/[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 1265 by ph10, Sun Mar 3 10:42:46 2013 UTC revision 1376 by ph10, Sat Oct 12 18:02:11 2013 UTC
# Line 56  possible. There are also some static sup Line 56  possible. There are also some static sup
56  #undef min  #undef min
57  #undef max  #undef max
58    
59  /* The md->capture_last field uses the lower 16 bits for the last captured  /* The md->capture_last field uses the lower 16 bits for the last captured
60  substring (which can never be greater than 65535) and a bit in the top half  substring (which can never be greater than 65535) and a bit in the top half
61  to mean "capture vector overflowed". This odd way of doing things was  to mean "capture vector overflowed". This odd way of doing things was
62  implemented when it was realized that preserving and restoring the overflow bit  implemented when it was realized that preserving and restoring the overflow bit
63  whenever the last capture number was saved/restored made for a neater  whenever the last capture number was saved/restored made for a neater
64  interface, and doing it this way saved on (a) another variable, which would  interface, and doing it this way saved on (a) another variable, which would
65  have increased the stack frame size (a big NO-NO in PCRE) and (b) another  have increased the stack frame size (a big NO-NO in PCRE) and (b) another
66  separate set of save/restore instructions. The following defines are used in  separate set of save/restore instructions. The following defines are used in
67  implementing this. */  implementing this. */
68    
69  #define CAPLMASK    0x0000ffff    /* The bits used for last_capture */  #define CAPLMASK    0x0000ffff    /* The bits used for last_capture */
# Line 87  defined PCRE_ERROR_xxx codes, which are Line 87  defined PCRE_ERROR_xxx codes, which are
87  negative to avoid the external error codes. */  negative to avoid the external error codes. */
88    
89  #define MATCH_ACCEPT       (-999)  #define MATCH_ACCEPT       (-999)
90  #define MATCH_COMMIT       (-998)  #define MATCH_KETRPOS      (-998)
91  #define MATCH_KETRPOS      (-997)  #define MATCH_ONCE         (-997)
92  #define MATCH_ONCE         (-996)  /* The next 5 must be kept together and in sequence so that a test that checks
93    for any one of them can use a range. */
94    #define MATCH_COMMIT       (-996)
95  #define MATCH_PRUNE        (-995)  #define MATCH_PRUNE        (-995)
96  #define MATCH_SKIP         (-994)  #define MATCH_SKIP         (-994)
97  #define MATCH_SKIP_ARG     (-993)  #define MATCH_SKIP_ARG     (-993)
98  #define MATCH_THEN         (-992)  #define MATCH_THEN         (-992)
99    #define MATCH_BACKTRACK_MAX MATCH_THEN
100    #define MATCH_BACKTRACK_MIN MATCH_COMMIT
101    
102  /* Maximum number of ints of offset to save on the stack for recursive calls.  /* Maximum number of ints of offset to save on the stack for recursive calls.
103  If the offset vector is bigger, malloc is used. This should be a multiple of 3,  If the offset vector is bigger, malloc is used. This should be a multiple of 3,
# Line 777  for (;;) Line 781  for (;;)
781      case OP_FAIL:      case OP_FAIL:
782      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
783    
     /* COMMIT overrides PRUNE, SKIP, and THEN */  
   
784      case OP_COMMIT:      case OP_COMMIT:
785      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
786        eptrb, RM52);        eptrb, RM52);
787      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE &&      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
         rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG &&  
         rrc != MATCH_THEN)  
       RRETURN(rrc);  
788      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
789    
     /* PRUNE overrides THEN */  
   
790      case OP_PRUNE:      case OP_PRUNE:
791      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
792        eptrb, RM51);        eptrb, RM51);
793      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
794      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
795    
796      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
# Line 803  for (;;) Line 800  for (;;)
800        eptrb, RM56);        eptrb, RM56);
801      if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&      if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
802           md->mark == NULL) md->mark = ecode + 2;           md->mark == NULL) md->mark = ecode + 2;
803      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
804      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
805    
     /* SKIP overrides PRUNE and THEN */  
   
806      case OP_SKIP:      case OP_SKIP:
807      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
808        eptrb, RM53);        eptrb, RM53);
809      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
       RRETURN(rrc);  
810      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
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. There is a flag that disables this opcode when re-matching a      nomatch_mark. When a pattern match ends with a SKIP_ARG for which there was
815      pattern that ended with a SKIP for which there was not a matching MARK. */      not a matching mark, we have to re-run the match, ignoring the SKIP_ARG
816        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
818        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. */
820    
821      case OP_SKIP_ARG:      case OP_SKIP_ARG:
822      if (md->ignore_skip_arg)      md->skip_arg_count++;
823        if (md->skip_arg_count <= md->ignore_skip_arg)
824        {        {
825        ecode += PRIV(OP_lengths)[*ecode] + ecode[1];        ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
826        break;        break;
827        }        }
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 && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
       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
835      with the md->ignore_skip_arg flag set. */      with md->ignore_skip_arg set to the value of md->skip_arg_count. */
836    
837      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
838      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
# Line 1276  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 1306  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. */
1319          if (cb.capture_last == 0) cb.capture_last = -1;          if (cb.capture_last == 0) cb.capture_last = -1;
1320          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1321          cb.mark             = md->nomatch_mark;          cb.mark             = md->nomatch_mark;
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 */
1334    
1335      /* Now see what the actual condition is */      condition = FALSE;
1336        switch(condcode = *ecode)
     if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */  
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 1525  for (;;) Line 1442  for (;;)
1442    
1443      else      else
1444        {        {
       ecode += 1 + LINK_SIZE;  
1445        }        }
1446      break;      break;
1447    
# Line 1605  for (;;) Line 1521  for (;;)
1521        }        }
1522      else condassert = FALSE;      else condassert = FALSE;
1523    
1524        /* 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
1531          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. */
1540    
1541        md->mark = save_mark;        md->mark = save_mark;
1542    
1543        /* A COMMIT failure must fail the entire assertion, without trying any        /* See comment in the code for capturing groups above about handling
1544        subsequent branches. */        THEN. */
1545    
1546        if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_THEN)
1547            {
1548            next = ecode + GET(ecode,1);
1549            if (md->start_match_ptr < next &&
1550                (*ecode == OP_ALT || *next == OP_ALT))
1551              rrc = MATCH_NOMATCH;
1552            }
1553    
1554        /* PCRE does not allow THEN to escape beyond an assertion; it        /* Anything other than NOMATCH causes the entire assertion to fail,
1555        is treated as NOMATCH. */        passing back the return code. This includes COMMIT, SKIP, PRUNE and an
1556          uncaptured THEN, which means they take their normal effect. This
1557          consistent approach does not always have exactly the same effect as in
1558          Perl. */
1559    
1560        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1561        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1562        }        }
1563      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);   /* Continue for next alternative */
1564    
1565        /* If we have tried all the alternative branches, the assertion has
1566        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 1634  for (;;) Line 1571  for (;;)
1571    
1572      if (condassert) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1573    
1574      /* Continue from after the assertion, updating the offsets high water      /* Continue from after a successful assertion, updating the offsets high
1575      mark, since extracts may have been taken during the assertion. */      water mark, since extracts may have been taken during the assertion. */
1576    
1577      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1578      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
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. Encountering SKIP,      /* Negative assertion: all branches must fail to match for the assertion to
1583      PRUNE, or COMMIT means we must assume failure without checking subsequent      succeed. */
     branches. */  
1584    
1585      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1586      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
# Line 1656  for (;;) Line 1592  for (;;)
1592        }        }
1593      else condassert = FALSE;      else condassert = FALSE;
1594    
1595        /* 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;        md->mark = save_mark;   /* Always restore the mark setting */
1601        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);  
1602        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        switch(rrc)
1603          {          {
1604          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          case MATCH_MATCH:            /* A successful match means */
1605            case MATCH_ACCEPT:           /* the assertion has failed. */
1606            RRETURN(MATCH_NOMATCH);
1607    
1608            case MATCH_NOMATCH:          /* Carry on with next branch */
1609          break;          break;
1610    
1611            /* See comment in the code for capturing groups above about handling
1612            THEN. */
1613    
1614            case MATCH_THEN:
1615            next = ecode + GET(ecode,1);
1616            if (md->start_match_ptr < next &&
1617                (*ecode == OP_ALT || *next == OP_ALT))
1618              {
1619              rrc = MATCH_NOMATCH;
1620              break;
1621              }
1622            /* Otherwise fall through. */
1623    
1624            /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole
1625            assertion to fail to match, without considering any more alternatives.
1626            Failing to match means the assertion is true. This is a consistent
1627            approach, but does not always have the same effect as in Perl. */
1628    
1629            case MATCH_COMMIT:
1630            case MATCH_SKIP:
1631            case MATCH_SKIP_ARG:
1632            case MATCH_PRUNE:
1633            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1634            goto NEG_ASSERT_TRUE;   /* Break out of alternation loop */
1635    
1636            /* Anything else is an error */
1637    
1638            default:
1639            RRETURN(rrc);
1640          }          }
1641    
1642        /* PCRE does not allow THEN to escape beyond an assertion; it is treated        /* Continue with next branch */
       as NOMATCH. */  
1643    
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
1644        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1645        }        }
1646      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1647    
1648      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      /* All branches in the assertion failed to match. */
1649    
1650      ecode += 1 + LINK_SIZE;      NEG_ASSERT_TRUE:
1651        if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1652        ecode += 1 + LINK_SIZE;                /* Continue with current branch */
1653      continue;      continue;
1654    
1655      /* Move the subject pointer back. This occurs only at the start of      /* Move the subject pointer back. This occurs only at the start of
# Line 1738  for (;;) Line 1710  for (;;)
1710        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1711        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1712        cb.capture_last     = md->capture_last & CAPLMASK;        cb.capture_last     = md->capture_last & CAPLMASK;
1713        /* Internal change requires this for API compatibility. */        /* Internal change requires this for API compatibility. */
1714        if (cb.capture_last == 0) cb.capture_last = -1;        if (cb.capture_last == 0) cb.capture_last = -1;
1715        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1716        cb.mark             = md->nomatch_mark;        cb.mark             = md->nomatch_mark;
1717        if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
# Line 1785  for (;;) Line 1757  for (;;)
1757        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1758    
1759        new_recursive.group_num = recno;        new_recursive.group_num = recno;
1760        new_recursive.saved_capture_last = md->capture_last;        new_recursive.saved_capture_last = md->capture_last;
1761        new_recursive.subject_position = eptr;        new_recursive.subject_position = eptr;
1762        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1763        md->recursive = &new_recursive;        md->recursive = &new_recursive;
# Line 1822  for (;;) Line 1794  for (;;)
1794            md, eptrb, RM6);            md, eptrb, RM6);
1795          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
1796              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1797          md->capture_last = new_recursive.saved_capture_last;          md->capture_last = new_recursive.saved_capture_last;
1798          md->recursive = new_recursive.prevrec;          md->recursive = new_recursive.prevrec;
1799          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1800            {            {
# Line 1839  for (;;) Line 1811  for (;;)
1811            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
1812            }            }
1813    
1814          /* PCRE does not allow THEN or COMMIT to escape beyond a recursion; it          /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a
1815          is treated as NOMATCH. */          recursion; they cause a NOMATCH for the entire recursion. These codes
1816            are defined in a range that can be tested for. */
1817    
1818          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&          if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX)
1819                   rrc != MATCH_COMMIT)            RRETURN(MATCH_NOMATCH);
1820    
1821            /* Any return code other than NOMATCH is an error. */
1822    
1823            if (rrc != MATCH_NOMATCH)
1824            {            {
1825            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1826            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 2597  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 2629  for (;;) Line 2611  for (;;)
2611              { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } }              { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } }
2612            }            }
2613          break;          break;
2614    
2615          case PT_UCNC:          case PT_UCNC:
2616          if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||          if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2617               c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||               c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2618               c >= 0xe000) == (op == OP_NOTPROP))               c >= 0xe000) == (op == OP_NOTPROP))
2619            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2620          break;          break;
2621    
2622          /* This should never occur */          /* This should never occur */
2623    
# Line 2683  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 2701  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 2710  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 3328  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. */      /* 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
3339        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
3341        time.
3342    
3343        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
3345        then done (no backing up). Otherwise, match at this position; anything
3346        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
3348        \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.
3350    
3351        The various UTF/non-UTF and caseful/caseless cases are handled separately,
3352        for speed. */
3353    
3354      REPEATCHAR:      REPEATCHAR:
3355  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
# Line 3412  for (;;) Line 3433  for (;;)
3433                }                }
3434              }              }
3435    
3436            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
   
3437            for(;;)            for(;;)
3438              {              {
3439                if (eptr == pp) goto TAIL_RECURSE;
3440              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3441              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
             if (eptr == pp) { RRETURN(MATCH_NOMATCH); }  
3442  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3443              eptr--;              eptr--;
3444              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3517  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 */
3541          if (possessive) continue;          for (;;)
   
         while (eptr >= pp)  
3542            {            {
3543              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 3576  for (;;) Line 3594  for (;;)
3594            if (fc != RAWUCHARTEST(eptr)) break;            if (fc != RAWUCHARTEST(eptr)) break;
3595            eptr++;            eptr++;
3596            }            }
3597          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
3598            for (;;)
         while (eptr >= pp)  
3599            {            {
3600              if (eptr == pp) goto TAIL_RECURSE;
3601            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
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 3756  for (;;) Line 3774  for (;;)
3774            }            }
3775          }          }
3776        else        else
3777  #endif  #endif  /* SUPPORT_UTF */
3778        /* Not UTF mode */        /* Not UTF mode */
3779          {          {
3780          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3794  for (;;) Line 3812  for (;;)
3812              }              }
3813            }            }
3814          else          else
3815  #endif  #endif  /*SUPPORT_UTF */
3816          /* Not UTF mode */          /* Not UTF mode */
3817            {            {
3818            for (fi = min;; fi++)            for (fi = min;; fi++)
# Line 3836  for (;;) Line 3854  for (;;)
3854              if (fc == d || (unsigned int)foc == d) break;              if (fc == d || (unsigned int)foc == d) break;
3855              eptr += len;              eptr += len;
3856              }              }
3857            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3858            for(;;)            for(;;)
3859              {              {
3860                if (eptr == pp) goto TAIL_RECURSE;
3861              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3862              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3863              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
3864              BACKCHAR(eptr);              BACKCHAR(eptr);
3865              }              }
3866            }            }
3867          else          else
3868  #endif  #endif  /* SUPPORT_UTF */
3869          /* Not UTF mode */          /* Not UTF mode */
3870            {            {
3871            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3859  for (;;) Line 3878  for (;;)
3878              if (fc == *eptr || foc == *eptr) break;              if (fc == *eptr || foc == *eptr) break;
3879              eptr++;              eptr++;
3880              }              }
3881            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3882            while (eptr >= pp)            for (;;)
3883              {              {
3884                if (eptr == pp) goto TAIL_RECURSE;
3885              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3886              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
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 3971  for (;;) Line 3989  for (;;)
3989              if (fc == d) break;              if (fc == d) break;
3990              eptr += len;              eptr += len;
3991              }              }
3992            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3993            for(;;)            for(;;)
3994              {              {
3995                if (eptr == pp) goto TAIL_RECURSE;
3996              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
3997              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3998              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
3999              BACKCHAR(eptr);              BACKCHAR(eptr);
4000              }              }
4001            }            }
# Line 3994  for (;;) Line 4013  for (;;)
4013              if (fc == *eptr) break;              if (fc == *eptr) break;
4014              eptr++;              eptr++;
4015              }              }
4016            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
4017            while (eptr >= pp)            for (;;)
4018              {              {
4019                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 4185  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 4194  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 4254  for (;;) Line 4270  for (;;)
4270                }                }
4271              }              }
4272            break;            break;
4273    
4274            case PT_UCNC:            case PT_UCNC:
4275            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4276              {              {
# Line 4268  for (;;) Line 4284  for (;;)
4284                   c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||                   c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
4285                   c >= 0xe000) == prop_fail_result)                   c >= 0xe000) == prop_fail_result)
4286                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4287              }              }
4288            break;            break;
4289    
4290            /* This should not occur */            /* This should not occur */
4291    
# Line 4933  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 4964  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 5016  for (;;) Line 5026  for (;;)
5026                }                }
5027              }              }
5028            /* Control never gets here */            /* Control never gets here */
5029    
5030            case PT_UCNC:            case PT_UCNC:
5031            for (fi = min;; fi++)            for (fi = min;; fi++)
5032              {              {
# Line 5032  for (;;) Line 5042  for (;;)
5042              if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||              if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
5043                   c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||                   c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
5044                   c >= 0xe000) == prop_fail_result)                   c >= 0xe000) == prop_fail_result)
5045                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
5046              }              }
5047            /* Control never gets here */            /* Control never gets here */
5048    
5049            /* This should never occur */            /* This should never occur */
# Line 5451  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 5461  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 5545  for (;;) Line 5551  for (;;)
5551                   c >= 0xe000) == prop_fail_result)                   c >= 0xe000) == prop_fail_result)
5552                break;                break;
5553              eptr += len;              eptr += len;
5554              }              }
5555            break;            break;
5556    
5557            default:            default:
# Line 5554  for (;;) Line 5560  for (;;)
5560    
5561          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5562    
5563          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
5564          for(;;)          for(;;)
5565            {            {
5566              if (eptr == pp) goto TAIL_RECURSE;
5567            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5568            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5569            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5570            if (utf) BACKCHAR(eptr);            if (utf) BACKCHAR(eptr);
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 5596  for (;;) Line 5603  for (;;)
5603    
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;          if (possessive) continue;    /* No backtracking */
5607    
5608          for(;;)          for(;;)
5609            {            {
5610              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            if (eptr-- == pp) break;        /* Stop if tried at original pos */  
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                BACKCHAR(eptr);
5625                GETCHAR(c, eptr);
5626                }
5627              rgb = UCD_GRAPHBREAK(c);
5628    
5629              for (;;)
5630              {              {
5631              if (!utf) c = *eptr; else              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 5877  for (;;) Line 5904  for (;;)
5904            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5905            }            }
5906    
5907          /* eptr is now past the end of the maximum run. If possessive, we are          if (possessive) continue;    /* No backtracking */
         done (no backing up). Otherwise, match at this position; anything other  
         than no match is immediately returned. For nomatch, back up one  
         character, unless we are matching \R and the last thing matched was  
         \r\n, in which case, back up two bytes. */  
   
         if (possessive) continue;  
5908          for(;;)          for(;;)
5909            {            {
5910              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            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5914            BACKCHAR(eptr);            BACKCHAR(eptr);
5915            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&
5916                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;
# Line 6126  for (;;) Line 6148  for (;;)
6148            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
6149            }            }
6150    
6151          /* eptr is now past the end of the maximum run. If possessive, we are          if (possessive) continue;    /* No backtracking */
6152          done (no backing up). Otherwise, match at this position; anything other          for (;;)
         than no match is immediately returned. For nomatch, back up one  
         character (byte), unless we are matching \R and the last thing matched  
         was \r\n, in which case, back up two bytes. */  
   
         if (possessive) continue;  
         while (eptr >= pp)  
6153            {            {
6154              if (eptr == pp) goto TAIL_RECURSE;
6155            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
6156            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
6157            eptr--;            eptr--;
# Line 6143  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 6342  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 6472  md->callout_data = NULL; Line 6486  md->callout_data = NULL;
6486    
6487  tables = re->tables;  tables = re->tables;
6488    
6489    /* The two limit values override the defaults, whatever their value. */
6490    
6491  if (extra_data != NULL)  if (extra_data != NULL)
6492    {    {
6493    register unsigned int flags = extra_data->flags;    register unsigned int flags = extra_data->flags;
# Line 6486  if (extra_data != NULL) Line 6502  if (extra_data != NULL)
6502    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;
6503    }    }
6504    
6505    /* Limits in the regex override only if they are smaller. */
6506    
6507    if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit)
6508      md->match_limit = re->limit_match;
6509    
6510    if ((re->flags & PCRE_RLSET) != 0 &&
6511        re->limit_recursion < md->match_limit_recursion)
6512      md->match_limit_recursion = re->limit_recursion;
6513    
6514  /* If the exec call supplied NULL for tables, use the inbuilt ones. This  /* If the exec call supplied NULL for tables, use the inbuilt ones. This
6515  is a feature that makes it possible to save compiled regex and re-use them  is a feature that makes it possible to save compiled regex and re-use them
6516  in other programs later. */  in other programs later. */
# Line 6511  end_subject = md->end_subject; Line 6536  end_subject = md->end_subject;
6536  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
6537  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6538  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6539  md->ignore_skip_arg = FALSE;  md->ignore_skip_arg = 0;
6540    
6541  /* Some options are unpacked into BOOL variables in the hope that testing  /* Some options are unpacked into BOOL variables in the hope that testing
6542  them will be faster than individual option bits. */  them will be faster than individual option bits. */
# Line 6893  for(;;) Line 6918  for(;;)
6918    md->match_call_count = 0;    md->match_call_count = 0;
6919    md->match_function_type = 0;    md->match_function_type = 0;
6920    md->end_offset_top = 0;    md->end_offset_top = 0;
6921      md->skip_arg_count = 0;
6922    rc = match(start_match, md->start_code, start_match, 2, md, NULL, 0);    rc = match(start_match, md->start_code, start_match, 2, md, NULL, 0);
6923    if (md->hitend && start_partial == NULL)    if (md->hitend && start_partial == NULL)
6924      {      {
6925      start_partial = md->start_used_ptr;      start_partial = md->start_used_ptr;
6926      match_partial = start_match;      match_partial = start_match;
6927      }      }
6928    
6929    switch(rc)    switch(rc)
6930      {      {
# Line 6911  for(;;) Line 6937  for(;;)
6937    
6938      case MATCH_SKIP_ARG:      case MATCH_SKIP_ARG:
6939      new_start_match = start_match;      new_start_match = start_match;
6940      md->ignore_skip_arg = TRUE;      md->ignore_skip_arg = md->skip_arg_count;
6941      break;      break;
6942    
6943      /* SKIP passes back the next starting point explicitly, but if it is the      /* SKIP passes back the next starting point explicitly, but if it is no
6944      same as the match we have just done, treat it as NOMATCH. */      greater than the match we have just done, treat it as NOMATCH. */
6945    
6946      case MATCH_SKIP:      case MATCH_SKIP:
6947      if (md->start_match_ptr != start_match)      if (md->start_match_ptr > start_match)
6948        {        {
6949        new_start_match = md->start_match_ptr;        new_start_match = md->start_match_ptr;
6950        break;        break;
# Line 6926  for(;;) Line 6952  for(;;)
6952      /* Fall through */      /* Fall through */
6953    
6954      /* NOMATCH and PRUNE advance by one character. THEN at this level acts      /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6955      exactly like PRUNE. Unset the ignore SKIP-with-argument flag. */      exactly like PRUNE. Unset ignore SKIP-with-argument. */
6956    
6957      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6958      case MATCH_PRUNE:      case MATCH_PRUNE:
6959      case MATCH_THEN:      case MATCH_THEN:
6960      md->ignore_skip_arg = FALSE;      md->ignore_skip_arg = 0;
6961      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6962  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6963      if (utf)      if (utf)
# Line 7032  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 7058  if (rc == MATCH_MATCH || rc == MATCH_ACC
7058    /* Set the return code to the number of captured strings, or 0 if there were    /* Set the return code to the number of captured strings, or 0 if there were
7059    too many to fit into the vector. */    too many to fit into the vector. */
7060    
7061    rc = ((md->capture_last & OVFLBIT) != 0 &&    rc = ((md->capture_last & OVFLBIT) != 0 &&
7062           md->end_offset_top >= arg_offset_max)?           md->end_offset_top >= arg_offset_max)?
7063      0 : md->end_offset_top/2;      0 : md->end_offset_top/2;
7064    
# Line 7098  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;
# Line 7106  if (start_partial != NULL) Line 7132  if (start_partial != NULL)
7132      {      {
7133      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);
7134      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);
7135      if (offsetcount > 2)      if (offsetcount > 2)
7136        offsets[2] = (int)(match_partial - (PCRE_PUCHAR)subject);        offsets[2] = (int)(match_partial - (PCRE_PUCHAR)subject);
7137      }      }
7138    rc = PCRE_ERROR_PARTIAL;    rc = PCRE_ERROR_PARTIAL;

Legend:
Removed from v.1265  
changed lines
  Added in v.1376

  ViewVC Help
Powered by ViewVC 1.1.5