/[pcre]/code/tags/pcre-8.37/pcre_jit_compile.c
ViewVC logotype

Diff of /code/tags/pcre-8.37/pcre_jit_compile.c

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

revision 1275 by zherczeg, Sun Mar 10 05:32:10 2013 UTC revision 1440 by zherczeg, Sat Jan 11 21:54:20 2014 UTC
# Line 168  typedef struct jit_arguments { Line 168  typedef struct jit_arguments {
168    pcre_uchar *mark_ptr;    pcre_uchar *mark_ptr;
169    void *callout_data;    void *callout_data;
170    /* Everything else after. */    /* Everything else after. */
171      pcre_uint32 limit_match;
172    int real_offset_count;    int real_offset_count;
173    int offset_count;    int offset_count;
   int call_limit;  
174    pcre_uint8 notbol;    pcre_uint8 notbol;
175    pcre_uint8 noteol;    pcre_uint8 noteol;
176    pcre_uint8 notempty;    pcre_uint8 notempty;
# Line 179  typedef struct jit_arguments { Line 179  typedef struct jit_arguments {
179    
180  typedef struct executable_functions {  typedef struct executable_functions {
181    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];
182      sljit_uw *read_only_data[JIT_NUMBER_OF_COMPILE_MODES];
183      sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];
184    PUBL(jit_callback) callback;    PUBL(jit_callback) callback;
185    void *userdata;    void *userdata;
186    pcre_uint32 top_bracket;    pcre_uint32 top_bracket;
187    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];    pcre_uint32 limit_match;
188  } executable_functions;  } executable_functions;
189    
190  typedef struct jump_list {  typedef struct jump_list {
# Line 196  typedef struct stub_list { Line 198  typedef struct stub_list {
198    struct stub_list *next;    struct stub_list *next;
199  } stub_list;  } stub_list;
200    
201    typedef struct label_addr_list {
202      struct sljit_label *label;
203      sljit_uw *addr;
204      struct label_addr_list *next;
205    } label_addr_list;
206    
207  enum frame_types {  enum frame_types {
208    no_frame = -1,    no_frame = -1,
209    no_stack = -2    no_stack = -2
210  };  };
211    
212  enum control_types {  enum control_types {
213    type_commit = 0,    type_mark = 0,
214    type_prune = 1,    type_then_trap = 1
   type_skip = 2  
215  };  };
216    
217  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);
218    
219  /* The following structure is the key data type for the recursive  /* The following structure is the key data type for the recursive
220  code generator. It is allocated by compile_matchingpath, and contains  code generator. It is allocated by compile_matchingpath, and contains
221  the aguments for compile_backtrackingpath. Must be the first member  the arguments for compile_backtrackingpath. Must be the first member
222  of its descendants. */  of its descendants. */
223  typedef struct backtrack_common {  typedef struct backtrack_common {
224    /* Concatenation stack. */    /* Concatenation stack. */
# Line 227  typedef struct backtrack_common { Line 234  typedef struct backtrack_common {
234  typedef struct assert_backtrack {  typedef struct assert_backtrack {
235    backtrack_common common;    backtrack_common common;
236    jump_list *condfailed;    jump_list *condfailed;
237    /* Less than 0 (-1) if a frame is not needed. */    /* Less than 0 if a frame is not needed. */
238    int framesize;    int framesize;
239    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
240    int private_data_ptr;    int private_data_ptr;
# Line 248  typedef struct bracket_backtrack { Line 255  typedef struct bracket_backtrack {
255      /* Both for OP_COND, OP_SCOND. */      /* Both for OP_COND, OP_SCOND. */
256      jump_list *condfailed;      jump_list *condfailed;
257      assert_backtrack *assert;      assert_backtrack *assert;
258      /* For OP_ONCE. -1 if not needed. */      /* For OP_ONCE. Less than 0 if not needed. */
259      int framesize;      int framesize;
260    } u;    } u;
261    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
# Line 283  typedef struct recurse_entry { Line 290  typedef struct recurse_entry {
290    /* Collects the calls until the function is not created. */    /* Collects the calls until the function is not created. */
291    jump_list *calls;    jump_list *calls;
292    /* Points to the starting opcode. */    /* Points to the starting opcode. */
293    int start;    sljit_sw start;
294  } recurse_entry;  } recurse_entry;
295    
296  typedef struct recurse_backtrack {  typedef struct recurse_backtrack {
# Line 291  typedef struct recurse_backtrack { Line 298  typedef struct recurse_backtrack {
298    BOOL inlined_pattern;    BOOL inlined_pattern;
299  } recurse_backtrack;  } recurse_backtrack;
300    
301  #define MAX_RANGE_SIZE 6  #define OP_THEN_TRAP OP_TABLE_LENGTH
302    
303    typedef struct then_trap_backtrack {
304      backtrack_common common;
305      /* If then_trap is not NULL, this structure contains the real
306      then_trap for the backtracking path. */
307      struct then_trap_backtrack *then_trap;
308      /* Points to the starting opcode. */
309      sljit_sw start;
310      /* Exit point for the then opcodes of this alternative. */
311      jump_list *quit;
312      /* Frame size of the current alternative. */
313      int framesize;
314    } then_trap_backtrack;
315    
316    #define MAX_RANGE_SIZE 4
317    
318  typedef struct compiler_common {  typedef struct compiler_common {
319    /* The sljit ceneric compiler. */    /* The sljit ceneric compiler. */
# Line 299  typedef struct compiler_common { Line 321  typedef struct compiler_common {
321    /* First byte code. */    /* First byte code. */
322    pcre_uchar *start;    pcre_uchar *start;
323    /* Maps private data offset to each opcode. */    /* Maps private data offset to each opcode. */
324    int *private_data_ptrs;    sljit_si *private_data_ptrs;
325      /* This read-only data is available during runtime. */
326      sljit_uw *read_only_data;
327      /* The total size of the read-only data. */
328      sljit_uw read_only_data_size;
329      /* The next free entry of the read_only_data. */
330      sljit_uw *read_only_data_ptr;
331    /* Tells whether the capturing bracket is optimized. */    /* Tells whether the capturing bracket is optimized. */
332    pcre_uint8 *optimized_cbracket;    pcre_uint8 *optimized_cbracket;
333      /* Tells whether the starting offset is a target of then. */
334      pcre_uint8 *then_offsets;
335      /* Current position where a THEN must jump. */
336      then_trap_backtrack *then_trap;
337    /* Starting offset of private data for capturing brackets. */    /* Starting offset of private data for capturing brackets. */
338    int cbra_ptr;    int cbra_ptr;
339    /* Output vector starting point. Must be divisible by 2. */    /* Output vector starting point. Must be divisible by 2. */
# Line 330  typedef struct compiler_common { Line 362  typedef struct compiler_common {
362    sljit_sw lcc;    sljit_sw lcc;
363    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */
364    int mode;    int mode;
365    /* \K is in the pattern. */    /* TRUE, when minlength is greater than 0. */
366      BOOL might_be_empty;
367      /* \K is found in the pattern. */
368    BOOL has_set_som;    BOOL has_set_som;
369      /* (*SKIP:arg) is found in the pattern. */
370      BOOL has_skip_arg;
371      /* (*THEN) is found in the pattern. */
372      BOOL has_then;
373    /* Needs to know the start position anytime. */    /* Needs to know the start position anytime. */
374    BOOL needs_start_ptr;    BOOL needs_start_ptr;
375    /* Currently in recurse or assert. */    /* Currently in recurse or negative assert. */
376    BOOL local_exit;    BOOL local_exit;
377      /* Currently in a positive assert. */
378      BOOL positive_assert;
379    /* Newline control. */    /* Newline control. */
380    int nltype;    int nltype;
381      pcre_uint32 nlmax;
382      pcre_uint32 nlmin;
383    int newline;    int newline;
384    int bsr_nltype;    int bsr_nltype;
385      pcre_uint32 bsr_nlmax;
386      pcre_uint32 bsr_nlmin;
387    /* Dollar endonly. */    /* Dollar endonly. */
388    int endonly;    int endonly;
389    /* Tables. */    /* Tables. */
390    sljit_sw ctypes;    sljit_sw ctypes;
   int digits[2 + MAX_RANGE_SIZE];  
391    /* Named capturing brackets. */    /* Named capturing brackets. */
392    sljit_uw name_table;    pcre_uchar *name_table;
393    sljit_sw name_count;    sljit_sw name_count;
394    sljit_sw name_entry_size;    sljit_sw name_entry_size;
395    
# Line 356  typedef struct compiler_common { Line 399  typedef struct compiler_common {
399    struct sljit_label *forced_quit_label;    struct sljit_label *forced_quit_label;
400    struct sljit_label *accept_label;    struct sljit_label *accept_label;
401    stub_list *stubs;    stub_list *stubs;
402      label_addr_list *label_addrs;
403    recurse_entry *entries;    recurse_entry *entries;
404    recurse_entry *currententry;    recurse_entry *currententry;
405    jump_list *partialmatch;    jump_list *partialmatch;
406    jump_list *quit;    jump_list *quit;
407      jump_list *positive_assert_quit;
408    jump_list *forced_quit;    jump_list *forced_quit;
409    jump_list *accept;    jump_list *accept;
410    jump_list *calllimit;    jump_list *calllimit;
# Line 378  typedef struct compiler_common { Line 423  typedef struct compiler_common {
423  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
424    BOOL use_ucp;    BOOL use_ucp;
425  #endif  #endif
 #ifndef COMPILE_PCRE32  
   jump_list *utfreadchar;  
 #endif  
426  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
427      jump_list *utfreadchar;
428      jump_list *utfreadchar16;
429    jump_list *utfreadtype8;    jump_list *utfreadtype8;
430  #endif  #endif
431  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
# Line 438  typedef struct compare_context { Line 482  typedef struct compare_context {
482  #define STACK_TOP     SLJIT_SCRATCH_REG2  #define STACK_TOP     SLJIT_SCRATCH_REG2
483  #define STACK_LIMIT   SLJIT_SAVED_REG3  #define STACK_LIMIT   SLJIT_SAVED_REG3
484  #define ARGUMENTS     SLJIT_SAVED_EREG1  #define ARGUMENTS     SLJIT_SAVED_EREG1
485  #define CALL_COUNT    SLJIT_SAVED_EREG2  #define COUNT_MATCH   SLJIT_SAVED_EREG2
486  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1
487    
488  /* Local space layout. */  /* Local space layout. */
# Line 449  typedef struct compare_context { Line 493  typedef struct compare_context {
493  #define POSSESSIVE0      (2 * sizeof(sljit_sw))  #define POSSESSIVE0      (2 * sizeof(sljit_sw))
494  #define POSSESSIVE1      (3 * sizeof(sljit_sw))  #define POSSESSIVE1      (3 * sizeof(sljit_sw))
495  /* Max limit of recursions. */  /* Max limit of recursions. */
496  #define CALL_LIMIT       (4 * sizeof(sljit_sw))  #define LIMIT_MATCH      (4 * sizeof(sljit_sw))
497  /* The output vector is stored on the stack, and contains pointers  /* The output vector is stored on the stack, and contains pointers
498  to characters. The vector data is divided into two groups: the first  to characters. The vector data is divided into two groups: the first
499  group contains the start / end character pointers, and the second is  group contains the start / end character pointers, and the second is
500  the start pointers when the end of the capturing group has not yet reached. */  the start pointers when the end of the capturing group has not yet reached. */
501  #define OVECTOR_START    (common->ovector_start)  #define OVECTOR_START    (common->ovector_start)
502  #define OVECTOR(i)       (OVECTOR_START + (i) * sizeof(sljit_sw))  #define OVECTOR(i)       (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw))
503  #define OVECTOR_PRIV(i)  (common->cbra_ptr + (i) * sizeof(sljit_sw))  #define OVECTOR_PRIV(i)  (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw))
504  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])
505    
506  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
# Line 498  the start pointers when the end of the c Line 542  the start pointers when the end of the c
542  #define GET_LOCAL_BASE(dst, dstw, offset) \  #define GET_LOCAL_BASE(dst, dstw, offset) \
543    sljit_get_local_base(compiler, (dst), (dstw), (offset))    sljit_get_local_base(compiler, (dst), (dstw), (offset))
544    
545    #define READ_CHAR_MAX 0x7fffffff
546    
547  static pcre_uchar* bracketend(pcre_uchar* cc)  static pcre_uchar* bracketend(pcre_uchar* cc)
548  {  {
549  SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));  SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
# Line 507  cc += 1 + LINK_SIZE; Line 553  cc += 1 + LINK_SIZE;
553  return cc;  return cc;
554  }  }
555    
556    static int no_alternatives(pcre_uchar* cc)
557    {
558    int count = 0;
559    SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
560    do
561      {
562      cc += GET(cc, 1);
563      count++;
564      }
565    while (*cc == OP_ALT);
566    SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS);
567    return count;
568    }
569    
570    static int ones_in_half_byte[16] = {
571      /* 0 */ 0, 1, 1, 2, /* 4 */ 1, 2, 2, 3,
572      /* 8 */ 1, 2, 2, 3, /* 12 */ 2, 3, 3, 4
573    };
574    
575  /* Functions whose might need modification for all new supported opcodes:  /* Functions whose might need modification for all new supported opcodes:
576   next_opcode   next_opcode
577   get_private_data_length   check_opcode_types
578   set_private_data_ptrs   set_private_data_ptrs
579   get_framesize   get_framesize
580   init_frame   init_frame
581   get_private_data_length_for_copy   get_private_data_copy_length
582   copy_private_data   copy_private_data
583   compile_matchingpath   compile_matchingpath
584   compile_backtrackingpath   compile_backtrackingpath
# Line 559  switch(*cc) Line 624  switch(*cc)
624    case OP_CRMINQUERY:    case OP_CRMINQUERY:
625    case OP_CRRANGE:    case OP_CRRANGE:
626    case OP_CRMINRANGE:    case OP_CRMINRANGE:
627      case OP_CRPOSSTAR:
628      case OP_CRPOSPLUS:
629      case OP_CRPOSQUERY:
630      case OP_CRPOSRANGE:
631    case OP_CLASS:    case OP_CLASS:
632    case OP_NCLASS:    case OP_NCLASS:
633    case OP_REF:    case OP_REF:
634    case OP_REFI:    case OP_REFI:
635      case OP_DNREF:
636      case OP_DNREFI:
637    case OP_RECURSE:    case OP_RECURSE:
638    case OP_CALLOUT:    case OP_CALLOUT:
639    case OP_ALT:    case OP_ALT:
# Line 588  switch(*cc) Line 659  switch(*cc)
659    case OP_SCBRAPOS:    case OP_SCBRAPOS:
660    case OP_SCOND:    case OP_SCOND:
661    case OP_CREF:    case OP_CREF:
662    case OP_NCREF:    case OP_DNCREF:
663    case OP_RREF:    case OP_RREF:
664    case OP_NRREF:    case OP_DNRREF:
665    case OP_DEF:    case OP_DEF:
666    case OP_BRAZERO:    case OP_BRAZERO:
667    case OP_BRAMINZERO:    case OP_BRAMINZERO:
668    case OP_BRAPOSZERO:    case OP_BRAPOSZERO:
669    case OP_PRUNE:    case OP_PRUNE:
670    case OP_SKIP:    case OP_SKIP:
671      case OP_THEN:
672    case OP_COMMIT:    case OP_COMMIT:
673    case OP_FAIL:    case OP_FAIL:
674    case OP_ACCEPT:    case OP_ACCEPT:
# Line 696  switch(*cc) Line 768  switch(*cc)
768    
769    case OP_MARK:    case OP_MARK:
770    case OP_PRUNE_ARG:    case OP_PRUNE_ARG:
771      case OP_SKIP_ARG:
772      case OP_THEN_ARG:
773    return cc + 1 + 2 + cc[1];    return cc + 1 + 2 + cc[1];
774    
775    default:    default:
776      /* All opcodes are supported now! */
777      SLJIT_ASSERT_STOP();
778    return NULL;    return NULL;
779    }    }
780  }  }
781    
782  #define CASE_ITERATOR_PRIVATE_DATA_1 \  static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)
     case OP_MINSTAR: \  
     case OP_MINPLUS: \  
     case OP_QUERY: \  
     case OP_MINQUERY: \  
     case OP_MINSTARI: \  
     case OP_MINPLUSI: \  
     case OP_QUERYI: \  
     case OP_MINQUERYI: \  
     case OP_NOTMINSTAR: \  
     case OP_NOTMINPLUS: \  
     case OP_NOTQUERY: \  
     case OP_NOTMINQUERY: \  
     case OP_NOTMINSTARI: \  
     case OP_NOTMINPLUSI: \  
     case OP_NOTQUERYI: \  
     case OP_NOTMINQUERYI:  
   
 #define CASE_ITERATOR_PRIVATE_DATA_2A \  
     case OP_STAR: \  
     case OP_PLUS: \  
     case OP_STARI: \  
     case OP_PLUSI: \  
     case OP_NOTSTAR: \  
     case OP_NOTPLUS: \  
     case OP_NOTSTARI: \  
     case OP_NOTPLUSI:  
   
 #define CASE_ITERATOR_PRIVATE_DATA_2B \  
     case OP_UPTO: \  
     case OP_MINUPTO: \  
     case OP_UPTOI: \  
     case OP_MINUPTOI: \  
     case OP_NOTUPTO: \  
     case OP_NOTMINUPTO: \  
     case OP_NOTUPTOI: \  
     case OP_NOTMINUPTOI:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \  
     case OP_TYPEMINSTAR: \  
     case OP_TYPEMINPLUS: \  
     case OP_TYPEQUERY: \  
     case OP_TYPEMINQUERY:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \  
     case OP_TYPESTAR: \  
     case OP_TYPEPLUS:  
   
 #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \  
     case OP_TYPEUPTO: \  
     case OP_TYPEMINUPTO:  
   
 static int get_class_iterator_size(pcre_uchar *cc)  
 {  
 switch(*cc)  
   {  
   case OP_CRSTAR:  
   case OP_CRPLUS:  
   return 2;  
   
   case OP_CRMINSTAR:  
   case OP_CRMINPLUS:  
   case OP_CRQUERY:  
   case OP_CRMINQUERY:  
   return 1;  
   
   case OP_CRRANGE:  
   case OP_CRMINRANGE:  
   if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))  
     return 0;  
   return 2;  
   
   default:  
   return 0;  
   }  
 }  
   
 static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  
783  {  {
784  int private_data_length = 0;  int count;
785  pcre_uchar *alternative;  pcre_uchar *slot;
 pcre_uchar *name;  
 pcre_uchar *end = NULL;  
 int space, size, i;  
 pcre_uint32 bracketlen;  
786    
787  /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */  /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */
788  while (cc < ccend)  while (cc < ccend)
789    {    {
   space = 0;  
   size = 0;  
   bracketlen = 0;  
790    switch(*cc)    switch(*cc)
791      {      {
792      case OP_SET_SOM:      case OP_SET_SOM:
793      common->has_set_som = TRUE;      common->has_set_som = TRUE;
794        common->might_be_empty = TRUE;
795      cc += 1;      cc += 1;
796      break;      break;
797    
# Line 808  while (cc < ccend) Line 801  while (cc < ccend)
801      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
802      break;      break;
803    
804      case OP_ASSERT:      case OP_BRA:
805      case OP_ASSERT_NOT:      case OP_CBRA:
     case OP_ASSERTBACK:  
     case OP_ASSERTBACK_NOT:  
     case OP_ONCE:  
     case OP_ONCE_NC:  
     case OP_BRAPOS:  
806      case OP_SBRA:      case OP_SBRA:
807      case OP_SBRAPOS:      case OP_SCBRA:
808      private_data_length += sizeof(sljit_sw);      count = no_alternatives(cc);
809      bracketlen = 1 + LINK_SIZE;      if (count > 4)
810          common->read_only_data_size += count * sizeof(sljit_uw);
811        cc += 1 + LINK_SIZE + (*cc == OP_CBRA || *cc == OP_SCBRA ? IMM2_SIZE : 0);
812      break;      break;
813    
814      case OP_CBRAPOS:      case OP_CBRAPOS:
815      case OP_SCBRAPOS:      case OP_SCBRAPOS:
     private_data_length += sizeof(sljit_sw);  
816      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;
817      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;      cc += 1 + LINK_SIZE + IMM2_SIZE;
818      break;      break;
819    
820      case OP_COND:      case OP_COND:
# Line 833  while (cc < ccend) Line 822  while (cc < ccend)
822      /* Only AUTO_CALLOUT can insert this opcode. We do      /* Only AUTO_CALLOUT can insert this opcode. We do
823         not intend to support this case. */         not intend to support this case. */
824      if (cc[1 + LINK_SIZE] == OP_CALLOUT)      if (cc[1 + LINK_SIZE] == OP_CALLOUT)
825        return -1;        return FALSE;
826        cc += 1 + LINK_SIZE;
     if (*cc == OP_COND)  
       {  
       /* Might be a hidden SCOND. */  
       alternative = cc + GET(cc, 1);  
       if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)  
         private_data_length += sizeof(sljit_sw);  
       }  
     else  
       private_data_length += sizeof(sljit_sw);  
     bracketlen = 1 + LINK_SIZE;  
827      break;      break;
828    
829      case OP_CREF:      case OP_CREF:
830      i = GET2(cc, 1);      common->optimized_cbracket[GET2(cc, 1)] = 0;
     common->optimized_cbracket[i] = 0;  
831      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
832      break;      break;
833    
834      case OP_NCREF:      case OP_DNREF:
835      bracketlen = GET2(cc, 1);      case OP_DNREFI:
836      name = (pcre_uchar *)common->name_table;      case OP_DNCREF:
837      alternative = name;      count = GET2(cc, 1 + IMM2_SIZE);
838      for (i = 0; i < common->name_count; i++)      slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
839        {      while (count-- > 0)
       if (GET2(name, 0) == bracketlen) break;  
       name += common->name_entry_size;  
       }  
     SLJIT_ASSERT(i != common->name_count);  
   
     for (i = 0; i < common->name_count; i++)  
840        {        {
841        if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0)        common->optimized_cbracket[GET2(slot, 0)] = 0;
842          common->optimized_cbracket[GET2(alternative, 0)] = 0;        slot += common->name_entry_size;
       alternative += common->name_entry_size;  
843        }        }
844      bracketlen = 0;      cc += 1 + 2 * IMM2_SIZE;
     cc += 1 + IMM2_SIZE;  
     break;  
   
     case OP_BRA:  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_CBRA:  
     case OP_SCBRA:  
     bracketlen = 1 + LINK_SIZE + IMM2_SIZE;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_1  
     space = 1;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2A  
     space = 2;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2B  
     space = 2;  
     size = -(2 + IMM2_SIZE);  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_1  
     space = 1;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2A  
     if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI)  
       space = 2;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2B  
     if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI)  
       space = 2;  
     size = 1 + IMM2_SIZE;  
     break;  
   
     case OP_CLASS:  
     case OP_NCLASS:  
     size += 1 + 32 / sizeof(pcre_uchar);  
     space = get_class_iterator_size(cc + size);  
845      break;      break;
846    
 #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  
     case OP_XCLASS:  
     size = GET(cc, 1);  
     space = get_class_iterator_size(cc + size);  
     break;  
 #endif  
   
847      case OP_RECURSE:      case OP_RECURSE:
848      /* Set its value only once. */      /* Set its value only once. */
849      if (common->recursive_head_ptr == 0)      if (common->recursive_head_ptr == 0)
# Line 947  while (cc < ccend) Line 863  while (cc < ccend)
863      cc += 2 + 2 * LINK_SIZE;      cc += 2 + 2 * LINK_SIZE;
864      break;      break;
865    
866        case OP_THEN_ARG:
867        common->has_then = TRUE;
868        common->control_head_ptr = 1;
869        /* Fall through. */
870    
871      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
872      common->needs_start_ptr = TRUE;      common->needs_start_ptr = TRUE;
     common->control_head_ptr = 1;  
873      /* Fall through. */      /* Fall through. */
874    
875      case OP_MARK:      case OP_MARK:
# Line 961  while (cc < ccend) Line 881  while (cc < ccend)
881      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
882      break;      break;
883    
884        case OP_THEN:
885        common->has_then = TRUE;
886        common->control_head_ptr = 1;
887        /* Fall through. */
888    
889      case OP_PRUNE:      case OP_PRUNE:
890      case OP_SKIP:      case OP_SKIP:
891      common->needs_start_ptr = TRUE;      common->needs_start_ptr = TRUE;
892      /* Fall through. */      cc += 1;
893        break;
894    
895      case OP_COMMIT:      case OP_SKIP_ARG:
896      common->control_head_ptr = 1;      common->control_head_ptr = 1;
897      cc += 1;      common->has_skip_arg = TRUE;
898        cc += 1 + 2 + cc[1];
899      break;      break;
900    
901      default:      default:
902      cc = next_opcode(common, cc);      cc = next_opcode(common, cc);
903      if (cc == NULL)      if (cc == NULL)
904        return -1;        return FALSE;
905      break;      break;
906      }      }
907      }
908    return TRUE;
909    }
910    
911    if (space > 0 && cc >= end)  static int get_class_iterator_size(pcre_uchar *cc)
912      private_data_length += sizeof(sljit_sw) * space;  {
913    switch(*cc)
914      {
915      case OP_CRSTAR:
916      case OP_CRPLUS:
917      return 2;
918    
919    if (size != 0)    case OP_CRMINSTAR:
920      {    case OP_CRMINPLUS:
921      if (size < 0)    case OP_CRQUERY:
922        {    case OP_CRMINQUERY:
923        cc += -size;    return 1;
 #ifdef SUPPORT_UTF  
       if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
       }  
     else  
       cc += size;  
     }  
924    
925    if (bracketlen != 0)    case OP_CRRANGE:
926      {    case OP_CRMINRANGE:
927      if (cc >= end)    if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))
928        {      return 0;
929        end = bracketend(cc);    return 2;
930        if (end[-1 - LINK_SIZE] == OP_KET)  
931          end = NULL;    default:
932        }    return 0;
     cc += bracketlen;  
     }  
933    }    }
 return private_data_length;  
934  }  }
935    
936  static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend)  static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin)
937  {  {
938  pcre_uchar *cc = common->start;  pcre_uchar *end = bracketend(begin);
939  pcre_uchar *alternative;  pcre_uchar *next;
940  pcre_uchar *end = NULL;  pcre_uchar *next_end;
941  int space, size, bracketlen;  pcre_uchar *max_end;
942    pcre_uchar type;
943    sljit_sw length = end - begin;
944    int min, max, i;
945    
946    /* Detect fixed iterations first. */
947    if (end[-(1 + LINK_SIZE)] != OP_KET)
948      return FALSE;
949    
950    /* Already detected repeat. */
951    if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0)
952      return TRUE;
953    
954    next = end;
955    min = 1;
956    while (1)
957      {
958      if (*next != *begin)
959        break;
960      next_end = bracketend(next);
961      if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0)
962        break;
963      next = next_end;
964      min++;
965      }
966    
967    if (min == 2)
968      return FALSE;
969    
970    max = 0;
971    max_end = next;
972    if (*next == OP_BRAZERO || *next == OP_BRAMINZERO)
973      {
974      type = *next;
975      while (1)
976        {
977        if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin)
978          break;
979        next_end = bracketend(next + 2 + LINK_SIZE);
980        if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0)
981          break;
982        next = next_end;
983        max++;
984        }
985    
986      if (next[0] == type && next[1] == *begin && max >= 1)
987        {
988        next_end = bracketend(next + 1);
989        if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0)
990          {
991          for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE)
992            if (*next_end != OP_KET)
993              break;
994    
995          if (i == max)
996            {
997            common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end;
998            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO;
999            /* +2 the original and the last. */
1000            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2;
1001            if (min == 1)
1002              return TRUE;
1003            min--;
1004            max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE);
1005            }
1006          }
1007        }
1008      }
1009    
1010    if (min >= 3)
1011      {
1012      common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end;
1013      common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT;
1014      common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min;
1015      return TRUE;
1016      }
1017    
1018    return FALSE;
1019    }
1020    
1021    #define CASE_ITERATOR_PRIVATE_DATA_1 \
1022        case OP_MINSTAR: \
1023        case OP_MINPLUS: \
1024        case OP_QUERY: \
1025        case OP_MINQUERY: \
1026        case OP_MINSTARI: \
1027        case OP_MINPLUSI: \
1028        case OP_QUERYI: \
1029        case OP_MINQUERYI: \
1030        case OP_NOTMINSTAR: \
1031        case OP_NOTMINPLUS: \
1032        case OP_NOTQUERY: \
1033        case OP_NOTMINQUERY: \
1034        case OP_NOTMINSTARI: \
1035        case OP_NOTMINPLUSI: \
1036        case OP_NOTQUERYI: \
1037        case OP_NOTMINQUERYI:
1038    
1039    #define CASE_ITERATOR_PRIVATE_DATA_2A \
1040        case OP_STAR: \
1041        case OP_PLUS: \
1042        case OP_STARI: \
1043        case OP_PLUSI: \
1044        case OP_NOTSTAR: \
1045        case OP_NOTPLUS: \
1046        case OP_NOTSTARI: \
1047        case OP_NOTPLUSI:
1048    
1049    #define CASE_ITERATOR_PRIVATE_DATA_2B \
1050        case OP_UPTO: \
1051        case OP_MINUPTO: \
1052        case OP_UPTOI: \
1053        case OP_MINUPTOI: \
1054        case OP_NOTUPTO: \
1055        case OP_NOTMINUPTO: \
1056        case OP_NOTUPTOI: \
1057        case OP_NOTMINUPTOI:
1058    
1059    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \
1060        case OP_TYPEMINSTAR: \
1061        case OP_TYPEMINPLUS: \
1062        case OP_TYPEQUERY: \
1063        case OP_TYPEMINQUERY:
1064    
1065    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \
1066        case OP_TYPESTAR: \
1067        case OP_TYPEPLUS:
1068    
1069    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \
1070        case OP_TYPEUPTO: \
1071        case OP_TYPEMINUPTO:
1072    
1073    static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend)
1074    {
1075    pcre_uchar *cc = common->start;
1076    pcre_uchar *alternative;
1077    pcre_uchar *end = NULL;
1078    int private_data_ptr = *private_data_start;
1079    int space, size, bracketlen;
1080    
1081  while (cc < ccend)  while (cc < ccend)
1082    {    {
1083    space = 0;    space = 0;
1084    size = 0;    size = 0;
1085    bracketlen = 0;    bracketlen = 0;
1086      if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE)
1087        return;
1088    
1089      if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND)
1090        if (detect_repeat(common, cc))
1091          {
1092          /* These brackets are converted to repeats, so no global
1093          based single character repeat is allowed. */
1094          if (cc >= end)
1095            end = bracketend(cc);
1096          }
1097    
1098    switch(*cc)    switch(*cc)
1099      {      {
1100        case OP_KET:
1101        if (common->private_data_ptrs[cc + 1 - common->start] != 0)
1102          {
1103          common->private_data_ptrs[cc - common->start] = private_data_ptr;
1104          private_data_ptr += sizeof(sljit_sw);
1105          cc += common->private_data_ptrs[cc + 1 - common->start];
1106          }
1107        cc += 1 + LINK_SIZE;
1108        break;
1109    
1110      case OP_ASSERT:      case OP_ASSERT:
1111      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1112      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1115  while (cc < ccend) Line 1200  while (cc < ccend)
1200      break;      break;
1201      }      }
1202    
1203      /* Character iterators, which are not inside a repeated bracket,
1204         gets a private slot instead of allocating it on the stack. */
1205    if (space > 0 && cc >= end)    if (space > 0 && cc >= end)
1206      {      {
1207      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
# Line 1145  while (cc < ccend) Line 1232  while (cc < ccend)
1232      cc += bracketlen;      cc += bracketlen;
1233      }      }
1234    }    }
1235    *private_data_start = private_data_ptr;
1236  }  }
1237    
1238  /* Returns with a frame_types (always < 0) if no need for frame. */  /* Returns with a frame_types (always < 0) if no need for frame. */
1239  static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive)  static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL* needs_control_head)
1240  {  {
 pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE);  
1241  int length = 0;  int length = 0;
1242  int possessive = 0;  int possessive = 0;
1243  BOOL stack_restore = FALSE;  BOOL stack_restore = FALSE;
# Line 1159  BOOL setmark_found = recursive; Line 1246  BOOL setmark_found = recursive;
1246  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
1247  BOOL capture_last_found = FALSE;  BOOL capture_last_found = FALSE;
1248    
1249  if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))  #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
1250    SLJIT_ASSERT(common->control_head_ptr != 0);
1251    *needs_control_head = TRUE;
1252    #else
1253    *needs_control_head = FALSE;
1254    #endif
1255    
1256    if (ccend == NULL)
1257    {    {
1258    possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;    ccend = bracketend(cc) - (1 + LINK_SIZE);
1259    /* This is correct regardless of common->capture_last_ptr. */    if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))
1260    capture_last_found = TRUE;      {
1261        possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;
1262        /* This is correct regardless of common->capture_last_ptr. */
1263        capture_last_found = TRUE;
1264        }
1265      cc = next_opcode(common, cc);
1266    }    }
1267    
 cc = next_opcode(common, cc);  
1268  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1269  while (cc < ccend)  while (cc < ccend)
1270    switch(*cc)    switch(*cc)
# Line 1184  while (cc < ccend) Line 1282  while (cc < ccend)
1282    
1283      case OP_MARK:      case OP_MARK:
1284      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1285        case OP_THEN_ARG:
1286      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1287      stack_restore = TRUE;      stack_restore = TRUE;
1288      if (!setmark_found)      if (!setmark_found)
# Line 1191  while (cc < ccend) Line 1290  while (cc < ccend)
1290        length += 2;        length += 2;
1291        setmark_found = TRUE;        setmark_found = TRUE;
1292        }        }
1293        if (common->control_head_ptr != 0)
1294          *needs_control_head = TRUE;
1295      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
1296      break;      break;
1297    
# Line 1310  if (length > 0) Line 1411  if (length > 0)
1411  return stack_restore ? no_frame : no_stack;  return stack_restore ? no_frame : no_stack;
1412  }  }
1413    
1414  static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive)  static void init_frame(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, int stackpos, int stacktop, BOOL recursive)
1415  {  {
1416  DEFINE_COMPILER;  DEFINE_COMPILER;
 pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE);  
1417  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1418  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1419  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
# Line 1325  SLJIT_UNUSED_ARG(stacktop); Line 1425  SLJIT_UNUSED_ARG(stacktop);
1425  SLJIT_ASSERT(stackpos >= stacktop + 2);  SLJIT_ASSERT(stackpos >= stacktop + 2);
1426    
1427  stackpos = STACK(stackpos);  stackpos = STACK(stackpos);
1428  if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))  if (ccend == NULL)
1429    cc = next_opcode(common, cc);    {
1430      ccend = bracketend(cc) - (1 + LINK_SIZE);
1431      if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))
1432        cc = next_opcode(common, cc);
1433      }
1434    
1435  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1436  while (cc < ccend)  while (cc < ccend)
1437    switch(*cc)    switch(*cc)
# Line 1347  while (cc < ccend) Line 1452  while (cc < ccend)
1452    
1453      case OP_MARK:      case OP_MARK:
1454      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1455        case OP_THEN_ARG:
1456      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1457      if (!setmark_found)      if (!setmark_found)
1458        {        {
# Line 1427  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st Line 1533  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st
1533  SLJIT_ASSERT(stackpos == STACK(stacktop));  SLJIT_ASSERT(stackpos == STACK(stacktop));
1534  }  }
1535    
1536  static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  static SLJIT_INLINE int get_private_data_copy_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL needs_control_head)
1537  {  {
1538  int private_data_length = common->control_head_ptr ? 3 : 2;  int private_data_length = needs_control_head ? 3 : 2;
1539  int size;  int size;
1540  pcre_uchar *alternative;  pcre_uchar *alternative;
1541  /* Calculate the sum of the private machine words. */  /* Calculate the sum of the private machine words. */
# Line 1438  while (cc < ccend) Line 1544  while (cc < ccend)
1544    size = 0;    size = 0;
1545    switch(*cc)    switch(*cc)
1546      {      {
1547        case OP_KET:
1548        if (PRIVATE_DATA(cc) != 0)
1549          private_data_length++;
1550        cc += 1 + LINK_SIZE;
1551        break;
1552    
1553      case OP_ASSERT:      case OP_ASSERT:
1554      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1555      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1542  return private_data_length; Line 1654  return private_data_length;
1654  }  }
1655    
1656  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,
1657    BOOL save, int stackptr, int stacktop)    BOOL save, int stackptr, int stacktop, BOOL needs_control_head)
1658  {  {
1659  DEFINE_COMPILER;  DEFINE_COMPILER;
1660  int srcw[2];  int srcw[2];
# Line 1563  stacktop = STACK(stacktop - 1); Line 1675  stacktop = STACK(stacktop - 1);
1675    
1676  if (!save)  if (!save)
1677    {    {
1678    stackptr += (common->control_head_ptr ? 2 : 1) * sizeof(sljit_sw);    stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw);
1679    if (stackptr < stacktop)    if (stackptr < stacktop)
1680      {      {
1681      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);
# Line 1588  do Line 1700  do
1700      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);
1701      count = 1;      count = 1;
1702      srcw[0] = common->recursive_head_ptr;      srcw[0] = common->recursive_head_ptr;
1703      if (common->control_head_ptr != 0)      if (needs_control_head)
1704        {        {
1705          SLJIT_ASSERT(common->control_head_ptr != 0);
1706        count = 2;        count = 2;
1707        srcw[1] = common->control_head_ptr;        srcw[1] = common->control_head_ptr;
1708        }        }
# Line 1605  do Line 1718  do
1718    
1719      switch(*cc)      switch(*cc)
1720        {        {
1721          case OP_KET:
1722          if (PRIVATE_DATA(cc) != 0)
1723            {
1724            count = 1;
1725            srcw[0] = PRIVATE_DATA(cc);
1726            }
1727          cc += 1 + LINK_SIZE;
1728          break;
1729    
1730        case OP_ASSERT:        case OP_ASSERT:
1731        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
1732        case OP_ASSERTBACK:        case OP_ASSERTBACK:
# Line 1851  if (save) Line 1973  if (save)
1973  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));
1974  }  }
1975    
1976    static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset)
1977    {
1978    pcre_uchar *end = bracketend(cc);
1979    BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT;
1980    
1981    /* Assert captures then. */
1982    if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT)
1983      current_offset = NULL;
1984    /* Conditional block does not. */
1985    if (*cc == OP_COND || *cc == OP_SCOND)
1986      has_alternatives = FALSE;
1987    
1988    cc = next_opcode(common, cc);
1989    if (has_alternatives)
1990      current_offset = common->then_offsets + (cc - common->start);
1991    
1992    while (cc < end)
1993      {
1994      if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND))
1995        cc = set_then_offsets(common, cc, current_offset);
1996      else
1997        {
1998        if (*cc == OP_ALT && has_alternatives)
1999          current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start);
2000        if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL)
2001          *current_offset = 1;
2002        cc = next_opcode(common, cc);
2003        }
2004      }
2005    
2006    return end;
2007    }
2008    
2009  #undef CASE_ITERATOR_PRIVATE_DATA_1  #undef CASE_ITERATOR_PRIVATE_DATA_1
2010  #undef CASE_ITERATOR_PRIVATE_DATA_2A  #undef CASE_ITERATOR_PRIVATE_DATA_2A
2011  #undef CASE_ITERATOR_PRIVATE_DATA_2B  #undef CASE_ITERATOR_PRIVATE_DATA_2B
# Line 1914  while (list_item) Line 2069  while (list_item)
2069  common->stubs = NULL;  common->stubs = NULL;
2070  }  }
2071    
2072  static SLJIT_INLINE void decrease_call_count(compiler_common *common)  static void add_label_addr(compiler_common *common)
2073    {
2074    DEFINE_COMPILER;
2075    label_addr_list *label_addr;
2076    
2077    label_addr = sljit_alloc_memory(compiler, sizeof(label_addr_list));
2078    if (label_addr == NULL)
2079      return;
2080    label_addr->label = LABEL();
2081    label_addr->addr = common->read_only_data_ptr;
2082    label_addr->next = common->label_addrs;
2083    common->label_addrs = label_addr;
2084    common->read_only_data_ptr++;
2085    }
2086    
2087    static SLJIT_INLINE void count_match(compiler_common *common)
2088  {  {
2089  DEFINE_COMPILER;  DEFINE_COMPILER;
2090    
2091  OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1);  OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
2092  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));
2093  }  }
2094    
# Line 1998  else Line 2168  else
2168  OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);
2169  if (common->mark_ptr != 0)  if (common->mark_ptr != 0)
2170    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0);
2171  SLJIT_ASSERT(common->control_head_ptr != 0);  if (common->control_head_ptr != 0)
2172  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0);
2173  OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));  OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));
2174  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr);
2175  OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base));  OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base));
2176  }  }
2177    
2178  static sljit_sw do_check_control_chain(sljit_sw *current)  static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg)
2179  {  {
2180  sljit_sw return_value = 0;  while (current != NULL)
   
 SLJIT_ASSERT(current != NULL);  
 do  
2181    {    {
2182    switch (current[-2])    switch (current[-2])
2183      {      {
2184      case type_commit:      case type_then_trap:
     /* Commit overwrites all. */  
     return -1;  
   
     case type_prune:  
2185      break;      break;
2186    
2187      case type_skip:      case type_mark:
2188      /* Overwrites prune, but not other skips. */      if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0)
2189      if (return_value == 0)        return current[-4];
       return_value = current[-3];  
2190      break;      break;
2191    
2192      default:      default:
# Line 2033  do Line 2195  do
2195      }      }
2196    current = (sljit_sw*)current[-1];    current = (sljit_sw*)current[-1];
2197    }    }
2198  while (current != NULL);  return -1;
 return return_value;  
2199  }  }
2200    
2201  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)
# Line 2106  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJI Line 2267  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJI
2267  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));
2268    
2269  jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3);  jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3);
2270  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + (int)sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0);
2271  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2272  OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);  OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);
2273  #endif  #endif
# Line 2274  return (bit < 256) ? ((0 << 8) | bit) : Line 2435  return (bit < 256) ? ((0 << 8) | bit) :
2435    
2436  static void check_partial(compiler_common *common, BOOL force)  static void check_partial(compiler_common *common, BOOL force)
2437  {  {
2438  /* Checks whether a partial matching is occured. Does not modify registers. */  /* Checks whether a partial matching is occurred. Does not modify registers. */
2439  DEFINE_COMPILER;  DEFINE_COMPILER;
2440  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
2441    
# Line 2361  else Line 2522  else
2522  JUMPHERE(jump);  JUMPHERE(jump);
2523  }  }
2524    
2525  static void read_char(compiler_common *common)  static void peek_char(compiler_common *common, pcre_uint32 max)
2526  {  {
2527  /* Reads the character into TMP1, updates STR_PTR.  /* Reads the character into TMP1, keeps STR_PTR.
2528  Does not check STR_END. TMP2 Destroyed. */  Does not check STR_END. TMP2 Destroyed. */
2529  DEFINE_COMPILER;  DEFINE_COMPILER;
2530  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2531  struct sljit_jump *jump;  struct sljit_jump *jump;
2532  #endif  #endif
2533    
2534    SLJIT_UNUSED_ARG(max);
2535    
2536  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2537  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2538  if (common->utf)  if (common->utf)
2539    {    {
2540  #if defined COMPILE_PCRE8    if (max < 128) return;
2541    
2542    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2543  #elif defined COMPILE_PCRE16    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
   jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);  
 #endif /* COMPILE_PCRE[8|16] */  
2544    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2545      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2546    JUMPHERE(jump);    JUMPHERE(jump);
2547    }    }
2548  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2549  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
2550    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2551    if (common->utf)
2552      {
2553      if (max < 0xd800) return;
2554    
2555      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2556      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2557      /* TMP2 contains the high surrogate. */
2558      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2559      OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2560      OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2561      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2562      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2563      JUMPHERE(jump);
2564      }
2565    #endif
2566  }  }
2567    
2568  static void peek_char(compiler_common *common)  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2569    
2570    static BOOL is_char7_bitset(const pcre_uint8 *bitset, BOOL nclass)
2571  {  {
2572  /* Reads the character into TMP1, keeps STR_PTR.  /* Tells whether the character codes below 128 are enough
2573  Does not check STR_END. TMP2 Destroyed. */  to determine a match. */
2574  DEFINE_COMPILER;  const pcre_uint8 value = nclass ? 0xff : 0;
2575  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  const pcre_uint8* end = bitset + 32;
 struct sljit_jump *jump;  
 #endif  
2576    
2577  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  bitset += 16;
2578  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  do
 if (common->utf)  
2579    {    {
2580  #if defined COMPILE_PCRE8    if (*bitset++ != value)
2581    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);      return FALSE;
 #elif defined COMPILE_PCRE16  
   jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);  
 #endif /* COMPILE_PCRE[8|16] */  
   add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));  
   OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);  
   JUMPHERE(jump);  
2582    }    }
2583  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  while (bitset < end);
2584    return TRUE;
2585  }  }
2586    
2587  static void read_char8_type(compiler_common *common)  static void read_char7_type(compiler_common *common, BOOL full_read)
2588  {  {
2589  /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */  /* Reads the precise character type of a character into TMP1, if the character
2590    is less than 128. Otherwise it returns with zero. Does not check STR_END. The
2591    full_read argument tells whether characters above max are accepted or not. */
2592  DEFINE_COMPILER;  DEFINE_COMPILER;
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
2593  struct sljit_jump *jump;  struct sljit_jump *jump;
 #endif  
2594    
2595  #ifdef SUPPORT_UTF  SLJIT_ASSERT(common->utf);
2596  if (common->utf)  
2597    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2598    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2599    
2600    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2601    
2602    if (full_read)
2603    {    {
   OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
   OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
 #if defined COMPILE_PCRE8  
   /* This can be an extra read in some situations, but hopefully  
   it is needed in most cases. */  
   OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
2604    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2605    add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));    OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
   JUMPHERE(jump);  
 #elif defined COMPILE_PCRE16  
   OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  
   jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
   OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
   JUMPHERE(jump);  
   /* Skip low surrogate if necessary. */  
   OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00);  
   OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800);  
   OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);  
   OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);  
2606    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
 #elif defined COMPILE_PCRE32  
   OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  
   jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
   OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
2607    JUMPHERE(jump);    JUMPHERE(jump);
 #endif /* COMPILE_PCRE[8|16|32] */  
   return;  
2608    }    }
 #endif /* SUPPORT_UTF */  
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
 #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
 /* The ctypes array contains only 256 values. */  
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  
 jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
 #endif  
 OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
 #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
 JUMPHERE(jump);  
 #endif  
2609  }  }
2610    
2611  static void skip_char_back(compiler_common *common)  #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2612    
2613    static void read_char_range(compiler_common *common, pcre_uint32 min, pcre_uint32 max, BOOL update_str_ptr)
2614  {  {
2615  /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */  /* Reads the precise value of a character into TMP1, if the character is
2616    between min and max (c >= min && c <= max). Otherwise it returns with a value
2617    outside the range. Does not check STR_END. */
2618  DEFINE_COMPILER;  DEFINE_COMPILER;
2619  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2620  #if defined COMPILE_PCRE8  struct sljit_jump *jump;
2621  struct sljit_label *label;  #endif
2622    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2623    struct sljit_jump *jump2;
2624    #endif
2625    
2626    SLJIT_UNUSED_ARG(update_str_ptr);
2627    SLJIT_UNUSED_ARG(min);
2628    SLJIT_UNUSED_ARG(max);
2629    SLJIT_ASSERT(min <= max);
2630    
2631    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2632    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2633    
2634    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2635  if (common->utf)  if (common->utf)
2636    {    {
2637    label = LABEL();    if (max < 128 && !update_str_ptr) return;
2638    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));  
2639    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2640    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);    if (min >= 0x10000)
2641        {
2642        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xf0);
2643        if (update_str_ptr)
2644          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2645        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2646        jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x7);
2647        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2648        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2649        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2650        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2651        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2652        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2653        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2654        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
2655        if (!update_str_ptr)
2656          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
2657        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2658        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2659        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2660        JUMPHERE(jump2);
2661        if (update_str_ptr)
2662          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2663        }
2664      else if (min >= 0x800 && max <= 0xffff)
2665        {
2666        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xe0);
2667        if (update_str_ptr)
2668          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2669        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2670        jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xf);
2671        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2672        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2673        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2674        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2675        if (!update_str_ptr)
2676          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2677        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2678        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2679        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2680        JUMPHERE(jump2);
2681        if (update_str_ptr)
2682          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2683        }
2684      else if (max >= 0x800)
2685        add_jump(compiler, (max < 0x10000) ? &common->utfreadchar16 : &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2686      else if (max < 128)
2687        {
2688        OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2689        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2690        }
2691      else
2692        {
2693        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2694        if (!update_str_ptr)
2695          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2696        else
2697          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2698        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2699        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2700        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2701        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2702        if (update_str_ptr)
2703          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2704        }
2705      JUMPHERE(jump);
2706      }
2707    #endif
2708    
2709    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2710    if (common->utf)
2711      {
2712      if (max >= 0x10000)
2713        {
2714        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2715        jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2716        /* TMP2 contains the high surrogate. */
2717        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2718        OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2719        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2720        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2721        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2722        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2723        JUMPHERE(jump);
2724        return;
2725        }
2726    
2727      if (max < 0xd800 && !update_str_ptr) return;
2728    
2729      /* Skip low surrogate if necessary. */
2730      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2731      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2732      if (update_str_ptr)
2733        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2734      if (max >= 0xd800)
2735        OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000);
2736      JUMPHERE(jump);
2737      }
2738    #endif
2739    }
2740    
2741    static SLJIT_INLINE void read_char(compiler_common *common)
2742    {
2743    read_char_range(common, 0, READ_CHAR_MAX, TRUE);
2744    }
2745    
2746    static void read_char8_type(compiler_common *common, BOOL update_str_ptr)
2747    {
2748    /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */
2749    DEFINE_COMPILER;
2750    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2751    struct sljit_jump *jump;
2752    #endif
2753    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2754    struct sljit_jump *jump2;
2755    #endif
2756    
2757    SLJIT_UNUSED_ARG(update_str_ptr);
2758    
2759    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2760    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2761    
2762    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2763    if (common->utf)
2764      {
2765      /* This can be an extra read in some situations, but hopefully
2766      it is needed in most cases. */
2767      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2768      jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2769      if (!update_str_ptr)
2770        {
2771        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2772        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2773        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2774        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2775        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2776        OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
2777        OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2778        jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2779        OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2780        JUMPHERE(jump2);
2781        }
2782      else
2783        add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));
2784      JUMPHERE(jump);
2785      return;
2786      }
2787    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2788    
2789    #if !defined COMPILE_PCRE8
2790    /* The ctypes array contains only 256 values. */
2791    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2792    jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2793    #endif
2794    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2795    #if !defined COMPILE_PCRE8
2796    JUMPHERE(jump);
2797    #endif
2798    
2799    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2800    if (common->utf && update_str_ptr)
2801      {
2802      /* Skip low surrogate if necessary. */
2803      OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
2804      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2805      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2806      JUMPHERE(jump);
2807      }
2808    #endif /* SUPPORT_UTF && COMPILE_PCRE16 */
2809    }
2810    
2811    static void skip_char_back(compiler_common *common)
2812    {
2813    /* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */
2814    DEFINE_COMPILER;
2815    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2816    #if defined COMPILE_PCRE8
2817    struct sljit_label *label;
2818    
2819    if (common->utf)
2820      {
2821      label = LABEL();
2822      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
2823      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2824      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);
2825    CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label);    CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label);
2826    return;    return;
2827    }    }
# Line 2499  if (common->utf) Line 2843  if (common->utf)
2843  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2844  }  }
2845    
2846  static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue)  static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpifmatch)
2847  {  {
2848  /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */  /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */
2849  DEFINE_COMPILER;  DEFINE_COMPILER;
2850    struct sljit_jump *jump;
2851    
2852  if (nltype == NLTYPE_ANY)  if (nltype == NLTYPE_ANY)
2853    {    {
2854    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
2855    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(jumpifmatch ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
2856    }    }
2857  else if (nltype == NLTYPE_ANYCRLF)  else if (nltype == NLTYPE_ANYCRLF)
2858    {    {
2859    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR);    if (jumpifmatch)
2860    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);      {
2861    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL);      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR));
2862    OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
2863    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));      }
2864      else
2865        {
2866        jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
2867        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
2868        JUMPHERE(jump);
2869        }
2870    }    }
2871  else  else
2872    {    {
2873    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);
2874    add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));    add_jump(compiler, backtracks, CMP(jumpifmatch ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));
2875    }    }
2876  }  }
2877    
# Line 2530  else Line 2881  else
2881  static void do_utfreadchar(compiler_common *common)  static void do_utfreadchar(compiler_common *common)
2882  {  {
2883  /* Fast decoding a UTF-8 character. TMP1 contains the first byte  /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2884  of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */  of the character (>= 0xc0). Return char value in TMP1, length in TMP2. */
2885  DEFINE_COMPILER;  DEFINE_COMPILER;
2886  struct sljit_jump *jump;  struct sljit_jump *jump;
2887    
2888  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2889    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2890    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2891    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2892    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2893    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2894    
2895  /* Searching for the first zero. */  /* Searching for the first zero. */
2896  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20);  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800);
2897  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2898  /* Two byte sequence. */  /* Two byte sequence. */
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
2899  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2900  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2));
2901    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2902    
2903    JUMPHERE(jump);
2904    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2905    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2906  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2907  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2908  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
 JUMPHERE(jump);  
2909    
2910  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10);  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2911  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2912  /* Three byte sequence. */  /* Three byte sequence. */
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
 OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f);  
 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12);  
 OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));  
2913  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2914  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2));  
2915  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2916    
2917  /* Four byte sequence. */  /* Four byte sequence. */
2918  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  JUMPHERE(jump);
2919  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07);  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
2920  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18);  OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2921    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2922    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
2923  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12);  
2924  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2925  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(4));
2926    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2927    }
2928    
2929    static void do_utfreadchar16(compiler_common *common)
2930    {
2931    /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2932    of the character (>= 0xc0). Return value in TMP1. */
2933    DEFINE_COMPILER;
2934    struct sljit_jump *jump;
2935    
2936    sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2937    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2938    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2939    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2940  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  
2941  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2942  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3));  
2943  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));  /* Searching for the first zero. */
2944    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800);
2945    jump = JUMP(SLJIT_C_NOT_ZERO);
2946    /* Two byte sequence. */
2947    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2948    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2949    
2950    JUMPHERE(jump);
2951    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400);
2952    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_NOT_ZERO);
2953    /* This code runs only in 8 bit mode. No need to shift the value. */
2954    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2955    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2956    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2957    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2958  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2959  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2960  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));  /* Three byte sequence. */
2961    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2962  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2963  }  }
2964    
# Line 2601  jump = JUMP(SLJIT_C_NOT_ZERO); Line 2978  jump = JUMP(SLJIT_C_NOT_ZERO);
2978  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2979  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2980  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);
2981    /* The upper 5 bits are known at this point. */
2982    compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x3);
2983  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2984  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2985  OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);  OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
 compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
2986  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2987  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2988    
2989  JUMPHERE(compare);  JUMPHERE(compare);
2990  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2991  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2992    
2993  /* We only have types for characters less than 256. */  /* We only have types for characters less than 256. */
 OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);  
 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);  
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
 }  
   
 #elif defined COMPILE_PCRE16  
   
 static void do_utfreadchar(compiler_common *common)  
 {  
 /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char  
 of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */  
 DEFINE_COMPILER;  
 struct sljit_jump *jump;  
   
 sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  
 jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00);  
 /* Do nothing, only return. */  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
   
2994  JUMPHERE(jump);  JUMPHERE(jump);
2995  /* Combine two 16 bit characters. */  OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2996  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2997  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
 OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);  
 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10);  
 OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff);  
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));  
 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);  
2998  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2999  }  }
3000    
3001  #endif /* COMPILE_PCRE[8|16] */  #endif /* COMPILE_PCRE8 */
3002    
3003  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
3004    
# Line 2722  if (firstline) Line 3073  if (firstline)
3073      mainloop = LABEL();      mainloop = LABEL();
3074      /* Continual stores does not cause data dependency. */      /* Continual stores does not cause data dependency. */
3075      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0);
3076      read_char(common);      read_char_range(common, common->nlmin, common->nlmax, TRUE);
3077      check_newlinechar(common, common->nltype, &newline, TRUE);      check_newlinechar(common, common->nltype, &newline, TRUE);
3078      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);
3079      JUMPHERE(end);      JUMPHERE(end);
# Line 2798  if (newlinecheck) Line 3149  if (newlinecheck)
3149  return mainloop;  return mainloop;
3150  }  }
3151    
3152  #define MAX_N_CHARS 3  static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, int max_chars)
   
 static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)  
3153  {  {
3154  DEFINE_COMPILER;  /* Recursive function, which scans prefix literals. */
3155  struct sljit_label *start;  int len, repeat, len_save, consumed = 0;
3156  struct sljit_jump *quit;  pcre_uint32 caseless, chr, mask;
3157  pcre_uint32 chars[MAX_N_CHARS * 2];  pcre_uchar *alternative, *cc_save;
3158  pcre_uchar *cc = common->start + 1 + LINK_SIZE;  BOOL last, any;
 int location = 0;  
 pcre_int32 len, c, bit, caseless;  
 int must_stop;  
   
 /* We do not support alternatives now. */  
 if (*(common->start + GET(common->start, 1)) == OP_ALT)  
   return FALSE;  
3159    
3160    repeat = 1;
3161  while (TRUE)  while (TRUE)
3162    {    {
3163      last = TRUE;
3164      any = FALSE;
3165    caseless = 0;    caseless = 0;
3166    must_stop = 1;    switch (*cc)
   switch(*cc)  
3167      {      {
     case OP_CHAR:  
     must_stop = 0;  
     cc++;  
     break;  
   
3168      case OP_CHARI:      case OP_CHARI:
3169      caseless = 1;      caseless = 1;
3170      must_stop = 0;      case OP_CHAR:
3171        last = FALSE;
3172      cc++;      cc++;
3173      break;      break;
3174    
# Line 2847  while (TRUE) Line 3187  while (TRUE)
3187      cc++;      cc++;
3188      continue;      continue;
3189    
3190        case OP_ASSERT:
3191        case OP_ASSERT_NOT:
3192        case OP_ASSERTBACK:
3193        case OP_ASSERTBACK_NOT:
3194        cc = bracketend(cc);
3195        continue;
3196    
3197      case OP_PLUS:      case OP_PLUS:
3198      case OP_MINPLUS:      case OP_MINPLUS:
3199      case OP_POSPLUS:      case OP_POSPLUS:
3200      cc++;      cc++;
3201      break;      break;
3202    
3203        case OP_EXACTI:
3204        caseless = 1;
3205      case OP_EXACT:      case OP_EXACT:
3206        repeat = GET2(cc, 1);
3207        last = FALSE;
3208      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3209      break;      break;
3210    
# Line 2864  while (TRUE) Line 3215  while (TRUE)
3215      cc++;      cc++;
3216      break;      break;
3217    
3218      case OP_EXACTI:      case OP_KET:
3219      caseless = 1;      cc += 1 + LINK_SIZE;
3220      cc += 1 + IMM2_SIZE;      continue;
3221    
3222        case OP_ALT:
3223        cc += GET(cc, 1);
3224        continue;
3225    
3226        case OP_ONCE:
3227        case OP_ONCE_NC:
3228        case OP_BRA:
3229        case OP_BRAPOS:
3230        case OP_CBRA:
3231        case OP_CBRAPOS:
3232        alternative = cc + GET(cc, 1);
3233        while (*alternative == OP_ALT)
3234          {
3235          max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, max_chars);
3236          if (max_chars == 0)
3237            return consumed;
3238          alternative += GET(alternative, 1);
3239          }
3240    
3241        if (*cc == OP_CBRA || *cc == OP_CBRAPOS)
3242          cc += IMM2_SIZE;
3243        cc += 1 + LINK_SIZE;
3244        continue;
3245    
3246        case OP_CLASS:
3247    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3248        if (common->utf && !is_char7_bitset((const pcre_uint8 *)(cc + 1), FALSE)) return consumed;
3249    #endif
3250        any = TRUE;
3251        cc += 1 + 32 / sizeof(pcre_uchar);
3252      break;      break;
3253    
3254      default:      case OP_NCLASS:
3255      must_stop = 2;  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3256        if (common->utf) return consumed;
3257    #endif
3258        any = TRUE;
3259        cc += 1 + 32 / sizeof(pcre_uchar);
3260      break;      break;
3261    
3262    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3263        case OP_XCLASS:
3264    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3265        if (common->utf) return consumed;
3266    #endif
3267        any = TRUE;
3268        cc += GET(cc, 1);
3269        break;
3270    #endif
3271    
3272        case OP_DIGIT:
3273    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3274        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_digit, FALSE))
3275          return consumed;
3276    #endif
3277        any = TRUE;
3278        cc++;
3279        break;
3280    
3281        case OP_WHITESPACE:
3282    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3283        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_space, FALSE))
3284          return consumed;
3285    #endif
3286        any = TRUE;
3287        cc++;
3288        break;
3289    
3290        case OP_WORDCHAR:
3291    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3292        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_word, FALSE))
3293          return consumed;
3294    #endif
3295        any = TRUE;
3296        cc++;
3297        break;
3298    
3299        case OP_NOT_DIGIT:
3300        case OP_NOT_WHITESPACE:
3301        case OP_NOT_WORDCHAR:
3302        case OP_ANY:
3303        case OP_ALLANY:
3304    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3305        if (common->utf) return consumed;
3306    #endif
3307        any = TRUE;
3308        cc++;
3309        break;
3310    
3311    #ifdef SUPPORT_UCP
3312        case OP_NOTPROP:
3313        case OP_PROP:
3314    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3315        if (common->utf) return consumed;
3316    #endif
3317        any = TRUE;
3318        cc += 1 + 2;
3319        break;
3320    #endif
3321    
3322        case OP_TYPEEXACT:
3323        repeat = GET2(cc, 1);
3324        cc += 1 + IMM2_SIZE;
3325        continue;
3326    
3327        default:
3328        return consumed;
3329        }
3330    
3331      if (any)
3332        {
3333    #if defined COMPILE_PCRE8
3334        mask = 0xff;
3335    #elif defined COMPILE_PCRE16
3336        mask = 0xffff;
3337    #elif defined COMPILE_PCRE32
3338        mask = 0xffffffff;
3339    #else
3340        SLJIT_ASSERT_STOP();
3341    #endif
3342    
3343        do
3344          {
3345          chars[0] = mask;
3346          chars[1] = mask;
3347    
3348          consumed++;
3349          if (--max_chars == 0)
3350            return consumed;
3351          chars += 2;
3352          }
3353        while (--repeat > 0);
3354    
3355        repeat = 1;
3356        continue;
3357        }
3358    
3359      len = 1;
3360    #ifdef SUPPORT_UTF
3361      if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3362    #endif
3363    
3364      if (caseless != 0 && char_has_othercase(common, cc))
3365        {
3366        caseless = char_get_othercase_bit(common, cc);
3367        if (caseless == 0)
3368          return consumed;
3369    #ifdef COMPILE_PCRE8
3370        caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));
3371    #else
3372        if ((caseless & 0x100) != 0)
3373          caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));
3374        else
3375          caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));
3376    #endif
3377        }
3378      else
3379        caseless = 0;
3380    
3381      len_save = len;
3382      cc_save = cc;
3383      while (TRUE)
3384        {
3385        do
3386          {
3387          chr = *cc;
3388    #ifdef COMPILE_PCRE32
3389          if (SLJIT_UNLIKELY(chr == NOTACHAR))
3390            return consumed;
3391    #endif
3392          mask = 0;
3393          if ((pcre_uint32)len == (caseless & 0xff))
3394            {
3395            mask = caseless >> 8;
3396            chr |= mask;
3397            }
3398    
3399    #ifdef COMPILE_PCRE32
3400          if (chars[0] == NOTACHAR && chars[1] == 0)
3401    #else
3402          if (chars[0] == NOTACHAR)
3403    #endif
3404            {
3405            chars[0] = chr;
3406            chars[1] = mask;
3407            }
3408          else
3409            {
3410            mask |= chars[0] ^ chr;
3411            chr |= mask;
3412            chars[0] = chr;
3413            chars[1] |= mask;
3414            }
3415    
3416          len--;
3417          consumed++;
3418          if (--max_chars == 0)
3419            return consumed;
3420          chars += 2;
3421          cc++;
3422          }
3423        while (len > 0);
3424    
3425        if (--repeat == 0)
3426          break;
3427    
3428        len = len_save;
3429        cc = cc_save;
3430        }
3431    
3432      repeat = 1;
3433      if (last)
3434        return consumed;
3435      }
3436    }
3437    
3438    #define MAX_N_CHARS 16
3439    #define MIN_RANGE_SIZE 4
3440    
3441    static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)
3442    {
3443    DEFINE_COMPILER;
3444    struct sljit_label *start;
3445    struct sljit_jump *quit;
3446    pcre_uint32 chars[MAX_N_CHARS * 2];
3447    pcre_uint8 ones[MAX_N_CHARS];
3448    int offsets[3];
3449    pcre_uint32 mask, byte;
3450    int i, max, from;
3451    int range_right = -1, range_len = -1;
3452    sljit_ub *update_table = NULL;
3453    BOOL in_range;
3454    
3455    /* This is even TRUE, if both are NULL. */
3456    SLJIT_ASSERT(common->read_only_data_ptr == common->read_only_data);
3457    
3458    for (i = 0; i < MAX_N_CHARS; i++)
3459      {
3460      chars[i << 1] = NOTACHAR;
3461      chars[(i << 1) + 1] = 0;
3462      }
3463    
3464    max = scan_prefix(common, common->start, chars, MAX_N_CHARS);
3465    
3466    if (max <= 1)
3467      return FALSE;
3468    
3469    for (i = 0; i < max; i++)
3470      {
3471      mask = chars[(i << 1) + 1];
3472      ones[i] = ones_in_half_byte[mask & 0xf];
3473      mask >>= 4;
3474      while (mask != 0)
3475        {
3476        ones[i] += ones_in_half_byte[mask & 0xf];
3477        mask >>= 4;
3478        }
3479      }
3480    
3481    in_range = FALSE;
3482    for (i = 0; i <= max; i++)
3483      {
3484      if (i < max && ones[i] <= 1)
3485        {
3486        if (!in_range)
3487          {
3488          in_range = TRUE;
3489          from = i;
3490          }
3491        }
3492      else if (in_range)
3493        {
3494        if (range_len == -1 || (i - from) > range_len)
3495          {
3496          range_len = i - from;
3497          range_right = i - 1;
3498          }
3499        in_range = FALSE;
3500        }
3501      }
3502    
3503    if (range_len >= MIN_RANGE_SIZE)
3504      {
3505      /* Since no data is consumed (see the assert in the beginning
3506      of this function), this space can be reallocated. */
3507      if (common->read_only_data)
3508        SLJIT_FREE(common->read_only_data);
3509    
3510      common->read_only_data_size += 256;
3511      common->read_only_data = (sljit_uw *)SLJIT_MALLOC(common->read_only_data_size);
3512      if (common->read_only_data == NULL)
3513        return TRUE;
3514    
3515      update_table = (sljit_ub *)common->read_only_data;
3516      common->read_only_data_ptr = (sljit_uw *)(update_table + 256);
3517      memset(update_table, IN_UCHARS(range_len), 256);
3518    
3519      for (i = 0; i < range_len; i++)
3520        {
3521        byte = chars[(range_right - i) << 1] & 0xff;
3522        if (update_table[byte] > IN_UCHARS(i))
3523          update_table[byte] = IN_UCHARS(i);
3524        mask = chars[((range_right - i) << 1) + 1] & 0xff;
3525        if (mask != 0)
3526          {
3527          byte ^= mask;
3528          if (update_table[byte] > IN_UCHARS(i))
3529            update_table[byte] = IN_UCHARS(i);
3530          }
3531      }      }
3532      }
3533    
3534    if (must_stop == 2)  offsets[0] = -1;
3535        break;  /* Scan forward. */
3536    for (i = 0; i < max; i++)
3537      if (ones[i] <= 2) {
3538        offsets[0] = i;
3539        break;
3540      }
3541    
3542    len = 1;  if (offsets[0] == -1)
3543  #ifdef SUPPORT_UTF    return FALSE;
   if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);  
 #endif  
3544    
3545    if (caseless && char_has_othercase(common, cc))  /* Scan backward. */
3546    offsets[1] = -1;
3547    for (i = max - 1; i > offsets[0]; i--)
3548      if (ones[i] <= 2 && i != range_right)
3549      {      {
3550      caseless = char_get_othercase_bit(common, cc);      offsets[1] = i;
3551      if (caseless == 0)      break;
       return FALSE;  
 #ifdef COMPILE_PCRE8  
     caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));  
 #else  
     if ((caseless & 0x100) != 0)  
       caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));  
     else  
       caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));  
 #endif  
3552      }      }
   else  
     caseless = 0;  
3553    
3554    while (len > 0 && location < MAX_N_CHARS * 2)  /* This case is handled better by fast_forward_first_char. */
3555      {  if (offsets[1] == -1 && offsets[0] == 0)
3556      c = *cc;    return FALSE;
3557      bit = 0;  
3558      if (len == (caseless & 0xff))  offsets[2] = -1;
3559    if (offsets[1] >= 0 && range_right == -1)
3560      {
3561      /* Scan from middle. */
3562      for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++)
3563        if (ones[i] <= 2)
3564        {        {
3565        bit = caseless >> 8;        offsets[2] = i;
3566        c |= bit;        break;
3567        }        }
3568    
3569      chars[location] = c;    if (offsets[2] == -1)
3570      chars[location + 1] = bit;      {
3571        for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--)
3572      len--;        if (ones[i] <= 2)
3573      location += 2;          {
3574      cc++;          offsets[2] = i;
3575            break;
3576            }
3577      }      }
   
   if (location >= MAX_N_CHARS * 2 || must_stop != 0)  
     break;  
3578    }    }
3579    
3580  /* At least two characters are required. */  SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1]));
3581  if (location < 2 * 2)  SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2]));
3582      return FALSE;  
3583    chars[0] = chars[offsets[0] << 1];
3584    chars[1] = chars[(offsets[0] << 1) + 1];
3585    if (offsets[2] >= 0)
3586      {
3587      chars[2] = chars[offsets[2] << 1];
3588      chars[3] = chars[(offsets[2] << 1) + 1];
3589      }
3590    if (offsets[1] >= 0)
3591      {
3592      chars[4] = chars[offsets[1] << 1];
3593      chars[5] = chars[(offsets[1] << 1) + 1];
3594      }
3595    
3596    max -= 1;
3597  if (firstline)  if (firstline)
3598    {    {
3599    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
3600      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3601    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
3602    OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3603      quit = CMP(SLJIT_C_LESS_EQUAL, STR_END, 0, TMP1, 0);
3604      OP1(SLJIT_MOV, STR_END, 0, TMP1, 0);
3605      JUMPHERE(quit);
3606    }    }
3607  else  else
3608    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3609    
3610    #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3611    if (range_len >= MIN_RANGE_SIZE)
3612      OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table);
3613    #endif
3614    
3615  start = LABEL();  start = LABEL();
3616  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3617    
3618  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  if (range_len >= MIN_RANGE_SIZE)
3619  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));    {
3620    #if defined COMPILE_PCRE8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN)
3621      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right));
3622    #else
3623      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right + 1) - 1);
3624    #endif
3625    
3626    #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3627      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0);
3628    #else
3629      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table);
3630    #endif
3631      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3632      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start);
3633      }
3634    
3635    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0]));
3636    if (offsets[1] >= 0)
3637      OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1]));
3638  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3639    
3640  if (chars[1] != 0)  if (chars[1] != 0)
3641    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);
3642  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);
3643  if (location > 2 * 2)  if (offsets[2] >= 0)
3644    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1));
3645  if (chars[3] != 0)  
3646    OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]);  if (offsets[1] >= 0)
 CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start);  
 if (location > 2 * 2)  
3647    {    {
3648    if (chars[5] != 0)    if (chars[5] != 0)
3649      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]);      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]);
3650    CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start);    CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start);
3651      }
3652    
3653    if (offsets[2] >= 0)
3654      {
3655      if (chars[3] != 0)
3656        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]);
3657      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start);
3658    }    }
3659  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3660    
3661  JUMPHERE(quit);  JUMPHERE(quit);
3662    
3663  if (firstline)  if (firstline)
3664      {
3665      if (range_len >= MIN_RANGE_SIZE)
3666        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3667    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3668      if (range_len >= MIN_RANGE_SIZE)
3669        {
3670        quit = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0);
3671        OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0);
3672        JUMPHERE(quit);
3673        }
3674      }
3675  else  else
3676    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3677  return TRUE;  return TRUE;
3678  }  }
3679    
3680  #undef MAX_N_CHARS  #undef MAX_N_CHARS
3681    #undef MIN_RANGE_SIZE
3682    
3683  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)
3684  {  {
# Line 3081  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ Line 3794  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_
3794  skip_char_back(common);  skip_char_back(common);
3795    
3796  loop = LABEL();  loop = LABEL();
3797  read_char(common);  read_char_range(common, common->nlmin, common->nlmax, TRUE);
3798  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3799  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
3800    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
# Line 3110  if (firstline) Line 3823  if (firstline)
3823    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3824  }  }
3825    
3826  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks);  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks);
3827    
3828  static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline)  static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, pcre_uint8 *start_bits, BOOL firstline)
3829  {  {
3830  DEFINE_COMPILER;  DEFINE_COMPILER;
3831  struct sljit_label *start;  struct sljit_label *start;
3832  struct sljit_jump *quit;  struct sljit_jump *quit;
3833  struct sljit_jump *found = NULL;  struct sljit_jump *found = NULL;
3834  jump_list *matches = NULL;  jump_list *matches = NULL;
 pcre_uint8 inverted_start_bits[32];  
 int i;  
3835  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3836  struct sljit_jump *jump;  struct sljit_jump *jump;
3837  #endif  #endif
3838    
 for (i = 0; i < 32; ++i)  
   inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]);  
   
3839  if (firstline)  if (firstline)
3840    {    {
3841    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
# Line 3143  if (common->utf) Line 3851  if (common->utf)
3851    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
3852  #endif  #endif
3853    
3854  if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches))  if (!check_class_ranges(common, start_bits, (start_bits[31] & 0x80) != 0, TRUE, &matches))
3855    {    {
3856  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3857    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);
# Line 3152  if (!check_class_ranges(common, inverted Line 3860  if (!check_class_ranges(common, inverted
3860  #endif  #endif
3861    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
3862    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
3863    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
3864    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
3865    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
3866    found = JUMP(SLJIT_C_NOT_ZERO);    found = JUMP(SLJIT_C_NOT_ZERO);
# Line 3355  JUMPHERE(skipread); Line 4063  JUMPHERE(skipread);
4063    
4064  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
4065  check_str_end(common, &skipread_list);  check_str_end(common, &skipread_list);
4066  peek_char(common);  peek_char(common, READ_CHAR_MAX);
4067    
4068  /* Testing char type. This is a code duplication. */  /* Testing char type. This is a code duplication. */
4069  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3401  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE Line 4109  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE
4109  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);
4110  }  }
4111    
4112  /*  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
   range format:  
   
   ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range).  
   ranges[1] = first bit (0 or 1)  
   ranges[2-length] = position of the bit change (when the current bit is not equal to the previous)  
 */  
   
 static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch)  
4113  {  {
4114  DEFINE_COMPILER;  DEFINE_COMPILER;
4115  struct sljit_jump *jump;  int ranges[MAX_RANGE_SIZE];
   
 if (ranges[0] < 0)  
   return FALSE;  
   
 switch(ranges[0])  
   {  
   case 1:  
   if (readch)  
     read_char(common);  
   add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));  
   return TRUE;  
   
   case 2:  
   if (readch)  
     read_char(common);  
   OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);  
   add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));  
   return TRUE;  
   
   case 4:  
   if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5])  
     {  
     if (readch)  
       read_char(common);  
     if (ranges[1] != 0)  
       {  
       add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));  
       add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4]));  
       }  
     else  
       {  
       jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]);  
       add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4]));  
       JUMPHERE(jump);  
       }  
     return TRUE;  
     }  
   if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2]))  
     {  
     if (readch)  
       read_char(common);  
     OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]);  
     OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]);  
     add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4]));  
     return TRUE;  
     }  
   return FALSE;  
   
   default:  
   return FALSE;  
   }  
 }  
   
 static void get_ctype_ranges(compiler_common *common, int flag, int *ranges)  
 {  
 int i, bit, length;  
 const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes;  
   
 bit = ctypes[0] & flag;  
 ranges[0] = -1;  
 ranges[1] = bit != 0 ? 1 : 0;  
 length = 0;  
   
 for (i = 1; i < 256; i++)  
   if ((ctypes[i] & flag) != bit)  
     {  
     if (length >= MAX_RANGE_SIZE)  
       return;  
     ranges[2 + length] = i;  
     length++;  
     bit ^= flag;  
     }  
   
 if (bit != 0)  
   {  
   if (length >= MAX_RANGE_SIZE)  
     return;  
   ranges[2 + length] = 256;  
   length++;  
   }  
 ranges[0] = length;  
 }  
   
 static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks)  
 {  
 int ranges[2 + MAX_RANGE_SIZE];  
4116  pcre_uint8 bit, cbit, all;  pcre_uint8 bit, cbit, all;
4117  int i, byte, length = 0;  int i, byte, length = 0;
4118    
4119  bit = bits[0] & 0x1;  bit = bits[0] & 0x1;
4120  ranges[1] = bit;  /* All bits will be zero or one (since bit is zero or one). */
 /* Can be 0 or 255. */  
4121  all = -bit;  all = -bit;
4122    
4123  for (i = 0; i < 256; )  for (i = 0; i < 256; )
# Line 3519  for (i = 0; i < 256; ) Line 4132  for (i = 0; i < 256; )
4132        {        {
4133        if (length >= MAX_RANGE_SIZE)        if (length >= MAX_RANGE_SIZE)
4134          return FALSE;          return FALSE;
4135        ranges[2 + length] = i;        ranges[length] = i;
4136        length++;        length++;
4137        bit = cbit;        bit = cbit;
4138        all = -cbit;        all = -cbit;
# Line 3532  if (((bit == 0) && nclass) || ((bit == 1 Line 4145  if (((bit == 0) && nclass) || ((bit == 1
4145    {    {
4146    if (length >= MAX_RANGE_SIZE)    if (length >= MAX_RANGE_SIZE)
4147      return FALSE;      return FALSE;
4148    ranges[2 + length] = 256;    ranges[length] = 256;
4149    length++;    length++;
4150    }    }
 ranges[0] = length;  
4151    
4152  return check_ranges(common, ranges, backtracks, FALSE);  if (length < 0 || length > 4)
4153      return FALSE;
4154    
4155    bit = bits[0] & 0x1;
4156    if (invert) bit ^= 0x1;
4157    
4158    /* No character is accepted. */
4159    if (length == 0 && bit == 0)
4160      add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4161    
4162    switch(length)
4163      {
4164      case 0:
4165      /* When bit != 0, all characters are accepted. */
4166      return TRUE;
4167    
4168      case 1:
4169      add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4170      return TRUE;
4171    
4172      case 2:
4173      if (ranges[0] + 1 != ranges[1])
4174        {
4175        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4176        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4177        }
4178      else
4179        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4180      return TRUE;
4181    
4182      case 3:
4183      if (bit != 0)
4184        {
4185        add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4186        if (ranges[0] + 1 != ranges[1])
4187          {
4188          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4189          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4190          }
4191        else
4192          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4193        return TRUE;
4194        }
4195    
4196      add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[0]));
4197      if (ranges[1] + 1 != ranges[2])
4198        {
4199        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]);
4200        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4201        }
4202      else
4203        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1]));
4204      return TRUE;
4205    
4206      case 4:
4207      if ((ranges[1] - ranges[0]) == (ranges[3] - ranges[2])
4208          && (ranges[0] | (ranges[2] - ranges[0])) == ranges[2]
4209          && is_powerof2(ranges[2] - ranges[0]))
4210        {
4211        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[0]);
4212        if (ranges[2] + 1 != ranges[3])
4213          {
4214          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
4215          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4216          }
4217        else
4218          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4219        return TRUE;
4220        }
4221    
4222      if (bit != 0)
4223        {
4224        i = 0;
4225        if (ranges[0] + 1 != ranges[1])
4226          {
4227          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4228          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4229          i = ranges[0];
4230          }
4231        else
4232          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4233    
4234        if (ranges[2] + 1 != ranges[3])
4235          {
4236          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - i);
4237          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4238          }
4239        else
4240          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2] - i));
4241        return TRUE;
4242        }
4243    
4244      OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4245      add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[0]));
4246      if (ranges[1] + 1 != ranges[2])
4247        {
4248        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]);
4249        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4250        }
4251      else
4252        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4253      return TRUE;
4254    
4255      default:
4256      SLJIT_ASSERT_STOP();
4257      return FALSE;
4258      }
4259  }  }
4260    
4261  static void check_anynewline(compiler_common *common)  static void check_anynewline(compiler_common *common)
# Line 3904  return cc; Line 4622  return cc;
4622  #define SET_TYPE_OFFSET(value) \  #define SET_TYPE_OFFSET(value) \
4623    if ((value) != typeoffset) \    if ((value) != typeoffset) \
4624      { \      { \
4625      if ((value) > typeoffset) \      if ((value) < typeoffset) \
       OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \  
     else \  
4626        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \
4627        else \
4628          OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \
4629      } \      } \
4630    typeoffset = (value);    typeoffset = (value);
4631    
4632  #define SET_CHAR_OFFSET(value) \  #define SET_CHAR_OFFSET(value) \
4633    if ((value) != charoffset) \    if ((value) != charoffset) \
4634      { \      { \
4635      if ((value) > charoffset) \      if ((value) < charoffset) \
4636        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(charoffset - (value))); \
4637      else \      else \
4638        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)((value) - charoffset)); \
4639      } \      } \
4640    charoffset = (value);    charoffset = (value);
4641    
# Line 3925  static void compile_xclass_matchingpath( Line 4643  static void compile_xclass_matchingpath(
4643  {  {
4644  DEFINE_COMPILER;  DEFINE_COMPILER;
4645  jump_list *found = NULL;  jump_list *found = NULL;
4646  jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks;  jump_list **list = (cc[0] & XCL_NOT) == 0 ? &found : backtracks;
4647  pcre_int32 c, charoffset;  sljit_uw c, charoffset, max = 256, min = READ_CHAR_MAX;
 const pcre_uint32 *other_cases;  
4648  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
4649  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
4650  int compares, invertcmp, numberofcmps;  int compares, invertcmp, numberofcmps;
4651    #if defined SUPPORT_UTF && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16)
4652    BOOL utf = common->utf;
4653    #endif
4654    
4655  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4656  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4657  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4658  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4659  pcre_int32 typeoffset;  const pcre_uint32 *other_cases;
4660    sljit_uw typeoffset;
4661  #endif  #endif
4662    
4663  /* Although SUPPORT_UTF must be defined, we are  /* Scanning the necessary info. */
4664     not necessary in utf mode even in 8 bit mode. */  cc++;
4665  detect_partial_match(common, backtracks);  ccbegin = cc;
4666  read_char(common);  compares = 0;
4667    if (cc[-1] & XCL_MAP)
 if ((*cc++ & XCL_MAP) != 0)  
4668    {    {
4669    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    min = 0;
 #ifndef COMPILE_PCRE8  
   jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
     jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
 #endif  
   
   if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list))  
     {  
     OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);  
     OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);  
     OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);  
     OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);  
     OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);  
     add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));  
     }  
   
 #ifndef COMPILE_PCRE8  
   JUMPHERE(jump);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
     JUMPHERE(jump);  
 #endif  
   OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);  
 #ifdef SUPPORT_UCP  
   charsaved = TRUE;  
 #endif  
4670    cc += 32 / sizeof(pcre_uchar);    cc += 32 / sizeof(pcre_uchar);
4671    }    }
4672    
 /* Scanning the necessary info. */  
 ccbegin = cc;  
 compares = 0;  
4673  while (*cc != XCL_END)  while (*cc != XCL_END)
4674    {    {
4675    compares++;    compares++;
4676    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4677      {      {
4678      cc += 2;      cc ++;
4679  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4680      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c > max) max = c;
4681  #endif      if (c < min) min = c;
4682  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4683      needschar = TRUE;      needschar = TRUE;
4684  #endif  #endif
4685      }      }
4686    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4687      {      {
4688      cc += 2;      cc ++;
4689  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4690      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c < min) min = c;
4691  #endif      GETCHARINCTEST(c, cc);
4692      cc++;      if (c > max) max = c;
 #ifdef SUPPORT_UTF  
     if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
4693  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4694      needschar = TRUE;      needschar = TRUE;
4695  #endif  #endif
# Line 4011  while (*cc != XCL_END) Line 4699  while (*cc != XCL_END)
4699      {      {
4700      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);
4701      cc++;      cc++;
4702        if (*cc == PT_CLIST)
4703          {
4704          other_cases = PRIV(ucd_caseless_sets) + cc[1];
4705          while (*other_cases != NOTACHAR)
4706            {
4707            if (*other_cases > max) max = *other_cases;
4708            if (*other_cases < min) min = *other_cases;
4709            other_cases++;
4710            }
4711          }
4712        else
4713          {
4714          max = READ_CHAR_MAX;
4715          min = 0;
4716          }
4717    
4718      switch(*cc)      switch(*cc)
4719        {        {
4720        case PT_ANY:        case PT_ANY:
# Line 4030  while (*cc != XCL_END) Line 4734  while (*cc != XCL_END)
4734        case PT_SPACE:        case PT_SPACE:
4735        case PT_PXSPACE:        case PT_PXSPACE:
4736        case PT_WORD:        case PT_WORD:
4737          case PT_PXGRAPH:
4738          case PT_PXPRINT:
4739          case PT_PXPUNCT:
4740        needstype = TRUE;        needstype = TRUE;
4741        needschar = TRUE;        needschar = TRUE;
4742        break;        break;
# Line 4043  while (*cc != XCL_END) Line 4750  while (*cc != XCL_END)
4750        SLJIT_ASSERT_STOP();        SLJIT_ASSERT_STOP();
4751        break;        break;
4752        }        }
4753      cc += 2;      cc += 2;
4754        }
4755    #endif
4756      }
4757    
4758    /* We are not necessary in utf mode even in 8 bit mode. */
4759    cc = ccbegin;
4760    detect_partial_match(common, backtracks);
4761    read_char_range(common, min, max, (cc[-1] & XCL_NOT) != 0);
4762    
4763    if ((cc[-1] & XCL_HASPROP) == 0)
4764      {
4765      if ((cc[-1] & XCL_MAP) != 0)
4766        {
4767        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4768        if (!check_class_ranges(common, (const pcre_uint8 *)cc, (((const pcre_uint8 *)cc)[31] & 0x80) != 0, TRUE, &found))
4769          {
4770          OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4771          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4772          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4773          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4774          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4775          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4776          }
4777    
4778        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4779        JUMPHERE(jump);
4780    
4781        cc += 32 / sizeof(pcre_uchar);
4782        }
4783      else
4784        {
4785        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, min);
4786        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, max - min));
4787      }      }
4788      }
4789    else if ((cc[-1] & XCL_MAP) != 0)
4790      {
4791      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4792    #ifdef SUPPORT_UCP
4793      charsaved = TRUE;
4794    #endif
4795      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4796        {
4797    #ifdef COMPILE_PCRE8
4798        SLJIT_ASSERT(common->utf);
4799  #endif  #endif
4800        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4801    
4802        OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4803        OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4804        OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4805        OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4806        OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4807        add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
4808    
4809        JUMPHERE(jump);
4810        }
4811    
4812      OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4813      cc += 32 / sizeof(pcre_uchar);
4814    }    }
4815    
4816  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4089  if (needstype || needsscript) Line 4854  if (needstype || needsscript)
4854  #endif  #endif
4855    
4856  /* Generating code. */  /* Generating code. */
 cc = ccbegin;  
4857  charoffset = 0;  charoffset = 0;
4858  numberofcmps = 0;  numberofcmps = 0;
4859  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4105  while (*cc != XCL_END) Line 4869  while (*cc != XCL_END)
4869    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4870      {      {
4871      cc ++;      cc ++;
4872  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4873    
4874      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4875        {        {
4876        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4877        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL);
4878        numberofcmps++;        numberofcmps++;
4879        }        }
4880      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4881        {        {
4882        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4883        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4884        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4885        numberofcmps = 0;        numberofcmps = 0;
4886        }        }
4887      else      else
4888        {        {
4889        jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset);        jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4890        numberofcmps = 0;        numberofcmps = 0;
4891        }        }
4892      }      }
4893    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4894      {      {
4895      cc ++;      cc ++;
4896  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4897      SET_CHAR_OFFSET(c);      SET_CHAR_OFFSET(c);
4898  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4899      if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4900      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4901        {        {
4902        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4903        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);
4904        numberofcmps++;        numberofcmps++;
4905        }        }
4906      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4907        {        {
4908        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4909        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4910        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4911        numberofcmps = 0;        numberofcmps = 0;
4912        }        }
4913      else      else
4914        {        {
4915        jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset);        jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4916        numberofcmps = 0;        numberofcmps = 0;
4917        }        }
4918      }      }
# Line 4217  while (*cc != XCL_END) Line 4961  while (*cc != XCL_END)
4961    
4962        case PT_SPACE:        case PT_SPACE:
4963        case PT_PXSPACE:        case PT_PXSPACE:
       if (*cc == PT_SPACE)  
         {  
         OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  
         jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 11 - charoffset);  
         }  
4964        SET_CHAR_OFFSET(9);        SET_CHAR_OFFSET(9);
4965        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd - 0x9);
4966        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4967        if (*cc == PT_SPACE)  
4968          JUMPHERE(jump);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9);
4969          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4970    
4971          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9);
4972          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4973    
4974        SET_TYPE_OFFSET(ucp_Zl);        SET_TYPE_OFFSET(ucp_Zl);
4975        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);
# Line 4235  while (*cc != XCL_END) Line 4978  while (*cc != XCL_END)
4978        break;        break;
4979    
4980        case PT_WORD:        case PT_WORD:
4981        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_UNDERSCORE - charoffset));
4982        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4983        /* Fall through. */        /* Fall through. */
4984    
# Line 4283  while (*cc != XCL_END) Line 5026  while (*cc != XCL_END)
5026          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]);
5027          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5028    
5029          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(other_cases[0] - charoffset));
5030          OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5031    
5032          other_cases += 3;          other_cases += 3;
5033          }          }
5034        else        else
5035          {          {
5036          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset));
5037          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5038          }          }
5039    
5040        while (*other_cases != NOTACHAR)        while (*other_cases != NOTACHAR)
5041          {          {
5042          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset));
5043          OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5044          }          }
5045        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5046        break;        break;
5047    
5048        case PT_UCNC:        case PT_UCNC:
5049        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_DOLLAR_SIGN - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_DOLLAR_SIGN - charoffset));
5050        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5051        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_COMMERCIAL_AT - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_COMMERCIAL_AT - charoffset));
5052        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5053        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_GRAVE_ACCENT - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_GRAVE_ACCENT - charoffset));
5054        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5055    
5056        SET_CHAR_OFFSET(0xa0);        SET_CHAR_OFFSET(0xa0);
5057        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd7ff - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(0xd7ff - charoffset));
5058        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5059        SET_CHAR_OFFSET(0);        SET_CHAR_OFFSET(0);
5060        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0);
5061        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL);
5062        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5063        break;        break;
5064    
5065          case PT_PXGRAPH:
5066          /* C and Z groups are the farthest two groups. */
5067          SET_TYPE_OFFSET(ucp_Ll);
5068          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5069          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5070    
5071          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5072    
5073          /* In case of ucp_Cf, we overwrite the result. */
5074          SET_CHAR_OFFSET(0x2066);
5075          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5076          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5077    
5078          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5079          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5080    
5081          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
5082          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5083    
5084          JUMPHERE(jump);
5085          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5086          break;
5087    
5088          case PT_PXPRINT:
5089          /* C and Z groups are the farthest two groups. */
5090          SET_TYPE_OFFSET(ucp_Ll);
5091          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5092          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5093    
5094          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
5095          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
5096    
5097          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5098    
5099          /* In case of ucp_Cf, we overwrite the result. */
5100          SET_CHAR_OFFSET(0x2066);
5101          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5102          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5103    
5104          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5105          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5106    
5107          JUMPHERE(jump);
5108          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5109          break;
5110    
5111          case PT_PXPUNCT:
5112          SET_TYPE_OFFSET(ucp_Sc);
5113          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc);
5114          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5115    
5116          SET_CHAR_OFFSET(0);
5117          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff);
5118          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5119    
5120          SET_TYPE_OFFSET(ucp_Pc);
5121          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc);
5122          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5123          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5124          break;
5125        }        }
5126      cc += 2;      cc += 2;
5127      }      }
# Line 4349  struct sljit_label *label; Line 5153  struct sljit_label *label;
5153  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5154  pcre_uchar propdata[5];  pcre_uchar propdata[5];
5155  #endif  #endif
5156  #endif  #endif /* SUPPORT_UTF */
5157    
5158  switch(type)  switch(type)
5159    {    {
# Line 4374  switch(type) Line 5178  switch(type)
5178    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
5179    case OP_DIGIT:    case OP_DIGIT:
5180    /* Digits are usually 0-9, so it is worth to optimize them. */    /* Digits are usually 0-9, so it is worth to optimize them. */
   if (common->digits[0] == -2)  
     get_ctype_ranges(common, ctype_digit, common->digits);  
5181    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5182    /* Flip the starting bit in the negative case. */  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5183    if (type == OP_NOT_DIGIT)    if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_digit, FALSE))
5184      common->digits[1] ^= 1;      read_char7_type(common, type == OP_NOT_DIGIT);
5185    if (!check_ranges(common, common->digits, backtracks, TRUE))    else
5186      {  #endif
5187      read_char8_type(common);      read_char8_type(common, type == OP_NOT_DIGIT);
5188      OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);      /* Flip the starting bit in the negative case. */
5189      add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);
5190      }    add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
   if (type == OP_NOT_DIGIT)  
     common->digits[1] ^= 1;  
5191    return cc;    return cc;
5192    
5193    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
5194    case OP_WHITESPACE:    case OP_WHITESPACE:
5195    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5196    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5197      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_space, FALSE))
5198        read_char7_type(common, type == OP_NOT_WHITESPACE);
5199      else
5200    #endif
5201        read_char8_type(common, type == OP_NOT_WHITESPACE);
5202    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space);
5203    add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
5204    return cc;    return cc;
# Line 4401  switch(type) Line 5206  switch(type)
5206    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
5207    case OP_WORDCHAR:    case OP_WORDCHAR:
5208    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5209    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5210      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_word, FALSE))
5211        read_char7_type(common, type == OP_NOT_WORDCHAR);
5212      else
5213    #endif
5214        read_char8_type(common, type == OP_NOT_WORDCHAR);
5215    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word);
5216    add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
5217    return cc;    return cc;
5218    
5219    case OP_ANY:    case OP_ANY:
5220    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5221    read_char(common);    read_char_range(common, common->nlmin, common->nlmax, TRUE);
5222    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
5223      {      {
5224      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);
# Line 4464  switch(type) Line 5274  switch(type)
5274  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5275    case OP_NOTPROP:    case OP_NOTPROP:
5276    case OP_PROP:    case OP_PROP:
5277    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
5278    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
5279    propdata[2] = cc[0];    propdata[2] = cc[0];
5280    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4476  switch(type) Line 5286  switch(type)
5286    
5287    case OP_ANYNL:    case OP_ANYNL:
5288    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5289    read_char(common);    read_char_range(common, common->bsr_nlmin, common->bsr_nlmax, FALSE);
5290    jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);    jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
5291    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
5292    end_list = NULL;    end_list = NULL;
# Line 4498  switch(type) Line 5308  switch(type)
5308    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
5309    case OP_HSPACE:    case OP_HSPACE:
5310    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5311    read_char(common);    read_char_range(common, 0x9, 0x3000, type == OP_NOT_HSPACE);
5312    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));
5313    add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
5314    return cc;    return cc;
# Line 4506  switch(type) Line 5316  switch(type)
5316    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
5317    case OP_VSPACE:    case OP_VSPACE:
5318    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5319    read_char(common);    read_char_range(common, 0xa, 0x2029, type == OP_NOT_VSPACE);
5320    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));
5321    add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
5322    return cc;    return cc;
# Line 4605  switch(type) Line 5415  switch(type)
5415      else      else
5416        {        {
5417        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);
5418        read_char(common);        read_char_range(common, common->nlmin, common->nlmax, TRUE);
5419        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0));        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0));
5420        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
5421        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4653  switch(type) Line 5463  switch(type)
5463    else    else
5464      {      {
5465      skip_char_back(common);      skip_char_back(common);
5466      read_char(common);      read_char_range(common, common->nlmin, common->nlmax, TRUE);
5467      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5468      }      }
5469    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4704  switch(type) Line 5514  switch(type)
5514      }      }
5515    else    else
5516      {      {
5517      peek_char(common);      peek_char(common, common->nlmax);
5518      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5519      }      }
5520    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4728  switch(type) Line 5538  switch(type)
5538  #endif  #endif
5539      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);
5540      }      }
5541    
5542    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
   read_char(common);  
5543  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5544    if (common->utf)    if (common->utf)
5545      {      {
# Line 4738  switch(type) Line 5548  switch(type)
5548    else    else
5549  #endif  #endif
5550      c = *cc;      c = *cc;
5551    
5552    if (type == OP_CHAR || !char_has_othercase(common, cc))    if (type == OP_CHAR || !char_has_othercase(common, cc))
5553      {      {
5554        read_char_range(common, c, c, FALSE);
5555      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c));      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c));
5556      return cc + length;      return cc + length;
5557      }      }
5558    oc = char_othercase(common, c);    oc = char_othercase(common, c);
5559      read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, FALSE);
5560    bit = c ^ oc;    bit = c ^ oc;
5561    if (is_powerof2(bit))    if (is_powerof2(bit))
5562      {      {
# Line 4751  switch(type) Line 5564  switch(type)
5564      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));
5565      return cc + length;      return cc + length;
5566      }      }
5567    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c);    jump[0] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c);
5568    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);    add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, oc));
5569    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc);    JUMPHERE(jump[0]);
   OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);  
   add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));  
5570    return cc + length;    return cc + length;
5571    
5572    case OP_NOT:    case OP_NOT:
# Line 4790  switch(type) Line 5601  switch(type)
5601  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
5602        {        {
5603        GETCHARLEN(c, cc, length);        GETCHARLEN(c, cc, length);
       read_char(common);  
5604        }        }
5605      }      }
5606    else    else
5607  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
     {  
     read_char(common);  
5608      c = *cc;      c = *cc;
     }  
5609    
5610    if (type == OP_NOT || !char_has_othercase(common, cc))    if (type == OP_NOT || !char_has_othercase(common, cc))
5611        {
5612        read_char_range(common, c, c, TRUE);
5613      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c));      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c));
5614        }
5615    else    else
5616      {      {
5617      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5618        read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, TRUE);
5619      bit = c ^ oc;      bit = c ^ oc;
5620      if (is_powerof2(bit))      if (is_powerof2(bit))
5621        {        {
# Line 4822  switch(type) Line 5633  switch(type)
5633    case OP_CLASS:    case OP_CLASS:
5634    case OP_NCLASS:    case OP_NCLASS:
5635    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5636    read_char(common);  
5637    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5638      bit = (common->utf && is_char7_bitset((const pcre_uint8 *)cc, type == OP_NCLASS)) ? 127 : 255;
5639      read_char_range(common, 0, bit, type == OP_NCLASS);
5640    #else
5641      read_char_range(common, 0, 255, type == OP_NCLASS);
5642    #endif
5643    
5644      if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5645      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5646    
5647  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5648    jump[0] = NULL;    jump[0] = NULL;
 #ifdef COMPILE_PCRE8  
   /* This check only affects 8 bit mode. In other modes, we  
   always need to compare the value with 255. */  
5649    if (common->utf)    if (common->utf)
 #endif /* COMPILE_PCRE8 */  
5650      {      {
5651      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, bit);
5652      if (type == OP_CLASS)      if (type == OP_CLASS)
5653        {        {
5654        add_jump(compiler, backtracks, jump[0]);        add_jump(compiler, backtracks, jump[0]);
5655        jump[0] = NULL;        jump[0] = NULL;
5656        }        }
5657      }      }
5658  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #elif !defined COMPILE_PCRE8
5659      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
5660      if (type == OP_CLASS)
5661        {
5662        add_jump(compiler, backtracks, jump[0]);
5663        jump[0] = NULL;
5664        }
5665    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
5666    
5667    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5668    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5669    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5670    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5671    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
5672    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
5673    
5674  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5675    if (jump[0] != NULL)    if (jump[0] != NULL)
5676      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5677  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif
5678    
5679    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5680    
5681  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 4956  if (context.length > 0) Line 5780  if (context.length > 0)
5780  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);
5781  }  }
5782    
 static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)  
 {  
 DEFINE_COMPILER;  
 int offset = GET2(cc, 1) << 1;  
   
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  
 if (!common->jscript_compat)  
   {  
   if (backtracks == NULL)  
     {  
     /* OVECTOR(1) contains the "string begin - 1" constant. */  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));  
     OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
     OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);  
     return JUMP(SLJIT_C_NOT_ZERO);  
     }  
   add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));  
   }  
 return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
 }  
   
5783  /* Forward definitions. */  /* Forward definitions. */
5784  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);
5785  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);
# Line 5010  static void compile_backtrackingpath(com Line 5812  static void compile_backtrackingpath(com
5812    
5813  #define BACKTRACK_AS(type) ((type *)backtrack)  #define BACKTRACK_AS(type) ((type *)backtrack)
5814    
5815  static pcre_uchar *compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)  static void compile_dnref_search(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)
5816    {
5817    /* The OVECTOR offset goes to TMP2. */
5818    DEFINE_COMPILER;
5819    int count = GET2(cc, 1 + IMM2_SIZE);
5820    pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
5821    unsigned int offset;
5822    jump_list *found = NULL;
5823    
5824    SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
5825    
5826    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
5827    
5828    count--;
5829    while (count-- > 0)
5830      {
5831      offset = GET2(slot, 0) << 1;
5832      GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5833      add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5834      slot += common->name_entry_size;
5835      }
5836    
5837    offset = GET2(slot, 0) << 1;
5838    GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5839    if (backtracks != NULL && !common->jscript_compat)
5840      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5841    
5842    set_jumps(found, LABEL());
5843    }
5844    
5845    static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
5846  {  {
5847  DEFINE_COMPILER;  DEFINE_COMPILER;
5848  int offset = GET2(cc, 1) << 1;  BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5849    int offset = 0;
5850  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
5851  struct sljit_jump *partial;  struct sljit_jump *partial;
5852  struct sljit_jump *nopartial;  struct sljit_jump *nopartial;
5853    
5854  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  if (ref)
5855  /* OVECTOR(1) contains the "string begin - 1" constant. */    {
5856  if (withchecks && !common->jscript_compat)    offset = GET2(cc, 1) << 1;
5857    add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5858      /* OVECTOR(1) contains the "string begin - 1" constant. */
5859      if (withchecks && !common->jscript_compat)
5860        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5861      }
5862    else
5863      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5864    
5865  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
5866  if (common->utf && *cc == OP_REFI)  if (common->utf && *cc == OP_REFI)
5867    {    {
5868    SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3);    SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3);
5869    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));    if (ref)
5870        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5871      else
5872        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5873    
5874    if (withchecks)    if (withchecks)
5875      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);
5876    
# Line 5052  if (common->utf && *cc == OP_REFI) Line 5895  if (common->utf && *cc == OP_REFI)
5895  else  else
5896  #endif /* SUPPORT_UTF && SUPPORT_UCP */  #endif /* SUPPORT_UTF && SUPPORT_UCP */
5897    {    {
5898    OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);    if (ref)
5899        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);
5900      else
5901        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
5902    
5903    if (withchecks)    if (withchecks)
5904      jump = JUMP(SLJIT_C_ZERO);      jump = JUMP(SLJIT_C_ZERO);
5905    
# Line 5089  if (jump != NULL) Line 5936  if (jump != NULL)
5936    else    else
5937      JUMPHERE(jump);      JUMPHERE(jump);
5938    }    }
 return cc + 1 + IMM2_SIZE;  
5939  }  }
5940    
5941  static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent)  static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent)
5942  {  {
5943  DEFINE_COMPILER;  DEFINE_COMPILER;
5944    BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5945  backtrack_common *backtrack;  backtrack_common *backtrack;
5946  pcre_uchar type;  pcre_uchar type;
5947    int offset = 0;
5948  struct sljit_label *label;  struct sljit_label *label;
5949  struct sljit_jump *zerolength;  struct sljit_jump *zerolength;
5950  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
# Line 5106  BOOL minimize; Line 5954  BOOL minimize;
5954    
5955  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);
5956    
5957    if (ref)
5958      offset = GET2(cc, 1) << 1;
5959    else
5960      cc += IMM2_SIZE;
5961  type = cc[1 + IMM2_SIZE];  type = cc[1 + IMM2_SIZE];
5962    
5963    SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even);
5964  minimize = (type & 0x1) != 0;  minimize = (type & 0x1) != 0;
5965  switch(type)  switch(type)
5966    {    {
# Line 5144  if (!minimize) Line 5998  if (!minimize)
5998    if (min == 0)    if (min == 0)
5999      {      {
6000      allocate_stack(common, 2);      allocate_stack(common, 2);
6001        if (ref)
6002          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6003      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6004      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
6005      /* Temporary release of STR_PTR. */      /* Temporary release of STR_PTR. */
6006      OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));      OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
6007      zerolength = compile_ref_checks(common, ccbegin, NULL);      /* Handles both invalid and empty cases. Since the minimum repeat,
6008        is zero the invalid case is basically the same as an empty case. */
6009        if (ref)
6010          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6011        else
6012          {
6013          compile_dnref_search(common, ccbegin, NULL);
6014          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6015          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
6016          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6017          }
6018      /* Restore if not zero length. */      /* Restore if not zero length. */
6019      OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));      OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
6020      }      }
6021    else    else
6022      {      {
6023      allocate_stack(common, 1);      allocate_stack(common, 1);
6024        if (ref)
6025          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6026      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
6027      zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);      if (ref)
6028          {
6029          add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
6030          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6031          }
6032        else
6033          {
6034          compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
6035          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6036          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
6037          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6038          }
6039      }      }
6040    
6041    if (min > 1 || max > 1)    if (min > 1 || max > 1)
6042      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);
6043    
6044    label = LABEL();    label = LABEL();
6045      if (!ref)
6046        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1);
6047    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);
6048    
6049    if (min > 1 || max > 1)    if (min > 1 || max > 1)
# Line 5193  if (!minimize) Line 6074  if (!minimize)
6074    JUMPHERE(zerolength);    JUMPHERE(zerolength);
6075    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
6076    
6077    decrease_call_count(common);    count_match(common);
6078    return cc;    return cc;
6079    }    }
6080    
6081  allocate_stack(common, 2);  allocate_stack(common, ref ? 2 : 3);
6082    if (ref)
6083      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6084  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
6085  if (type != OP_CRMINSTAR)  if (type != OP_CRMINSTAR)
6086    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
6087    
6088  if (min == 0)  if (min == 0)
6089    {    {
6090    zerolength = compile_ref_checks(common, ccbegin, NULL);    /* Handles both invalid and empty cases. Since the minimum repeat,
6091      is zero the invalid case is basically the same as an empty case. */
6092      if (ref)
6093        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6094      else
6095        {
6096        compile_dnref_search(common, ccbegin, NULL);
6097        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6098        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
6099        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6100        }
6101      /* Length is non-zero, we can match real repeats. */
6102    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6103    jump = JUMP(SLJIT_JUMP);    jump = JUMP(SLJIT_JUMP);
6104    }    }
6105  else  else
6106    zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);    {
6107      if (ref)
6108        {
6109        add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
6110        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6111        }
6112      else
6113        {
6114        compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
6115        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6116        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
6117        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6118        }
6119      }
6120    
6121  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
6122  if (max > 0)  if (max > 0)
6123    add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max));    add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max));
6124    
6125    if (!ref)
6126      OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
6127  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);
6128  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6129    
# Line 5232  if (jump != NULL) Line 6141  if (jump != NULL)
6141    JUMPHERE(jump);    JUMPHERE(jump);
6142  JUMPHERE(zerolength);  JUMPHERE(zerolength);
6143    
6144  decrease_call_count(common);  count_match(common);
6145  return cc;  return cc;
6146  }  }
6147    
# Line 5242  DEFINE_COMPILER; Line 6151  DEFINE_COMPILER;
6151  backtrack_common *backtrack;  backtrack_common *backtrack;
6152  recurse_entry *entry = common->entries;  recurse_entry *entry = common->entries;
6153  recurse_entry *prev = NULL;  recurse_entry *prev = NULL;
6154  int start = GET(cc, 1);  sljit_sw start = GET(cc, 1);
6155  pcre_uchar *start_cc;  pcre_uchar *start_cc;
6156    BOOL needs_control_head;
6157    
6158  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);
6159    
6160  /* Inlining simple patterns. */  /* Inlining simple patterns. */
6161  if (get_framesize(common, common->start + start, TRUE) == no_stack)  if (get_framesize(common, common->start + start, NULL, TRUE, &needs_control_head) == no_stack)
6162    {    {
6163    start_cc = common->start + start;    start_cc = common->start + start;
6164    compile_matchingpath(common, next_opcode(common, start_cc), bracketend(start_cc) - (1 + LINK_SIZE), backtrack);    compile_matchingpath(common, next_opcode(common, start_cc), bracketend(start_cc) - (1 + LINK_SIZE), backtrack);
# Line 5408  static pcre_uchar *compile_assert_matchi Line 6318  static pcre_uchar *compile_assert_matchi
6318  DEFINE_COMPILER;  DEFINE_COMPILER;
6319  int framesize;  int framesize;
6320  int extrasize;  int extrasize;
6321  BOOL needs_control_head = common->control_head_ptr != 0;  BOOL needs_control_head;
6322  int private_data_ptr;  int private_data_ptr;
6323  backtrack_common altbacktrack;  backtrack_common altbacktrack;
6324  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
# Line 5418  jump_list *tmp = NULL; Line 6328  jump_list *tmp = NULL;
6328  jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks;  jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks;
6329  jump_list **found;  jump_list **found;
6330  /* Saving previous accept variables. */  /* Saving previous accept variables. */
6331    BOOL save_local_exit = common->local_exit;
6332    BOOL save_positive_assert = common->positive_assert;
6333    then_trap_backtrack *save_then_trap = common->then_trap;
6334  struct sljit_label *save_quit_label = common->quit_label;  struct sljit_label *save_quit_label = common->quit_label;
6335  struct sljit_label *save_accept_label = common->accept_label;  struct sljit_label *save_accept_label = common->accept_label;
6336  jump_list *save_quit = common->quit;  jump_list *save_quit = common->quit;
6337    jump_list *save_positive_assert_quit = common->positive_assert_quit;
6338  jump_list *save_accept = common->accept;  jump_list *save_accept = common->accept;
 BOOL save_local_exit = common->local_exit;  
6339  struct sljit_jump *jump;  struct sljit_jump *jump;
6340  struct sljit_jump *brajump = NULL;  struct sljit_jump *brajump = NULL;
6341    
6342    /* Assert captures then. */
6343    common->then_trap = NULL;
6344    
6345  if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)  if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)
6346    {    {
6347    SLJIT_ASSERT(!conditional);    SLJIT_ASSERT(!conditional);
# Line 5434  if (*cc == OP_BRAZERO || *cc == OP_BRAMI Line 6350  if (*cc == OP_BRAZERO || *cc == OP_BRAMI
6350    }    }
6351  private_data_ptr = PRIVATE_DATA(cc);  private_data_ptr = PRIVATE_DATA(cc);
6352  SLJIT_ASSERT(private_data_ptr != 0);  SLJIT_ASSERT(private_data_ptr != 0);
6353  framesize = get_framesize(common, cc, FALSE);  framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head);
6354  backtrack->framesize = framesize;  backtrack->framesize = framesize;
6355  backtrack->private_data_ptr = private_data_ptr;  backtrack->private_data_ptr = private_data_ptr;
6356  opcode = *cc;  opcode = *cc;
# Line 5454  if (bra == OP_BRAMINZERO) Line 6370  if (bra == OP_BRAMINZERO)
6370  if (framesize < 0)  if (framesize < 0)
6371    {    {
6372    extrasize = needs_control_head ? 2 : 1;    extrasize = needs_control_head ? 2 : 1;
6373    if (framesize != no_stack)    if (framesize == no_frame)
6374      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0);
6375    allocate_stack(common, extrasize);    allocate_stack(common, extrasize);
6376    if (needs_control_head)    if (needs_control_head)
# Line 5484  else Line 6400  else
6400      }      }
6401    else    else
6402      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);
6403    init_frame(common, ccbegin, framesize + extrasize - 1, extrasize, FALSE);    init_frame(common, ccbegin, NULL, framesize + extrasize - 1, extrasize, FALSE);
6404    }    }
6405    
6406  memset(&altbacktrack, 0, sizeof(backtrack_common));  memset(&altbacktrack, 0, sizeof(backtrack_common));
6407  common->local_exit = TRUE;  if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6408  common->quit_label = NULL;    {
6409  common->quit = NULL;    /* Negative assert is stronger than positive assert. */
6410      common->local_exit = TRUE;
6411      common->quit_label = NULL;
6412      common->quit = NULL;
6413      common->positive_assert = FALSE;
6414      }
6415    else
6416      common->positive_assert = TRUE;
6417    common->positive_assert_quit = NULL;
6418    
6419  while (1)  while (1)
6420    {    {
6421    common->accept_label = NULL;    common->accept_label = NULL;
# Line 5505  while (1) Line 6430  while (1)
6430    compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack);    compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack);
6431    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
6432      {      {
6433      common->local_exit = save_local_exit;      if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6434      common->quit_label = save_quit_label;        {
6435          common->local_exit = save_local_exit;
6436          common->quit_label = save_quit_label;
6437          common->quit = save_quit;
6438          }
6439        common->positive_assert = save_positive_assert;
6440        common->then_trap = save_then_trap;
6441      common->accept_label = save_accept_label;      common->accept_label = save_accept_label;
6442      common->quit = save_quit;      common->positive_assert_quit = save_positive_assert_quit;
6443      common->accept = save_accept;      common->accept = save_accept;
6444      return NULL;      return NULL;
6445      }      }
# Line 5519  while (1) Line 6450  while (1)
6450    /* Reset stack. */    /* Reset stack. */
6451    if (framesize < 0)    if (framesize < 0)
6452      {      {
6453      if (framesize != no_stack)      if (framesize == no_frame)
6454        OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr);        OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr);
6455      else      else
6456        free_stack(common, extrasize);        free_stack(common, extrasize);
# Line 5573  while (1) Line 6504  while (1)
6504    compile_backtrackingpath(common, altbacktrack.top);    compile_backtrackingpath(common, altbacktrack.top);
6505    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
6506      {      {
6507      common->local_exit = save_local_exit;      if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6508      common->quit_label = save_quit_label;        {
6509          common->local_exit = save_local_exit;
6510          common->quit_label = save_quit_label;
6511          common->quit = save_quit;
6512          }
6513        common->positive_assert = save_positive_assert;
6514        common->then_trap = save_then_trap;
6515      common->accept_label = save_accept_label;      common->accept_label = save_accept_label;
6516      common->quit = save_quit;      common->positive_assert_quit = save_positive_assert_quit;
6517      common->accept = save_accept;      common->accept = save_accept;
6518      return NULL;      return NULL;
6519      }      }
# Line 5589  while (1) Line 6526  while (1)
6526    cc += GET(cc, 1);    cc += GET(cc, 1);
6527    }    }
6528    
6529    if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6530      {
6531      SLJIT_ASSERT(common->positive_assert_quit == NULL);
6532      /* Makes the check less complicated below. */
6533      common->positive_assert_quit = common->quit;
6534      }
6535    
6536  /* None of them matched. */  /* None of them matched. */
6537  if (common->quit != NULL)  if (common->positive_assert_quit != NULL)
6538    {    {
6539    jump = JUMP(SLJIT_JUMP);    jump = JUMP(SLJIT_JUMP);
6540    set_jumps(common->quit, LABEL());    set_jumps(common->positive_assert_quit, LABEL());
6541    SLJIT_ASSERT(framesize != no_stack);    SLJIT_ASSERT(framesize != no_stack);
6542    if (framesize < 0)    if (framesize < 0)
6543      OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, extrasize * sizeof(sljit_sw));      OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, extrasize * sizeof(sljit_sw));
# Line 5753  else Line 6697  else
6697      }      }
6698    }    }
6699    
6700  common->local_exit = save_local_exit;  if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6701  common->quit_label = save_quit_label;    {
6702      common->local_exit = save_local_exit;
6703      common->quit_label = save_quit_label;
6704      common->quit = save_quit;
6705      }
6706    common->positive_assert = save_positive_assert;
6707    common->then_trap = save_then_trap;
6708  common->accept_label = save_accept_label;  common->accept_label = save_accept_label;
6709  common->quit = save_quit;  common->positive_assert_quit = save_positive_assert_quit;
6710  common->accept = save_accept;  common->accept = save_accept;
6711  return cc + 1 + LINK_SIZE;  return cc + 1 + LINK_SIZE;
6712  }  }
6713    
6714  static sljit_sw SLJIT_CALL do_searchovector(sljit_uw refno, sljit_sw* locals, pcre_uchar *name_table)  static SLJIT_INLINE void match_once_common(compiler_common *common, pcre_uchar ket, int framesize, int private_data_ptr, BOOL has_alternatives, BOOL needs_control_head)
6715  {  {
6716  int condition = FALSE;  DEFINE_COMPILER;
6717  pcre_uchar *slotA = name_table;  int stacksize;
 pcre_uchar *slotB;  
 sljit_sw name_count = locals[LOCALS0 / sizeof(sljit_sw)];  
 sljit_sw name_entry_size = locals[LOCALS1 / sizeof(sljit_sw)];  
 sljit_sw no_capture;  
 int i;  
   
 locals += refno & 0xff;  
 refno >>= 8;  
 no_capture = locals[1];  
   
 for (i = 0; i < name_count; i++)  
   {  
   if (GET2(slotA, 0) == refno) break;  
   slotA += name_entry_size;  
   }  
6718    
6719  if (i < name_count)  if (framesize < 0)
6720    {    {
6721    /* Found a name for the number - there can be only one; duplicate names    if (framesize == no_frame)
6722    for different numbers are allowed, but not vice versa. First scan down      OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr);
6723    for duplicates. */    else
   
   slotB = slotA;  
   while (slotB > name_table)  
6724      {      {
6725      slotB -= name_entry_size;      stacksize = needs_control_head ? 1 : 0;
6726      if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)      if (ket != OP_KET || has_alternatives)
6727        {        stacksize++;
6728        condition = locals[GET2(slotB, 0) << 1] != no_capture;      free_stack(common, stacksize);
       if (condition) break;  
       }  
     else break;  
6729      }      }
6730    
6731    /* Scan up for duplicates */    if (needs_control_head)
6732    if (!condition)      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), (ket != OP_KET || has_alternatives) ? sizeof(sljit_sw) : 0);
6733    
6734      /* TMP2 which is set here used by OP_KETRMAX below. */
6735      if (ket == OP_KETRMAX)
6736        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), 0);
6737      else if (ket == OP_KETRMIN)
6738      {      {
6739      slotB = slotA;      /* Move the STR_PTR to the private_data_ptr. */
6740      for (i++; i < name_count; i++)      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), 0);
       {  
       slotB += name_entry_size;  
       if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0)  
         {  
         condition = locals[GET2(slotB, 0) << 1] != no_capture;  
         if (condition) break;  
         }  
       else break;  
       }  
6741      }      }
6742    }    }
6743  return condition;  else
 }  
   
 static sljit_sw SLJIT_CALL do_searchgroups(sljit_uw recno, sljit_uw* locals, pcre_uchar *name_table)  
 {  
 int condition = FALSE;  
 pcre_uchar *slotA = name_table;  
 pcre_uchar *slotB;  
 sljit_uw name_count = locals[LOCALS0 / sizeof(sljit_sw)];  
 sljit_uw name_entry_size = locals[LOCALS1 / sizeof(sljit_sw)];  
 sljit_uw group_num = locals[POSSESSIVE0 / sizeof(sljit_sw)];  
 sljit_uw i;  
   
 for (i = 0; i < name_count; i++)  
   {  
   if (GET2(slotA, 0) == recno) break;  
   slotA += name_entry_size;  
   }  
   
 if (i < name_count)