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

Diff of /code/trunk/pcre_study.c

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

revision 673 by ph10, Thu Aug 25 16:06:03 2011 UTC revision 708 by ph10, Fri Sep 23 11:03:03 2011 UTC
# Line 66  string of that length that matches. In U Line 66  string of that length that matches. In U
66  rather than bytes.  rather than bytes.
67    
68  Arguments:  Arguments:
69    code        pointer to start of group (the bracket)    code            pointer to start of group (the bracket)
70    startcode   pointer to start of the whole pattern    startcode       pointer to start of the whole pattern
71    options     the compiling options    options         the compiling options
72    had_accept  pointer to flag for (*ACCEPT) encountered    int             RECURSE depth
   int         RECURSE depth  
73    
74  Returns:   the minimum length  Returns:   the minimum length
75             -1 if \C was encountered             -1 if \C in UTF-8 mode or (*ACCEPT) was encountered
76             -2 internal error (missing capturing bracket)             -2 internal error (missing capturing bracket)
77             -3 internal error (opcode not listed)             -3 internal error (opcode not listed)
78  */  */
79    
80  static int  static int
81  find_minlength(const uschar *code, const uschar *startcode, int options,  find_minlength(const uschar *code, const uschar *startcode, int options,
82    BOOL *had_accept_ptr, int recurse_depth)    int recurse_depth)
83  {  {
84  int length = -1;  int length = -1;
85  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
# Line 128  for (;;) Line 127  for (;;)
127      case OP_BRAPOS:      case OP_BRAPOS:
128      case OP_SBRAPOS:      case OP_SBRAPOS:
129      case OP_ONCE:      case OP_ONCE:
130      d = find_minlength(cc, startcode, options, had_accept_ptr, recurse_depth);      d = find_minlength(cc, startcode, options, recurse_depth);
131      if (d < 0) return d;      if (d < 0) return d;
132      branchlength += d;      branchlength += d;
     if (*had_accept_ptr) return branchlength;  
133      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
134      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
135      break;      break;
136    
137      /* Reached end of a branch; if it's a ket it is the end of a nested      /* ACCEPT makes things far too complicated; we have to give up. */
     call. If it's ALT it is an alternation in a nested call. If it is END it's  
     the end of the outer call. All can be handled by the same code. If it is  
     ACCEPT, it is essentially the same as END, but we set a flag so that  
     counting stops. */  
138    
139      case OP_ACCEPT:      case OP_ACCEPT:
140      case OP_ASSERT_ACCEPT:      case OP_ASSERT_ACCEPT:
141      *had_accept_ptr = TRUE;      return -1;
142      /* Fall through */  
143        /* Reached end of a branch; if it's a ket it is the end of a nested
144        call. If it's ALT it is an alternation in a nested call. If it is END it's
145        the end of the outer call. All can be handled by the same code. If an
146        ACCEPT was previously encountered, use the length that was in force at that
147        time, and pass back the shortest ACCEPT length. */
148    
149      case OP_ALT:      case OP_ALT:
150      case OP_KET:      case OP_KET:
151      case OP_KETRMAX:      case OP_KETRMAX:
# Line 379  for (;;) Line 379  for (;;)
379          }          }
380        else        else
381          {          {
382          d = find_minlength(cs, startcode, options, had_accept_ptr,          d = find_minlength(cs, startcode, options, recurse_depth);
           recurse_depth);  
         *had_accept_ptr = FALSE;  
383          }          }
384        }        }
385      else d = 0;      else d = 0;
# Line 430  for (;;) Line 428  for (;;)
428        had_recurse = TRUE;        had_recurse = TRUE;
429      else      else
430        {        {
431        branchlength += find_minlength(cs, startcode, options, had_accept_ptr,        branchlength += find_minlength(cs, startcode, options, recurse_depth + 1);
         recurse_depth + 1);  
       *had_accept_ptr = FALSE;  
432        }        }
433      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
434      break;      break;
# Line 1228  pcre_study(const pcre *external_re, int Line 1224  pcre_study(const pcre *external_re, int
1224  {  {
1225  int min;  int min;
1226  BOOL bits_set = FALSE;  BOOL bits_set = FALSE;
 BOOL had_accept = FALSE;  
1227  uschar start_bits[32];  uschar start_bits[32];
1228  pcre_extra *extra;  pcre_extra *extra = NULL;
1229  pcre_study_data *study;  pcre_study_data *study;
1230  const uschar *tables;  const uschar *tables;
1231  uschar *code;  uschar *code;
# Line 1281  if ((re->options & PCRE_ANCHORED) == 0 & Line 1276  if ((re->options & PCRE_ANCHORED) == 0 &
1276    rc = set_start_bits(code, start_bits, (re->options & PCRE_UTF8) != 0,    rc = set_start_bits(code, start_bits, (re->options & PCRE_UTF8) != 0,
1277      &compile_block);      &compile_block);
1278    bits_set = rc == SSB_DONE;    bits_set = rc == SSB_DONE;
1279    if (rc == SSB_UNKNOWN) *errorptr = "internal error: opcode not recognized";    if (rc == SSB_UNKNOWN)
1280        {
1281        *errorptr = "internal error: opcode not recognized";
1282        return NULL;
1283        }
1284    }    }
1285    
1286  /* Find the minimum length of subject string. */  /* Find the minimum length of subject string. */
1287    
1288  switch(min = find_minlength(code, code, re->options, &had_accept, 0))  switch(min = find_minlength(code, code, re->options, 0))
1289    {    {
1290    case -2: *errorptr = "internal error: missing capturing bracket"; break;    case -2: *errorptr = "internal error: missing capturing bracket"; return NULL;
1291    case -3: *errorptr = "internal error: opcode not recognized"; break;    case -3: *errorptr = "internal error: opcode not recognized"; return NULL;
1292    default: break;    default: break;
1293    }    }
1294    
1295  /* Return NULL if there's been an (internal) error or if no optimization is  /* If a set of starting bytes has been identified, or if the minimum length is
1296  possible. A FALSE setting for bits_set is common when there are no obvious  greater than zero, or if JIT optimization has been requested, get a pcre_extra
1297  starting bytes. However a negative value of min occurs only when the pattern  block and a pcre_study_data block. The study data is put in the latter, which
1298  contains \C, in other words, it's an exceptional case nowadays. */  is pointed to by the former, which may also get additional data set later by
1299    the calling program. At the moment, the size of pcre_study_data is fixed. We
1300  if (*errorptr != NULL || (!bits_set && min < 0)) return NULL;  nevertheless save it in a field for returning via the pcre_fullinfo() function
1301    so that if it becomes variable in the future, we don't have to change that
1302  /* Get a pcre_extra block and a pcre_study_data block. The study data is put in  code. */
 the latter, which is pointed to by the former, which may also get additional  
 data set later by the calling program. At the moment, the size of  
 pcre_study_data is fixed. We nevertheless save it in a field for returning via  
 the pcre_fullinfo() function so that if it becomes variable in the future, we  
 don't have to change that code. */  
1303    
1304  extra = (pcre_extra *)(pcre_malloc)  if (bits_set || min > 0
1305    (sizeof(pcre_extra) + sizeof(pcre_study_data));  #ifdef SUPPORT_JIT
1306        || (options & PCRE_STUDY_JIT_COMPILE) != 0
1307  if (extra == NULL)  #endif
1308      )
1309    {    {
1310    *errorptr = "failed to get memory";    extra = (pcre_extra *)(pcre_malloc)
1311    return NULL;      (sizeof(pcre_extra) + sizeof(pcre_study_data));
1312    }    if (extra == NULL)
1313        {
1314        *errorptr = "failed to get memory";
1315        return NULL;
1316        }
1317    
1318  study = (pcre_study_data *)((char *)extra + sizeof(pcre_extra));    study = (pcre_study_data *)((char *)extra + sizeof(pcre_extra));
1319  extra->flags = PCRE_EXTRA_STUDY_DATA;    extra->flags = PCRE_EXTRA_STUDY_DATA;
1320  extra->study_data = study;    extra->study_data = study;
1321    
1322  study->size = sizeof(pcre_study_data);    study->size = sizeof(pcre_study_data);
1323  study->flags = 0;    study->flags = 0;
1324    
1325  if (bits_set)    if (bits_set)
1326    {      {
1327    study->flags |= PCRE_STUDY_MAPPED;      study->flags |= PCRE_STUDY_MAPPED;
1328    memcpy(study->start_bits, start_bits, sizeof(start_bits));      memcpy(study->start_bits, start_bits, sizeof(start_bits));
1329    }      }
1330    
1331  if (min >= 0)    /* Always set the minlength value in the block, because the JIT compiler
1332    {    makes use of it. However, don't set the bit unless the length is greater than
1333    study->flags |= PCRE_STUDY_MINLEN;    zero - the interpretive pcre_exec() and pcre_dfa_exec() needn't waste time
1334    study->minlength = min;    checking the zero case. */
1335    }  
1336      if (min > 0)
1337        {
1338        study->flags |= PCRE_STUDY_MINLEN;
1339        study->minlength = min;
1340        }
1341      else study->minlength = 0;
1342    
1343  /* If JIT support was compiled and requested, attempt the JIT compilation. */    /* If JIT support was compiled and requested, attempt the JIT compilation.
1344      If no starting bytes were found, and the minimum length is zero, and JIT
1345      compilation fails, abandon the extra block and return NULL. */
1346    
 extra->executable_jit = NULL;  
1347  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
1348  if ((options & PCRE_STUDY_JIT_COMPILE) != 0) _pcre_jit_compile(re, extra);    extra->executable_jit = NULL;
1349      if ((options & PCRE_STUDY_JIT_COMPILE) != 0) _pcre_jit_compile(re, extra);
1350      if (study->flags == 0 && (extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) == 0)
1351        {
1352        pcre_free_study(extra);
1353        extra = NULL;
1354        }
1355  #endif  #endif
1356      }
1357    
1358  return extra;  return extra;
1359  }  }
# Line 1360  PCRE_EXP_DEFN void Line 1373  PCRE_EXP_DEFN void
1373  pcre_free_study(pcre_extra *extra)  pcre_free_study(pcre_extra *extra)
1374  {  {
1375  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
1376  if ((extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 &&  if ((extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 &&
1377       extra->executable_jit != NULL)       extra->executable_jit != NULL)
1378    _pcre_jit_free(extra->executable_jit);    _pcre_jit_free(extra->executable_jit);
1379  #endif  #endif

Legend:
Removed from v.673  
changed lines
  Added in v.708

  ViewVC Help
Powered by ViewVC 1.1.5