline_mode; */  // do this once it's cached
340      if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
341                          PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF)) {
342        newline_mode = (pcre_options &
343                        (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
344                         PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF));
345      } else {
346        int newline;
347        pcre_config(PCRE_CONFIG_NEWLINE, &newline);
348        if (newline == 10)
349          newline_mode = PCRE_NEWLINE_LF;
350        else if (newline == 13)
351          newline_mode = PCRE_NEWLINE_CR;
352        else if (newline == 3338)
353          newline_mode = PCRE_NEWLINE_CRLF;
354        else if (newline == -1)
355          newline_mode = PCRE_NEWLINE_ANY;
356        else if (newline == -2)
357          newline_mode = PCRE_NEWLINE_ANYCRLF;
358        else
359          assert("" == "Unexpected return value from pcre_config(NEWLINE)");
360      }
361      return newline_mode;
362    }
363    
364  int RE::GlobalReplace(const StringPiece& rewrite,  int RE::GlobalReplace(const StringPiece& rewrite,
365                        string *str) const {                        string *str) const {
366    int count = 0;    int count = 0;
# Line 350  int RE::GlobalReplace(const StringPiece& Line 379  int RE::GlobalReplace(const StringPiece&
379      if (matchstart == matchend && matchstart == lastend) {      if (matchstart == matchend && matchstart == lastend) {
380        // advance one character if we matched an empty string at the same        // advance one character if we matched an empty string at the same
381        // place as the last match occurred        // place as the last match occurred
382        if (start < static_cast<int>(str->length()))        matchend = start + 1;
383          out.push_back((*str)[start]);        // If the current char is CR and we're in CRLF mode, skip LF too.
384        start++;        // Note it's better to call pcre_fullinfo() than to examine
385          // all_options(), since options_ could have changed bewteen
386          // compile-time and now, but this is simpler and safe enough.
387          // Modified by PH to add ANY and ANYCRLF.
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               NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANY ||
392               NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANYCRLF)
393              ) {
394            matchend++;
395          }
396          // We also need to advance more than one char if we're in utf8 mode.
397    #ifdef SUPPORT_UTF8
398          if (options_.utf8()) {
399            while (matchend < static_cast<int>(str->length()) &&
400                   ((*str)[matchend] & 0xc0) == 0x80)
401              matchend++;
402          }
403    #endif
404          if (matchend <= static_cast<int>(str->length()))
405            out.append(*str, start, matchend - start);
406          start = matchend;
407      } else {      } else {
408        out.append(*str, start, matchstart - start);        out.append(*str, start, matchstart - start);
409        Rewrite(&out, rewrite, *str, vec, matches);        Rewrite(&out, rewrite, *str, vec, matches);
# Line 378  bool RE::Extract(const StringPiece& rewr Line 429  bool RE::Extract(const StringPiece& rewr
429    int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);    int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
430    if (matches == 0)    if (matches == 0)
431      return false;      return false;
432    out->clear();    out->erase();
433    return Rewrite(out, rewrite, text, vec, matches);    return Rewrite(out, rewrite, text, vec, matches);
434  }  }
435    
436    /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
437      string result;
438    
439      // Escape any ascii character not in [A-Za-z_0-9].
440      //
441      // Note that it's legal to escape a character even if it has no
442      // special meaning in a regular expression -- so this function does
443      // that.  (This also makes it identical to the perl function of the
444      // same name; see `perldoc -f quotemeta`.)
445      for (int ii = 0; ii < unquoted.size(); ++ii) {
446        // Note that using 'isalnum' here raises the benchmark time from
447        // 32ns to 58ns:
448        if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
449            (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
450            (unquoted[ii] < '0' || unquoted[ii] > '9') &&
451            unquoted[ii] != '_' &&
452            // If this is the part of a UTF8 or Latin1 character, we need
453            // to copy this byte without escaping.  Experimentally this is
454            // what works correctly with the regexp library.
455            !(unquoted[ii] & 128)) {
456          result += '\\';
457        }
458        result += unquoted[ii];
459      }
460    
461      return result;
462    }
463    
464  /***** Actual matching and rewriting code *****/  /***** Actual matching and rewriting code *****/
465    
466  int RE::TryMatch(const StringPiece& text,  int RE::TryMatch(const StringPiece& text,
# Line 395  int RE::TryMatch(const StringPiece& text Line 474  int RE::TryMatch(const StringPiece& text
474      return 0;      return 0;
475    }    }
476    
477    pcre_extra extra = { 0 };    pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
478    if (options_.match_limit() > 0) {    if (options_.match_limit() > 0) {
479      extra.flags = PCRE_EXTRA_MATCH_LIMIT;      extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
480      extra.match_limit = options_.match_limit();      extra.match_limit = options_.match_limit();
481    }    }
482      if (options_.match_limit_recursion() > 0) {
483        extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
484        extra.match_limit_recursion = options_.match_limit_recursion();
485      }
486    int rc = pcre_exec(re,              // The regular expression object    int rc = pcre_exec(re,              // The regular expression object
487                       &extra,                       &extra,
488                       text.data(),                       (text.data() == NULL) ? "" : text.data(),
489                       text.size(),                       text.size(),
490                       startpos,                       startpos,
491                       (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,                       (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
# Line 424  int RE::TryMatch(const StringPiece& text Line 507  int RE::TryMatch(const StringPiece& text
507      rc = vecsize / 2;      rc = vecsize / 2;
508    }    }
509    
   if ((anchor == ANCHOR_BOTH) && (re_full_ == re_partial_)) {  
     // We need an extra check to make sure that the match extended  
     // to the end of the input string  
     assert(vec[0] == 0);                 // PCRE_ANCHORED forces starting match  
     if (vec[1] != text.size()) return 0; // Did not get ending match  
   }  
   
510    return rc;    return rc;
511  }  }
512    
# Line 449  bool RE::DoMatchImpl(const StringPiece& Line 525  bool RE::DoMatchImpl(const StringPiece&
525    
526    *consumed = vec[1];    *consumed = vec[1];
527    
528    if (args == NULL) {    if (n == 0 || args == NULL) {
529      // We are not interested in results      // We are not interested in results
530      return true;      return true;
531    }    }
532    
533      if (NumberOfCapturingGroups() < n) {
534        // RE has fewer capturing groups than number of arg pointers passed in
535        return false;
536      }
537    
538    // If we got here, we must have matched the whole pattern.    // If we got here, we must have matched the whole pattern.
539    // We do not need (can not do) any more checks on the value of 'matches' here    // We do not need (can not do) any more checks on the value of 'matches' here
540    // -- see the comment for TryMatch.    // -- see the comment for TryMatch.
# Line 517  bool RE::Rewrite(string *out, const Stri Line 598  bool RE::Rewrite(string *out, const Stri
598    
599  // Return the number of capturing subpatterns, or -1 if the  // Return the number of capturing subpatterns, or -1 if the
600  // regexp wasn't valid on construction.  // regexp wasn't valid on construction.
601  int RE::NumberOfCapturingGroups() {  int RE::NumberOfCapturingGroups() const {
602    if (re_partial_ == NULL) return -1;    if (re_partial_ == NULL) return -1;
603    
604    int result;    int result;
# Line 613  bool Arg::parse_ulong_radix(const char* Line 694  bool Arg::parse_ulong_radix(const char*
694    if (n == 0) return false;    if (n == 0) return false;
695    char buf[kMaxNumberLength+1];    char buf[kMaxNumberLength+1];
696    str = TerminateNumber(buf, str, n);    str = TerminateNumber(buf, str, n);
697      if (str[0] == '-') return false;    // strtoul() on a negative number?!
698    char* end;    char* end;
699    errno = 0;    errno = 0;
700    unsigned long r = strtoul(str, &end, radix);    unsigned long r = strtoul(str, &end, radix);
# Line 629  bool Arg::parse_short_radix(const char* Line 711  bool Arg::parse_short_radix(const char*
711    long r;    long r;
712    if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse    if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
713    if (r < SHRT_MIN || r > SHRT_MAX) return false;       // Out of range    if (r < SHRT_MIN || r > SHRT_MAX) return false;       // Out of range
714    *(reinterpret_cast<short*>(dest)) = r;    *(reinterpret_cast<short*>(dest)) = static_cast<short>(r);
715    return true;    return true;
716  }  }
717    
# Line 640  bool Arg::parse_ushort_radix(const char* Line 722  bool Arg::parse_ushort_radix(const char*
722    unsigned long r;    unsigned long r;
723    if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse    if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
724    if (r > USHRT_MAX) return false;                      // Out of range    if (r > USHRT_MAX) return false;                      // Out of range
725    *(reinterpret_cast<unsigned short*>(dest)) = r;    *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r);
726    return true;    return true;
727  }  }
728    
# Line 682  bool Arg::parse_longlong_radix(const cha Line 764  bool Arg::parse_longlong_radix(const cha
764    long long r = strtoq(str, &end, radix);    long long r = strtoq(str, &end, radix);
765  #elif defined HAVE_STRTOLL  #elif defined HAVE_STRTOLL
766    long long r = strtoll(str, &end, radix);    long long r = strtoll(str, &end, radix);
767    #elif defined HAVE__STRTOI64
768      long long r = _strtoi64(str, &end, radix);
769  #else  #else
770  #error parse_longlong_radix: cannot convert input to a long-long  #error parse_longlong_radix: cannot convert input to a long-long
771  #endif  #endif
# Line 702  bool Arg::parse_ulonglong_radix(const ch Line 786  bool Arg::parse_ulonglong_radix(const ch
786    if (n == 0) return false;    if (n == 0) return false;
787    char buf[kMaxNumberLength+1];    char buf[kMaxNumberLength+1];
788    str = TerminateNumber(buf, str, n);    str = TerminateNumber(buf, str, n);
789      if (str[0] == '-') return false;    // strtoull() on a negative number?!
790    char* end;    char* end;
791    errno = 0;    errno = 0;
792  #if defined HAVE_STRTOQ  #if defined HAVE_STRTOQ
793    unsigned long long r = strtouq(str, &end, radix);    unsigned long long r = strtouq(str, &end, radix);
794  #elif defined HAVE_STRTOLL  #elif defined HAVE_STRTOLL
795    unsigned long long r = strtoull(str, &end, radix);    unsigned long long r = strtoull(str, &end, radix);
796    #elif defined HAVE__STRTOI64
797      unsigned long long r = _strtoui64(str, &end, radix);
798  #else  #else
799  #error parse_ulonglong_radix: cannot convert input to a long-long  #error parse_ulonglong_radix: cannot convert input to a long-long
800  #endif  #endif
# Line 756  bool Arg::parse_float(const char* str, i Line 843  bool Arg::parse_float(const char* str, i
843      return parse_##name##_radix(str, n, dest, 0);                       \      return parse_##name##_radix(str, n, dest, 0);                       \
844    }    }
845    
846  DEFINE_INTEGER_PARSERS(short);  DEFINE_INTEGER_PARSERS(short)      /*                                   */
847  DEFINE_INTEGER_PARSERS(ushort);  DEFINE_INTEGER_PARSERS(ushort)     /*                                   */
848  DEFINE_INTEGER_PARSERS(int);  DEFINE_INTEGER_PARSERS(int)        /* Don't use semicolons after these  */
849  DEFINE_INTEGER_PARSERS(uint);  DEFINE_INTEGER_PARSERS(uint)       /* statements because they can cause */
850  DEFINE_INTEGER_PARSERS(long);  DEFINE_INTEGER_PARSERS(long)       /* compiler warnings if the checking */
851  DEFINE_INTEGER_PARSERS(ulong);  DEFINE_INTEGER_PARSERS(ulong)      /* level is turned up high enough.   */
852  DEFINE_INTEGER_PARSERS(longlong);  DEFINE_INTEGER_PARSERS(longlong)   /*                                   */
853  DEFINE_INTEGER_PARSERS(ulonglong);  DEFINE_INTEGER_PARSERS(ulonglong)  /*                                   */
854    
855  #undef DEFINE_INTEGER_PARSERS  #undef DEFINE_INTEGER_PARSERS
856    

Legend:
Removed from v.77  
changed lines
  Added in v.257

  ViewVC Help
Powered by ViewVC 1.1.5