/[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 1248 by ph10, Wed Feb 13 17:36:38 2013 UTC revision 1417 by ph10, Tue Dec 24 18:03:06 2013 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 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 103  because the offset vector is always a mu Line 107  because the offset vector is always a mu
107    
108  /* Min and max values for the common repeats; for the maxima, 0 => infinity */  /* Min and max values for the common repeats; for the maxima, 0 => infinity */
109    
110  static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };  static const char rep_min[] = { 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, };
111  static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };  static const char rep_max[] = { 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, };
112    
113  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
114  /*************************************************  /*************************************************
# Line 163  match_ref(int offset, register PCRE_PUCH Line 167  match_ref(int offset, register PCRE_PUCH
167  {  {
168  PCRE_PUCHAR eptr_start = eptr;  PCRE_PUCHAR eptr_start = eptr;
169  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];  register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset];
170  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && defined SUPPORT_UCP
171  BOOL utf = md->utf;  BOOL utf = md->utf;
172  #endif  #endif
173    
# Line 191  ASCII characters. */ Line 195  ASCII characters. */
195    
196  if (caseless)  if (caseless)
197    {    {
198  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && defined SUPPORT_UCP
 #ifdef SUPPORT_UCP  
199    if (utf)    if (utf)
200      {      {
201      /* Match characters up to the end of the reference. NOTE: the number of      /* Match characters up to the end of the reference. NOTE: the number of
# Line 226  if (caseless) Line 229  if (caseless)
229      }      }
230    else    else
231  #endif  #endif
 #endif  
232    
233    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there    /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
234    is no UCP support. */    is no UCP support. */
235      {      {
236      while (length-- > 0)      while (length-- > 0)
237        {        {
238        pcre_uchar cc, cp;        pcre_uint32 cc, cp;
239        if (eptr >= md->end_subject) return -2;   /* Partial match */        if (eptr >= md->end_subject) return -2;   /* Partial match */
240        cc = RAWUCHARTEST(eptr);        cc = RAWUCHARTEST(eptr);
241        cp = RAWUCHARTEST(p);        cp = RAWUCHARTEST(p);
# Line 777  for (;;) Line 779  for (;;)
779      case OP_FAIL:      case OP_FAIL:
780      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
781    
     /* COMMIT overrides PRUNE, SKIP, and THEN */  
   
782      case OP_COMMIT:      case OP_COMMIT:
783      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
784        eptrb, RM52);        eptrb, RM52);
785      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);  
786      RRETURN(MATCH_COMMIT);      RRETURN(MATCH_COMMIT);
787    
     /* PRUNE overrides THEN */  
   
788      case OP_PRUNE:      case OP_PRUNE:
789      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
790        eptrb, RM51);        eptrb, RM51);
791      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
792      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
793    
794      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
# Line 803  for (;;) Line 798  for (;;)
798        eptrb, RM56);        eptrb, RM56);
799      if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&      if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) &&
800           md->mark == NULL) md->mark = ecode + 2;           md->mark == NULL) md->mark = ecode + 2;
801      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
802      RRETURN(MATCH_PRUNE);      RRETURN(MATCH_PRUNE);
803    
     /* SKIP overrides PRUNE and THEN */  
   
804      case OP_SKIP:      case OP_SKIP:
805      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md,
806        eptrb, RM53);        eptrb, RM53);
807      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
       RRETURN(rrc);  
808      md->start_match_ptr = eptr;   /* Pass back current position */      md->start_match_ptr = eptr;   /* Pass back current position */
809      RRETURN(MATCH_SKIP);      RRETURN(MATCH_SKIP);
810    
811      /* Note that, for Perl compatibility, SKIP with an argument does NOT set      /* Note that, for Perl compatibility, SKIP with an argument does NOT set
812      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
813      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
814        that failed and any that precede it (either they also failed, or were not
815        triggered). To do this, we maintain a count of executed SKIP_ARGs. If a
816        SKIP_ARG gets to top level, the match is re-run with md->ignore_skip_arg
817        set to the count of the one that failed. */
818    
819      case OP_SKIP_ARG:      case OP_SKIP_ARG:
820      if (md->ignore_skip_arg)      md->skip_arg_count++;
821        if (md->skip_arg_count <= md->ignore_skip_arg)
822        {        {
823        ecode += PRIV(OP_lengths)[*ecode] + ecode[1];        ecode += PRIV(OP_lengths)[*ecode] + ecode[1];
824        break;        break;
825        }        }
826      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,      RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md,
827        eptrb, RM57);        eptrb, RM57);
828      if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN)      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
       RRETURN(rrc);  
829    
830      /* 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
831      returning the special MATCH_SKIP_ARG return code. This will either be      returning the special MATCH_SKIP_ARG return code. This will either be
832      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
833      with the md->ignore_skip_arg flag set. */      with md->ignore_skip_arg set to the value of md->skip_arg_count. */
834    
835      md->start_match_ptr = ecode + 2;      md->start_match_ptr = ecode + 2;
836      RRETURN(MATCH_SKIP_ARG);      RRETURN(MATCH_SKIP_ARG);
# Line 1175  for (;;) Line 1171  for (;;)
1171            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1172            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1173            matched_once = TRUE;            matched_once = TRUE;
1174              mstart = md->start_match_ptr;    /* In case \K changed it */
1175            continue;            continue;
1176            }            }
1177    
# Line 1247  for (;;) Line 1244  for (;;)
1244          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1245          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1246          matched_once = TRUE;          matched_once = TRUE;
1247            mstart = md->start_match_ptr;   /* In case \K reset it */
1248          continue;          continue;
1249          }          }
1250    
# 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];
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 1524  for (;;) Line 1442  for (;;)
1442    
1443      else      else
1444        {        {
       ecode += 1 + LINK_SIZE;  
1445        }        }
1446      break;      break;
1447    
# Line 1604  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 1633  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 1655  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 1737  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 1784  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 1821  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 1838  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            if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX)
1819              RRETURN(MATCH_NOMATCH);
1820    
1821            /* Any return code other than NOMATCH is an error. */
1822    
1823          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&          if (rrc != MATCH_NOMATCH)
                  rrc != MATCH_COMMIT)  
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 2029  for (;;) Line 2007  for (;;)
2007    
2008      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
2009        {        {
2010          md->start_match_ptr = mstart;    /* In case \K reset it */
2011        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
2012        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
2013        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
# Line 2596  for (;;) Line 2575  for (;;)
2575            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2576          break;          break;
2577    
2578          case PT_SPACE:    /* Perl space */          /* Perl space used to exclude VT, but from Perl 5.18 it is included,
2579          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          which means that Perl space and POSIX space are now identical. PCRE
2580               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;  
2581    
2582            case PT_SPACE:    /* Perl space */
2583          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2584          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          switch(c)
2585               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            {
2586               c == CHAR_FF || c == CHAR_CR)            HSPACE_CASES:
2587                 == (op == OP_NOTPROP))            VSPACE_CASES:
2588            RRETURN(MATCH_NOMATCH);            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2589              break;
2590    
2591              default:
2592              if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) ==
2593                (op == OP_NOTPROP)) RRETURN(MATCH_NOMATCH);
2594              break;
2595              }
2596          break;          break;
2597    
2598          case PT_WORD:          case PT_WORD:
# Line 2629  for (;;) Line 2613  for (;;)
2613            }            }
2614          break;          break;
2615    
2616            case PT_UCNC:
2617            if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2618                 c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2619                 c >= 0xe000) == (op == OP_NOTPROP))
2620              RRETURN(MATCH_NOMATCH);
2621            break;
2622    
2623          /* This should never occur */          /* This should never occur */
2624    
2625          default:          default:
# Line 2675  for (;;) Line 2666  for (;;)
2666      similar code to character type repeats - written out again for speed.      similar code to character type repeats - written out again for speed.
2667      However, if the referenced string is the empty string, always treat      However, if the referenced string is the empty string, always treat
2668      it as matched, any number of times (otherwise there could be infinite      it as matched, any number of times (otherwise there could be infinite
2669      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:  
2670    
2671      (a) In the default, Perl-compatible state, set the length negative;      (a) In the default, Perl-compatible state, set the length negative;
2672      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 2693  for (;;) Line 2676  for (;;)
2676      so that the back reference matches an empty string.      so that the back reference matches an empty string.
2677    
2678      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
2679      referenced subpattern. */      referenced subpattern.
2680    
2681        The OP_REF and OP_REFI opcodes are used for a reference to a numbered group
2682        or to a non-duplicated named group. For a duplicated named group, OP_DNREF
2683        and OP_DNREFI are used. In this case we must scan the list of groups to
2684        which the name refers, and use the first one that is set. */
2685    
2686        case OP_DNREF:
2687        case OP_DNREFI:
2688        caseless = op == OP_DNREFI;
2689          {
2690          int count = GET2(ecode, 1+IMM2_SIZE);
2691          pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
2692          ecode += 1 + 2*IMM2_SIZE;
2693    
2694          /* Setting the default length first and initializing 'offset' avoids
2695          compiler warnings in the REF_REPEAT code. */
2696    
2697          length = (md->jscript_compat)? 0 : -1;
2698          offset = 0;
2699    
2700          while (count-- > 0)
2701            {
2702            offset = GET2(slot, 0) << 1;
2703            if (offset < offset_top && md->offset_vector[offset] >= 0)
2704              {
2705              length = md->offset_vector[offset+1] - md->offset_vector[offset];
2706              break;
2707              }
2708            slot += md->name_entry_size;
2709            }
2710          }
2711        goto REF_REPEAT;
2712    
2713        case OP_REF:
2714        case OP_REFI:
2715        caseless = op == OP_REFI;
2716        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2717        ecode += 1 + IMM2_SIZE;
2718      if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2719        length = (md->jscript_compat)? 0 : -1;        length = (md->jscript_compat)? 0 : -1;
2720      else      else
# Line 2702  for (;;) Line 2722  for (;;)
2722    
2723      /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2724    
2725        REF_REPEAT:
2726      switch (*ecode)      switch (*ecode)
2727        {        {
2728        case OP_CRSTAR:        case OP_CRSTAR:
# Line 2850  for (;;) Line 2871  for (;;)
2871          case OP_CRMINPLUS:          case OP_CRMINPLUS:
2872          case OP_CRQUERY:          case OP_CRQUERY:
2873          case OP_CRMINQUERY:          case OP_CRMINQUERY:
2874            case OP_CRPOSSTAR:
2875            case OP_CRPOSPLUS:
2876            case OP_CRPOSQUERY:
2877          c = *ecode++ - OP_CRSTAR;          c = *ecode++ - OP_CRSTAR;
2878          minimize = (c & 1) != 0;          if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0;
2879            else possessive = TRUE;
2880          min = rep_min[c];                 /* Pick up values from tables; */          min = rep_min[c];                 /* Pick up values from tables; */
2881          max = rep_max[c];                 /* zero for max => infinity */          max = rep_max[c];                 /* zero for max => infinity */
2882          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 2859  for (;;) Line 2884  for (;;)
2884    
2885          case OP_CRRANGE:          case OP_CRRANGE:
2886          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2887            case OP_CRPOSRANGE:
2888          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2889            possessive = (*ecode == OP_CRPOSRANGE);
2890          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2891          max = GET2(ecode, 1 + IMM2_SIZE);          max = GET2(ecode, 1 + IMM2_SIZE);
2892          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 3001  for (;;) Line 3028  for (;;)
3028                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3029              eptr += len;              eptr += len;
3030              }              }
3031    
3032              if (possessive) continue;    /* No backtracking */
3033    
3034            for (;;)            for (;;)
3035              {              {
3036              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
# Line 3031  for (;;) Line 3061  for (;;)
3061                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3062              eptr++;              eptr++;
3063              }              }
3064    
3065              if (possessive) continue;    /* No backtracking */
3066    
3067            while (eptr >= pp)            while (eptr >= pp)
3068              {              {
3069              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
# Line 3046  for (;;) Line 3079  for (;;)
3079      /* Control never gets here */      /* Control never gets here */
3080    
3081    
3082      /* Match an extended character class. This opcode is encountered only      /* Match an extended character class. In the 8-bit library, this opcode is
3083      when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8      encountered only when UTF-8 mode mode is supported. In the 16-bit and
3084      mode, because Unicode properties are supported in non-UTF-8 mode. */      32-bit libraries, codepoints greater than 255 may be encountered even when
3085        UTF is not supported. */
3086    
3087  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3088      case OP_XCLASS:      case OP_XCLASS:
# Line 3064  for (;;) Line 3098  for (;;)
3098          case OP_CRMINPLUS:          case OP_CRMINPLUS:
3099          case OP_CRQUERY:          case OP_CRQUERY:
3100          case OP_CRMINQUERY:          case OP_CRMINQUERY:
3101            case OP_CRPOSSTAR:
3102            case OP_CRPOSPLUS:
3103            case OP_CRPOSQUERY:
3104          c = *ecode++ - OP_CRSTAR;          c = *ecode++ - OP_CRSTAR;
3105          minimize = (c & 1) != 0;          if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0;
3106            else possessive = TRUE;
3107          min = rep_min[c];                 /* Pick up values from tables; */          min = rep_min[c];                 /* Pick up values from tables; */
3108          max = rep_max[c];                 /* zero for max => infinity */          max = rep_max[c];                 /* zero for max => infinity */
3109          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 3073  for (;;) Line 3111  for (;;)
3111    
3112          case OP_CRRANGE:          case OP_CRRANGE:
3113          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3114            case OP_CRPOSRANGE:
3115          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3116            possessive = (*ecode == OP_CRPOSRANGE);
3117          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3118          max = GET2(ecode, 1 + IMM2_SIZE);          max = GET2(ecode, 1 + IMM2_SIZE);
3119          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 3145  for (;;) Line 3185  for (;;)
3185            if (!PRIV(xclass)(c, data, utf)) break;            if (!PRIV(xclass)(c, data, utf)) break;
3186            eptr += len;            eptr += len;
3187            }            }
3188    
3189            if (possessive) continue;    /* No backtracking */
3190    
3191          for(;;)          for(;;)
3192            {            {
3193            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
# Line 3215  for (;;) Line 3258  for (;;)
3258    
3259        if (fc < 128)        if (fc < 128)
3260          {          {
3261          pcre_uchar cc = RAWUCHAR(eptr);          pcre_uint32 cc = RAWUCHAR(eptr);
3262          if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH);          if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH);
3263          ecode++;          ecode++;
3264          eptr++;          eptr++;
# Line 3320  for (;;) Line 3363  for (;;)
3363      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3364      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3365    
3366      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. We first check
3367        for the minimum number of characters. If the minimum equals the maximum, we
3368        are done. Otherwise, if minimizing, check the rest of the pattern for a
3369        match; if there isn't one, advance up to the maximum, one character at a
3370        time.
3371    
3372        If maximizing, advance up to the maximum number of matching characters,
3373        until eptr is past the end of the maximum run. If possessive, we are
3374        then done (no backing up). Otherwise, match at this position; anything
3375        other than no match is immediately returned. For nomatch, back up one
3376        character, unless we are matching \R and the last thing matched was
3377        \r\n, in which case, back up two bytes. When we reach the first optional
3378        character position, we can save stack by doing a tail recurse.
3379    
3380        The various UTF/non-UTF and caseful/caseless cases are handled separately,
3381        for speed. */
3382    
3383      REPEATCHAR:      REPEATCHAR:
3384  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
# Line 3404  for (;;) Line 3462  for (;;)
3462                }                }
3463              }              }
3464    
3465            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
   
3466            for(;;)            for(;;)
3467              {              {
3468                if (eptr == pp) goto TAIL_RECURSE;
3469              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3470              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
             if (eptr == pp) { RRETURN(MATCH_NOMATCH); }  
3471  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3472              eptr--;              eptr--;
3473              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3509  for (;;) Line 3566  for (;;)
3566            if (fc != cc && foc != cc) break;            if (fc != cc && foc != cc) break;
3567            eptr++;            eptr++;
3568            }            }
3569            if (possessive) continue;       /* No backtracking */
3570          if (possessive) continue;          for (;;)
   
         while (eptr >= pp)  
3571            {            {
3572              if (eptr == pp) goto TAIL_RECURSE;
3573            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3574            eptr--;            eptr--;
3575            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3576            }            }
3577          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3578          }          }
       /* Control never gets here */  
3579        }        }
3580    
3581      /* Caseful comparisons (includes all multi-byte characters) */      /* Caseful comparisons (includes all multi-byte characters) */
# Line 3568  for (;;) Line 3623  for (;;)
3623            if (fc != RAWUCHARTEST(eptr)) break;            if (fc != RAWUCHARTEST(eptr)) break;
3624            eptr++;            eptr++;
3625            }            }
3626          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
3627            for (;;)
         while (eptr >= pp)  
3628            {            {
3629              if (eptr == pp) goto TAIL_RECURSE;
3630            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3631            eptr--;            eptr--;
3632            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3633            }            }
3634          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3635          }          }
3636        }        }
3637      /* Control never gets here */      /* Control never gets here */
# Line 3748  for (;;) Line 3803  for (;;)
3803            }            }
3804          }          }
3805        else        else
3806  #endif  #endif  /* SUPPORT_UTF */
3807        /* Not UTF mode */        /* Not UTF mode */
3808          {          {
3809          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3786  for (;;) Line 3841  for (;;)
3841              }              }
3842            }            }
3843          else          else
3844  #endif  #endif  /*SUPPORT_UTF */
3845          /* Not UTF mode */          /* Not UTF mode */
3846            {            {
3847            for (fi = min;; fi++)            for (fi = min;; fi++)
# Line 3828  for (;;) Line 3883  for (;;)
3883              if (fc == d || (unsigned int)foc == d) break;              if (fc == d || (unsigned int)foc == d) break;
3884              eptr += len;              eptr += len;
3885              }              }
3886            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3887            for(;;)            for(;;)
3888              {              {
3889                if (eptr == pp) goto TAIL_RECURSE;
3890              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3891              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3892              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
3893              BACKCHAR(eptr);              BACKCHAR(eptr);
3894              }              }
3895            }            }
3896          else          else
3897  #endif  #endif  /* SUPPORT_UTF */
3898          /* Not UTF mode */          /* Not UTF mode */
3899            {            {
3900            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3851  for (;;) Line 3907  for (;;)
3907              if (fc == *eptr || foc == *eptr) break;              if (fc == *eptr || foc == *eptr) break;
3908              eptr++;              eptr++;
3909              }              }
3910            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3911            while (eptr >= pp)            for (;;)
3912              {              {
3913                if (eptr == pp) goto TAIL_RECURSE;
3914              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3915              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3916              eptr--;              eptr--;
3917              }              }
3918            }            }
3919            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
3920          }          }
       /* Control never gets here */  
3921        }        }
3922    
3923      /* Caseful comparisons */      /* Caseful comparisons */
# Line 3963  for (;;) Line 4018  for (;;)
4018              if (fc == d) break;              if (fc == d) break;
4019              eptr += len;              eptr += len;
4020              }              }
4021            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
4022            for(;;)            for(;;)
4023              {              {
4024                if (eptr == pp) goto TAIL_RECURSE;
4025              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
4026              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4027              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
4028              BACKCHAR(eptr);              BACKCHAR(eptr);
4029              }              }
4030            }            }
# Line 3986  for (;;) Line 4042  for (;;)
4042              if (fc == *eptr) break;              if (fc == *eptr) break;
4043              eptr++;              eptr++;
4044              }              }
4045            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
4046            while (eptr >= pp)            for (;;)
4047              {              {
4048                if (eptr == pp) goto TAIL_RECURSE;
4049              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
4050              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4051              eptr--;              eptr--;
4052              }              }
4053            }            }
4054            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
4055          }          }
4056        }        }
4057      /* Control never gets here */      /* Control never gets here */
# Line 4177  for (;;) Line 4233  for (;;)
4233              }              }
4234            break;            break;
4235    
4236              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4237              which means that Perl space and POSIX space are now identical. PCRE
4238              was changed at release 8.34. */
4239    
4240            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4241              case PT_PXSPACE:  /* POSIX space */
4242            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4243              {              {
4244              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4186  for (;;) Line 4247  for (;;)
4247                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4248                }                }
4249              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4250              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)  
4251                {                {
4252                SCHECK_PARTIAL();                HSPACE_CASES:
4253                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
4254                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4255                  break;
4256    
4257                  default:
4258                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
4259                    RRETURN(MATCH_NOMATCH);
4260                  break;
4261                }                }
             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);  
4262              }              }
4263            break;            break;
4264    
# Line 4247  for (;;) Line 4300  for (;;)
4300              }              }
4301            break;            break;
4302    
4303              case PT_UCNC:
4304              for (i = 1; i <= min; i++)
4305                {
4306                if (eptr >= md->end_subject)
4307                  {
4308                  SCHECK_PARTIAL();
4309                  RRETURN(MATCH_NOMATCH);
4310                  }
4311                GETCHARINCTEST(c, eptr);
4312                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
4313                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
4314                     c >= 0xe000) == prop_fail_result)
4315                  RRETURN(MATCH_NOMATCH);
4316                }
4317              break;
4318    
4319            /* This should not occur */            /* This should not occur */
4320    
4321            default:            default:
# Line 4452  for (;;) Line 4521  for (;;)
4521          case OP_DIGIT:          case OP_DIGIT:
4522          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4523            {            {
4524            pcre_uchar cc;            pcre_uint32 cc;
   
4525            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4526              {              {
4527              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4470  for (;;) Line 4538  for (;;)
4538          case OP_NOT_WHITESPACE:          case OP_NOT_WHITESPACE:
4539          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4540            {            {
4541            pcre_uchar cc;            pcre_uint32 cc;
   
4542            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4543              {              {
4544              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4488  for (;;) Line 4555  for (;;)
4555          case OP_WHITESPACE:          case OP_WHITESPACE:
4556          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4557            {            {
4558            pcre_uchar cc;            pcre_uint32 cc;
   
4559            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4560              {              {
4561              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4506  for (;;) Line 4572  for (;;)
4572          case OP_NOT_WORDCHAR:          case OP_NOT_WORDCHAR:
4573          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4574            {            {
4575            pcre_uchar cc;            pcre_uint32 cc;
   
4576            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4577              {              {
4578              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4524  for (;;) Line 4589  for (;;)
4589          case OP_WORDCHAR:          case OP_WORDCHAR:
4590          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4591            {            {
4592            pcre_uchar cc;            pcre_uint32 cc;
   
4593            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4594              {              {
4595              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4914  for (;;) Line 4978  for (;;)
4978              }              }
4979            /* Control never gets here */            /* Control never gets here */
4980    
4981            case PT_SPACE:    /* Perl space */            /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4982            for (fi = min;; fi++)            which means that Perl space and POSIX space are now identical. PCRE
4983              {            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 */  
4984    
4985              case PT_SPACE:    /* Perl space */
4986            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4987            for (fi = min;; fi++)            for (fi = min;; fi++)
4988              {              {
# Line 4945  for (;;) Line 4995  for (;;)
4995                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4996                }                }
4997              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4998              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
4999                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                {
5000                     == prop_fail_result)                HSPACE_CASES:
5001                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
5002                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
5003                  break;
5004    
5005                  default:
5006                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
5007                    RRETURN(MATCH_NOMATCH);
5008                  break;
5009                  }
5010              }              }
5011            /* Control never gets here */            /* Control never gets here */
5012    
# Line 4998  for (;;) Line 5056  for (;;)
5056              }              }
5057            /* Control never gets here */            /* Control never gets here */
5058    
5059              case PT_UCNC:
5060              for (fi = min;; fi++)
5061                {
5062                RMATCH(eptr, ecode, offset_top, md, eptrb, RM60);
5063                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5064                if (fi >= max) RRETURN(MATCH_NOMATCH);
5065                if (eptr >= md->end_subject)
5066                  {
5067                  SCHECK_PARTIAL();
5068                  RRETURN(MATCH_NOMATCH);
5069                  }
5070                GETCHARINCTEST(c, eptr);
5071                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
5072                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
5073                     c >= 0xe000) == prop_fail_result)
5074                  RRETURN(MATCH_NOMATCH);
5075                }
5076              /* Control never gets here */
5077    
5078            /* This should never occur */            /* This should never occur */
5079            default:            default:
5080            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
# Line 5413  for (;;) Line 5490  for (;;)
5490              }              }
5491            break;            break;
5492    
5493              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
5494              which means that Perl space and POSIX space are now identical. PCRE
5495              was changed at release 8.34. */
5496    
5497            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
5498              case PT_PXSPACE:  /* POSIX space */
5499            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5500              {              {
5501              int len = 1;              int len = 1;
# Line 5423  for (;;) Line 5505  for (;;)
5505                break;                break;
5506                }                }
5507              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5508              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
5509                   c == CHAR_FF || c == CHAR_CR)                {
5510                   == prop_fail_result)                HSPACE_CASES:
5511                  VSPACE_CASES:
5512                  if (prop_fail_result) goto ENDLOOP99;  /* Break the loop */
5513                break;                break;
             eptr+= len;  
             }  
           break;  
5514    
5515            case PT_PXSPACE:  /* POSIX space */                default:
5516            for (i = min; i < max; i++)                if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
5517              {                  goto ENDLOOP99;   /* Break the loop */
             int len = 1;  
             if (eptr >= md->end_subject)  
               {  
               SCHECK_PARTIAL();  
5518                break;                break;
5519                }                }
             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;  
5520              eptr+= len;              eptr+= len;
5521              }              }
5522              ENDLOOP99:
5523            break;            break;
5524    
5525            case PT_WORD:            case PT_WORD:
# Line 5492  for (;;) Line 5565  for (;;)
5565            GOT_MAX:            GOT_MAX:
5566            break;            break;
5567    
5568              case PT_UCNC:
5569              for (i = min; i < max; i++)
5570                {
5571                int len = 1;
5572                if (eptr >= md->end_subject)
5573                  {
5574                  SCHECK_PARTIAL();
5575                  break;
5576                  }
5577                GETCHARLENTEST(c, eptr, len);
5578                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
5579                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
5580                     c >= 0xe000) == prop_fail_result)
5581                  break;
5582                eptr += len;
5583                }
5584              break;
5585    
5586            default:            default:
5587            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5588            }            }
5589    
5590          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5591    
5592          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
5593          for(;;)          for(;;)
5594            {            {
5595              if (eptr == pp) goto TAIL_RECURSE;
5596            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5597            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5598            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5599            if (utf) BACKCHAR(eptr);            if (utf) BACKCHAR(eptr);
5600            }            }
5601          }          }
5602    
5603        /* Match extended Unicode sequences. We will get here only if the        /* Match extended Unicode grapheme clusters. We will get here only if the
5604        support is in the binary; otherwise a compile-time error occurs. */        support is in the binary; otherwise a compile-time error occurs. */
5605    
5606        else if (ctype == OP_EXTUNI)        else if (ctype == OP_EXTUNI)
# Line 5540  for (;;) Line 5632  for (;;)
5632    
5633          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5634    
5635          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
5636    
5637          for(;;)          for(;;)
5638            {            {
5639              int lgb, rgb;
5640              PCRE_PUCHAR fptr;
5641    
5642              if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5643            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5644            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5645            if (eptr-- == pp) break;        /* Stop if tried at original pos */  
5646            for (;;)                        /* Move back over one extended */            /* Backtracking over an extended grapheme cluster involves inspecting
5647              the previous two characters (if present) to see if a break is
5648              permitted between them. */
5649    
5650              eptr--;
5651              if (!utf) c = *eptr; else
5652              {              {
5653              if (!utf) c = *eptr; else              BACKCHAR(eptr);
5654                GETCHAR(c, eptr);
5655                }
5656              rgb = UCD_GRAPHBREAK(c);
5657    
5658              for (;;)
5659                {
5660                if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5661                fptr = eptr - 1;
5662                if (!utf) c = *fptr; else
5663                {                {
5664                BACKCHAR(eptr);                BACKCHAR(fptr);
5665                GETCHAR(c, eptr);                GETCHAR(c, fptr);
5666                }                }
5667              if (UCD_CATEGORY(c) != ucp_M) break;              lgb = UCD_GRAPHBREAK(c);
5668              eptr--;              if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
5669                eptr = fptr;
5670                rgb = lgb;
5671              }              }
5672            }            }
5673          }          }
# Line 5821  for (;;) Line 5933  for (;;)
5933            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5934            }            }
5935    
5936          /* 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;  
5937          for(;;)          for(;;)
5938            {            {
5939              if (eptr == pp) goto TAIL_RECURSE;
5940            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5941            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5942            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5943            BACKCHAR(eptr);            BACKCHAR(eptr);
5944            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&
5945                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;
# Line 6070  for (;;) Line 6177  for (;;)
6177            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
6178            }            }
6179    
6180          /* eptr is now past the end of the maximum run. If possessive, we are          if (possessive) continue;    /* No backtracking */
6181          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)  
6182            {            {
6183              if (eptr == pp) goto TAIL_RECURSE;
6184            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
6185            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
6186            eptr--;            eptr--;
# Line 6087  for (;;) Line 6189  for (;;)
6189            }            }
6190          }          }
6191    
6192        /* Get here if we can't make it match with any permitted repetitions */        /* Control never gets here */
   
       RRETURN(MATCH_NOMATCH);  
6193        }        }
     /* Control never gets here */  
6194    
6195      /* 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
6196      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 6125  switch (frame->Xwhere) Line 6224  switch (frame->Xwhere)
6224    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)    LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64)
6225    LBL(65) LBL(66)    LBL(65) LBL(66)
6226  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
6227    LBL(21)    LBL(20) LBL(21)
6228  #endif  #endif
6229  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6230    LBL(16) LBL(18) LBL(20)    LBL(16) LBL(18)
6231    LBL(22) LBL(23) LBL(28) LBL(30)    LBL(22) LBL(23) LBL(28) LBL(30)
6232    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
6233  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 6286  const pcre_uint8 *start_bits = NULL; Line 6385  const pcre_uint8 *start_bits = NULL;
6385  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;
6386  PCRE_PUCHAR end_subject;  PCRE_PUCHAR end_subject;
6387  PCRE_PUCHAR start_partial = NULL;  PCRE_PUCHAR start_partial = NULL;
6388    PCRE_PUCHAR match_partial = NULL;
6389  PCRE_PUCHAR req_char_ptr = start_match - 1;  PCRE_PUCHAR req_char_ptr = start_match - 1;
6390    
6391  const pcre_study_data *study;  const pcre_study_data *study;
# Line 6415  md->callout_data = NULL; Line 6515  md->callout_data = NULL;
6515    
6516  tables = re->tables;  tables = re->tables;
6517    
6518    /* The two limit values override the defaults, whatever their value. */
6519    
6520  if (extra_data != NULL)  if (extra_data != NULL)
6521    {    {
6522    register unsigned int flags = extra_data->flags;    register unsigned int flags = extra_data->flags;
# Line 6429  if (extra_data != NULL) Line 6531  if (extra_data != NULL)
6531    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;
6532    }    }
6533    
6534    /* Limits in the regex override only if they are smaller. */
6535    
6536    if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit)
6537      md->match_limit = re->limit_match;
6538    
6539    if ((re->flags & PCRE_RLSET) != 0 &&
6540        re->limit_recursion < md->match_limit_recursion)
6541      md->match_limit_recursion = re->limit_recursion;
6542    
6543  /* 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
6544  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
6545  in other programs later. */  in other programs later. */
# Line 6454  end_subject = md->end_subject; Line 6565  end_subject = md->end_subject;
6565  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
6566  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6567  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6568  md->ignore_skip_arg = FALSE;  md->ignore_skip_arg = 0;
6569    
6570  /* Some options are unpacked into BOOL variables in the hope that testing  /* Some options are unpacked into BOOL variables in the hope that testing
6571  them will be faster than individual option bits. */  them will be faster than individual option bits. */
# Line 6728  for(;;) Line 6839  for(;;)
6839  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
6840          if (c > 255) c = 255;          if (c > 255) c = 255;
6841  #endif  #endif
6842          if ((start_bits[c/8] & (1 << (c&7))) == 0)          if ((start_bits[c/8] & (1 << (c&7))) != 0) break;
6843            {          start_match++;
           start_match++;  
 #if defined SUPPORT_UTF && defined COMPILE_PCRE8  
           /* In non 8-bit mode, the iteration will stop for  
           characters > 255 at the beginning or not stop at all. */  
           if (utf)  
             ACROSSCHAR(start_match < end_subject, *start_match,  
               start_match++);  
 #endif  
           }  
         else break;  
6844          }          }
6845        }        }
6846      }   /* Starting optimizations */      }   /* Starting optimizations */
# Line 6836  for(;;) Line 6937  for(;;)
6937    md->match_call_count = 0;    md->match_call_count = 0;
6938    md->match_function_type = 0;    md->match_function_type = 0;
6939    md->end_offset_top = 0;    md->end_offset_top = 0;
6940      md->skip_arg_count = 0;
6941    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);
6942    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL)
6943        {
6944        start_partial = md->start_used_ptr;
6945        match_partial = start_match;
6946        }
6947    
6948    switch(rc)    switch(rc)
6949      {      {
# Line 6850  for(;;) Line 6956  for(;;)
6956    
6957      case MATCH_SKIP_ARG:      case MATCH_SKIP_ARG:
6958      new_start_match = start_match;      new_start_match = start_match;
6959      md->ignore_skip_arg = TRUE;      md->ignore_skip_arg = md->skip_arg_count;
6960      break;      break;
6961    
6962      /* 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
6963      same as the match we have just done, treat it as NOMATCH. */      greater than the match we have just done, treat it as NOMATCH. */
6964    
6965      case MATCH_SKIP:      case MATCH_SKIP:
6966      if (md->start_match_ptr != start_match)      if (md->start_match_ptr > start_match)
6967        {        {
6968        new_start_match = md->start_match_ptr;        new_start_match = md->start_match_ptr;
6969        break;        break;
# Line 6865  for(;;) Line 6971  for(;;)
6971      /* Fall through */      /* Fall through */
6972    
6973      /* NOMATCH and PRUNE advance by one character. THEN at this level acts      /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6974      exactly like PRUNE. Unset the ignore SKIP-with-argument flag. */      exactly like PRUNE. Unset ignore SKIP-with-argument. */
6975    
6976      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6977      case MATCH_PRUNE:      case MATCH_PRUNE:
6978      case MATCH_THEN:      case MATCH_THEN:
6979      md->ignore_skip_arg = FALSE;      md->ignore_skip_arg = 0;
6980      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6981  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6982      if (utf)      if (utf)
# Line 6971  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 7077  if (rc == MATCH_MATCH || rc == MATCH_ACC
7077    /* 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
7078    too many to fit into the vector. */    too many to fit into the vector. */
7079    
7080    rc = ((md->capture_last & OVFLBIT) != 0 &&    rc = ((md->capture_last & OVFLBIT) != 0 &&
7081           md->end_offset_top >= arg_offset_max)?           md->end_offset_top >= arg_offset_max)?
7082      0 : md->end_offset_top/2;      0 : md->end_offset_top/2;
7083    
# Line 7037  if (rc != MATCH_NOMATCH && rc != PCRE_ER Line 7143  if (rc != MATCH_NOMATCH && rc != PCRE_ER
7143    
7144  /* Handle partial matches - disable any mark data */  /* Handle partial matches - disable any mark data */
7145    
7146  if (start_partial != NULL)  if (match_partial != NULL)
7147    {    {
7148    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));
7149    md->mark = NULL;    md->mark = NULL;
# Line 7045  if (start_partial != NULL) Line 7151  if (start_partial != NULL)
7151      {      {
7152      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);
7153      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);
7154        if (offsetcount > 2)
7155          offsets[2] = (int)(match_partial - (PCRE_PUCHAR)subject);
7156      }      }
7157    rc = PCRE_ERROR_PARTIAL;    rc = PCRE_ERROR_PARTIAL;
7158    }    }

Legend:
Removed from v.1248  
changed lines
  Added in v.1417

  ViewVC Help
Powered by ViewVC 1.1.5