/[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 1313 by ph10, Wed Apr 24 12:07:09 2013 UTC revision 1376 by ph10, Sat Oct 12 18:02:11 2013 UTC
# Line 811  for (;;) Line 811  for (;;)
811      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
812    
813      /* Note that, for Perl compatibility, SKIP with an argument does NOT set      /* Note that, for Perl compatibility, SKIP with an argument does NOT set
814      nomatch_mark. When a pattern match ends with a SKIP_ARG for which there was      nomatch_mark. When a pattern match ends with a SKIP_ARG for which there was
815      not a matching mark, we have to re-run the match, ignoring the SKIP_ARG      not a matching mark, we have to re-run the match, ignoring the SKIP_ARG
816      that failed and any that preceed it (either they also failed, or were not      that failed and any that precede it (either they also failed, or were not
817      triggered). To do this, we maintain a count of executed SKIP_ARGs. If a      triggered). To do this, we maintain a count of executed SKIP_ARGs. If a
818      SKIP_ARG gets to top level, the match is re-run with md->ignore_skip_arg      SKIP_ARG gets to top level, the match is re-run with md->ignore_skip_arg
819      set to the count of the one that failed. */      set to the count of the one that failed. */
820    
# Line 828  for (;;) Line 828  for (;;)
828      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
829        eptrb, RM57);        eptrb, RM57);
830      if (rrc != MATCH_NOMATCH) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
831    
832      /* Pass back the current skip name by overloading md->start_match_ptr and      /* Pass back the current skip name by overloading md->start_match_ptr and
833      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
834      caught by a matching MARK, or get to the top, where it causes a rematch      caught by a matching MARK, or get to the top, where it causes a rematch
# Line 1274  for (;;) Line 1274  for (;;)
1274    
1275      /* Control never reaches here. */      /* Control never reaches here. */
1276    
1277      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than two
1278      two branches. If the condition is false, skipping the first branch takes us      branches. If the condition is false, skipping the first branch takes us
1279      past the end if there is only one branch, but that's OK because that is      past the end of the item if there is only one branch, but that's exactly
1280      exactly what going to the ket would do. */      what we want. */
1281    
1282      case OP_COND:      case OP_COND:
1283      case OP_SCOND:      case OP_SCOND:
1284      codelink = GET(ecode, 1);  
1285        /* The variable codelink will be added to ecode when the condition is
1286        false, to get to the second branch. Setting it to the offset to the ALT
1287        or KET, then incrementing ecode achieves this effect. We now have ecode
1288        pointing to the condition or callout. */
1289    
1290        codelink = GET(ecode, 1);   /* Offset to the second branch */
1291        ecode += 1 + LINK_SIZE;     /* From this opcode */
1292    
1293      /* Because of the way auto-callout works during compile, a callout item is      /* Because of the way auto-callout works during compile, a callout item is
1294      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1295    
1296      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (*ecode == OP_CALLOUT)
1297        {        {
1298        if (PUBL(callout) != NULL)        if (PUBL(callout) != NULL)
1299          {          {
1300          PUBL(callout_block) cb;          PUBL(callout_block) cb;
1301          cb.version          = 2;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1302          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[1];
1303          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1304  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1305          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1304  for (;;) Line 1311  for (;;)
1311          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1312          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1313          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
1314          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, 2);
1315          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1316          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1317          cb.capture_last     = md->capture_last & CAPLMASK;          cb.capture_last     = md->capture_last & CAPLMASK;
1318          /* Internal change requires this for API compatibility. */          /* Internal change requires this for API compatibility. */
# Line 1315  for (;;) Line 1322  for (;;)
1322          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1323          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1324          }          }
1325    
1326          /* Advance ecode past the callout, so it now points to the condition. We
1327          must adjust codelink so that the value of ecode+codelink is unchanged. */
1328    
1329        ecode += PRIV(OP_lengths)[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1330        codelink -= PRIV(OP_lengths)[OP_CALLOUT];        codelink -= PRIV(OP_lengths)[OP_CALLOUT];
1331        }        }
1332    
1333      condcode = ecode[LINK_SIZE+1];      /* Test the various possible conditions */
   
     /* Now see what the actual condition is */  
1334    
1335      if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */      condition = FALSE;
1336        switch(condcode = *ecode)
1337        {        {
1338        if (md->recursive == NULL)                /* Not recursing => FALSE */        case OP_RREF:         /* Numbered group recursion test */
1339          if (md->recursive != NULL)     /* Not recursing => FALSE */
1340          {          {
1341          condition = FALSE;          unsigned int recno = GET2(ecode, 1);   /* Recursion group number*/
         ecode += GET(ecode, 1);  
         }  
       else  
         {  
         unsigned int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/  
1342          condition = (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1343            }
1344          break;
1345    
1346          /* If the test is for recursion into a specific subpattern, and it is        case OP_DNRREF:       /* Duplicate named group recursion test */
1347          false, but the test was set up by name, scan the table to see if the        if (md->recursive != NULL)
1348          name refers to any other numbers, and test them. The condition is true          {
1349          if any one is set. */          int count = GET2(ecode, 1 + IMM2_SIZE);
1350            pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
1351          if (!condition && condcode == OP_NRREF)          while (count-- > 0)
1352            {            {
1353            pcre_uchar *slotA = md->name_table;            unsigned int recno = GET2(slot, 0);
1354            for (i = 0; i < md->name_count; i++)            condition = recno == md->recursive->group_num;
1355              {            if (condition) break;
1356              if (GET2(slotA, 0) == recno) break;            slot += md->name_entry_size;
             slotA += md->name_entry_size;  
             }  
   
           /* Found a name for the number - there can be only one; duplicate  
           names for different numbers are allowed, but not vice versa. First  
           scan down for duplicates. */  
   
           if (i < md->name_count)  
             {  
             pcre_uchar *slotB = slotA;  
             while (slotB > md->name_table)  
               {  
               slotB -= md->name_entry_size;  
               if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)  
                 {  
                 condition = GET2(slotB, 0) == md->recursive->group_num;  
                 if (condition) break;  
                 }  
               else break;  
               }  
   
             /* Scan up for duplicates */  
   
             if (!condition)  
               {  
               slotB = slotA;  
               for (i++; i < md->name_count; i++)  
                 {  
                 slotB += md->name_entry_size;  
                 if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)  
                   {  
                   condition = GET2(slotB, 0) == md->recursive->group_num;  
                   if (condition) break;  
                   }  
                 else break;  
                 }  
               }  
             }  
1357            }            }
   
         /* Chose branch according to the condition */  
   
         ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);  
1358          }          }
1359        }        break;
1360    
1361      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */        case OP_CREF:         /* Numbered group used test */
1362        {        offset = GET2(ecode, 1) << 1;  /* Doubled ref number */
       offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */  
1363        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1364          break;
1365    
1366        /* If the numbered capture is unset, but the reference was by name,        case OP_DNCREF:      /* Duplicate named group used test */
       scan the table to see if the name refers to any other numbers, and test  
       them. The condition is true if any one is set. This is tediously similar  
       to the code above, but not close enough to try to amalgamate. */  
   
       if (!condition && condcode == OP_NCREF)  
1367          {          {
1368          unsigned int refno = offset >> 1;          int count = GET2(ecode, 1 + IMM2_SIZE);
1369          pcre_uchar *slotA = md->name_table;          pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
1370            while (count-- > 0)
1371          for (i = 0; i < md->name_count; i++)            {
1372            {            offset = GET2(slot, 0) << 1;
1373            if (GET2(slotA, 0) == refno) break;            condition = offset < offset_top && md->offset_vector[offset] >= 0;
1374            slotA += md->name_entry_size;            if (condition) break;
1375            }            slot += md->name_entry_size;
   
         /* Found a name for the number - there can be only one; duplicate names  
         for different numbers are allowed, but not vice versa. First scan down  
         for duplicates. */  
   
         if (i < md->name_count)  
           {  
           pcre_uchar *slotB = slotA;  
           while (slotB > md->name_table)  
             {  
             slotB -= md->name_entry_size;  
             if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)  
               {  
               offset = GET2(slotB, 0) << 1;  
               condition = offset < offset_top &&  
                 md->offset_vector[offset] >= 0;  
               if (condition) break;  
               }  
             else break;  
             }  
   
           /* Scan up for duplicates */  
   
           if (!condition)  
             {  
             slotB = slotA;  
             for (i++; i < md->name_count; i++)  
               {  
               slotB += md->name_entry_size;  
               if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)  
                 {  
                 offset = GET2(slotB, 0) << 1;  
                 condition = offset < offset_top &&  
                   md->offset_vector[offset] >= 0;  
                 if (condition) break;  
                 }  
               else break;  
               }  
             }  
1376            }            }
1377          }          }
1378          break;
1379    
1380        /* Chose branch according to the condition */        case OP_DEF:     /* DEFINE - always false */
1381          break;
       ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);  
       }  
   
     else if (condcode == OP_DEF)     /* DEFINE - always false */  
       {  
       condition = FALSE;  
       ecode += GET(ecode, 1);  
       }  
1382    
1383      /* The condition is an assertion. Call match() to evaluate it - setting        /* The condition is an assertion. Call match() to evaluate it - setting
1384      md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of        md->match_function_type to MATCH_CONDASSERT causes it to stop at the end
1385      an assertion. */        of an assertion. */
1386    
1387      else        default:
       {  
1388        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1389        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode, offset_top, md, NULL, RM3);
1390        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1391          {          {
1392          if (md->end_offset_top > offset_top)          if (md->end_offset_top > offset_top)
1393            offset_top = md->end_offset_top;  /* Captures may have happened */            offset_top = md->end_offset_top;  /* Captures may have happened */
1394          condition = TRUE;          condition = TRUE;
1395          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);  
1396            /* Advance ecode past the assertion to the start of the first branch,
1397            but adjust it so that the general choosing code below works. */
1398    
1399            ecode += GET(ecode, 1);
1400          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1401            ecode += 1 + LINK_SIZE - PRIV(OP_lengths)[condcode];
1402          }          }
1403    
1404        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1405        assertion; it is therefore treated as NOMATCH. */        assertion; it is therefore treated as NOMATCH. Any other return is an
1406          error. */
1407    
1408        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1409          {          {
1410          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1411          }          }
1412        else        break;
         {  
         condition = FALSE;  
         ecode += codelink;  
         }  
1413        }        }
1414    
1415      /* We are now at the branch that is to be obeyed. As there is only one, can      /* Choose branch according to the condition */
1416      use tail recursion to avoid using another stack frame, except when there is  
1417      unlimited repeat of a possibly empty group. In the latter case, a recursive      ecode += condition? PRIV(OP_lengths)[condcode] : codelink;
     call to match() is always required, unless the second alternative doesn't  
     exist, in which case we can just plough on. Note that, for compatibility  
     with Perl, the | in a conditional group is NOT treated as creating two  
     alternatives. If a THEN is encountered in the branch, it propagates out to  
     the enclosing alternative (unless nested in a deeper set of alternatives,  
     of course). */  
1418    
1419      if (condition || *ecode == OP_ALT)      /* We are now at the branch that is to be obeyed. As there is only one, we
1420        can use tail recursion to avoid using another stack frame, except when
1421        there is unlimited repeat of a possibly empty group. In the latter case, a
1422        recursive call to match() is always required, unless the second alternative
1423        doesn't exist, in which case we can just plough on. Note that, for
1424        compatibility with Perl, the | in a conditional group is NOT treated as
1425        creating two alternatives. If a THEN is encountered in the branch, it
1426        propagates out to the enclosing alternative (unless nested in a deeper set
1427        of alternatives, of course). */
1428    
1429        if (condition || ecode[-(1+LINK_SIZE)] == OP_ALT)
1430        {        {
1431        if (op != OP_SCOND)        if (op != OP_SCOND)
1432          {          {
         ecode += 1 + LINK_SIZE;  
1433          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1434          }          }
1435    
1436        md->match_function_type = MATCH_CBEGROUP;        md->match_function_type = MATCH_CBEGROUP;
1437        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);        RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);
1438        RRETURN(rrc);        RRETURN(rrc);
1439        }        }
1440    
# Line 1523  for (;;) Line 1442  for (;;)
1442    
1443      else      else
1444        {        {
       ecode += 1 + LINK_SIZE;  
1445        }        }
1446      break;      break;
1447    
# Line 1604  for (;;) Line 1522  for (;;)
1522      else condassert = FALSE;      else condassert = FALSE;
1523    
1524      /* Loop for each branch */      /* Loop for each branch */
1525    
1526      do      do
1527        {        {
1528        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1529    
1530        /* A match means that the assertion is true; break out of the loop        /* A match means that the assertion is true; break out of the loop
1531        that matches its alternatives. */        that matches its alternatives. */
1532    
1533        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1534          {          {
1535          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1536          break;          break;
1537          }          }
1538    
1539        /* If not matched, restore the previous mark setting. */        /* If not matched, restore the previous mark setting. */
1540    
1541        md->mark = save_mark;        md->mark = save_mark;
1542    
1543        /* See comment in the code for capturing groups above about handling        /* See comment in the code for capturing groups above about handling
# Line 1632  for (;;) Line 1550  for (;;)
1550              (*ecode == OP_ALT || *next == OP_ALT))              (*ecode == OP_ALT || *next == OP_ALT))
1551            rrc = MATCH_NOMATCH;            rrc = MATCH_NOMATCH;
1552          }          }
1553    
1554        /* Anything other than NOMATCH causes the entire assertion to fail,        /* Anything other than NOMATCH causes the entire assertion to fail,
1555        passing back the return code. This includes COMMIT, SKIP, PRUNE and an        passing back the return code. This includes COMMIT, SKIP, PRUNE and an
1556        uncaptured THEN, which means they take their normal effect. This        uncaptured THEN, which means they take their normal effect. This
# Line 1643  for (;;) Line 1561  for (;;)
1561        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1562        }        }
1563      while (*ecode == OP_ALT);   /* Continue for next alternative */      while (*ecode == OP_ALT);   /* Continue for next alternative */
1564    
1565      /* If we have tried all the alternative branches, the assertion has      /* If we have tried all the alternative branches, the assertion has
1566      failed. If not, we broke out after a match. */      failed. If not, we broke out after a match. */
1567    
1568      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1569    
# Line 1661  for (;;) Line 1579  for (;;)
1579      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1580      continue;      continue;
1581    
1582      /* Negative assertion: all branches must fail to match for the assertion to      /* Negative assertion: all branches must fail to match for the assertion to
1583      succeed. */      succeed. */
1584    
1585      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
# Line 1675  for (;;) Line 1593  for (;;)
1593      else condassert = FALSE;      else condassert = FALSE;
1594    
1595      /* Loop for each alternative branch. */      /* Loop for each alternative branch. */
1596    
1597      do      do
1598        {        {
1599        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1600        md->mark = save_mark;   /* Always restore the mark setting */        md->mark = save_mark;   /* Always restore the mark setting */
1601    
1602        switch(rrc)        switch(rrc)
1603          {          {
1604          case MATCH_MATCH:            /* A successful match means */          case MATCH_MATCH:            /* A successful match means */
1605          case MATCH_ACCEPT:           /* the assertion has failed. */          case MATCH_ACCEPT:           /* the assertion has failed. */
1606          RRETURN(MATCH_NOMATCH);          RRETURN(MATCH_NOMATCH);
1607    
1608          case MATCH_NOMATCH:          /* Carry on with next branch */          case MATCH_NOMATCH:          /* Carry on with next branch */
1609          break;          break;
1610    
1611          /* See comment in the code for capturing groups above about handling          /* See comment in the code for capturing groups above about handling
1612          THEN. */          THEN. */
# Line 1697  for (;;) Line 1615  for (;;)
1615          next = ecode + GET(ecode,1);          next = ecode + GET(ecode,1);
1616          if (md->start_match_ptr < next &&          if (md->start_match_ptr < next &&
1617              (*ecode == OP_ALT || *next == OP_ALT))              (*ecode == OP_ALT || *next == OP_ALT))
1618            {            {
1619            rrc = MATCH_NOMATCH;            rrc = MATCH_NOMATCH;
1620            break;            break;
1621            }            }
1622          /* Otherwise fall through. */          /* Otherwise fall through. */
1623    
1624          /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole          /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole
1625          assertion to fail to match, without considering any more alternatives.          assertion to fail to match, without considering any more alternatives.
1626          Failing to match means the assertion is true. This is a consistent          Failing to match means the assertion is true. This is a consistent
# Line 1710  for (;;) Line 1628  for (;;)
1628    
1629          case MATCH_COMMIT:          case MATCH_COMMIT:
1630          case MATCH_SKIP:          case MATCH_SKIP:
1631          case MATCH_SKIP_ARG:          case MATCH_SKIP_ARG:
1632          case MATCH_PRUNE:          case MATCH_PRUNE:
1633          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1634          goto NEG_ASSERT_TRUE;   /* Break out of alternation loop */          goto NEG_ASSERT_TRUE;   /* Break out of alternation loop */
1635    
1636          /* Anything else is an error */          /* Anything else is an error */
1637    
1638          default:          default:
1639          RRETURN(rrc);          RRETURN(rrc);
1640          }          }
1641    
1642        /* Continue with next branch */        /* Continue with next branch */
1643    
1644        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1645        }        }
1646      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1647    
1648      /* All branches in the assertion failed to match. */      /* All branches in the assertion failed to match. */
1649    
1650      NEG_ASSERT_TRUE:      NEG_ASSERT_TRUE:
1651      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1652      ecode += 1 + LINK_SIZE;                /* Continue with current branch */      ecode += 1 + LINK_SIZE;                /* Continue with current branch */
# Line 1896  for (;;) Line 1814  for (;;)
1814          /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a          /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a
1815          recursion; they cause a NOMATCH for the entire recursion. These codes          recursion; they cause a NOMATCH for the entire recursion. These codes
1816          are defined in a range that can be tested for. */          are defined in a range that can be tested for. */
1817    
1818          if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX)          if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX)
1819            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
1820    
1821          /* Any return code other than NOMATCH is an error. */          /* Any return code other than NOMATCH is an error. */
1822    
1823          if (rrc != MATCH_NOMATCH)          if (rrc != MATCH_NOMATCH)
# Line 2656  for (;;) Line 2574  for (;;)
2574            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2575          break;          break;
2576    
2577          case PT_SPACE:    /* Perl space */          /* Perl space used to exclude VT, but from Perl 5.18 it is included,
2578          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          which means that Perl space and POSIX space are now identical. PCRE
2579               c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)          was changed at release 8.34. */
                == (op == OP_NOTPROP))  
           RRETURN(MATCH_NOMATCH);  
         break;  
2580    
2581            case PT_SPACE:    /* Perl space */
2582          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2583          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          switch(c)
2584               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            {
2585               c == CHAR_FF || c == CHAR_CR)            HSPACE_CASES:
2586                 == (op == OP_NOTPROP))            VSPACE_CASES:
2587            RRETURN(MATCH_NOMATCH);            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2588              break;
2589    
2590              default:
2591              if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) ==
2592                (op == OP_NOTPROP)) RRETURN(MATCH_NOMATCH);
2593              break;
2594              }
2595          break;          break;
2596    
2597          case PT_WORD:          case PT_WORD:
# Line 2742  for (;;) Line 2665  for (;;)
2665      similar code to character type repeats - written out again for speed.      similar code to character type repeats - written out again for speed.
2666      However, if the referenced string is the empty string, always treat      However, if the referenced string is the empty string, always treat
2667      it as matched, any number of times (otherwise there could be infinite      it as matched, any number of times (otherwise there could be infinite
2668      loops). */      loops). If the reference is unset, there are two possibilities:
   
     case OP_REF:  
     case OP_REFI:  
     caseless = op == OP_REFI;  
     offset = GET2(ecode, 1) << 1;               /* Doubled ref number */  
     ecode += 1 + IMM2_SIZE;  
   
     /* If the reference is unset, there are two possibilities:  
2669    
2670      (a) In the default, Perl-compatible state, set the length negative;      (a) In the default, Perl-compatible state, set the length negative;
2671      this ensures that every attempt at a match fails. We can't just fail      this ensures that every attempt at a match fails. We can't just fail
# Line 2760  for (;;) Line 2675  for (;;)
2675      so that the back reference matches an empty string.      so that the back reference matches an empty string.
2676    
2677      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
2678      referenced subpattern. */      referenced subpattern.
2679    
2680        The OP_REF and OP_REFI opcodes are used for a reference to a numbered group
2681        or to a non-duplicated named group. For a duplicated named group, OP_DNREF
2682        and OP_DNREFI are used. In this case we must scan the list of groups to
2683        which the name refers, and use the first one that is set. */
2684    
2685        case OP_DNREF:
2686        case OP_DNREFI:
2687        caseless = op == OP_DNREFI;
2688          {
2689          int count = GET2(ecode, 1+IMM2_SIZE);
2690          pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
2691          ecode += 1 + 2*IMM2_SIZE;
2692    
2693          while (count-- > 0)
2694            {
2695            offset = GET2(slot, 0) << 1;
2696            if (offset < offset_top && md->offset_vector[offset] >= 0) break;
2697            slot += md->name_entry_size;
2698            }
2699          if (count < 0)
2700            length = (md->jscript_compat)? 0 : -1;
2701          else
2702            length = md->offset_vector[offset+1] - md->offset_vector[offset];
2703          }
2704        goto REF_REPEAT;
2705    
2706        case OP_REF:
2707        case OP_REFI:
2708        caseless = op == OP_REFI;
2709        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2710        ecode += 1 + IMM2_SIZE;
2711      if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2712        length = (md->jscript_compat)? 0 : -1;        length = (md->jscript_compat)? 0 : -1;
2713      else      else
# Line 2769  for (;;) Line 2715  for (;;)
2715    
2716      /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2717    
2718        REF_REPEAT:
2719      switch (*ecode)      switch (*ecode)
2720        {        {
2721        case OP_CRSTAR:        case OP_CRSTAR:
# Line 3387  for (;;) Line 3334  for (;;)
3334      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3335      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3336    
3337      /* Common code for all repeated single-character matches. We first check      /* Common code for all repeated single-character matches. We first check
3338      for the minimum number of characters. If the minimum equals the maximum, we      for the minimum number of characters. If the minimum equals the maximum, we
3339      are done. Otherwise, if minimizing, check the rest of the pattern for a      are done. Otherwise, if minimizing, check the rest of the pattern for a
3340      match; if there isn't one, advance up to the maximum, one character at a      match; if there isn't one, advance up to the maximum, one character at a
3341      time.      time.
3342    
3343      If maximizing, advance up to the maximum number of matching characters,      If maximizing, advance up to the maximum number of matching characters,
3344      until eptr is past the end of the maximum run. If possessive, we are      until eptr is past the end of the maximum run. If possessive, we are
3345      then done (no backing up). Otherwise, match at this position; anything      then done (no backing up). Otherwise, match at this position; anything
3346      other than no match is immediately returned. For nomatch, back up one      other than no match is immediately returned. For nomatch, back up one
3347      character, unless we are matching \R and the last thing matched was      character, unless we are matching \R and the last thing matched was
3348      \r\n, in which case, back up two bytes. When we reach the first optional      \r\n, in which case, back up two bytes. When we reach the first optional
3349      character position, we can save stack by doing a tail recurse.      character position, we can save stack by doing a tail recurse.
3350    
3351      The various UTF/non-UTF and caseful/caseless cases are handled separately,      The various UTF/non-UTF and caseful/caseless cases are handled separately,
3352      for speed. */      for speed. */
3353    
# Line 3590  for (;;) Line 3537  for (;;)
3537            if (fc != cc && foc != cc) break;            if (fc != cc && foc != cc) break;
3538            eptr++;            eptr++;
3539            }            }
   
3540          if (possessive) continue;       /* No backtracking */          if (possessive) continue;       /* No backtracking */
3541          for (;;)          for (;;)
3542            {            {
3543            if (eptr == pp) goto TAIL_RECURSE;            if (eptr == pp) goto TAIL_RECURSE;
3544            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3545            eptr--;            eptr--;
3546            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3547            }            }
3548          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3549          }          }
       /* Control never gets here */  
3550        }        }
3551    
3552      /* Caseful comparisons (includes all multi-byte characters) */      /* Caseful comparisons (includes all multi-byte characters) */
# Line 3657  for (;;) Line 3602  for (;;)
3602            eptr--;            eptr--;
3603            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3604            }            }
3605          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3606          }          }
3607        }        }
3608      /* Control never gets here */      /* Control never gets here */
# Line 3942  for (;;) Line 3887  for (;;)
3887              eptr--;              eptr--;
3888              }              }
3889            }            }
3890            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
3891          }          }
       /* Control never gets here */  
3892        }        }
3893    
3894      /* Caseful comparisons */      /* Caseful comparisons */
# Line 4073  for (;;) Line 4016  for (;;)
4016            if (possessive) continue;    /* No backtracking */            if (possessive) continue;    /* No backtracking */
4017            for (;;)            for (;;)
4018              {              {
4019              if (eptr == pp) goto TAIL_RECURSE;              if (eptr == pp) goto TAIL_RECURSE;
4020              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
4021              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4022              eptr--;              eptr--;
4023              }              }
4024            }            }
4025            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
4026          }          }
4027        }        }
4028      /* Control never gets here */      /* Control never gets here */
# Line 4262  for (;;) Line 4204  for (;;)
4204              }              }
4205            break;            break;
4206    
4207              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4208              which means that Perl space and POSIX space are now identical. PCRE
4209              was changed at release 8.34. */
4210    
4211            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4212              case PT_PXSPACE:  /* POSIX space */
4213            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4214              {              {
4215              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4271  for (;;) Line 4218  for (;;)
4218                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4219                }                }
4220              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4221              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
                  c == CHAR_FF || c == CHAR_CR)  
                    == prop_fail_result)  
               RRETURN(MATCH_NOMATCH);  
             }  
           break;  
   
           case PT_PXSPACE:  /* POSIX space */  
           for (i = 1; i <= min; i++)  
             {  
             if (eptr >= md->end_subject)  
4222                {                {
4223                SCHECK_PARTIAL();                HSPACE_CASES:
4224                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
4225                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4226                  break;
4227    
4228                  default:
4229                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
4230                    RRETURN(MATCH_NOMATCH);
4231                  break;
4232                }                }
             GETCHARINCTEST(c, eptr);  
             if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
                  c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)  
                    == prop_fail_result)  
               RRETURN(MATCH_NOMATCH);  
4233              }              }
4234            break;            break;
4235    
# Line 5010  for (;;) Line 4949  for (;;)
4949              }              }
4950            /* Control never gets here */            /* Control never gets here */
4951    
4952            case PT_SPACE:    /* Perl space */            /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4953            for (fi = min;; fi++)            which means that Perl space and POSIX space are now identical. PCRE
4954              {            was changed at release 8.34. */
             RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);  
             if (rrc != MATCH_NOMATCH) RRETURN(rrc);  
             if (fi >= max) RRETURN(MATCH_NOMATCH);  
             if (eptr >= md->end_subject)  
               {  
               SCHECK_PARTIAL();  
               RRETURN(MATCH_NOMATCH);  
               }  
             GETCHARINCTEST(c, eptr);  
             if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
                  c == CHAR_FF || c == CHAR_CR)  
                    == prop_fail_result)  
               RRETURN(MATCH_NOMATCH);  
             }  
           /* Control never gets here */  
4955    
4956              case PT_SPACE:    /* Perl space */
4957            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4958            for (fi = min;; fi++)            for (fi = min;; fi++)
4959              {              {
# Line 5041  for (;;) Line 4966  for (;;)
4966                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4967                }                }
4968              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4969              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
4970                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                {
4971                     == prop_fail_result)                HSPACE_CASES:
4972                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
4973                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4974                  break;
4975    
4976                  default:
4977                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
4978                    RRETURN(MATCH_NOMATCH);
4979                  break;
4980                  }
4981              }              }
4982            /* Control never gets here */            /* Control never gets here */
4983    
# Line 5528  for (;;) Line 5461  for (;;)
5461              }              }
5462            break;            break;
5463    
5464              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
5465              which means that Perl space and POSIX space are now identical. PCRE
5466              was changed at release 8.34. */
5467    
5468            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
5469              case PT_PXSPACE:  /* POSIX space */
5470            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5471              {              {
5472              int len = 1;              int len = 1;
# Line 5538  for (;;) Line 5476  for (;;)
5476                break;                break;
5477                }                }
5478              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5479              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
5480                   c == CHAR_FF || c == CHAR_CR)                {
5481                   == prop_fail_result)                HSPACE_CASES:
5482                  VSPACE_CASES:
5483                  if (prop_fail_result) goto ENDLOOP99;  /* Break the loop */
5484                break;                break;
             eptr+= len;  
             }  
           break;  
5485    
5486            case PT_PXSPACE:  /* POSIX space */                default:
5487            for (i = min; i < max; i++)                if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
5488              {                  goto ENDLOOP99;   /* Break the loop */
             int len = 1;  
             if (eptr >= md->end_subject)  
               {  
               SCHECK_PARTIAL();  
5489                break;                break;
5490                }                }
             GETCHARLENTEST(c, eptr, len);  
             if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||  
                  c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)  
                  == prop_fail_result)  
               break;  
5491              eptr+= len;              eptr+= len;
5492              }              }
5493              ENDLOOP99:
5494            break;            break;
5495    
5496            case PT_WORD:            case PT_WORD:
# Line 5642  for (;;) Line 5571  for (;;)
5571            }            }
5572          }          }
5573    
5574        /* Match extended Unicode sequences. We will get here only if the        /* Match extended Unicode grapheme clusters. We will get here only if the
5575        support is in the binary; otherwise a compile-time error occurs. */        support is in the binary; otherwise a compile-time error occurs. */
5576    
5577        else if (ctype == OP_EXTUNI)        else if (ctype == OP_EXTUNI)
# Line 5675  for (;;) Line 5604  for (;;)
5604          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5605    
5606          if (possessive) continue;    /* No backtracking */          if (possessive) continue;    /* No backtracking */
5607    
5608          for(;;)          for(;;)
5609            {            {
5610            if (eptr == pp) goto TAIL_RECURSE;            int lgb, rgb;
5611              PCRE_PUCHAR fptr;
5612    
5613              if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5614            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5615            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5616            eptr--;  
5617            for (;;)                        /* Move back over one extended */            /* Backtracking over an extended grapheme cluster involves inspecting
5618              the previous two characters (if present) to see if a break is
5619              permitted between them. */
5620    
5621              eptr--;
5622              if (!utf) c = *eptr; else
5623              {              {
5624              if (!utf) c = *eptr; else              BACKCHAR(eptr);
5625                GETCHAR(c, eptr);
5626                }
5627              rgb = UCD_GRAPHBREAK(c);
5628    
5629              for (;;)
5630                {
5631                if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5632                fptr = eptr - 1;
5633                if (!utf) c = *fptr; else
5634                {                {
5635                BACKCHAR(eptr);                BACKCHAR(fptr);
5636                GETCHAR(c, eptr);                GETCHAR(c, fptr);
5637                }                }
5638              if (UCD_CATEGORY(c) != ucp_M) break;              lgb = UCD_GRAPHBREAK(c);
5639              eptr--;              if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
5640                eptr = fptr;
5641                rgb = lgb;
5642              }              }
5643            }            }
5644          }          }
# Line 5958  for (;;) Line 5907  for (;;)
5907          if (possessive) continue;    /* No backtracking */          if (possessive) continue;    /* No backtracking */
5908          for(;;)          for(;;)
5909            {            {
5910            if (eptr == pp) goto TAIL_RECURSE;            if (eptr == pp) goto TAIL_RECURSE;
5911            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5912            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5913            eptr--;            eptr--;
# Line 6211  for (;;) Line 6160  for (;;)
6160            }            }
6161          }          }
6162    
6163        /* Get here if we can't make it match with any permitted repetitions */        /* Control never gets here */
   
       RRETURN(MATCH_NOMATCH);  
6164        }        }
     /* Control never gets here */  
6165    
6166      /* There's been some horrible disaster. Arrival here can only mean there is      /* There's been some horrible disaster. Arrival here can only mean there is
6167      something seriously wrong in the code above or the OP_xxx definitions. */      something seriously wrong in the code above or the OP_xxx definitions. */
# Line 6410  const pcre_uint8 *start_bits = NULL; Line 6356  const pcre_uint8 *start_bits = NULL;
6356  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;
6357  PCRE_PUCHAR end_subject;  PCRE_PUCHAR end_subject;
6358  PCRE_PUCHAR start_partial = NULL;  PCRE_PUCHAR start_partial = NULL;
6359  PCRE_PUCHAR match_partial;  PCRE_PUCHAR match_partial = NULL;
6360  PCRE_PUCHAR req_char_ptr = start_match - 1;  PCRE_PUCHAR req_char_ptr = start_match - 1;
6361    
6362  const pcre_study_data *study;  const pcre_study_data *study;
# Line 6511  if (extra_data != NULL Line 6457  if (extra_data != NULL
6457      && extra_data->executable_jit != NULL      && extra_data->executable_jit != NULL
6458      && (options & ~PUBLIC_JIT_EXEC_OPTIONS) == 0)      && (options & ~PUBLIC_JIT_EXEC_OPTIONS) == 0)
6459    {    {
   /* A facility for setting the match limit in the regex was added; this puts  
   a value in the compiled block. (Similarly for recursion limit, but the JIT  
   does not make use of that.) Because the regex is not passed to jit_exec, we  
   fudge up an alternative extra block, because we must not modify the extra  
   block that the user has passed. */  
   
 #if defined COMPILE_PCRE8  
   pcre_extra extra_data_copy;  
 #elif defined COMPILE_PCRE16  
   pcre16_extra extra_data_copy;  
 #elif defined COMPILE_PCRE32  
   pcre32_extra extra_data_copy;  
 #endif  
   
   if ((re->flags & PCRE_MLSET) != 0 &&  
       ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0 ||  
        re->limit_match < extra_data->match_limit))  
     {  
     extra_data_copy = *extra_data;  
     extra_data_copy.match_limit = re->limit_match;  
     extra_data_copy.flags |= PCRE_EXTRA_MATCH_LIMIT;  
     extra_data = &extra_data_copy;  
     }  
   
6460    rc = PRIV(jit_exec)(extra_data, (const pcre_uchar *)subject, length,    rc = PRIV(jit_exec)(extra_data, (const pcre_uchar *)subject, length,
6461         start_offset, options, offsets, offsetcount);         start_offset, options, offsets, offsetcount);
6462    
# Line 6581  if (extra_data != NULL) Line 6503  if (extra_data != NULL)
6503    }    }
6504    
6505  /* Limits in the regex override only if they are smaller. */  /* Limits in the regex override only if they are smaller. */
6506    
6507  if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit)  if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit)
6508    md->match_limit = re->limit_match;    md->match_limit = re->limit_match;
6509    
6510  if ((re->flags & PCRE_RLSET) != 0 &&  if ((re->flags & PCRE_RLSET) != 0 &&
6511      re->limit_recursion < md->match_limit_recursion)      re->limit_recursion < md->match_limit_recursion)
6512    md->match_limit_recursion = re->limit_recursion;    md->match_limit_recursion = re->limit_recursion;
6513    
# Line 7202  if (rc != MATCH_NOMATCH && rc != PCRE_ER Line 7124  if (rc != MATCH_NOMATCH && rc != PCRE_ER
7124    
7125  /* Handle partial matches - disable any mark data */  /* Handle partial matches - disable any mark data */
7126    
7127  if (start_partial != NULL)  if (match_partial != NULL)
7128    {    {
7129    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));
7130    md->mark = NULL;    md->mark = NULL;

Legend:
Removed from v.1313  
changed lines
  Added in v.1376

  ViewVC Help
Powered by ViewVC 1.1.5