/[pcre]/code/trunk/pcre_exec.c
ViewVC logotype

Diff of /code/trunk/pcre_exec.c

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

revision 553 by ph10, Fri Oct 22 15:57:50 2010 UTC revision 598 by ph10, Sat May 7 15:37:31 2011 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2011 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 132  while (length-- > 0) Line 132  while (length-- > 0)
132  *          Match a back-reference                *  *          Match a back-reference                *
133  *************************************************/  *************************************************/
134    
135  /* If a back reference hasn't been set, the length that is passed is greater  /* Normally, if a back reference hasn't been set, the length that is passed is
136  than the number of characters left in the string, so the match fails.  negative, so the match always fails. However, in JavaScript compatibility mode,
137    the length passed is zero. Note that in caseless UTF-8 mode, the number of
138    subject bytes matched may be different to the number of reference bytes.
139    
140  Arguments:  Arguments:
141    offset      index into the offset vector    offset      index into the offset vector
142    eptr        points into the subject    eptr        pointer into the subject
143    length      length to be matched    length      length of reference to be matched (number of bytes)
144    md          points to match data block    md          points to match data block
145    ims         the ims flags    ims         the ims flags
146    
147  Returns:      TRUE if matched  Returns:      < 0 if not matched, otherwise the number of subject bytes matched
148  */  */
149    
150  static BOOL  static int
151  match_ref(int offset, register USPTR eptr, int length, match_data *md,  match_ref(int offset, register USPTR eptr, int length, match_data *md,
152    unsigned long int ims)    unsigned long int ims)
153  {  {
154  USPTR p = md->start_subject + md->offset_vector[offset];  USPTR eptr_start = eptr;
155    register USPTR p = md->start_subject + md->offset_vector[offset];
156    
157  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
158  if (eptr >= md->end_subject)  if (eptr >= md->end_subject)
# Line 164  pchars(p, length, FALSE, md); Line 167  pchars(p, length, FALSE, md);
167  printf("\n");  printf("\n");
168  #endif  #endif
169    
170  /* Always fail if not enough characters left */  /* Always fail if reference not set (and not JavaScript compatible). */
171    
172  if (length > md->end_subject - eptr) return FALSE;  if (length < 0) return -1;
173    
174  /* Separate the caseless case for speed. In UTF-8 mode we can only do this  /* Separate the caseless case for speed. In UTF-8 mode we can only do this
175  properly if Unicode properties are supported. Otherwise, we can check only  properly if Unicode properties are supported. Otherwise, we can check only
# Line 178  if ((ims & PCRE_CASELESS) != 0) Line 181  if ((ims & PCRE_CASELESS) != 0)
181  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
182    if (md->utf8)    if (md->utf8)
183      {      {
184      USPTR endptr = eptr + length;      /* Match characters up to the end of the reference. NOTE: the number of
185      while (eptr < endptr)      bytes matched may differ, because there are some characters whose upper and
186        lower case versions code as different numbers of bytes. For example, U+023A
187        (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8);
188        a sequence of 3 of the former uses 6 bytes, as does a sequence of two of
189        the latter. It is important, therefore, to check the length along the
190        reference, not along the subject (earlier code did this wrong). */
191    
192        USPTR endptr = p + length;
193        while (p < endptr)
194        {        {
195        int c, d;        int c, d;
196          if (eptr >= md->end_subject) return -1;
197        GETCHARINC(c, eptr);        GETCHARINC(c, eptr);
198        GETCHARINC(d, p);        GETCHARINC(d, p);
199        if (c != d && c != UCD_OTHERCASE(d)) return FALSE;        if (c != d && c != UCD_OTHERCASE(d)) return -1;
200        }        }
201      }      }
202    else    else
# Line 193  if ((ims & PCRE_CASELESS) != 0) Line 205  if ((ims & PCRE_CASELESS) != 0)
205    
206    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
207    is no UCP support. */    is no UCP support. */
208        {
209    while (length-- > 0)      if (eptr + length > md->end_subject) return -1;
210      { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }      while (length-- > 0)
211          { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; }
212        }
213    }    }
214    
215  /* In the caseful case, we can just compare the bytes, whether or not we  /* In the caseful case, we can just compare the bytes, whether or not we
216  are in UTF-8 mode. */  are in UTF-8 mode. */
217    
218  else  else
219    { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }    {
220      if (eptr + length > md->end_subject) return -1;
221      while (length-- > 0) if (*p++ != *eptr++) return -1;
222      }
223    
224  return TRUE;  return eptr - eptr_start;
225  }  }
226    
227    
# Line 294  argument of match(), which never changes Line 311  argument of match(), which never changes
311    
312  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\  #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\
313    {\    {\
314    heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\    heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\
315    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\    if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\
316    frame->Xwhere = rw; \    frame->Xwhere = rw; \
317    newframe->Xeptr = ra;\    newframe->Xeptr = ra;\
# Line 489  heap storage. Set up the top-level frame Line 506  heap storage. Set up the top-level frame
506  heap whenever RMATCH() does a "recursion". See the macro definitions above. */  heap whenever RMATCH() does a "recursion". See the macro definitions above. */
507    
508  #ifdef NO_RECURSE  #ifdef NO_RECURSE
509  heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));  heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));
510  if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);  if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
511  frame->Xprevframe = NULL;            /* Marks the top level */  frame->Xprevframe = NULL;            /* Marks the top level */
512    
# Line 1705  for (;;) Line 1722  for (;;)
1722        if (eptr < md->end_subject)        if (eptr < md->end_subject)
1723          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }          { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); }
1724        else        else
1725          {          {
1726          if (md->noteol) MRRETURN(MATCH_NOMATCH);          if (md->noteol) MRRETURN(MATCH_NOMATCH);
1727          SCHECK_PARTIAL();          SCHECK_PARTIAL();
1728          }          }
1729        ecode++;        ecode++;
# Line 1717  for (;;) Line 1734  for (;;)
1734        if (md->noteol) MRRETURN(MATCH_NOMATCH);        if (md->noteol) MRRETURN(MATCH_NOMATCH);
1735        if (!md->endonly) goto ASSERT_NL_OR_EOS;        if (!md->endonly) goto ASSERT_NL_OR_EOS;
1736        }        }
1737    
1738      /* ... else fall through for endonly */      /* ... else fall through for endonly */
1739    
1740      /* End of subject assertion (\z) */      /* End of subject assertion (\z) */
# Line 1735  for (;;) Line 1752  for (;;)
1752      if (eptr < md->end_subject &&      if (eptr < md->end_subject &&
1753          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))          (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
1754        MRRETURN(MATCH_NOMATCH);        MRRETURN(MATCH_NOMATCH);
1755    
1756      /* Either at end of string or \n before end. */      /* Either at end of string or \n before end. */
1757    
1758      SCHECK_PARTIAL();      SCHECK_PARTIAL();
1759      ecode++;      ecode++;
1760      break;      break;
# Line 2252  for (;;) Line 2269  for (;;)
2269      loops). */      loops). */
2270    
2271      case OP_REF:      case OP_REF:
2272        {      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2273        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */      ecode += 3;
       ecode += 3;  
2274    
2275        /* If the reference is unset, there are two possibilities:      /* If the reference is unset, there are two possibilities:
2276    
2277        (a) In the default, Perl-compatible state, set the length to be longer      (a) In the default, Perl-compatible state, set the length negative;
2278        than the amount of subject left; this ensures that every attempt at a      this ensures that every attempt at a match fails. We can't just fail
2279        match fails. We can't just fail here, because of the possibility of      here, because of the possibility of quantifiers with zero minima.
       quantifiers with zero minima.  
2280    
2281        (b) If the JavaScript compatibility flag is set, set the length to zero      (b) If the JavaScript compatibility flag is set, set the length to zero
2282        so that the back reference matches an empty string.      so that the back reference matches an empty string.
2283    
2284        Otherwise, set the length to the length of what was matched by the      Otherwise, set the length to the length of what was matched by the
2285        referenced subpattern. */      referenced subpattern. */
2286    
2287        if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2288          length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1);        length = (md->jscript_compat)? 0 : -1;
2289        else      else
2290          length = md->offset_vector[offset+1] - md->offset_vector[offset];        length = md->offset_vector[offset+1] - md->offset_vector[offset];
2291    
2292        /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2293    
2294        switch (*ecode)      switch (*ecode)
2295          {        {
2296          case OP_CRSTAR:        case OP_CRSTAR:
2297          case OP_CRMINSTAR:        case OP_CRMINSTAR:
2298          case OP_CRPLUS:        case OP_CRPLUS:
2299          case OP_CRMINPLUS:        case OP_CRMINPLUS:
2300          case OP_CRQUERY:        case OP_CRQUERY:
2301          case OP_CRMINQUERY:        case OP_CRMINQUERY:
2302          c = *ecode++ - OP_CRSTAR;        c = *ecode++ - OP_CRSTAR;
2303          minimize = (c & 1) != 0;        minimize = (c & 1) != 0;
2304          min = rep_min[c];                 /* Pick up values from tables; */        min = rep_min[c];                 /* Pick up values from tables; */
2305          max = rep_max[c];                 /* zero for max => infinity */        max = rep_max[c];                 /* zero for max => infinity */
2306          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2307          break;        break;
2308    
2309          case OP_CRRANGE:        case OP_CRRANGE:
2310          case OP_CRMINRANGE:        case OP_CRMINRANGE:
2311          minimize = (*ecode == OP_CRMINRANGE);        minimize = (*ecode == OP_CRMINRANGE);
2312          min = GET2(ecode, 1);        min = GET2(ecode, 1);
2313          max = GET2(ecode, 3);        max = GET2(ecode, 3);
2314          if (max == 0) max = INT_MAX;        if (max == 0) max = INT_MAX;
2315          ecode += 5;        ecode += 5;
2316          break;        break;
2317    
2318          default:               /* No repeat follows */        default:               /* No repeat follows */
2319          if (!match_ref(offset, eptr, length, md, ims))        if ((length = match_ref(offset, eptr, length, md, ims)) < 0)
2320            {          {
2321            CHECK_PARTIAL();          CHECK_PARTIAL();
2322            MRRETURN(MATCH_NOMATCH);          MRRETURN(MATCH_NOMATCH);
           }  
         eptr += length;  
         continue;              /* With the main loop */  
2323          }          }
2324          eptr += length;
2325          continue;              /* With the main loop */
2326          }
2327    
2328        /* If the length of the reference is zero, just continue with the      /* Handle repeated back references. If the length of the reference is
2329        main loop. */      zero, just continue with the main loop. */
2330    
2331        if (length == 0) continue;      if (length == 0) continue;
2332    
2333        /* First, ensure the minimum number of matches are present. We get back      /* First, ensure the minimum number of matches are present. We get back
2334        the length of the reference string explicitly rather than passing the      the length of the reference string explicitly rather than passing the
2335        address of eptr, so that eptr can be a register variable. */      address of eptr, so that eptr can be a register variable. */
2336    
2337        for (i = 1; i <= min; i++)      for (i = 1; i <= min; i++)
2338          {
2339          int slength;
2340          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)
2341          {          {
2342          if (!match_ref(offset, eptr, length, md, ims))          CHECK_PARTIAL();
2343            {          MRRETURN(MATCH_NOMATCH);
           CHECK_PARTIAL();  
           MRRETURN(MATCH_NOMATCH);  
           }  
         eptr += length;  
2344          }          }
2345          eptr += slength;
2346          }
2347    
2348        /* If min = max, continue at the same level without recursion.      /* If min = max, continue at the same level without recursion.
2349        They are not both allowed to be zero. */      They are not both allowed to be zero. */
2350    
2351        if (min == max) continue;      if (min == max) continue;
2352    
2353        /* If minimizing, keep trying and advancing the pointer */      /* If minimizing, keep trying and advancing the pointer */
2354    
2355        if (minimize)      if (minimize)
2356          {
2357          for (fi = min;; fi++)
2358          {          {
2359          for (fi = min;; fi++)          int slength;
2360            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);
2361            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2362            if (fi >= max) MRRETURN(MATCH_NOMATCH);
2363            if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)
2364            {            {
2365            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14);            CHECK_PARTIAL();
2366            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            MRRETURN(MATCH_NOMATCH);
           if (fi >= max) MRRETURN(MATCH_NOMATCH);  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             MRRETURN(MATCH_NOMATCH);  
             }  
           eptr += length;  
2367            }            }
2368          /* Control never gets here */          eptr += slength;
2369          }          }
2370          /* Control never gets here */
2371          }
2372    
2373        /* If maximizing, find the longest string and work backwards */      /* If maximizing, find the longest string and work backwards */
2374    
2375        else      else
2376          {
2377          pp = eptr;
2378          for (i = min; i < max; i++)
2379          {          {
2380          pp = eptr;          int slength;
2381          for (i = min; i < max; i++)          if ((slength = match_ref(offset, eptr, length, md, ims)) < 0)
           {  
           if (!match_ref(offset, eptr, length, md, ims))  
             {  
             CHECK_PARTIAL();  
             break;  
             }  
           eptr += length;  
           }  
         while (eptr >= pp)  
2382            {            {
2383            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);            CHECK_PARTIAL();
2384            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            break;
           eptr -= length;  
2385            }            }
2386          MRRETURN(MATCH_NOMATCH);          eptr += slength;
2387            }
2388          while (eptr >= pp)
2389            {
2390            RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15);
2391            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
2392            eptr -= length;
2393          }          }
2394          MRRETURN(MATCH_NOMATCH);
2395        }        }
2396      /* Control never gets here */      /* Control never gets here */
2397    
# Line 5634  if ((options & ~PUBLIC_EXEC_OPTIONS) != Line 5651  if ((options & ~PUBLIC_EXEC_OPTIONS) !=
5651  if (re == NULL || subject == NULL ||  if (re == NULL || subject == NULL ||
5652     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;     (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL;
5653  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;  if (offsetcount < 0) return PCRE_ERROR_BADCOUNT;
5654    if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET;
5655    
5656  /* This information is for finding all the numbers associated with a given  /* This information is for finding all the numbers associated with a given
5657  name, for condition testing. */  name, for condition testing. */
# Line 5794  defined (though never set). So there's n Line 5812  defined (though never set). So there's n
5812  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)  if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0)
5813    return PCRE_ERROR_BADPARTIAL;    return PCRE_ERROR_BADPARTIAL;
5814    
5815  /* Check a UTF-8 string if required. Unfortunately there's no way of passing  /* Check a UTF-8 string if required. Pass back the character offset and error
5816  back the character offset. */  code if a results vector is available. */
5817    
5818  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5819  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0)
5820    {    {
5821    if (_pcre_valid_utf8((USPTR)subject, length) >= 0)    int errorcode;
5822      return PCRE_ERROR_BADUTF8;    int tb = _pcre_valid_utf8((USPTR)subject, length, &errorcode);
5823    if (start_offset > 0 && start_offset < length)    if (tb >= 0)
5824      {      {
5825      int tb = ((USPTR)subject)[start_offset];      if (offsetcount >= 2)
     if (tb > 127)  
5826        {        {
5827        tb &= 0xc0;        offsets[0] = tb;
5828        if (tb != 0 && tb != 0xc0) return PCRE_ERROR_BADUTF8_OFFSET;        offsets[1] = errorcode;
5829        }        }
5830        return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)?
5831          PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8;
5832        }
5833      if (start_offset > 0 && start_offset < length)
5834        {
5835        tb = ((USPTR)subject)[start_offset] & 0xc0;
5836        if (tb == 0x80) return PCRE_ERROR_BADUTF8_OFFSET;
5837      }      }
5838    }    }
5839  #endif  #endif
# Line 5937  for(;;) Line 5961  for(;;)
5961    /* There are some optimizations that avoid running the match if a known    /* There are some optimizations that avoid running the match if a known
5962    starting point is not found, or if a known later character is not present.    starting point is not found, or if a known later character is not present.
5963    However, there is an option that disables these, for testing and for ensuring    However, there is an option that disables these, for testing and for ensuring
5964    that all callouts do actually occur. */    that all callouts do actually occur. The option can be set in the regex by
5965      (*NO_START_OPT) or passed in match-time options. */
5966    
5967    if ((options & PCRE_NO_START_OPTIMIZE) == 0)    if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0)
5968      {      {
5969      /* Advance to a unique first byte if there is one. */      /* Advance to a unique first byte if there is one. */
5970    

Legend:
Removed from v.553  
changed lines
  Added in v.598

  ViewVC Help
Powered by ViewVC 1.1.5