/[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 1238 by ph10, Sat Jan 5 16:27:59 2013 UTC revision 1397 by ph10, Mon Nov 11 18:33:23 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
60    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
62    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
64    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
66    separate set of save/restore instructions. The following defines are used in
67    implementing this. */
68    
69    #define CAPLMASK    0x0000ffff    /* The bits used for last_capture */
70    #define OVFLMASK    0xffff0000    /* The bits used for the overflow flag */
71    #define OVFLBIT     0x00010000    /* The bit that is set for overflow */
72    
73  /* Values for setting in md->match_function_type to indicate two special types  /* Values for setting in md->match_function_type to indicate two special types
74  of call to match(). We do it this way to save on using another stack variable,  of call to match(). We do it this way to save on using another stack variable,
75  as stack usage is to be discouraged. */  as stack usage is to be discouraged. */
# Line 73  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 89  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 219  if (caseless) Line 237  if (caseless)
237      {      {
238      while (length-- > 0)      while (length-- > 0)
239        {        {
240        pcre_uchar cc, cp;        pcre_uint32 cc, cp;
241        if (eptr >= md->end_subject) return -2;   /* Partial match */        if (eptr >= md->end_subject) return -2;   /* Partial match */
242        cc = RAWUCHARTEST(eptr);        cc = RAWUCHARTEST(eptr);
243        cp = RAWUCHARTEST(p);        cp = RAWUCHARTEST(p);
# Line 294  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM Line 312  enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM
312         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,         RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
313         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,         RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
314         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,         RM51,  RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60,
315         RM61,  RM62, RM63, RM64, RM65, RM66, RM67 };         RM61,  RM62, RM63, RM64, RM65, RM66, RM67, RM68 };
316    
317  /* These versions of the macros use the stack, as normal. There are debugging  /* These versions of the macros use the stack, as normal. There are debugging
318  versions and production versions. Note that the "rw" argument of RMATCH isn't  versions and production versions. Note that the "rw" argument of RMATCH isn't
# Line 419  typedef struct heapframe { Line 437  typedef struct heapframe {
437    unsigned int Xnumber;    unsigned int Xnumber;
438    int Xoffset;    int Xoffset;
439    unsigned int Xop;    unsigned int Xop;
440    int Xsave_capture_last;    pcre_int32 Xsave_capture_last;
441    int Xsave_offset1, Xsave_offset2, Xsave_offset3;    int Xsave_offset1, Xsave_offset2, Xsave_offset3;
442    int Xstacksave[REC_STACK_SAVE_MAX];    int Xstacksave[REC_STACK_SAVE_MAX];
443    
# Line 635  int min; Line 653  int min;
653  unsigned int number;  unsigned int number;
654  int offset;  int offset;
655  unsigned int op;  unsigned int op;
656  int save_capture_last;  pcre_int32 save_capture_last;
657  int save_offset1, save_offset2, save_offset3;  int save_offset1, save_offset2, save_offset3;
658  int stacksave[REC_STACK_SAVE_MAX];  int stacksave[REC_STACK_SAVE_MAX];
659    
# Line 763  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 789  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 1066  for (;;) Line 1078  for (;;)
1078        /* In all other cases, we have to make another call to match(). */        /* In all other cases, we have to make another call to match(). */
1079    
1080        save_mark = md->mark;        save_mark = md->mark;
1081          save_capture_last = md->capture_last;
1082        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,        RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb,
1083          RM2);          RM2);
1084    
# Line 1097  for (;;) Line 1110  for (;;)
1110        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1111        md->mark = save_mark;        md->mark = save_mark;
1112        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1113          md->capture_last = save_capture_last;
1114        }        }
1115    
1116      RRETURN(MATCH_NOMATCH);      RRETURN(MATCH_NOMATCH);
# Line 1159  for (;;) Line 1173  for (;;)
1173            ecode = md->start_code + code_offset;            ecode = md->start_code + code_offset;
1174            save_capture_last = md->capture_last;            save_capture_last = md->capture_last;
1175            matched_once = TRUE;            matched_once = TRUE;
1176              mstart = md->start_match_ptr;    /* In case \K changed it */
1177            continue;            continue;
1178            }            }
1179    
# Line 1218  for (;;) Line 1233  for (;;)
1233      POSSESSIVE_NON_CAPTURE:      POSSESSIVE_NON_CAPTURE:
1234      matched_once = FALSE;      matched_once = FALSE;
1235      code_offset = (int)(ecode - md->start_code);      code_offset = (int)(ecode - md->start_code);
1236        save_capture_last = md->capture_last;
1237    
1238      for (;;)      for (;;)
1239        {        {
# Line 1230  for (;;) Line 1246  for (;;)
1246          eptr = md->end_match_ptr;          eptr = md->end_match_ptr;
1247          ecode = md->start_code + code_offset;          ecode = md->start_code + code_offset;
1248          matched_once = TRUE;          matched_once = TRUE;
1249            mstart = md->start_match_ptr;   /* In case \K reset it */
1250          continue;          continue;
1251          }          }
1252    
# Line 1247  for (;;) Line 1264  for (;;)
1264        if (rrc != MATCH_NOMATCH) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1265        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1266        if (*ecode != OP_ALT) break;        if (*ecode != OP_ALT) break;
1267          md->capture_last = save_capture_last;
1268        }        }
1269    
1270      if (matched_once || allow_zero)      if (matched_once || allow_zero)
# Line 1258  for (;;) Line 1276  for (;;)
1276    
1277      /* Control never reaches here. */      /* Control never reaches here. */
1278    
1279      /* Conditional group: compilation checked that there are no more than      /* Conditional group: compilation checked that there are no more than two
1280      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
1281      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
1282      exactly what going to the ket would do. */      what we want. */
1283    
1284      case OP_COND:      case OP_COND:
1285      case OP_SCOND:      case OP_SCOND:
1286      codelink = GET(ecode, 1);  
1287        /* The variable codelink will be added to ecode when the condition is
1288        false, to get to the second branch. Setting it to the offset to the ALT
1289        or KET, then incrementing ecode achieves this effect. We now have ecode
1290        pointing to the condition or callout. */
1291    
1292        codelink = GET(ecode, 1);   /* Offset to the second branch */
1293        ecode += 1 + LINK_SIZE;     /* From this opcode */
1294    
1295      /* 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
1296      inserted between OP_COND and an assertion condition. */      inserted between OP_COND and an assertion condition. */
1297    
1298      if (ecode[LINK_SIZE+1] == OP_CALLOUT)      if (*ecode == OP_CALLOUT)
1299        {        {
1300        if (PUBL(callout) != NULL)        if (PUBL(callout) != NULL)
1301          {          {
1302          PUBL(callout_block) cb;          PUBL(callout_block) cb;
1303          cb.version          = 2;   /* Version 1 of the callout block */          cb.version          = 2;   /* Version 1 of the callout block */
1304          cb.callout_number   = ecode[LINK_SIZE+2];          cb.callout_number   = ecode[1];
1305          cb.offset_vector    = md->offset_vector;          cb.offset_vector    = md->offset_vector;
1306  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1307          cb.subject          = (PCRE_SPTR)md->start_subject;          cb.subject          = (PCRE_SPTR)md->start_subject;
# Line 1288  for (;;) Line 1313  for (;;)
1313          cb.subject_length   = (int)(md->end_subject - md->start_subject);          cb.subject_length   = (int)(md->end_subject - md->start_subject);
1314          cb.start_match      = (int)(mstart - md->start_subject);          cb.start_match      = (int)(mstart - md->start_subject);
1315          cb.current_position = (int)(eptr - md->start_subject);          cb.current_position = (int)(eptr - md->start_subject);
1316          cb.pattern_position = GET(ecode, LINK_SIZE + 3);          cb.pattern_position = GET(ecode, 2);
1317          cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);          cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1318          cb.capture_top      = offset_top/2;          cb.capture_top      = offset_top/2;
1319          cb.capture_last     = md->capture_last;          cb.capture_last     = md->capture_last & CAPLMASK;
1320            /* Internal change requires this for API compatibility. */
1321            if (cb.capture_last == 0) cb.capture_last = -1;
1322          cb.callout_data     = md->callout_data;          cb.callout_data     = md->callout_data;
1323          cb.mark             = md->nomatch_mark;          cb.mark             = md->nomatch_mark;
1324          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);          if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
1325          if (rrc < 0) RRETURN(rrc);          if (rrc < 0) RRETURN(rrc);
1326          }          }
1327    
1328          /* Advance ecode past the callout, so it now points to the condition. We
1329          must adjust codelink so that the value of ecode+codelink is unchanged. */
1330    
1331        ecode += PRIV(OP_lengths)[OP_CALLOUT];        ecode += PRIV(OP_lengths)[OP_CALLOUT];
1332          codelink -= PRIV(OP_lengths)[OP_CALLOUT];
1333        }        }
1334    
1335      condcode = ecode[LINK_SIZE+1];      /* Test the various possible conditions */
   
     /* Now see what the actual condition is */  
1336    
1337      if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */      condition = FALSE;
1338        switch(condcode = *ecode)
1339        {        {
1340        if (md->recursive == NULL)                /* Not recursing => FALSE */        case OP_RREF:         /* Numbered group recursion test */
1341          if (md->recursive != NULL)     /* Not recursing => FALSE */
1342          {          {
1343          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*/  
1344          condition = (recno == RREF_ANY || recno == md->recursive->group_num);          condition = (recno == RREF_ANY || recno == md->recursive->group_num);
1345            }
1346          break;
1347    
1348          /* If the test is for recursion into a specific subpattern, and it is        case OP_DNRREF:       /* Duplicate named group recursion test */
1349          false, but the test was set up by name, scan the table to see if the        if (md->recursive != NULL)
1350          name refers to any other numbers, and test them. The condition is true          {
1351          if any one is set. */          int count = GET2(ecode, 1 + IMM2_SIZE);
1352            pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
1353          if (!condition && condcode == OP_NRREF)          while (count-- > 0)
1354            {            {
1355            pcre_uchar *slotA = md->name_table;            unsigned int recno = GET2(slot, 0);
1356            for (i = 0; i < md->name_count; i++)            condition = recno == md->recursive->group_num;
1357              {            if (condition) break;
1358              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;  
                 }  
               }  
             }  
1359            }            }
   
         /* Chose branch according to the condition */  
   
         ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);  
1360          }          }
1361        }        break;
1362    
1363      else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */        case OP_CREF:         /* Numbered group used test */
1364        {        offset = GET2(ecode, 1) << 1;  /* Doubled ref number */
       offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */  
1365        condition = offset < offset_top && md->offset_vector[offset] >= 0;        condition = offset < offset_top && md->offset_vector[offset] >= 0;
1366          break;
1367    
1368        /* 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)  
1369          {          {
1370          unsigned int refno = offset >> 1;          int count = GET2(ecode, 1 + IMM2_SIZE);
1371          pcre_uchar *slotA = md->name_table;          pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
1372            while (count-- > 0)
1373          for (i = 0; i < md->name_count; i++)            {
1374            {            offset = GET2(slot, 0) << 1;
1375            if (GET2(slotA, 0) == refno) break;            condition = offset < offset_top && md->offset_vector[offset] >= 0;
1376            slotA += md->name_entry_size;            if (condition) break;
1377            }            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;  
               }  
             }  
1378            }            }
1379          }          }
1380          break;
1381    
1382        /* Chose branch according to the condition */        case OP_DEF:     /* DEFINE - always false */
1383          break;
       ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1);  
       }  
   
     else if (condcode == OP_DEF)     /* DEFINE - always false */  
       {  
       condition = FALSE;  
       ecode += GET(ecode, 1);  
       }  
1384    
1385      /* The condition is an assertion. Call match() to evaluate it - setting        /* The condition is an assertion. Call match() to evaluate it - setting
1386      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
1387      an assertion. */        of an assertion. */
1388    
1389      else        default:
       {  
1390        md->match_function_type = MATCH_CONDASSERT;        md->match_function_type = MATCH_CONDASSERT;
1391        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3);        RMATCH(eptr, ecode, offset_top, md, NULL, RM3);
1392        if (rrc == MATCH_MATCH)        if (rrc == MATCH_MATCH)
1393          {          {
1394          if (md->end_offset_top > offset_top)          if (md->end_offset_top > offset_top)
1395            offset_top = md->end_offset_top;  /* Captures may have happened */            offset_top = md->end_offset_top;  /* Captures may have happened */
1396          condition = TRUE;          condition = TRUE;
1397          ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);  
1398            /* Advance ecode past the assertion to the start of the first branch,
1399            but adjust it so that the general choosing code below works. */
1400    
1401            ecode += GET(ecode, 1);
1402          while (*ecode == OP_ALT) ecode += GET(ecode, 1);          while (*ecode == OP_ALT) ecode += GET(ecode, 1);
1403            ecode += 1 + LINK_SIZE - PRIV(OP_lengths)[condcode];
1404          }          }
1405    
1406        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an        /* PCRE doesn't allow the effect of (*THEN) to escape beyond an
1407        assertion; it is therefore treated as NOMATCH. */        assertion; it is therefore treated as NOMATCH. Any other return is an
1408          error. */
1409    
1410        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
1411          {          {
1412          RRETURN(rrc);         /* Need braces because of following else */          RRETURN(rrc);         /* Need braces because of following else */
1413          }          }
1414        else        break;
         {  
         condition = FALSE;  
         ecode += codelink;  
         }  
1415        }        }
1416    
1417      /* We are now at the branch that is to be obeyed. As there is only one, can      /* Choose branch according to the condition */
     use tail recursion to avoid using another stack frame, except when there is  
     unlimited repeat of a possibly empty group. In the latter case, a recursive  
     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)      ecode += condition? PRIV(OP_lengths)[condcode] : codelink;
1420    
1421        /* We are now at the branch that is to be obeyed. As there is only one, we
1422        can use tail recursion to avoid using another stack frame, except when
1423        there is unlimited repeat of a possibly empty group. In the latter case, a
1424        recursive call to match() is always required, unless the second alternative
1425        doesn't exist, in which case we can just plough on. Note that, for
1426        compatibility with Perl, the | in a conditional group is NOT treated as
1427        creating two alternatives. If a THEN is encountered in the branch, it
1428        propagates out to the enclosing alternative (unless nested in a deeper set
1429        of alternatives, of course). */
1430    
1431        if (condition || ecode[-(1+LINK_SIZE)] == OP_ALT)
1432        {        {
1433        if (op != OP_SCOND)        if (op != OP_SCOND)
1434          {          {
         ecode += 1 + LINK_SIZE;  
1435          goto TAIL_RECURSE;          goto TAIL_RECURSE;
1436          }          }
1437    
1438        md->match_function_type = MATCH_CBEGROUP;        md->match_function_type = MATCH_CBEGROUP;
1439        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49);        RMATCH(eptr, ecode, offset_top, md, eptrb, RM49);
1440        RRETURN(rrc);        RRETURN(rrc);
1441        }        }
1442    
# Line 1504  for (;;) Line 1444  for (;;)
1444    
1445      else      else
1446        {        {
       ecode += 1 + LINK_SIZE;  
1447        }        }
1448      break;      break;
1449    
# Line 1513  for (;;) Line 1452  for (;;)
1452      to close any currently open capturing brackets. */      to close any currently open capturing brackets. */
1453    
1454      case OP_CLOSE:      case OP_CLOSE:
1455      number = GET2(ecode, 1);      number = GET2(ecode, 1);   /* Must be less than 65536 */
1456      offset = number << 1;      offset = number << 1;
1457    
1458  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 1521  for (;;) Line 1460  for (;;)
1460        printf("\n");        printf("\n");
1461  #endif  #endif
1462    
1463      md->capture_last = number;      md->capture_last = (md->capture_last & OVFLMASK) | number;
1464      if (offset >= md->offset_max) md->offset_overflow = TRUE; else      if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1465        {        {
1466        md->offset_vector[offset] =        md->offset_vector[offset] =
1467          md->offset_vector[md->offset_end - number];          md->offset_vector[md->offset_end - number];
# Line 1584  for (;;) Line 1523  for (;;)
1523        }        }
1524      else condassert = FALSE;      else condassert = FALSE;
1525    
1526        /* Loop for each branch */
1527    
1528      do      do
1529        {        {
1530        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4);
1531    
1532          /* A match means that the assertion is true; break out of the loop
1533          that matches its alternatives. */
1534    
1535        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1536          {          {
1537          mstart = md->start_match_ptr;   /* In case \K reset it */          mstart = md->start_match_ptr;   /* In case \K reset it */
1538          break;          break;
1539          }          }
1540    
1541          /* If not matched, restore the previous mark setting. */
1542    
1543        md->mark = save_mark;        md->mark = save_mark;
1544    
1545        /* A COMMIT failure must fail the entire assertion, without trying any        /* See comment in the code for capturing groups above about handling
1546        subsequent branches. */        THEN. */
1547    
1548        if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH);        if (rrc == MATCH_THEN)
1549            {
1550            next = ecode + GET(ecode,1);
1551            if (md->start_match_ptr < next &&
1552                (*ecode == OP_ALT || *next == OP_ALT))
1553              rrc = MATCH_NOMATCH;
1554            }
1555    
1556        /* PCRE does not allow THEN to escape beyond an assertion; it        /* Anything other than NOMATCH causes the entire assertion to fail,
1557        is treated as NOMATCH. */        passing back the return code. This includes COMMIT, SKIP, PRUNE and an
1558          uncaptured THEN, which means they take their normal effect. This
1559          consistent approach does not always have exactly the same effect as in
1560          Perl. */
1561    
1562        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);        if (rrc != MATCH_NOMATCH) RRETURN(rrc);
1563        ecode += GET(ecode, 1);        ecode += GET(ecode, 1);
1564        }        }
1565      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);   /* Continue for next alternative */
1566    
1567        /* If we have tried all the alternative branches, the assertion has
1568        failed. If not, we broke out after a match. */
1569    
1570      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);      if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);
1571    
# Line 1613  for (;;) Line 1573  for (;;)
1573    
1574      if (condassert) RRETURN(MATCH_MATCH);      if (condassert) RRETURN(MATCH_MATCH);
1575    
1576      /* Continue from after the assertion, updating the offsets high water      /* Continue from after a successful assertion, updating the offsets high
1577      mark, since extracts may have been taken during the assertion. */      water mark, since extracts may have been taken during the assertion. */
1578    
1579      do ecode += GET(ecode,1); while (*ecode == OP_ALT);      do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1580      ecode += 1 + LINK_SIZE;      ecode += 1 + LINK_SIZE;
1581      offset_top = md->end_offset_top;      offset_top = md->end_offset_top;
1582      continue;      continue;
1583    
1584      /* Negative assertion: all branches must fail to match. Encountering SKIP,      /* Negative assertion: all branches must fail to match for the assertion to
1585      PRUNE, or COMMIT means we must assume failure without checking subsequent      succeed. */
     branches. */  
1586    
1587      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1588      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
# Line 1635  for (;;) Line 1594  for (;;)
1594        }        }
1595      else condassert = FALSE;      else condassert = FALSE;
1596    
1597        /* Loop for each alternative branch. */
1598    
1599      do      do
1600        {        {
1601        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);        RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5);
1602        md->mark = save_mark;        md->mark = save_mark;   /* Always restore the mark setting */
1603        if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH);  
1604        if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT)        switch(rrc)
1605          {          {
1606          do ecode += GET(ecode,1); while (*ecode == OP_ALT);          case MATCH_MATCH:            /* A successful match means */
1607            case MATCH_ACCEPT:           /* the assertion has failed. */
1608            RRETURN(MATCH_NOMATCH);
1609    
1610            case MATCH_NOMATCH:          /* Carry on with next branch */
1611          break;          break;
1612    
1613            /* See comment in the code for capturing groups above about handling
1614            THEN. */
1615    
1616            case MATCH_THEN:
1617            next = ecode + GET(ecode,1);
1618            if (md->start_match_ptr < next &&
1619                (*ecode == OP_ALT || *next == OP_ALT))
1620              {
1621              rrc = MATCH_NOMATCH;
1622              break;
1623              }
1624            /* Otherwise fall through. */
1625    
1626            /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole
1627            assertion to fail to match, without considering any more alternatives.
1628            Failing to match means the assertion is true. This is a consistent
1629            approach, but does not always have the same effect as in Perl. */
1630    
1631            case MATCH_COMMIT:
1632            case MATCH_SKIP:
1633            case MATCH_SKIP_ARG:
1634            case MATCH_PRUNE:
1635            do ecode += GET(ecode,1); while (*ecode == OP_ALT);
1636            goto NEG_ASSERT_TRUE;   /* Break out of alternation loop */
1637    
1638            /* Anything else is an error */
1639    
1640            default:
1641            RRETURN(rrc);
1642          }          }
1643    
1644        /* PCRE does not allow THEN to escape beyond an assertion; it is treated        /* Continue with next branch */
       as NOMATCH. */  
1645    
       if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);  
1646        ecode += GET(ecode,1);        ecode += GET(ecode,1);
1647        }        }
1648      while (*ecode == OP_ALT);      while (*ecode == OP_ALT);
1649    
1650      if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */      /* All branches in the assertion failed to match. */
1651    
1652      ecode += 1 + LINK_SIZE;      NEG_ASSERT_TRUE:
1653        if (condassert) RRETURN(MATCH_MATCH);  /* Condition assertion */
1654        ecode += 1 + LINK_SIZE;                /* Continue with current branch */
1655      continue;      continue;
1656    
1657      /* 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 1716  for (;;) Line 1711  for (;;)
1711        cb.pattern_position = GET(ecode, 2);        cb.pattern_position = GET(ecode, 2);
1712        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);        cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
1713        cb.capture_top      = offset_top/2;        cb.capture_top      = offset_top/2;
1714        cb.capture_last     = md->capture_last;        cb.capture_last     = md->capture_last & CAPLMASK;
1715          /* Internal change requires this for API compatibility. */
1716          if (cb.capture_last == 0) cb.capture_last = -1;
1717        cb.callout_data     = md->callout_data;        cb.callout_data     = md->callout_data;
1718        cb.mark             = md->nomatch_mark;        cb.mark             = md->nomatch_mark;
1719        if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);        if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH);
# Line 1762  for (;;) Line 1759  for (;;)
1759        /* Add to "recursing stack" */        /* Add to "recursing stack" */
1760    
1761        new_recursive.group_num = recno;        new_recursive.group_num = recno;
1762          new_recursive.saved_capture_last = md->capture_last;
1763        new_recursive.subject_position = eptr;        new_recursive.subject_position = eptr;
1764        new_recursive.prevrec = md->recursive;        new_recursive.prevrec = md->recursive;
1765        md->recursive = &new_recursive;        md->recursive = &new_recursive;
# Line 1785  for (;;) Line 1783  for (;;)
1783              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1784    
1785        /* OK, now we can do the recursion. After processing each alternative,        /* OK, now we can do the recursion. After processing each alternative,
1786        restore the offset data. If there were nested recursions, md->recursive        restore the offset data and the last captured value. If there were nested
1787        might be changed, so reset it before looping. */        recursions, md->recursive might be changed, so reset it before looping.
1788          */
1789    
1790        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));        DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
1791        cbegroup = (*callpat >= OP_SBRA);        cbegroup = (*callpat >= OP_SBRA);
# Line 1797  for (;;) Line 1796  for (;;)
1796            md, eptrb, RM6);            md, eptrb, RM6);
1797          memcpy(md->offset_vector, new_recursive.offset_save,          memcpy(md->offset_vector, new_recursive.offset_save,
1798              new_recursive.saved_max * sizeof(int));              new_recursive.saved_max * sizeof(int));
1799            md->capture_last = new_recursive.saved_capture_last;
1800          md->recursive = new_recursive.prevrec;          md->recursive = new_recursive.prevrec;
1801          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)          if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT)
1802            {            {
# Line 1813  for (;;) Line 1813  for (;;)
1813            goto RECURSION_MATCHED;        /* Exit loop; end processing */            goto RECURSION_MATCHED;        /* Exit loop; end processing */
1814            }            }
1815    
1816          /* 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
1817          is treated as NOMATCH. */          recursion; they cause a NOMATCH for the entire recursion. These codes
1818            are defined in a range that can be tested for. */
1819    
1820            if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX)
1821              RRETURN(MATCH_NOMATCH);
1822    
1823          else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN &&          /* Any return code other than NOMATCH is an error. */
1824                   rrc != MATCH_COMMIT)  
1825            if (rrc != MATCH_NOMATCH)
1826            {            {
1827            DPRINTF(("Recursion gave error %d\n", rrc));            DPRINTF(("Recursion gave error %d\n", rrc));
1828            if (new_recursive.offset_save != stacksave)            if (new_recursive.offset_save != stacksave)
# Line 1947  for (;;) Line 1952  for (;;)
1952    
1953        /* Deal with capturing */        /* Deal with capturing */
1954    
1955        md->capture_last = number;        md->capture_last = (md->capture_last & OVFLMASK) | number;
1956        if (offset >= md->offset_max) md->offset_overflow = TRUE; else        if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else
1957          {          {
1958          /* If offset is greater than offset_top, it means that we are          /* If offset is greater than offset_top, it means that we are
1959          "skipping" a capturing group, and that group's offsets must be marked          "skipping" a capturing group, and that group's offsets must be marked
# Line 2004  for (;;) Line 2009  for (;;)
2009    
2010      if (*ecode == OP_KETRPOS)      if (*ecode == OP_KETRPOS)
2011        {        {
2012          md->start_match_ptr = mstart;    /* In case \K reset it */
2013        md->end_match_ptr = eptr;        md->end_match_ptr = eptr;
2014        md->end_offset_top = offset_top;        md->end_offset_top = offset_top;
2015        RRETURN(MATCH_KETRPOS);        RRETURN(MATCH_KETRPOS);
# Line 2571  for (;;) Line 2577  for (;;)
2577            RRETURN(MATCH_NOMATCH);            RRETURN(MATCH_NOMATCH);
2578          break;          break;
2579    
2580          case PT_SPACE:    /* Perl space */          /* Perl space used to exclude VT, but from Perl 5.18 it is included,
2581          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          which means that Perl space and POSIX space are now identical. PCRE
2582               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;  
2583    
2584            case PT_SPACE:    /* Perl space */
2585          case PT_PXSPACE:  /* POSIX space */          case PT_PXSPACE:  /* POSIX space */
2586          if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||          switch(c)
2587               c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            {
2588               c == CHAR_FF || c == CHAR_CR)            HSPACE_CASES:
2589                 == (op == OP_NOTPROP))            VSPACE_CASES:
2590            RRETURN(MATCH_NOMATCH);            if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
2591              break;
2592    
2593              default:
2594              if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) ==
2595                (op == OP_NOTPROP)) RRETURN(MATCH_NOMATCH);
2596              break;
2597              }
2598          break;          break;
2599    
2600          case PT_WORD:          case PT_WORD:
# Line 2604  for (;;) Line 2615  for (;;)
2615            }            }
2616          break;          break;
2617    
2618            case PT_UCNC:
2619            if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
2620                 c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
2621                 c >= 0xe000) == (op == OP_NOTPROP))
2622              RRETURN(MATCH_NOMATCH);
2623            break;
2624    
2625          /* This should never occur */          /* This should never occur */
2626    
2627          default:          default:
# Line 2650  for (;;) Line 2668  for (;;)
2668      similar code to character type repeats - written out again for speed.      similar code to character type repeats - written out again for speed.
2669      However, if the referenced string is the empty string, always treat      However, if the referenced string is the empty string, always treat
2670      it as matched, any number of times (otherwise there could be infinite      it as matched, any number of times (otherwise there could be infinite
2671      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:  
2672    
2673      (a) In the default, Perl-compatible state, set the length negative;      (a) In the default, Perl-compatible state, set the length negative;
2674      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 2668  for (;;) Line 2678  for (;;)
2678      so that the back reference matches an empty string.      so that the back reference matches an empty string.
2679    
2680      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
2681      referenced subpattern. */      referenced subpattern.
2682    
2683        The OP_REF and OP_REFI opcodes are used for a reference to a numbered group
2684        or to a non-duplicated named group. For a duplicated named group, OP_DNREF
2685        and OP_DNREFI are used. In this case we must scan the list of groups to
2686        which the name refers, and use the first one that is set. */
2687    
2688        case OP_DNREF:
2689        case OP_DNREFI:
2690        caseless = op == OP_DNREFI;
2691          {
2692          int count = GET2(ecode, 1+IMM2_SIZE);
2693          pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size;
2694          ecode += 1 + 2*IMM2_SIZE;
2695    
2696          while (count-- > 0)
2697            {
2698            offset = GET2(slot, 0) << 1;
2699            if (offset < offset_top && md->offset_vector[offset] >= 0) break;
2700            slot += md->name_entry_size;
2701            }
2702          if (count < 0)
2703            length = (md->jscript_compat)? 0 : -1;
2704          else
2705            length = md->offset_vector[offset+1] - md->offset_vector[offset];
2706          }
2707        goto REF_REPEAT;
2708    
2709        case OP_REF:
2710        case OP_REFI:
2711        caseless = op == OP_REFI;
2712        offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
2713        ecode += 1 + IMM2_SIZE;
2714      if (offset >= offset_top || md->offset_vector[offset] < 0)      if (offset >= offset_top || md->offset_vector[offset] < 0)
2715        length = (md->jscript_compat)? 0 : -1;        length = (md->jscript_compat)? 0 : -1;
2716      else      else
# Line 2677  for (;;) Line 2718  for (;;)
2718    
2719      /* Set up for repetition, or handle the non-repeated case */      /* Set up for repetition, or handle the non-repeated case */
2720    
2721        REF_REPEAT:
2722      switch (*ecode)      switch (*ecode)
2723        {        {
2724        case OP_CRSTAR:        case OP_CRSTAR:
# Line 2825  for (;;) Line 2867  for (;;)
2867          case OP_CRMINPLUS:          case OP_CRMINPLUS:
2868          case OP_CRQUERY:          case OP_CRQUERY:
2869          case OP_CRMINQUERY:          case OP_CRMINQUERY:
2870            case OP_CRPOSSTAR:
2871            case OP_CRPOSPLUS:
2872            case OP_CRPOSQUERY:
2873          c = *ecode++ - OP_CRSTAR;          c = *ecode++ - OP_CRSTAR;
2874          minimize = (c & 1) != 0;          if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0;
2875            else possessive = TRUE;
2876          min = rep_min[c];                 /* Pick up values from tables; */          min = rep_min[c];                 /* Pick up values from tables; */
2877          max = rep_max[c];                 /* zero for max => infinity */          max = rep_max[c];                 /* zero for max => infinity */
2878          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 2834  for (;;) Line 2880  for (;;)
2880    
2881          case OP_CRRANGE:          case OP_CRRANGE:
2882          case OP_CRMINRANGE:          case OP_CRMINRANGE:
2883            case OP_CRPOSRANGE:
2884          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
2885            possessive = (*ecode == OP_CRPOSRANGE);
2886          min = GET2(ecode, 1);          min = GET2(ecode, 1);
2887          max = GET2(ecode, 1 + IMM2_SIZE);          max = GET2(ecode, 1 + IMM2_SIZE);
2888          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 2976  for (;;) Line 3024  for (;;)
3024                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3025              eptr += len;              eptr += len;
3026              }              }
3027    
3028              if (possessive) continue;    /* No backtracking */
3029    
3030            for (;;)            for (;;)
3031              {              {
3032              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM18);
# Line 3006  for (;;) Line 3057  for (;;)
3057                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;                if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break;
3058              eptr++;              eptr++;
3059              }              }
3060    
3061              if (possessive) continue;    /* No backtracking */
3062    
3063            while (eptr >= pp)            while (eptr >= pp)
3064              {              {
3065              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM19);
# Line 3039  for (;;) Line 3093  for (;;)
3093          case OP_CRMINPLUS:          case OP_CRMINPLUS:
3094          case OP_CRQUERY:          case OP_CRQUERY:
3095          case OP_CRMINQUERY:          case OP_CRMINQUERY:
3096            case OP_CRPOSSTAR:
3097            case OP_CRPOSPLUS:
3098            case OP_CRPOSQUERY:
3099          c = *ecode++ - OP_CRSTAR;          c = *ecode++ - OP_CRSTAR;
3100          minimize = (c & 1) != 0;          if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0;
3101            else possessive = TRUE;
3102          min = rep_min[c];                 /* Pick up values from tables; */          min = rep_min[c];                 /* Pick up values from tables; */
3103          max = rep_max[c];                 /* zero for max => infinity */          max = rep_max[c];                 /* zero for max => infinity */
3104          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 3048  for (;;) Line 3106  for (;;)
3106    
3107          case OP_CRRANGE:          case OP_CRRANGE:
3108          case OP_CRMINRANGE:          case OP_CRMINRANGE:
3109            case OP_CRPOSRANGE:
3110          minimize = (*ecode == OP_CRMINRANGE);          minimize = (*ecode == OP_CRMINRANGE);
3111            possessive = (*ecode == OP_CRPOSRANGE);
3112          min = GET2(ecode, 1);          min = GET2(ecode, 1);
3113          max = GET2(ecode, 1 + IMM2_SIZE);          max = GET2(ecode, 1 + IMM2_SIZE);
3114          if (max == 0) max = INT_MAX;          if (max == 0) max = INT_MAX;
# Line 3120  for (;;) Line 3180  for (;;)
3180            if (!PRIV(xclass)(c, data, utf)) break;            if (!PRIV(xclass)(c, data, utf)) break;
3181            eptr += len;            eptr += len;
3182            }            }
3183    
3184            if (possessive) continue;    /* No backtracking */
3185    
3186          for(;;)          for(;;)
3187            {            {
3188            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM21);
# Line 3190  for (;;) Line 3253  for (;;)
3253    
3254        if (fc < 128)        if (fc < 128)
3255          {          {
3256          pcre_uchar cc = RAWUCHAR(eptr);          pcre_uint32 cc = RAWUCHAR(eptr);
3257          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);
3258          ecode++;          ecode++;
3259          eptr++;          eptr++;
# Line 3295  for (;;) Line 3358  for (;;)
3358      max = rep_max[c];                 /* zero for max => infinity */      max = rep_max[c];                 /* zero for max => infinity */
3359      if (max == 0) max = INT_MAX;      if (max == 0) max = INT_MAX;
3360    
3361      /* Common code for all repeated single-character matches. */      /* Common code for all repeated single-character matches. We first check
3362        for the minimum number of characters. If the minimum equals the maximum, we
3363        are done. Otherwise, if minimizing, check the rest of the pattern for a
3364        match; if there isn't one, advance up to the maximum, one character at a
3365        time.
3366    
3367        If maximizing, advance up to the maximum number of matching characters,
3368        until eptr is past the end of the maximum run. If possessive, we are
3369        then done (no backing up). Otherwise, match at this position; anything
3370        other than no match is immediately returned. For nomatch, back up one
3371        character, unless we are matching \R and the last thing matched was
3372        \r\n, in which case, back up two bytes. When we reach the first optional
3373        character position, we can save stack by doing a tail recurse.
3374    
3375        The various UTF/non-UTF and caseful/caseless cases are handled separately,
3376        for speed. */
3377    
3378      REPEATCHAR:      REPEATCHAR:
3379  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
# Line 3379  for (;;) Line 3457  for (;;)
3457                }                }
3458              }              }
3459    
3460            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
   
3461            for(;;)            for(;;)
3462              {              {
3463                if (eptr == pp) goto TAIL_RECURSE;
3464              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM23);
3465              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
             if (eptr == pp) { RRETURN(MATCH_NOMATCH); }  
3466  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3467              eptr--;              eptr--;
3468              BACKCHAR(eptr);              BACKCHAR(eptr);
# Line 3484  for (;;) Line 3561  for (;;)
3561            if (fc != cc && foc != cc) break;            if (fc != cc && foc != cc) break;
3562            eptr++;            eptr++;
3563            }            }
3564            if (possessive) continue;       /* No backtracking */
3565          if (possessive) continue;          for (;;)
   
         while (eptr >= pp)  
3566            {            {
3567              if (eptr == pp) goto TAIL_RECURSE;
3568            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM25);
3569            eptr--;            eptr--;
3570            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3571            }            }
3572          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3573          }          }
       /* Control never gets here */  
3574        }        }
3575    
3576      /* Caseful comparisons (includes all multi-byte characters) */      /* Caseful comparisons (includes all multi-byte characters) */
# Line 3543  for (;;) Line 3618  for (;;)
3618            if (fc != RAWUCHARTEST(eptr)) break;            if (fc != RAWUCHARTEST(eptr)) break;
3619            eptr++;            eptr++;
3620            }            }
3621          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
3622            for (;;)
         while (eptr >= pp)  
3623            {            {
3624              if (eptr == pp) goto TAIL_RECURSE;
3625            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM27);
3626            eptr--;            eptr--;
3627            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3628            }            }
3629          RRETURN(MATCH_NOMATCH);          /* Control never gets here */
3630          }          }
3631        }        }
3632      /* Control never gets here */      /* Control never gets here */
# Line 3723  for (;;) Line 3798  for (;;)
3798            }            }
3799          }          }
3800        else        else
3801  #endif  #endif  /* SUPPORT_UTF */
3802        /* Not UTF mode */        /* Not UTF mode */
3803          {          {
3804          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
# Line 3761  for (;;) Line 3836  for (;;)
3836              }              }
3837            }            }
3838          else          else
3839  #endif  #endif  /*SUPPORT_UTF */
3840          /* Not UTF mode */          /* Not UTF mode */
3841            {            {
3842            for (fi = min;; fi++)            for (fi = min;; fi++)
# Line 3803  for (;;) Line 3878  for (;;)
3878              if (fc == d || (unsigned int)foc == d) break;              if (fc == d || (unsigned int)foc == d) break;
3879              eptr += len;              eptr += len;
3880              }              }
3881            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3882            for(;;)            for(;;)
3883              {              {
3884                if (eptr == pp) goto TAIL_RECURSE;
3885              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM30);
3886              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3887              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
3888              BACKCHAR(eptr);              BACKCHAR(eptr);
3889              }              }
3890            }            }
3891          else          else
3892  #endif  #endif  /* SUPPORT_UTF */
3893          /* Not UTF mode */          /* Not UTF mode */
3894            {            {
3895            for (i = min; i < max; i++)            for (i = min; i < max; i++)
# Line 3826  for (;;) Line 3902  for (;;)
3902              if (fc == *eptr || foc == *eptr) break;              if (fc == *eptr || foc == *eptr) break;
3903              eptr++;              eptr++;
3904              }              }
3905            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
3906            while (eptr >= pp)            for (;;)
3907              {              {
3908                if (eptr == pp) goto TAIL_RECURSE;
3909              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM31);
3910              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
3911              eptr--;              eptr--;
3912              }              }
3913            }            }
3914            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
3915          }          }
       /* Control never gets here */  
3916        }        }
3917    
3918      /* Caseful comparisons */      /* Caseful comparisons */
# Line 3938  for (;;) Line 4013  for (;;)
4013              if (fc == d) break;              if (fc == d) break;
4014              eptr += len;              eptr += len;
4015              }              }
4016            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
4017            for(;;)            for(;;)
4018              {              {
4019                if (eptr == pp) goto TAIL_RECURSE;
4020              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM34);
4021              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4022              if (eptr-- == pp) break;        /* Stop if tried at original pos */              eptr--;
4023              BACKCHAR(eptr);              BACKCHAR(eptr);
4024              }              }
4025            }            }
# Line 3961  for (;;) Line 4037  for (;;)
4037              if (fc == *eptr) break;              if (fc == *eptr) break;
4038              eptr++;              eptr++;
4039              }              }
4040            if (possessive) continue;            if (possessive) continue;    /* No backtracking */
4041            while (eptr >= pp)            for (;;)
4042              {              {
4043                if (eptr == pp) goto TAIL_RECURSE;
4044              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);              RMATCH(eptr, ecode, offset_top, md, eptrb, RM35);
4045              if (rrc != MATCH_NOMATCH) RRETURN(rrc);              if (rrc != MATCH_NOMATCH) RRETURN(rrc);
4046              eptr--;              eptr--;
4047              }              }
4048            }            }
4049            /* Control never gets here */
         RRETURN(MATCH_NOMATCH);  
4050          }          }
4051        }        }
4052      /* Control never gets here */      /* Control never gets here */
# Line 4152  for (;;) Line 4228  for (;;)
4228              }              }
4229            break;            break;
4230    
4231              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4232              which means that Perl space and POSIX space are now identical. PCRE
4233              was changed at release 8.34. */
4234    
4235            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
4236              case PT_PXSPACE:  /* POSIX space */
4237            for (i = 1; i <= min; i++)            for (i = 1; i <= min; i++)
4238              {              {
4239              if (eptr >= md->end_subject)              if (eptr >= md->end_subject)
# Line 4161  for (;;) Line 4242  for (;;)
4242                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4243                }                }
4244              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4245              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)  
4246                {                {
4247                SCHECK_PARTIAL();                HSPACE_CASES:
4248                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
4249                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4250                  break;
4251    
4252                  default:
4253                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
4254                    RRETURN(MATCH_NOMATCH);
4255                  break;
4256                }                }
             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);  
4257              }              }
4258            break;            break;
4259    
# Line 4222  for (;;) Line 4295  for (;;)
4295              }              }
4296            break;            break;
4297    
4298              case PT_UCNC:
4299              for (i = 1; i <= min; i++)
4300                {
4301                if (eptr >= md->end_subject)
4302                  {
4303                  SCHECK_PARTIAL();
4304                  RRETURN(MATCH_NOMATCH);
4305                  }
4306                GETCHARINCTEST(c, eptr);
4307                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
4308                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
4309                     c >= 0xe000) == prop_fail_result)
4310                  RRETURN(MATCH_NOMATCH);
4311                }
4312              break;
4313    
4314            /* This should not occur */            /* This should not occur */
4315    
4316            default:            default:
# Line 4427  for (;;) Line 4516  for (;;)
4516          case OP_DIGIT:          case OP_DIGIT:
4517          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4518            {            {
4519            pcre_uchar cc;            pcre_uint32 cc;
   
4520            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4521              {              {
4522              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4445  for (;;) Line 4533  for (;;)
4533          case OP_NOT_WHITESPACE:          case OP_NOT_WHITESPACE:
4534          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4535            {            {
4536            pcre_uchar cc;            pcre_uint32 cc;
   
4537            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4538              {              {
4539              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4463  for (;;) Line 4550  for (;;)
4550          case OP_WHITESPACE:          case OP_WHITESPACE:
4551          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4552            {            {
4553            pcre_uchar cc;            pcre_uint32 cc;
   
4554            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4555              {              {
4556              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4481  for (;;) Line 4567  for (;;)
4567          case OP_NOT_WORDCHAR:          case OP_NOT_WORDCHAR:
4568          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4569            {            {
4570            pcre_uchar cc;            pcre_uint32 cc;
   
4571            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4572              {              {
4573              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4499  for (;;) Line 4584  for (;;)
4584          case OP_WORDCHAR:          case OP_WORDCHAR:
4585          for (i = 1; i <= min; i++)          for (i = 1; i <= min; i++)
4586            {            {
4587            pcre_uchar cc;            pcre_uint32 cc;
   
4588            if (eptr >= md->end_subject)            if (eptr >= md->end_subject)
4589              {              {
4590              SCHECK_PARTIAL();              SCHECK_PARTIAL();
# Line 4889  for (;;) Line 4973  for (;;)
4973              }              }
4974            /* Control never gets here */            /* Control never gets here */
4975    
4976            case PT_SPACE:    /* Perl space */            /* Perl space used to exclude VT, but from Perl 5.18 it is included,
4977            for (fi = min;; fi++)            which means that Perl space and POSIX space are now identical. PCRE
4978              {            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 */  
4979    
4980              case PT_SPACE:    /* Perl space */
4981            case PT_PXSPACE:  /* POSIX space */            case PT_PXSPACE:  /* POSIX space */
4982            for (fi = min;; fi++)            for (fi = min;; fi++)
4983              {              {
# Line 4920  for (;;) Line 4990  for (;;)
4990                RRETURN(MATCH_NOMATCH);                RRETURN(MATCH_NOMATCH);
4991                }                }
4992              GETCHARINCTEST(c, eptr);              GETCHARINCTEST(c, eptr);
4993              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
4994                   c == CHAR_VT || c == CHAR_FF || c == CHAR_CR)                {
4995                     == prop_fail_result)                HSPACE_CASES:
4996                RRETURN(MATCH_NOMATCH);                VSPACE_CASES:
4997                  if (prop_fail_result) RRETURN(MATCH_NOMATCH);
4998                  break;
4999    
5000                  default:
5001                  if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
5002                    RRETURN(MATCH_NOMATCH);
5003                  break;
5004                  }
5005              }              }
5006            /* Control never gets here */            /* Control never gets here */
5007    
# Line 4973  for (;;) Line 5051  for (;;)
5051              }              }
5052            /* Control never gets here */            /* Control never gets here */
5053    
5054              case PT_UCNC:
5055              for (fi = min;; fi++)
5056                {
5057                RMATCH(eptr, ecode, offset_top, md, eptrb, RM68);
5058                if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5059                if (fi >= max) RRETURN(MATCH_NOMATCH);
5060                if (eptr >= md->end_subject)
5061                  {
5062                  SCHECK_PARTIAL();
5063                  RRETURN(MATCH_NOMATCH);
5064                  }
5065                GETCHARINCTEST(c, eptr);
5066                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
5067                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
5068                     c >= 0xe000) == prop_fail_result)
5069                  RRETURN(MATCH_NOMATCH);
5070                }
5071              /* Control never gets here */
5072    
5073            /* This should never occur */            /* This should never occur */
5074            default:            default:
5075            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
# Line 5388  for (;;) Line 5485  for (;;)
5485              }              }
5486            break;            break;
5487    
5488              /* Perl space used to exclude VT, but from Perl 5.18 it is included,
5489              which means that Perl space and POSIX space are now identical. PCRE
5490              was changed at release 8.34. */
5491    
5492            case PT_SPACE:    /* Perl space */            case PT_SPACE:    /* Perl space */
5493              case PT_PXSPACE:  /* POSIX space */
5494            for (i = min; i < max; i++)            for (i = min; i < max; i++)
5495              {              {
5496              int len = 1;              int len = 1;
# Line 5398  for (;;) Line 5500  for (;;)
5500                break;                break;
5501                }                }
5502              GETCHARLENTEST(c, eptr, len);              GETCHARLENTEST(c, eptr, len);
5503              if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL ||              switch(c)
5504                   c == CHAR_FF || c == CHAR_CR)                {
5505                   == prop_fail_result)                HSPACE_CASES:
5506                  VSPACE_CASES:
5507                  if (prop_fail_result) goto ENDLOOP99;  /* Break the loop */
5508                break;                break;
             eptr+= len;  
             }  
           break;  
5509    
5510            case PT_PXSPACE:  /* POSIX space */                default:
5511            for (i = min; i < max; i++)                if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result)
5512              {                  goto ENDLOOP99;   /* Break the loop */
             int len = 1;  
             if (eptr >= md->end_subject)  
               {  
               SCHECK_PARTIAL();  
5513                break;                break;
5514                }                }
             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;  
5515              eptr+= len;              eptr+= len;
5516              }              }
5517              ENDLOOP99:
5518            break;            break;
5519    
5520            case PT_WORD:            case PT_WORD:
# Line 5467  for (;;) Line 5560  for (;;)
5560            GOT_MAX:            GOT_MAX:
5561            break;            break;
5562    
5563              case PT_UCNC:
5564              for (i = min; i < max; i++)
5565                {
5566                int len = 1;
5567                if (eptr >= md->end_subject)
5568                  {
5569                  SCHECK_PARTIAL();
5570                  break;
5571                  }
5572                GETCHARLENTEST(c, eptr, len);
5573                if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
5574                     c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
5575                     c >= 0xe000) == prop_fail_result)
5576                  break;
5577                eptr += len;
5578                }
5579              break;
5580    
5581            default:            default:
5582            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5583            }            }
5584    
5585          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5586    
5587          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
5588          for(;;)          for(;;)
5589            {            {
5590              if (eptr == pp) goto TAIL_RECURSE;
5591            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM44);
5592            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5593            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5594            if (utf) BACKCHAR(eptr);            if (utf) BACKCHAR(eptr);
5595            }            }
5596          }          }
5597    
5598        /* Match extended Unicode sequences. We will get here only if the        /* Match extended Unicode grapheme clusters. We will get here only if the
5599        support is in the binary; otherwise a compile-time error occurs. */        support is in the binary; otherwise a compile-time error occurs. */
5600    
5601        else if (ctype == OP_EXTUNI)        else if (ctype == OP_EXTUNI)
# Line 5515  for (;;) Line 5627  for (;;)
5627    
5628          /* eptr is now past the end of the maximum run */          /* eptr is now past the end of the maximum run */
5629    
5630          if (possessive) continue;          if (possessive) continue;    /* No backtracking */
5631    
5632          for(;;)          for(;;)
5633            {            {
5634              int lgb, rgb;
5635              PCRE_PUCHAR fptr;
5636    
5637              if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5638            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
5639            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5640            if (eptr-- == pp) break;        /* Stop if tried at original pos */  
5641            for (;;)                        /* Move back over one extended */            /* Backtracking over an extended grapheme cluster involves inspecting
5642              the previous two characters (if present) to see if a break is
5643              permitted between them. */
5644    
5645              eptr--;
5646              if (!utf) c = *eptr; else
5647              {              {
5648              if (!utf) c = *eptr; else              BACKCHAR(eptr);
5649                GETCHAR(c, eptr);
5650                }
5651              rgb = UCD_GRAPHBREAK(c);
5652    
5653              for (;;)
5654                {
5655                if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
5656                fptr = eptr - 1;
5657                if (!utf) c = *fptr; else
5658                {                {
5659                BACKCHAR(eptr);                BACKCHAR(fptr);
5660                GETCHAR(c, eptr);                GETCHAR(c, fptr);
5661                }                }
5662              if (UCD_CATEGORY(c) != ucp_M) break;              lgb = UCD_GRAPHBREAK(c);
5663              eptr--;              if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
5664                eptr = fptr;
5665                rgb = lgb;
5666              }              }
5667            }            }
5668          }          }
# Line 5796  for (;;) Line 5928  for (;;)
5928            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
5929            }            }
5930    
5931          /* 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;  
5932          for(;;)          for(;;)
5933            {            {
5934              if (eptr == pp) goto TAIL_RECURSE;
5935            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM46);
5936            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
5937            if (eptr-- == pp) break;        /* Stop if tried at original pos */            eptr--;
5938            BACKCHAR(eptr);            BACKCHAR(eptr);
5939            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&            if (ctype == OP_ANYNL && eptr > pp  && RAWUCHAR(eptr) == CHAR_NL &&
5940                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;                RAWUCHAR(eptr - 1) == CHAR_CR) eptr--;
# Line 6045  for (;;) Line 6172  for (;;)
6172            RRETURN(PCRE_ERROR_INTERNAL);            RRETURN(PCRE_ERROR_INTERNAL);
6173            }            }
6174    
6175          /* eptr is now past the end of the maximum run. If possessive, we are          if (possessive) continue;    /* No backtracking */
6176          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)  
6177            {            {
6178              if (eptr == pp) goto TAIL_RECURSE;
6179            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);            RMATCH(eptr, ecode, offset_top, md, eptrb, RM47);
6180            if (rrc != MATCH_NOMATCH) RRETURN(rrc);            if (rrc != MATCH_NOMATCH) RRETURN(rrc);
6181            eptr--;            eptr--;
# Line 6062  for (;;) Line 6184  for (;;)
6184            }            }
6185          }          }
6186    
6187        /* Get here if we can't make it match with any permitted repetitions */        /* Control never gets here */
   
       RRETURN(MATCH_NOMATCH);  
6188        }        }
     /* Control never gets here */  
6189    
6190      /* 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
6191      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 6108  switch (frame->Xwhere) Line 6227  switch (frame->Xwhere)
6227    LBL(32) LBL(34) LBL(42) LBL(46)    LBL(32) LBL(34) LBL(42) LBL(46)
6228  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
6229    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)    LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45)
6230    LBL(59) LBL(60) LBL(61) LBL(62) LBL(67)    LBL(59) LBL(60) LBL(61) LBL(62) LBL(67) LBL(68)
6231  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
6232  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
6233    default:    default:
# Line 6261  const pcre_uint8 *start_bits = NULL; Line 6380  const pcre_uint8 *start_bits = NULL;
6380  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;  PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset;
6381  PCRE_PUCHAR end_subject;  PCRE_PUCHAR end_subject;
6382  PCRE_PUCHAR start_partial = NULL;  PCRE_PUCHAR start_partial = NULL;
6383    PCRE_PUCHAR match_partial = NULL;
6384  PCRE_PUCHAR req_char_ptr = start_match - 1;  PCRE_PUCHAR req_char_ptr = start_match - 1;
6385    
6386  const pcre_study_data *study;  const pcre_study_data *study;
# Line 6390  md->callout_data = NULL; Line 6510  md->callout_data = NULL;
6510    
6511  tables = re->tables;  tables = re->tables;
6512    
6513    /* The two limit values override the defaults, whatever their value. */
6514    
6515  if (extra_data != NULL)  if (extra_data != NULL)
6516    {    {
6517    register unsigned int flags = extra_data->flags;    register unsigned int flags = extra_data->flags;
# Line 6404  if (extra_data != NULL) Line 6526  if (extra_data != NULL)
6526    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;    if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables;
6527    }    }
6528    
6529    /* Limits in the regex override only if they are smaller. */
6530    
6531    if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit)
6532      md->match_limit = re->limit_match;
6533    
6534    if ((re->flags & PCRE_RLSET) != 0 &&
6535        re->limit_recursion < md->match_limit_recursion)
6536      md->match_limit_recursion = re->limit_recursion;
6537    
6538  /* 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
6539  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
6540  in other programs later. */  in other programs later. */
# Line 6429  end_subject = md->end_subject; Line 6560  end_subject = md->end_subject;
6560  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
6561  md->use_ucp = (re->options & PCRE_UCP) != 0;  md->use_ucp = (re->options & PCRE_UCP) != 0;
6562  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;  md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0;
6563  md->ignore_skip_arg = FALSE;  md->ignore_skip_arg = 0;
6564    
6565  /* Some options are unpacked into BOOL variables in the hope that testing  /* Some options are unpacked into BOOL variables in the hope that testing
6566  them will be faster than individual option bits. */  them will be faster than individual option bits. */
# Line 6539  if (re->top_backref > 0 && re->top_backr Line 6670  if (re->top_backref > 0 && re->top_backr
6670    DPRINTF(("Got memory to hold back references\n"));    DPRINTF(("Got memory to hold back references\n"));
6671    }    }
6672  else md->offset_vector = offsets;  else md->offset_vector = offsets;
   
6673  md->offset_end = ocount;  md->offset_end = ocount;
6674  md->offset_max = (2*ocount)/3;  md->offset_max = (2*ocount)/3;
6675  md->offset_overflow = FALSE;  md->capture_last = 0;
 md->capture_last = -1;  
6676    
6677  /* Reset the working variable associated with each extraction. These should  /* Reset the working variable associated with each extraction. These should
6678  never be used unless previously set, but they get saved and restored, and so we  never be used unless previously set, but they get saved and restored, and so we
# Line 6813  for(;;) Line 6942  for(;;)
6942    md->match_call_count = 0;    md->match_call_count = 0;
6943    md->match_function_type = 0;    md->match_function_type = 0;
6944    md->end_offset_top = 0;    md->end_offset_top = 0;
6945      md->skip_arg_count = 0;
6946    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);
6947    if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr;    if (md->hitend && start_partial == NULL)
6948        {
6949        start_partial = md->start_used_ptr;
6950        match_partial = start_match;
6951        }
6952    
6953    switch(rc)    switch(rc)
6954      {      {
# Line 6827  for(;;) Line 6961  for(;;)
6961    
6962      case MATCH_SKIP_ARG:      case MATCH_SKIP_ARG:
6963      new_start_match = start_match;      new_start_match = start_match;
6964      md->ignore_skip_arg = TRUE;      md->ignore_skip_arg = md->skip_arg_count;
6965      break;      break;
6966    
6967      /* 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
6968      same as the match we have just done, treat it as NOMATCH. */      greater than the match we have just done, treat it as NOMATCH. */
6969    
6970      case MATCH_SKIP:      case MATCH_SKIP:
6971      if (md->start_match_ptr != start_match)      if (md->start_match_ptr > start_match)
6972        {        {
6973        new_start_match = md->start_match_ptr;        new_start_match = md->start_match_ptr;
6974        break;        break;
# Line 6842  for(;;) Line 6976  for(;;)
6976      /* Fall through */      /* Fall through */
6977    
6978      /* NOMATCH and PRUNE advance by one character. THEN at this level acts      /* NOMATCH and PRUNE advance by one character. THEN at this level acts
6979      exactly like PRUNE. Unset the ignore SKIP-with-argument flag. */      exactly like PRUNE. Unset ignore SKIP-with-argument. */
6980    
6981      case MATCH_NOMATCH:      case MATCH_NOMATCH:
6982      case MATCH_PRUNE:      case MATCH_PRUNE:
6983      case MATCH_THEN:      case MATCH_THEN:
6984      md->ignore_skip_arg = FALSE;      md->ignore_skip_arg = 0;
6985      new_start_match = start_match + 1;      new_start_match = start_match + 1;
6986  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
6987      if (utf)      if (utf)
# Line 6940  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 7074  if (rc == MATCH_MATCH || rc == MATCH_ACC
7074          (arg_offset_max - 2) * sizeof(int));          (arg_offset_max - 2) * sizeof(int));
7075        DPRINTF(("Copied offsets from temporary memory\n"));        DPRINTF(("Copied offsets from temporary memory\n"));
7076        }        }
7077      if (md->end_offset_top > arg_offset_max) md->offset_overflow = TRUE;      if (md->end_offset_top > arg_offset_max) md->capture_last |= OVFLBIT;
7078      DPRINTF(("Freeing temporary memory\n"));      DPRINTF(("Freeing temporary memory\n"));
7079      (PUBL(free))(md->offset_vector);      (PUBL(free))(md->offset_vector);
7080      }      }
# Line 6948  if (rc == MATCH_MATCH || rc == MATCH_ACC Line 7082  if (rc == MATCH_MATCH || rc == MATCH_ACC
7082    /* 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
7083    too many to fit into the vector. */    too many to fit into the vector. */
7084    
7085    rc = (md->offset_overflow && md->end_offset_top >= arg_offset_max)?    rc = ((md->capture_last & OVFLBIT) != 0 &&
7086             md->end_offset_top >= arg_offset_max)?
7087      0 : md->end_offset_top/2;      0 : md->end_offset_top/2;
7088    
7089    /* If there is space in the offset vector, set any unused pairs at the end of    /* If there is space in the offset vector, set any unused pairs at the end of
# Line 7013  if (rc != MATCH_NOMATCH && rc != PCRE_ER Line 7148  if (rc != MATCH_NOMATCH && rc != PCRE_ER
7148    
7149  /* Handle partial matches - disable any mark data */  /* Handle partial matches - disable any mark data */
7150    
7151  if (start_partial != NULL)  if (match_partial != NULL)
7152    {    {
7153    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));    DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n"));
7154    md->mark = NULL;    md->mark = NULL;
# Line 7021  if (start_partial != NULL) Line 7156  if (start_partial != NULL)
7156      {      {
7157      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);      offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject);
7158      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);      offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject);
7159        if (offsetcount > 2)
7160          offsets[2] = (int)(match_partial - (PCRE_PUCHAR)subject);
7161      }      }
7162    rc = PCRE_ERROR_PARTIAL;    rc = PCRE_ERROR_PARTIAL;
7163    }    }

Legend:
Removed from v.1238  
changed lines
  Added in v.1397

  ViewVC Help
Powered by ViewVC 1.1.5