/[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 1448 by zherczeg, Mon Jan 13 21:25:53 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  #define MAX_N_CHARS 16
3153    #define MAX_N_BYTES 8
3154    
3155  static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)  static SLJIT_INLINE void add_prefix_byte(pcre_uint8 byte, pcre_uint8 *bytes)
3156  {  {
3157  DEFINE_COMPILER;  pcre_uint8 len = bytes[0];
3158  struct sljit_label *start;  int i;
 struct sljit_jump *quit;  
 pcre_uint32 chars[MAX_N_CHARS * 2];  
 pcre_uchar *cc = common->start + 1 + LINK_SIZE;  
 int location = 0;  
 pcre_int32 len, c, bit, caseless;  
 int must_stop;  
3159    
3160  /* We do not support alternatives now. */  if (len == 255)
3161  if (*(common->start + GET(common->start, 1)) == OP_ALT)    return;
3162    return FALSE;  
3163    if (len == 0)
3164      {
3165      bytes[0] = 1;
3166      bytes[1] = byte;
3167      return;
3168      }
3169    
3170    for (i = len; i > 0; i--)
3171      if (bytes[i] == byte)
3172        return;
3173    
3174    if (len >= MAX_N_BYTES - 1)
3175      {
3176      bytes[0] = 255;
3177      return;
3178      }
3179    
3180    len++;
3181    bytes[len] = byte;
3182    bytes[0] = len;
3183    }
3184    
3185    static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, pcre_uint8 *bytes, int max_chars)
3186    {
3187    /* Recursive function, which scans prefix literals. */
3188    BOOL last, any, caseless;
3189    int len, repeat, len_save, consumed = 0;
3190    pcre_uint32 chr, mask;
3191    pcre_uchar *alternative, *cc_save, *oc;
3192    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3193    pcre_uchar othercase[8];
3194    #elif defined SUPPORT_UTF && defined COMPILE_PCRE16
3195    pcre_uchar othercase[2];
3196    #else
3197    pcre_uchar othercase[1];
3198    #endif
3199    
3200    repeat = 1;
3201  while (TRUE)  while (TRUE)
3202    {    {
3203    caseless = 0;    last = TRUE;
3204    must_stop = 1;    any = FALSE;
3205    switch(*cc)    caseless = FALSE;
3206      switch (*cc)
3207      {      {
     case OP_CHAR:  
     must_stop = 0;  
     cc++;  
     break;  
   
3208      case OP_CHARI:      case OP_CHARI:
3209      caseless = 1;      caseless = TRUE;
3210      must_stop = 0;      case OP_CHAR:
3211        last = FALSE;
3212      cc++;      cc++;
3213      break;      break;
3214    
# Line 2847  while (TRUE) Line 3227  while (TRUE)
3227      cc++;      cc++;
3228      continue;      continue;
3229    
3230        case OP_ASSERT:
3231        case OP_ASSERT_NOT:
3232        case OP_ASSERTBACK:
3233        case OP_ASSERTBACK_NOT:
3234        cc = bracketend(cc);
3235        continue;
3236    
3237        case OP_PLUSI:
3238        case OP_MINPLUSI:
3239        case OP_POSPLUSI:
3240        caseless = TRUE;
3241      case OP_PLUS:      case OP_PLUS:
3242      case OP_MINPLUS:      case OP_MINPLUS:
3243      case OP_POSPLUS:      case OP_POSPLUS:
3244      cc++;      cc++;
3245      break;      break;
3246    
3247        case OP_EXACTI:
3248        caseless = TRUE;
3249      case OP_EXACT:      case OP_EXACT:
3250        repeat = GET2(cc, 1);
3251        last = FALSE;
3252      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3253      break;      break;
3254    
3255      case OP_PLUSI:      case OP_QUERYI:
3256      case OP_MINPLUSI:      case OP_MINQUERYI:
3257      case OP_POSPLUSI:      case OP_POSQUERYI:
3258      caseless = 1;      caseless = TRUE;
3259        case OP_QUERY:
3260        case OP_MINQUERY:
3261        case OP_POSQUERY:
3262        len = 1;
3263      cc++;      cc++;
3264    #ifdef SUPPORT_UTF
3265        if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3266    #endif
3267        max_chars = scan_prefix(common, cc + len, chars, bytes, max_chars);
3268        if (max_chars == 0)
3269          return consumed;
3270        last = FALSE;
3271      break;      break;
3272    
3273      case OP_EXACTI:      case OP_KET:
3274      caseless = 1;      cc += 1 + LINK_SIZE;
3275        continue;
3276    
3277        case OP_ALT:
3278        cc += GET(cc, 1);
3279        continue;
3280    
3281        case OP_ONCE:
3282        case OP_ONCE_NC:
3283        case OP_BRA:
3284        case OP_BRAPOS:
3285        case OP_CBRA:
3286        case OP_CBRAPOS:
3287        alternative = cc + GET(cc, 1);
3288        while (*alternative == OP_ALT)
3289          {
3290          max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, bytes, max_chars);
3291          if (max_chars == 0)
3292            return consumed;
3293          alternative += GET(alternative, 1);
3294          }
3295    
3296        if (*cc == OP_CBRA || *cc == OP_CBRAPOS)
3297          cc += IMM2_SIZE;
3298        cc += 1 + LINK_SIZE;
3299        continue;
3300    
3301        case OP_CLASS:
3302    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3303        if (common->utf && !is_char7_bitset((const pcre_uint8 *)(cc + 1), FALSE)) return consumed;
3304    #endif
3305        any = TRUE;
3306        cc += 1 + 32 / sizeof(pcre_uchar);
3307        break;
3308    
3309        case OP_NCLASS:
3310    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3311        if (common->utf) return consumed;
3312    #endif
3313        any = TRUE;
3314        cc += 1 + 32 / sizeof(pcre_uchar);
3315        break;
3316    
3317    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3318        case OP_XCLASS:
3319    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3320        if (common->utf) return consumed;
3321    #endif
3322        any = TRUE;
3323        cc += GET(cc, 1);
3324        break;
3325    #endif
3326    
3327        case OP_DIGIT:
3328    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3329        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_digit, FALSE))
3330          return consumed;
3331    #endif
3332        any = TRUE;
3333        cc++;
3334        break;
3335    
3336        case OP_WHITESPACE:
3337    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3338        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_space, FALSE))
3339          return consumed;
3340    #endif
3341        any = TRUE;
3342        cc++;
3343        break;
3344    
3345        case OP_WORDCHAR:
3346    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3347        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_word, FALSE))
3348          return consumed;
3349    #endif
3350        any = TRUE;
3351        cc++;
3352        break;
3353    
3354        case OP_NOT:
3355        case OP_NOTI:
3356        cc++;
3357        /* Fall through. */
3358        case OP_NOT_DIGIT:
3359        case OP_NOT_WHITESPACE:
3360        case OP_NOT_WORDCHAR:
3361        case OP_ANY:
3362        case OP_ALLANY:
3363    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3364        if (common->utf) return consumed;
3365    #endif
3366        any = TRUE;
3367        cc++;
3368        break;
3369    
3370    #ifdef SUPPORT_UCP
3371        case OP_NOTPROP:
3372        case OP_PROP:
3373    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3374        if (common->utf) return consumed;
3375    #endif
3376        any = TRUE;
3377        cc += 1 + 2;
3378        break;
3379    #endif
3380    
3381        case OP_TYPEEXACT:
3382        repeat = GET2(cc, 1);
3383      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3384        continue;
3385    
3386        case OP_NOTEXACT:
3387        case OP_NOTEXACTI:
3388    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3389        if (common->utf) return consumed;
3390    #endif
3391        any = TRUE;
3392        repeat = GET2(cc, 1);
3393        cc += 1 + IMM2_SIZE + 1;
3394      break;      break;
3395    
3396      default:      default:
3397      must_stop = 2;      return consumed;
     break;  
3398      }      }
3399    
3400    if (must_stop == 2)    if (any)
3401        {
3402    #if defined COMPILE_PCRE8
3403        mask = 0xff;
3404    #elif defined COMPILE_PCRE16
3405        mask = 0xffff;
3406    #elif defined COMPILE_PCRE32
3407        mask = 0xffffffff;
3408    #else
3409        SLJIT_ASSERT_STOP();
3410    #endif
3411    
3412        do
3413          {
3414          chars[0] = mask;
3415          chars[1] = mask;
3416          bytes[0] = 255;
3417    
3418          consumed++;
3419          if (--max_chars == 0)
3420            return consumed;
3421          chars += 2;
3422          bytes += MAX_N_BYTES;
3423          }
3424        while (--repeat > 0);
3425    
3426        repeat = 1;
3427        continue;
3428        }
3429    
3430      len = 1;
3431    #ifdef SUPPORT_UTF
3432      if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3433    #endif
3434    
3435      if (caseless && char_has_othercase(common, cc))
3436        {
3437    #ifdef SUPPORT_UTF
3438        if (common->utf)
3439          {
3440          GETCHAR(chr, cc);
3441          if ((int)PRIV(ord2utf)(char_othercase(common, chr), othercase) != len)
3442            return consumed;
3443          }
3444        else
3445    #endif
3446          {
3447          chr = *cc;
3448          othercase[0] = TABLE_GET(chr, common->fcc, chr);
3449          }
3450        }
3451      else
3452        caseless = FALSE;
3453    
3454      len_save = len;
3455      cc_save = cc;
3456      while (TRUE)
3457        {
3458        oc = othercase;
3459        do
3460          {
3461          chr = *cc;
3462    #ifdef COMPILE_PCRE32
3463          if (SLJIT_UNLIKELY(chr == NOTACHAR))
3464            return consumed;
3465    #endif
3466          add_prefix_byte((pcre_uint8)chr, bytes);
3467    
3468          mask = 0;
3469          if (caseless)
3470            {
3471            add_prefix_byte((pcre_uint8)*oc, bytes);
3472            mask = *cc ^ *oc;
3473            chr |= mask;
3474            }
3475    
3476    #ifdef COMPILE_PCRE32
3477          if (chars[0] == NOTACHAR && chars[1] == 0)
3478    #else
3479          if (chars[0] == NOTACHAR)
3480    #endif
3481            {
3482            chars[0] = chr;
3483            chars[1] = mask;
3484            }
3485          else
3486            {
3487            mask |= chars[0] ^ chr;
3488            chr |= mask;
3489            chars[0] = chr;
3490            chars[1] |= mask;
3491            }
3492    
3493          len--;
3494          consumed++;
3495          if (--max_chars == 0)
3496            return consumed;
3497          chars += 2;
3498          bytes += MAX_N_BYTES;
3499          cc++;
3500          oc++;
3501          }
3502        while (len > 0);
3503    
3504        if (--repeat == 0)
3505          break;
3506    
3507        len = len_save;
3508        cc = cc_save;
3509        }
3510    
3511      repeat = 1;
3512      if (last)
3513        return consumed;
3514      }
3515    }
3516    
3517    static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)
3518    {
3519    DEFINE_COMPILER;
3520    struct sljit_label *start;
3521    struct sljit_jump *quit;
3522    pcre_uint32 chars[MAX_N_CHARS * 2];
3523    pcre_uint8 bytes[MAX_N_CHARS * MAX_N_BYTES];
3524    pcre_uint8 ones[MAX_N_CHARS];
3525    int offsets[3];
3526    pcre_uint32 mask;
3527    pcre_uint8 *byte_set, *byte_set_end;
3528    int i, max, from;
3529    int range_right = -1, range_len = 3 - 1;
3530    sljit_ub *update_table = NULL;
3531    BOOL in_range;
3532    
3533    /* This is even TRUE, if both are NULL. */
3534    SLJIT_ASSERT(common->read_only_data_ptr == common->read_only_data);
3535    
3536    for (i = 0; i < MAX_N_CHARS; i++)
3537      {
3538      chars[i << 1] = NOTACHAR;
3539      chars[(i << 1) + 1] = 0;
3540      bytes[i * MAX_N_BYTES] = 0;
3541      }
3542    
3543    max = scan_prefix(common, common->start, chars, bytes, MAX_N_CHARS);
3544    
3545    if (max <= 1)
3546      return FALSE;
3547    
3548    for (i = 0; i < max; i++)
3549      {
3550      mask = chars[(i << 1) + 1];
3551      ones[i] = ones_in_half_byte[mask & 0xf];
3552      mask >>= 4;
3553      while (mask != 0)
3554        {
3555        ones[i] += ones_in_half_byte[mask & 0xf];
3556        mask >>= 4;
3557        }
3558      }
3559    
3560    in_range = FALSE;
3561    for (i = 0; i <= max; i++)
3562      {
3563      if (in_range && (i - from) > range_len && (bytes[(i - 1) * MAX_N_BYTES] <= 4))
3564        {
3565        range_len = i - from;
3566        range_right = i - 1;
3567        }
3568    
3569      if (i < max && bytes[i * MAX_N_BYTES] < 255)
3570        {
3571        if (!in_range)
3572          {
3573          in_range = TRUE;
3574          from = i;
3575          }
3576        }
3577      else if (in_range)
3578        in_range = FALSE;
3579      }
3580    
3581    if (range_right >= 0)
3582      {
3583      /* Since no data is consumed (see the assert in the beginning
3584      of this function), this space can be reallocated. */
3585      if (common->read_only_data)
3586        SLJIT_FREE(common->read_only_data);
3587    
3588      common->read_only_data_size += 256;
3589      common->read_only_data = (sljit_uw *)SLJIT_MALLOC(common->read_only_data_size);
3590      if (common->read_only_data == NULL)
3591        return TRUE;
3592    
3593      update_table = (sljit_ub *)common->read_only_data;
3594      common->read_only_data_ptr = (sljit_uw *)(update_table + 256);
3595      memset(update_table, IN_UCHARS(range_len), 256);
3596    
3597      for (i = 0; i < range_len; i++)
3598        {
3599        byte_set = bytes + ((range_right - i) * MAX_N_BYTES);
3600        SLJIT_ASSERT(byte_set[0] > 0 && byte_set[0] < 255);
3601        byte_set_end = byte_set + byte_set[0];
3602        byte_set++;
3603        while (byte_set <= byte_set_end)
3604          {
3605          if (update_table[*byte_set] > IN_UCHARS(i))
3606            update_table[*byte_set] = IN_UCHARS(i);
3607          byte_set++;
3608          }
3609        }
3610      }
3611    
3612    offsets[0] = -1;
3613    /* Scan forward. */
3614    for (i = 0; i < max; i++)
3615      if (ones[i] <= 2) {
3616        offsets[0] = i;
3617        break;
3618      }
3619    
3620    if (offsets[0] < 0 && range_right < 0)
3621      return FALSE;
3622    
3623    if (offsets[0] >= 0)
3624      {
3625      /* Scan backward. */
3626      offsets[1] = -1;
3627      for (i = max - 1; i > offsets[0]; i--)
3628        if (ones[i] <= 2 && i != range_right)
3629          {
3630          offsets[1] = i;
3631        break;        break;
3632          }
3633    
3634    len = 1;    /* This case is handled better by fast_forward_first_char. */
3635  #ifdef SUPPORT_UTF    if (offsets[1] == -1 && offsets[0] == 0 && range_right < 0)
3636    if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);      return FALSE;
 #endif  
3637    
3638    if (caseless && char_has_othercase(common, cc))    offsets[2] = -1;
3639      {    /* We only search for a middle character if there is no range check. */
3640      caseless = char_get_othercase_bit(common, cc);    if (offsets[1] >= 0 && range_right == -1)
3641      if (caseless == 0)      {
3642        return FALSE;      /* Scan from middle. */
3643  #ifdef COMPILE_PCRE8      for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++)
3644      caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));        if (ones[i] <= 2)
3645  #else          {
3646      if ((caseless & 0x100) != 0)          offsets[2] = i;
3647        caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));          break;
3648      else          }
       caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));  
 #endif  
     }  
   else  
     caseless = 0;  
3649    
3650    while (len > 0 && location < MAX_N_CHARS * 2)      if (offsets[2] == -1)
     {  
     c = *cc;  
     bit = 0;  
     if (len == (caseless & 0xff))  
3651        {        {
3652        bit = caseless >> 8;        for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--)
3653        c |= bit;          if (ones[i] <= 2)
3654              {
3655              offsets[2] = i;
3656              break;
3657              }
3658        }        }
3659        }
3660    
3661      chars[location] = c;    SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1]));
3662      chars[location + 1] = bit;    SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2]));
3663    
3664      len--;    chars[0] = chars[offsets[0] << 1];
3665      location += 2;    chars[1] = chars[(offsets[0] << 1) + 1];
3666      cc++;    if (offsets[2] >= 0)
3667        {
3668        chars[2] = chars[offsets[2] << 1];
3669        chars[3] = chars[(offsets[2] << 1) + 1];
3670        }
3671      if (offsets[1] >= 0)
3672        {
3673        chars[4] = chars[offsets[1] << 1];
3674        chars[5] = chars[(offsets[1] << 1) + 1];
3675      }      }
   
   if (location >= MAX_N_CHARS * 2 || must_stop != 0)  
     break;  
3676    }    }
3677    
3678  /* At least two characters are required. */  max -= 1;
 if (location < 2 * 2)  
     return FALSE;  
   
3679  if (firstline)  if (firstline)
3680    {    {
3681    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
3682      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3683    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
3684    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));
3685      quit = CMP(SLJIT_C_LESS_EQUAL, STR_END, 0, TMP1, 0);
3686      OP1(SLJIT_MOV, STR_END, 0, TMP1, 0);
3687      JUMPHERE(quit);
3688    }    }
3689  else  else
3690    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));
3691    
3692    #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3693    if (range_right >= 0)
3694      OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table);
3695    #endif
3696    
3697  start = LABEL();  start = LABEL();
3698  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3699    
3700  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  SLJIT_ASSERT(range_right >= 0 || offsets[0] >= 0);
3701  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
3702  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  if (range_right >= 0)
3703  if (chars[1] != 0)    {
3704    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);  #if defined COMPILE_PCRE8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN)
3705  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right));
3706  if (location > 2 * 2)  #else
3707    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right + 1) - 1);
3708  if (chars[3] != 0)  #endif
3709    OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]);  
3710  CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start);  #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3711  if (location > 2 * 2)    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0);
3712    {  #else
3713    if (chars[5] != 0)    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table);
3714      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]);  #endif
3715    CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3716      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start);
3717      }
3718    
3719    if (offsets[0] >= 0)
3720      {
3721      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0]));
3722      if (offsets[1] >= 0)
3723        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1]));
3724      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3725    
3726      if (chars[1] != 0)
3727        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);
3728      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);
3729      if (offsets[2] >= 0)
3730        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1));
3731    
3732      if (offsets[1] >= 0)
3733        {
3734        if (chars[5] != 0)
3735          OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]);
3736        CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start);
3737        }
3738    
3739      if (offsets[2] >= 0)
3740        {
3741        if (chars[3] != 0)
3742          OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]);
3743        CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start);
3744        }
3745      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3746    }    }
 OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
3747    
3748  JUMPHERE(quit);  JUMPHERE(quit);
3749    
3750  if (firstline)  if (firstline)
3751      {
3752      if (range_right >= 0)
3753        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3754    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3755      if (range_right >= 0)
3756        {
3757        quit = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0);
3758        OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0);
3759        JUMPHERE(quit);
3760        }
3761      }
3762  else  else
3763    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));
3764  return TRUE;  return TRUE;
3765  }  }
3766    
3767  #undef MAX_N_CHARS  #undef MAX_N_CHARS
3768    #undef MAX_N_BYTES
3769    
3770  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)
3771  {  {
# Line 3081  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ Line 3881  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_
3881  skip_char_back(common);  skip_char_back(common);
3882    
3883  loop = LABEL();  loop = LABEL();
3884  read_char(common);  read_char_range(common, common->nlmin, common->nlmax, TRUE);
3885  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3886  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
3887    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 3910  if (firstline)
3910    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3911  }  }
3912    
3913  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);
3914    
3915  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)
3916  {  {
3917  DEFINE_COMPILER;  DEFINE_COMPILER;
3918  struct sljit_label *start;  struct sljit_label *start;
3919  struct sljit_jump *quit;  struct sljit_jump *quit;
3920  struct sljit_jump *found = NULL;  struct sljit_jump *found = NULL;
3921  jump_list *matches = NULL;  jump_list *matches = NULL;
 pcre_uint8 inverted_start_bits[32];  
 int i;  
3922  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3923  struct sljit_jump *jump;  struct sljit_jump *jump;
3924  #endif  #endif
3925    
 for (i = 0; i < 32; ++i)  
   inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]);  
   
3926  if (firstline)  if (firstline)
3927    {    {
3928    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
# Line 3143  if (common->utf) Line 3938  if (common->utf)
3938    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
3939  #endif  #endif
3940    
3941  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))
3942    {    {
3943  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3944    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 3947  if (!check_class_ranges(common, inverted
3947  #endif  #endif
3948    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
3949    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
3950    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
3951    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
3952    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);
3953    found = JUMP(SLJIT_C_NOT_ZERO);    found = JUMP(SLJIT_C_NOT_ZERO);
# Line 3355  JUMPHERE(skipread); Line 4150  JUMPHERE(skipread);
4150    
4151  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
4152  check_str_end(common, &skipread_list);  check_str_end(common, &skipread_list);
4153  peek_char(common);  peek_char(common, READ_CHAR_MAX);
4154    
4155  /* Testing char type. This is a code duplication. */  /* Testing char type. This is a code duplication. */
4156  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3401  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE Line 4196  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE
4196  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);
4197  }  }
4198    
4199  /*  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)  
4200  {  {
4201  DEFINE_COMPILER;  DEFINE_COMPILER;
4202  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];  
4203  pcre_uint8 bit, cbit, all;  pcre_uint8 bit, cbit, all;
4204  int i, byte, length = 0;  int i, byte, length = 0;
4205    
4206  bit = bits[0] & 0x1;  bit = bits[0] & 0x1;
4207  ranges[1] = bit;  /* All bits will be zero or one (since bit is zero or one). */
 /* Can be 0 or 255. */  
4208  all = -bit;  all = -bit;
4209    
4210  for (i = 0; i < 256; )  for (i = 0; i < 256; )
# Line 3519  for (i = 0; i < 256; ) Line 4219  for (i = 0; i < 256; )
4219        {        {
4220        if (length >= MAX_RANGE_SIZE)        if (length >= MAX_RANGE_SIZE)
4221          return FALSE;          return FALSE;
4222        ranges[2 + length] = i;        ranges[length] = i;
4223        length++;        length++;
4224        bit = cbit;        bit = cbit;
4225        all = -cbit;        all = -cbit;
# Line 3532  if (((bit == 0) && nclass) || ((bit == 1 Line 4232  if (((bit == 0) && nclass) || ((bit == 1
4232    {    {
4233    if (length >= MAX_RANGE_SIZE)    if (length >= MAX_RANGE_SIZE)
4234      return FALSE;      return FALSE;
4235    ranges[2 + length] = 256;    ranges[length] = 256;
4236    length++;    length++;
4237    }    }
 ranges[0] = length;  
4238    
4239  return check_ranges(common, ranges, backtracks, FALSE);  if (length < 0 || length > 4)
4240      return FALSE;
4241    
4242    bit = bits[0] & 0x1;
4243    if (invert) bit ^= 0x1;
4244    
4245    /* No character is accepted. */
4246    if (length == 0 && bit == 0)
4247      add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4248    
4249    switch(length)
4250      {
4251      case 0:
4252      /* When bit != 0, all characters are accepted. */
4253      return TRUE;
4254    
4255      case 1:
4256      add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4257      return TRUE;
4258    
4259      case 2:
4260      if (ranges[0] + 1 != ranges[1])
4261        {
4262        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4263        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4264        }
4265      else
4266        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4267      return TRUE;
4268    
4269      case 3:
4270      if (bit != 0)
4271        {
4272        add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4273        if (ranges[0] + 1 != ranges[1])
4274          {
4275          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4276          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4277          }
4278        else
4279          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4280        return TRUE;
4281        }
4282    
4283      add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[0]));
4284      if (ranges[1] + 1 != ranges[2])
4285        {
4286        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]);
4287        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4288        }
4289      else
4290        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1]));
4291      return TRUE;
4292    
4293      case 4:
4294      if ((ranges[1] - ranges[0]) == (ranges[3] - ranges[2])
4295          && (ranges[0] | (ranges[2] - ranges[0])) == ranges[2]
4296          && is_powerof2(ranges[2] - ranges[0]))
4297        {
4298        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[0]);
4299        if (ranges[2] + 1 != ranges[3])
4300          {
4301          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
4302          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4303          }
4304        else
4305          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4306        return TRUE;
4307        }
4308    
4309      if (bit != 0)
4310        {
4311        i = 0;
4312        if (ranges[0] + 1 != ranges[1])
4313          {
4314          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4315          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4316          i = ranges[0];
4317          }
4318        else
4319          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4320    
4321        if (ranges[2] + 1 != ranges[3])
4322          {
4323          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - i);
4324          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4325          }
4326        else
4327          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2] - i));
4328        return TRUE;
4329        }
4330    
4331      OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4332      add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[0]));
4333      if (ranges[1] + 1 != ranges[2])
4334        {
4335        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]);
4336        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4337        }
4338      else
4339        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4340      return TRUE;
4341    
4342      default:
4343      SLJIT_ASSERT_STOP();
4344      return FALSE;
4345      }
4346  }  }
4347    
4348  static void check_anynewline(compiler_common *common)  static void check_anynewline(compiler_common *common)
# Line 3904  return cc; Line 4709  return cc;
4709  #define SET_TYPE_OFFSET(value) \  #define SET_TYPE_OFFSET(value) \
4710    if ((value) != typeoffset) \    if ((value) != typeoffset) \
4711      { \      { \
4712      if ((value) > typeoffset) \      if ((value) < typeoffset) \
       OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \  
     else \  
4713        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \
4714        else \
4715          OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \
4716      } \      } \
4717    typeoffset = (value);    typeoffset = (value);
4718    
4719  #define SET_CHAR_OFFSET(value) \  #define SET_CHAR_OFFSET(value) \
4720    if ((value) != charoffset) \    if ((value) != charoffset) \
4721      { \      { \
4722      if ((value) > charoffset) \      if ((value) < charoffset) \
4723        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(charoffset - (value))); \
4724      else \      else \
4725        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)((value) - charoffset)); \
4726      } \      } \
4727    charoffset = (value);    charoffset = (value);
4728    
# Line 3925  static void compile_xclass_matchingpath( Line 4730  static void compile_xclass_matchingpath(
4730  {  {
4731  DEFINE_COMPILER;  DEFINE_COMPILER;
4732  jump_list *found = NULL;  jump_list *found = NULL;
4733  jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks;  jump_list **list = (cc[0] & XCL_NOT) == 0 ? &found : backtracks;
4734  pcre_int32 c, charoffset;  sljit_uw c, charoffset, max = 256, min = READ_CHAR_MAX;
 const pcre_uint32 *other_cases;  
4735  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
4736  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
4737  int compares, invertcmp, numberofcmps;  int compares, invertcmp, numberofcmps;
4738    #if defined SUPPORT_UTF && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16)
4739    BOOL utf = common->utf;
4740    #endif
4741    
4742  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4743  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4744  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4745  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4746  pcre_int32 typeoffset;  const pcre_uint32 *other_cases;
4747    sljit_uw typeoffset;
4748  #endif  #endif
4749    
4750  /* Although SUPPORT_UTF must be defined, we are  /* Scanning the necessary info. */
4751     not necessary in utf mode even in 8 bit mode. */  cc++;
4752  detect_partial_match(common, backtracks);  ccbegin = cc;
4753  read_char(common);  compares = 0;
4754    if (cc[-1] & XCL_MAP)
 if ((*cc++ & XCL_MAP) != 0)  
4755    {    {
4756    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  
4757    cc += 32 / sizeof(pcre_uchar);    cc += 32 / sizeof(pcre_uchar);
4758    }    }
4759    
 /* Scanning the necessary info. */  
 ccbegin = cc;  
 compares = 0;  
4760  while (*cc != XCL_END)  while (*cc != XCL_END)
4761    {    {
4762    compares++;    compares++;
4763    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4764      {      {
4765      cc += 2;      cc ++;
4766  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4767      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c > max) max = c;
4768  #endif      if (c < min) min = c;
4769  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4770      needschar = TRUE;      needschar = TRUE;
4771  #endif  #endif
4772      }      }
4773    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4774      {      {
4775      cc += 2;      cc ++;
4776  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4777      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c < min) min = c;
4778  #endif      GETCHARINCTEST(c, cc);
4779      cc++;      if (c > max) max = c;
 #ifdef SUPPORT_UTF  
     if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
4780  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4781      needschar = TRUE;      needschar = TRUE;
4782  #endif  #endif
# Line 4011  while (*cc != XCL_END) Line 4786  while (*cc != XCL_END)
4786      {      {
4787      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);
4788      cc++;      cc++;
4789        if (*cc == PT_CLIST)
4790          {
4791          other_cases = PRIV(ucd_caseless_sets) + cc[1];
4792          while (*other_cases != NOTACHAR)
4793            {
4794            if (*other_cases > max) max = *other_cases;
4795            if (*other_cases < min) min = *other_cases;
4796            other_cases++;
4797            }
4798          }
4799        else
4800          {
4801          max = READ_CHAR_MAX;
4802          min = 0;
4803          }
4804    
4805      switch(*cc)      switch(*cc)
4806        {        {
4807        case PT_ANY:        case PT_ANY:
# Line 4030  while (*cc != XCL_END) Line 4821  while (*cc != XCL_END)
4821        case PT_SPACE:        case PT_SPACE:
4822        case PT_PXSPACE:        case PT_PXSPACE:
4823        case PT_WORD:        case PT_WORD:
4824          case PT_PXGRAPH:
4825          case PT_PXPRINT:
4826          case PT_PXPUNCT:
4827        needstype = TRUE;        needstype = TRUE;
4828        needschar = TRUE;        needschar = TRUE;
4829        break;        break;
# Line 4043  while (*cc != XCL_END) Line 4837  while (*cc != XCL_END)
4837        SLJIT_ASSERT_STOP();        SLJIT_ASSERT_STOP();
4838        break;        break;
4839        }        }
4840      cc += 2;      cc += 2;
4841        }
4842    #endif
4843      }
4844    
4845    /* We are not necessary in utf mode even in 8 bit mode. */
4846    cc = ccbegin;
4847    detect_partial_match(common, backtracks);
4848    read_char_range(common, min, max, (cc[-1] & XCL_NOT) != 0);
4849    
4850    if ((cc[-1] & XCL_HASPROP) == 0)
4851      {
4852      if ((cc[-1] & XCL_MAP) != 0)
4853        {
4854        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4855        if (!check_class_ranges(common, (const pcre_uint8 *)cc, (((const pcre_uint8 *)cc)[31] & 0x80) != 0, TRUE, &found))
4856          {
4857          OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4858          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4859          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4860          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4861          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4862          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4863          }
4864    
4865        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4866        JUMPHERE(jump);
4867    
4868        cc += 32 / sizeof(pcre_uchar);
4869        }
4870      else
4871        {
4872        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, min);
4873        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, max - min));
4874      }      }
4875      }
4876    else if ((cc[-1] & XCL_MAP) != 0)
4877      {
4878      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4879    #ifdef SUPPORT_UCP
4880      charsaved = TRUE;
4881    #endif
4882      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4883        {
4884    #ifdef COMPILE_PCRE8
4885        SLJIT_ASSERT(common->utf);
4886  #endif  #endif
4887        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4888    
4889        OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4890        OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4891        OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4892        OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4893        OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4894        add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
4895    
4896        JUMPHERE(jump);
4897        }
4898    
4899      OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4900      cc += 32 / sizeof(pcre_uchar);
4901    }    }
4902    
4903  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4089  if (needstype || needsscript) Line 4941  if (needstype || needsscript)
4941  #endif  #endif
4942    
4943  /* Generating code. */  /* Generating code. */
 cc = ccbegin;  
4944  charoffset = 0;  charoffset = 0;
4945  numberofcmps = 0;  numberofcmps = 0;
4946  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4105  while (*cc != XCL_END) Line 4956  while (*cc != XCL_END)
4956    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4957      {      {
4958      cc ++;      cc ++;
4959  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4960    
4961      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4962        {        {
4963        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));
4964        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);
4965        numberofcmps++;        numberofcmps++;
4966        }        }
4967      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4968        {        {
4969        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));
4970        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);
4971        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4972        numberofcmps = 0;        numberofcmps = 0;
4973        }        }
4974      else      else
4975        {        {
4976        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));
4977        numberofcmps = 0;        numberofcmps = 0;
4978        }        }
4979      }      }
4980    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4981      {      {
4982      cc ++;      cc ++;
4983  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4984      SET_CHAR_OFFSET(c);      SET_CHAR_OFFSET(c);
4985  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4986      if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4987      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4988        {        {
4989        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));
4990        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);
4991        numberofcmps++;        numberofcmps++;
4992        }        }
4993      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4994        {        {
4995        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));
4996        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);
4997        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4998        numberofcmps = 0;        numberofcmps = 0;
4999        }        }
5000      else      else
5001        {        {
5002        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));
5003        numberofcmps = 0;        numberofcmps = 0;
5004        }        }
5005      }      }
# Line 4217  while (*cc != XCL_END) Line 5048  while (*cc != XCL_END)
5048    
5049        case PT_SPACE:        case PT_SPACE:
5050        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);  
         }  
5051        SET_CHAR_OFFSET(9);        SET_CHAR_OFFSET(9);
5052        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);
5053        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);
5054        if (*cc == PT_SPACE)  
5055          JUMPHERE(jump);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9);
5056          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5057    
5058          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9);
5059          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5060    
5061        SET_TYPE_OFFSET(ucp_Zl);        SET_TYPE_OFFSET(ucp_Zl);
5062        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 5065  while (*cc != XCL_END)
5065        break;        break;
5066    
5067        case PT_WORD:        case PT_WORD:
5068        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));
5069        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5070        /* Fall through. */        /* Fall through. */
5071    
# Line 4283  while (*cc != XCL_END) Line 5113  while (*cc != XCL_END)
5113          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]);
5114          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5115    
5116          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));
5117          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);
5118    
5119          other_cases += 3;          other_cases += 3;
5120          }          }
5121        else        else
5122          {          {
5123          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));
5124          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5125          }          }
5126    
5127        while (*other_cases != NOTACHAR)        while (*other_cases != NOTACHAR)
5128          {          {
5129          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));
5130          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);
5131          }          }
5132        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5133        break;        break;
5134    
5135        case PT_UCNC:        case PT_UCNC:
5136        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));
5137        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5138        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));
5139        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5140        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));
5141        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5142    
5143        SET_CHAR_OFFSET(0xa0);        SET_CHAR_OFFSET(0xa0);
5144        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));
5145        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5146        SET_CHAR_OFFSET(0);        SET_CHAR_OFFSET(0);
5147        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);
5148        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);
5149        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5150        break;        break;
5151    
5152          case PT_PXGRAPH:
5153          /* C and Z groups are the farthest two groups. */
5154          SET_TYPE_OFFSET(ucp_Ll);
5155          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5156          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5157    
5158          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5159    
5160          /* In case of ucp_Cf, we overwrite the result. */
5161          SET_CHAR_OFFSET(0x2066);
5162          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5163          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5164    
5165          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5166          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5167    
5168          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
5169          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5170    
5171          JUMPHERE(jump);
5172          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5173          break;
5174    
5175          case PT_PXPRINT:
5176          /* C and Z groups are the farthest two groups. */
5177          SET_TYPE_OFFSET(ucp_Ll);
5178          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5179          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5180    
5181          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
5182          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
5183    
5184          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5185    
5186          /* In case of ucp_Cf, we overwrite the result. */
5187          SET_CHAR_OFFSET(0x2066);
5188          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5189          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5190    
5191          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5192          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5193    
5194          JUMPHERE(jump);
5195          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5196          break;
5197    
5198          case PT_PXPUNCT:
5199          SET_TYPE_OFFSET(ucp_Sc);
5200          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc);
5201          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5202    
5203          SET_CHAR_OFFSET(0);
5204          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff);
5205          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5206    
5207          SET_TYPE_OFFSET(ucp_Pc);
5208          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc);
5209          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5210          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5211          break;
5212        }        }
5213      cc += 2;      cc += 2;
5214      }      }
# Line 4349  struct sljit_label *label; Line 5240  struct sljit_label *label;
5240  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5241  pcre_uchar propdata[5];  pcre_uchar propdata[5];
5242  #endif  #endif
5243  #endif  #endif /* SUPPORT_UTF */
5244    
5245  switch(type)  switch(type)
5246    {    {
# Line 4374  switch(type) Line 5265  switch(type)
5265    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
5266    case OP_DIGIT:    case OP_DIGIT:
5267    /* 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);  
5268    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5269    /* Flip the starting bit in the negative case. */  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5270    if (type == OP_NOT_DIGIT)    if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_digit, FALSE))
5271      common->digits[1] ^= 1;      read_char7_type(common, type == OP_NOT_DIGIT);
5272    if (!check_ranges(common, common->digits, backtracks, TRUE))    else
5273      {  #endif
5274      read_char8_type(common);      read_char8_type(common, type == OP_NOT_DIGIT);
5275      OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);      /* Flip the starting bit in the negative case. */
5276      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);
5277      }    add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
   if (type == OP_NOT_DIGIT)  
     common->digits[1] ^= 1;  
5278    return cc;    return cc;
5279    
5280    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
5281    case OP_WHITESPACE:    case OP_WHITESPACE:
5282    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5283    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5284      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_space, FALSE))
5285        read_char7_type(common, type == OP_NOT_WHITESPACE);
5286      else
5287    #endif
5288        read_char8_type(common, type == OP_NOT_WHITESPACE);
5289    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);
5290    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));
5291    return cc;    return cc;
# Line 4401  switch(type) Line 5293  switch(type)
5293    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
5294    case OP_WORDCHAR:    case OP_WORDCHAR:
5295    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5296    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5297      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_word, FALSE))
5298        read_char7_type(common, type == OP_NOT_WORDCHAR);
5299      else
5300    #endif
5301        read_char8_type(common, type == OP_NOT_WORDCHAR);
5302    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);
5303    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));
5304    return cc;    return cc;
5305    
5306    case OP_ANY:    case OP_ANY:
5307    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5308    read_char(common);    read_char_range(common, common->nlmin, common->nlmax, TRUE);
5309    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
5310      {      {
5311      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 5361  switch(type)
5361  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5362    case OP_NOTPROP:    case OP_NOTPROP:
5363    case OP_PROP:    case OP_PROP:
5364    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
5365    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
5366    propdata[2] = cc[0];    propdata[2] = cc[0];
5367    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4476  switch(type) Line 5373  switch(type)
5373    
5374    case OP_ANYNL:    case OP_ANYNL:
5375    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5376    read_char(common);    read_char_range(common, common->bsr_nlmin, common->bsr_nlmax, FALSE);
5377    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);
5378    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
5379    end_list = NULL;    end_list = NULL;
# Line 4498  switch(type) Line 5395  switch(type)
5395    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
5396    case OP_HSPACE:    case OP_HSPACE:
5397    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5398    read_char(common);    read_char_range(common, 0x9, 0x3000, type == OP_NOT_HSPACE);
5399    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));
5400    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));
5401    return cc;    return cc;
# Line 4506  switch(type) Line 5403  switch(type)
5403    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
5404    case OP_VSPACE:    case OP_VSPACE:
5405    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5406    read_char(common);    read_char_range(common, 0xa, 0x2029, type == OP_NOT_VSPACE);
5407    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));
5408    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));
5409    return cc;    return cc;
# Line 4605  switch(type) Line 5502  switch(type)
5502      else      else
5503        {        {
5504        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);
5505        read_char(common);        read_char_range(common, common->nlmin, common->nlmax, TRUE);
5506        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));
5507        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
5508        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4653  switch(type) Line 5550  switch(type)
5550    else    else
5551      {      {
5552      skip_char_back(common);      skip_char_back(common);
5553      read_char(common);      read_char_range(common, common->nlmin, common->nlmax, TRUE);
5554      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5555      }      }
5556    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4704  switch(type) Line 5601  switch(type)
5601      }      }
5602    else    else
5603      {      {
5604      peek_char(common);      peek_char(common, common->nlmax);
5605      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5606      }      }
5607    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4728  switch(type) Line 5625  switch(type)
5625  #endif  #endif
5626      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);
5627      }      }
5628    
5629    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
   read_char(common);  
5630  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5631    if (common->utf)    if (common->utf)
5632      {      {
# Line 4738  switch(type) Line 5635  switch(type)
5635    else    else
5636  #endif  #endif
5637      c = *cc;      c = *cc;
5638    
5639    if (type == OP_CHAR || !char_has_othercase(common, cc))    if (type == OP_CHAR || !char_has_othercase(common, cc))
5640      {      {
5641        read_char_range(common, c, c, FALSE);
5642      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));
5643      return cc + length;      return cc + length;
5644      }      }
5645    oc = char_othercase(common, c);    oc = char_othercase(common, c);
5646      read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, FALSE);
5647    bit = c ^ oc;    bit = c ^ oc;
5648    if (is_powerof2(bit))    if (is_powerof2(bit))
5649      {      {
# Line 4751  switch(type) Line 5651  switch(type)
5651      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));
5652      return cc + length;      return cc + length;
5653      }      }
5654    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);
5655    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));
5656    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));  
5657    return cc + length;    return cc + length;
5658    
5659    case OP_NOT:    case OP_NOT:
# Line 4790  switch(type) Line 5688  switch(type)
5688  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
5689        {        {
5690        GETCHARLEN(c, cc, length);        GETCHARLEN(c, cc, length);
       read_char(common);  
5691        }        }
5692      }      }
5693    else    else
5694  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
     {  
     read_char(common);  
5695      c = *cc;      c = *cc;
     }  
5696    
5697    if (type == OP_NOT || !char_has_othercase(common, cc))    if (type == OP_NOT || !char_has_othercase(common, cc))
5698        {
5699        read_char_range(common, c, c, TRUE);
5700      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));
5701        }
5702    else    else
5703      {      {
5704      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5705        read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, TRUE);
5706      bit = c ^ oc;      bit = c ^ oc;
5707      if (is_powerof2(bit))      if (is_powerof2(bit))
5708        {        {
# Line 4822  switch(type) Line 5720  switch(type)
5720    case OP_CLASS:    case OP_CLASS:
5721    case OP_NCLASS:    case OP_NCLASS:
5722    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5723    read_char(common);  
5724    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5725      bit = (common->utf && is_char7_bitset((const pcre_uint8 *)cc, type == OP_NCLASS)) ? 127 : 255;
5726      read_char_range(common, 0, bit, type == OP_NCLASS);
5727    #else
5728      read_char_range(common, 0, 255, type == OP_NCLASS);
5729    #endif
5730    
5731      if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5732      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5733    
5734  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5735    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. */  
5736    if (common->utf)    if (common->utf)
 #endif /* COMPILE_PCRE8 */  
5737      {      {
5738      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, bit);
5739      if (type == OP_CLASS)      if (type == OP_CLASS)
5740        {        {
5741        add_jump(compiler, backtracks, jump[0]);        add_jump(compiler, backtracks, jump[0]);
5742        jump[0] = NULL;        jump[0] = NULL;
5743        }        }
5744      }      }
5745  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #elif !defined COMPILE_PCRE8
5746      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
5747      if (type == OP_CLASS)
5748        {
5749        add_jump(compiler, backtracks, jump[0]);
5750        jump[0] = NULL;
5751        }
5752    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
5753    
5754    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5755    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5756    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5757    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5758    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);
5759    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
5760    
5761  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5762    if (jump[0] != NULL)    if (jump[0] != NULL)
5763      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5764  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif
5765    
5766    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5767    
5768  #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 5867  if (context.length > 0)
5867  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);
5868  }  }
5869    
 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));  
 }  
   
5870  /* Forward definitions. */  /* Forward definitions. */
5871  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);
5872  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 5899  static void compile_backtrackingpath(com
5899    
5900  #define BACKTRACK_AS(type) ((type *)backtrack)  #define BACKTRACK_AS(type) ((type *)backtrack)
5901    
5902  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)
5903    {
5904    /* The OVECTOR offset goes to TMP2. */
5905    DEFINE_COMPILER;
5906    int count = GET2(cc, 1 + IMM2_SIZE);
5907    pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
5908    unsigned int offset;
5909    jump_list *found = NULL;
5910    
5911    SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
5912    
5913    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
5914    
5915    count--;
5916    while (count-- > 0)
5917      {
5918      offset = GET2(slot, 0) << 1;
5919      GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5920      add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5921      slot += common->name_entry_size;
5922      }
5923    
5924    offset = GET2(slot, 0) << 1;
5925    GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5926    if (backtracks != NULL && !common->jscript_compat)
5927      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5928    
5929    set_jumps(found, LABEL());
5930    }
5931    
5932    static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
5933  {  {
5934  DEFINE_COMPILER;  DEFINE_COMPILER;
5935  int offset = GET2(cc, 1) << 1;  BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5936    int offset = 0;
5937  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
5938  struct sljit_jump *partial;  struct sljit_jump *partial;
5939  struct sljit_jump *nopartial;  struct sljit_jump *nopartial;
5940    
5941  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  if (ref)
5942  /* OVECTOR(1) contains the "string begin - 1" constant. */    {
5943  if (withchecks && !common->jscript_compat)    offset = GET2(cc, 1) << 1;
5944    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));
5945      /* OVECTOR(1) contains the "string begin - 1" constant. */
5946      if (withchecks && !common->jscript_compat)
5947        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5948      }
5949    else
5950      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5951    
5952  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
5953  if (common->utf && *cc == OP_REFI)  if (common->utf && *cc == OP_REFI)
5954    {    {
5955    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);
5956    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));    if (ref)
5957        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5958      else
5959        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5960    
5961    if (withchecks)    if (withchecks)
5962      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);
5963    
# Line 5052  if (common->utf && *cc == OP_REFI) Line 5982  if (common->utf && *cc == OP_REFI)
5982  else  else
5983  #endif /* SUPPORT_UTF && SUPPORT_UCP */  #endif /* SUPPORT_UTF && SUPPORT_UCP */
5984    {    {
5985    OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);    if (ref)
5986        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);
5987      else
5988        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
5989    
5990    if (withchecks)    if (withchecks)
5991      jump = JUMP(SLJIT_C_ZERO);      jump = JUMP(SLJIT_C_ZERO);
5992    
# Line 5089  if (jump != NULL) Line 6023  if (jump != NULL)
6023    else    else
6024      JUMPHERE(jump);      JUMPHERE(jump);
6025    }    }
 return cc + 1 + IMM2_SIZE;  
6026  }  }
6027    
6028  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)
6029  {  {
6030  DEFINE_COMPILER;  DEFINE_COMPILER;
6031    BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
6032  backtrack_common *backtrack;  backtrack_common *backtrack;
6033  pcre_uchar type;  pcre_uchar type;
6034    int offset = 0;
6035  struct sljit_label *label;  struct sljit_label *label;
6036  struct sljit_jump *zerolength;  struct sljit_jump *zerolength;
6037  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
# Line 5106  BOOL minimize; Line 6041  BOOL minimize;
6041    
6042  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);
6043    
6044    if (ref)
6045      offset = GET2(cc, 1) << 1;
6046    else
6047      cc += IMM2_SIZE;
6048  type = cc[1 + IMM2_SIZE];  type = cc[1 + IMM2_SIZE];
6049    
6050    SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even);
6051  minimize = (type & 0x1) != 0;  minimize = (type & 0x1) != 0;
6052  switch(type)  switch(type)
6053    {    {
# Line 5144  if (!minimize) Line 6085  if (!minimize)
6085    if (min == 0)    if (min == 0)
6086      {      {
6087      allocate_stack(common, 2);      allocate_stack(common, 2);
6088        if (ref)
6089          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6090      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6091      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
6092      /* Temporary release of STR_PTR. */      /* Temporary release of STR_PTR. */
6093      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));
6094      zerolength = compile_ref_checks(common, ccbegin, NULL);      /* Handles both invalid and empty cases. Since the minimum repeat,
6095        is zero the invalid case is basically the same as an empty case. */
6096        if (ref)
6097          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6098        else
6099          {
6100          compile_dnref_search(common, ccbegin, NULL);
6101          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6102          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
6103          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6104          }
6105      /* Restore if not zero length. */      /* Restore if not zero length. */
6106      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));
6107      }      }
6108    else    else
6109      {      {
6110      allocate_stack(common, 1);      allocate_stack(common, 1);
6111        if (ref)
6112          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6113      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
6114      zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);      if (ref)
6115          {
6116          add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
6117          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6118          }
6119        else
6120          {
6121          compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
6122          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6123          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
6124          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6125          }
6126      }      }
6127    
6128    if (min > 1 || max > 1)    if (min > 1 || max > 1)
6129      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);
6130    
6131    label = LABEL();    label = LABEL();
6132      if (!ref)
6133        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1);
6134    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);
6135    
6136    if (min > 1 || max > 1)    if (min > 1 || max > 1)
# Line 5193  if (!minimize) Line 6161  if (!minimize)
6161    JUMPHERE(zerolength);    JUMPHERE(zerolength);
6162    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
6163    
6164    decrease_call_count(common);    count_match(common);
6165    return cc;    return cc;
6166    }    }
6167    
6168  allocate_stack(common, 2);  allocate_stack(common, ref ? 2 : 3);
6169    if (ref)
6170      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6171  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
6172  if (type != OP_CRMINSTAR)  if (type != OP_CRMINSTAR)
6173    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
6174    
6175  if (min == 0)  if (min == 0)
6176    {    {
6177    zerolength = compile_ref_checks(common, ccbegin, NULL);    /* Handles both invalid and empty cases. Since the minimum repeat,
6178      is zero the invalid case is basically the same as an empty case. */
6179      if (ref)
6180        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6181      else
6182        {
6183        compile_dnref_search(common, ccbegin, NULL);
6184        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6185        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
6186        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6187        }
6188      /* Length is non-zero, we can match real repeats. */
6189    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6190    jump = JUMP(SLJIT_JUMP);    jump = JUMP(SLJIT_JUMP);
6191    }    }
6192  else  else
6193    zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);    {
6194      if (ref)
6195        {
6196        add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
6197        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6198        }
6199      else
6200        {
6201        compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
6202        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6203        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
6204        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6205        }
6206      }
6207    
6208  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
6209  if (max > 0)  if (max > 0)
6210    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));
6211    
6212    if (!ref)
6213      OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
6214  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);
6215  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6216    
# Line 5232  if (jump != NULL) Line 6228  if (jump != NULL)
6228    JUMPHERE(jump);    JUMPHERE(jump);
6229  JUMPHERE(zerolength);  JUMPHERE(zerolength);
6230    
6231  decrease_call_count(common);  count_match(common);
6232  return cc;  return cc;
6233  }  }
6234    
# Line 5242  DEFINE_COMPILER; Line 6238  DEFINE_COMPILER;
6238  backtrack_common *backtrack;  backtrack_common *backtrack;
6239  recurse_entry *entry = common->entries;  recurse_entry *entry = common->entries;
6240  recurse_entry *prev = NULL;  recurse_entry *prev = NULL;
6241  int start = GET(cc, 1);  sljit_sw start = GET(cc, 1);
6242  pcre_uchar *start_cc;  pcre_uchar *start_cc;
6243    BOOL needs_control_head;
6244    
6245  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);
6246    
6247  /* Inlining simple patterns. */  /* Inlining simple patterns. */
6248  if (get_framesize(common, common->start + start, TRUE) == no_stack)  if (get_framesize(common, common->start + start, NULL, TRUE, &needs_control_head) == no_stack)
6249    {    {
6250    start_cc = common->start + start;    start_cc = common->start + start;
6251    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 6405  static pcre_uchar *compile_assert_matchi
6405  DEFINE_COMPILER;  DEFINE_COMPILER;
6406  int framesize;  int framesize;
6407  int extrasize;  int extrasize;
6408  BOOL needs_control_head = common->control_head_ptr != 0;  BOOL needs_control_head;
6409  int private_data_ptr;  int private_data_ptr;
6410  backtrack_common altbacktrack;  backtrack_common altbacktrack;
6411  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
# Line 5418  jump_list *tmp = NULL; Line 6415  jump_list *tmp = NULL;
6415  jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks;  jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks;
6416  jump_list **found;  jump_list **found;
6417  /* Saving previous accept variables. */  /* Saving previous accept variables. */
6418    BOOL save_local_exit = common->local_exit;
6419    BOOL save_positive_assert = common->positive_assert;
6420    then_trap_backtrack *save_then_trap = common->then_trap;
6421  struct sljit_label *save_quit_label = common->quit_label;  struct sljit_label *save_quit_label = common->quit_label;
6422  struct sljit_label *save_accept_label = common->accept_label;  struct sljit_label *save_accept_label = common->accept_label;
6423  jump_list *save_quit = common->quit;  jump_list *save_quit = common->quit;
6424    jump_list *save_positive_assert_quit = common->positive_assert_quit;
6425  jump_list *save_accept = common->accept;  jump_list *save_accept = common->accept;
 BOOL save_local_exit = common->local_exit;  
6426  struct sljit_jump *jump;  struct sljit_jump *jump;
6427  struct sljit_jump *brajump = NULL;  struct sljit_jump *brajump = NULL;
6428    
6429    /* Assert captures then. */
6430    common->then_trap = NULL;
6431    
6432  if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)  if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)
6433    {    {
6434    SLJIT_ASSERT(!conditional);    SLJIT_ASSERT(!conditional);
# Line 5434  if (*cc == OP_BRAZERO || *cc == OP_BRAMI Line 6437  if (*cc == OP_BRAZERO || *cc == OP_BRAMI
6437    }    }
6438  private_data_ptr = PRIVATE_DATA(cc);  private_data_ptr = PRIVATE_DATA(cc);
6439  SLJIT_ASSERT(private_data_ptr != 0);  SLJIT_ASSERT(private_data_ptr != 0);
6440  framesize = get_framesize(common, cc, FALSE);  framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head);
6441  backtrack->framesize = framesize;  backtrack->framesize = framesize;
6442  backtrack->private_data_ptr = private_data_ptr;  backtrack->private_data_ptr = private_data_ptr;
6443  opcode = *cc;  opcode = *cc;
# Line 5454  if (bra == OP_BRAMINZERO) Line 6457  if (bra == OP_BRAMINZERO)
6457  if (framesize < 0)  if (framesize < 0)
6458    {    {
6459    extrasize = needs_control_head ? 2 : 1;    extrasize = needs_control_head ? 2 : 1;
6460    if (framesize != no_stack)    if (framesize == no_frame)
6461      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);
6462    allocate_stack(common, extrasize);    allocate_stack(common, extrasize);
6463    if (needs_control_head)    if (needs_control_head)
# Line 5484  else Line 6487  else
6487      }      }
6488    else    else
6489      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);
6490    init_frame(common, ccbegin, framesize + extrasize - 1, extrasize, FALSE);    init_frame(common, ccbegin, NULL, framesize + extrasize - 1, extrasize, FALSE);
6491    }    }
6492    
6493  memset(&altbacktrack, 0, sizeof(backtrack_common));  memset(&altbacktrack, 0, sizeof(backtrack_common));
6494  common->local_exit = TRUE;  if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6495  common->quit_label = NULL;    {
6496  common->quit = NULL;    /* Negative assert is stronger than positive assert. */
6497      common->local_exit = TRUE;
6498      common->quit_label = NULL;
6499      common->quit = NULL;
6500      common->positive_assert = FALSE;
6501      }
6502    else
6503      common->positive_assert = TRUE;
6504    common->positive_assert_quit = NULL;
6505    
6506  while (1)  while (1)
6507    {    {
6508    common->accept_label = NULL;    common->accept_label = NULL;
# Line 5505  while (1) Line 6517  while (1)
6517    compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack);    compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack);
6518    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
6519      {      {
6520      common->local_exit = save_local_exit;      if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6521      common->quit_label = save_quit_label;        {
6522          common->local_exit = save_local_exit;
6523          common->quit_label = save_quit_label;
6524          common->quit = save_quit;
6525          }
6526        common->positive_assert = save_positive_assert;
6527        common->then_trap = save_then_trap;
6528      common->accept_label = save_accept_label;      common->accept_label = save_accept_label;
6529      common->quit = save_quit;      common->positive_assert_quit = save_positive_assert_quit;
6530      common->accept = save_accept;      common->accept = save_accept;
6531      return NULL;      return NULL;
6532      }      }
# Line 5519  while (1) Line 6537  while (1)
6537    /* Reset stack. */    /* Reset stack. */
6538    if (framesize < 0)    if (framesize < 0)
6539      {      {
6540      if (framesize != no_stack)      if (framesize == no_frame)
6541        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);
6542      else      else
6543        free_stack(common, extrasize);        free_stack(common, extrasize);
# Line 5573  while (1) Line 6591  while (1)
6591    compile_backtrackingpath(common, altbacktrack.top);    compile_backtrackingpath(common, altbacktrack.top);
6592    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))    if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
6593      {      {
6594      common->local_exit = save_local_exit;      if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
6595      common->quit_label = save_quit_label;        {
6596          common->local_exit = save_local_exit;
6597          common->quit_label = save_quit_label;
6598          common->quit = save_quit;
6599          }
6600        common->positive_assert = save_positive_assert;