/[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 1247 by zherczeg, Mon Feb 11 21:37:46 2013 UTC revision 1424 by zherczeg, Tue Dec 31 11:22:31 2013 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
10    
11    The machine code generator part (this module) was written by Zoltan Herczeg    The machine code generator part (this module) was written by Zoltan Herczeg
12                        Copyright (c) 2010-2012                        Copyright (c) 2010-2013
13    
14  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
15  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 71  system files. */ Line 71  system files. */
71     2 - Enable capture_last_ptr (includes option 1). */     2 - Enable capture_last_ptr (includes option 1). */
72  /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */  /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */
73    
74    /* 1 - Always have a control head. */
75    /* #define DEBUG_FORCE_CONTROL_HEAD 1 */
76    
77  /* Allocate memory for the regex stack on the real machine stack.  /* Allocate memory for the regex stack on the real machine stack.
78  Fast, but limited size. */  Fast, but limited size. */
79  #define MACHINE_STACK_SIZE 32768  #define MACHINE_STACK_SIZE 32768
# Line 165  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;
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 178  typedef struct executable_functions { Line 182  typedef struct executable_functions {
182    PUBL(jit_callback) callback;    PUBL(jit_callback) callback;
183    void *userdata;    void *userdata;
184    pcre_uint32 top_bracket;    pcre_uint32 top_bracket;
185      pcre_uint32 limit_match;
186    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];
187  } executable_functions;  } executable_functions;
188    
# Line 186  typedef struct jump_list { Line 191  typedef struct jump_list {
191    struct jump_list *next;    struct jump_list *next;
192  } jump_list;  } jump_list;
193    
 enum stub_types { stack_alloc };  
   
194  typedef struct stub_list {  typedef struct stub_list {
   enum stub_types type;  
   int data;  
195    struct sljit_jump *start;    struct sljit_jump *start;
196    struct sljit_label *quit;    struct sljit_label *quit;
197    struct stub_list *next;    struct stub_list *next;
198  } stub_list;  } stub_list;
199    
200    enum frame_types {
201      no_frame = -1,
202      no_stack = -2
203    };
204    
205    enum control_types {
206      type_mark = 0,
207      type_then_trap = 1
208    };
209    
210  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);
211    
212  /* The following structure is the key data type for the recursive  /* The following structure is the key data type for the recursive
213  code generator. It is allocated by compile_matchingpath, and contains  code generator. It is allocated by compile_matchingpath, and contains
214  the aguments for compile_backtrackingpath. Must be the first member  the arguments for compile_backtrackingpath. Must be the first member
215  of its descendants. */  of its descendants. */
216  typedef struct backtrack_common {  typedef struct backtrack_common {
217    /* Concatenation stack. */    /* Concatenation stack. */
# Line 216  typedef struct backtrack_common { Line 227  typedef struct backtrack_common {
227  typedef struct assert_backtrack {  typedef struct assert_backtrack {
228    backtrack_common common;    backtrack_common common;
229    jump_list *condfailed;    jump_list *condfailed;
230    /* Less than 0 (-1) if a frame is not needed. */    /* Less than 0 if a frame is not needed. */
231    int framesize;    int framesize;
232    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
233    int private_data_ptr;    int private_data_ptr;
# Line 237  typedef struct bracket_backtrack { Line 248  typedef struct bracket_backtrack {
248      /* Both for OP_COND, OP_SCOND. */      /* Both for OP_COND, OP_SCOND. */
249      jump_list *condfailed;      jump_list *condfailed;
250      assert_backtrack *assert;      assert_backtrack *assert;
251      /* For OP_ONCE. -1 if not needed. */      /* For OP_ONCE. Less than 0 if not needed. */
252      int framesize;      int framesize;
253    } u;    } u;
254    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
# Line 272  typedef struct recurse_entry { Line 283  typedef struct recurse_entry {
283    /* Collects the calls until the function is not created. */    /* Collects the calls until the function is not created. */
284    jump_list *calls;    jump_list *calls;
285    /* Points to the starting opcode. */    /* Points to the starting opcode. */
286    int start;    sljit_sw start;
287  } recurse_entry;  } recurse_entry;
288    
289  typedef struct recurse_backtrack {  typedef struct recurse_backtrack {
290    backtrack_common common;    backtrack_common common;
291      BOOL inlined_pattern;
292  } recurse_backtrack;  } recurse_backtrack;
293    
294  #define MAX_RANGE_SIZE 6  #define OP_THEN_TRAP OP_TABLE_LENGTH
295    
296    typedef struct then_trap_backtrack {
297      backtrack_common common;
298      /* If then_trap is not NULL, this structure contains the real
299      then_trap for the backtracking path. */
300      struct then_trap_backtrack *then_trap;
301      /* Points to the starting opcode. */
302      sljit_sw start;
303      /* Exit point for the then opcodes of this alternative. */
304      jump_list *quit;
305      /* Frame size of the current alternative. */
306      int framesize;
307    } then_trap_backtrack;
308    
309    #define MAX_RANGE_SIZE 4
310    
311  typedef struct compiler_common {  typedef struct compiler_common {
312      /* The sljit ceneric compiler. */
313    struct sljit_compiler *compiler;    struct sljit_compiler *compiler;
314      /* First byte code. */
315    pcre_uchar *start;    pcre_uchar *start;
   
316    /* Maps private data offset to each opcode. */    /* Maps private data offset to each opcode. */
317    int *private_data_ptrs;    sljit_si *private_data_ptrs;
318    /* Tells whether the capturing bracket is optimized. */    /* Tells whether the capturing bracket is optimized. */
319    pcre_uint8 *optimized_cbracket;    pcre_uint8 *optimized_cbracket;
320      /* Tells whether the starting offset is a target of then. */
321      pcre_uint8 *then_offsets;
322      /* Current position where a THEN must jump. */
323      then_trap_backtrack *then_trap;
324    /* Starting offset of private data for capturing brackets. */    /* Starting offset of private data for capturing brackets. */
325    int cbraptr;    int cbra_ptr;
326    /* OVector starting point. Must be divisible by 2. */    /* Output vector starting point. Must be divisible by 2. */
327    int ovector_start;    int ovector_start;
328    /* Last known position of the requested byte. */    /* Last known position of the requested byte. */
329    int req_char_ptr;    int req_char_ptr;
# Line 305  typedef struct compiler_common { Line 337  typedef struct compiler_common {
337    int first_line_end;    int first_line_end;
338    /* Points to the marked string. */    /* Points to the marked string. */
339    int mark_ptr;    int mark_ptr;
340      /* Recursive control verb management chain. */
341      int control_head_ptr;
342    /* Points to the last matched capture block index. */    /* Points to the last matched capture block index. */
343    int capture_last_ptr;    int capture_last_ptr;
344      /* Points to the starting position of the current match. */
345      int start_ptr;
346    
347    /* Flipped and lower case tables. */    /* Flipped and lower case tables. */
348    const pcre_uint8 *fcc;    const pcre_uint8 *fcc;
349    sljit_sw lcc;    sljit_sw lcc;
350    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */
351    int mode;    int mode;
352      /* \K is found in the pattern. */
353      BOOL has_set_som;
354      /* (*SKIP:arg) is found in the pattern. */
355      BOOL has_skip_arg;
356      /* (*THEN) is found in the pattern. */
357      BOOL has_then;
358      /* Needs to know the start position anytime. */
359      BOOL needs_start_ptr;
360      /* Currently in recurse or negative assert. */
361      BOOL local_exit;
362      /* Currently in a positive assert. */
363      BOOL positive_assert;
364    /* Newline control. */    /* Newline control. */
365    int nltype;    int nltype;
366      pcre_uint32 nlmax;
367    int newline;    int newline;
368    int bsr_nltype;    int bsr_nltype;
369      pcre_uint32 bsr_nlmax;
370    /* Dollar endonly. */    /* Dollar endonly. */
371    int endonly;    int endonly;
   BOOL has_set_som;  
372    /* Tables. */    /* Tables. */
373    sljit_sw ctypes;    sljit_sw ctypes;
   int digits[2 + MAX_RANGE_SIZE];  
374    /* Named capturing brackets. */    /* Named capturing brackets. */
375    sljit_uw name_table;    pcre_uchar *name_table;
376    sljit_sw name_count;    sljit_sw name_count;
377    sljit_sw name_entry_size;    sljit_sw name_entry_size;
378    
# Line 338  typedef struct compiler_common { Line 386  typedef struct compiler_common {
386    recurse_entry *currententry;    recurse_entry *currententry;
387    jump_list *partialmatch;    jump_list *partialmatch;
388    jump_list *quit;    jump_list *quit;
389      jump_list *positive_assert_quit;
390    jump_list *forced_quit;    jump_list *forced_quit;
391    jump_list *accept;    jump_list *accept;
392    jump_list *calllimit;    jump_list *calllimit;
# Line 349  typedef struct compiler_common { Line 398  typedef struct compiler_common {
398    jump_list *vspace;    jump_list *vspace;
399    jump_list *casefulcmp;    jump_list *casefulcmp;
400    jump_list *caselesscmp;    jump_list *caselesscmp;
401      jump_list *reset_match;
402    BOOL jscript_compat;    BOOL jscript_compat;
403  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
404    BOOL utf;    BOOL utf;
405  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
406    BOOL use_ucp;    BOOL use_ucp;
407  #endif  #endif
 #ifndef COMPILE_PCRE32  
   jump_list *utfreadchar;  
 #endif  
408  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
409      jump_list *utfreadchar;
410      jump_list *utfreadchar16;
411    jump_list *utfreadtype8;    jump_list *utfreadtype8;
412  #endif  #endif
413  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
# Line 415  typedef struct compare_context { Line 464  typedef struct compare_context {
464  #define STACK_TOP     SLJIT_SCRATCH_REG2  #define STACK_TOP     SLJIT_SCRATCH_REG2
465  #define STACK_LIMIT   SLJIT_SAVED_REG3  #define STACK_LIMIT   SLJIT_SAVED_REG3
466  #define ARGUMENTS     SLJIT_SAVED_EREG1  #define ARGUMENTS     SLJIT_SAVED_EREG1
467  #define CALL_COUNT    SLJIT_SAVED_EREG2  #define COUNT_MATCH   SLJIT_SAVED_EREG2
468  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1
469    
470  /* Local space layout. */  /* Local space layout. */
# Line 426  typedef struct compare_context { Line 475  typedef struct compare_context {
475  #define POSSESSIVE0      (2 * sizeof(sljit_sw))  #define POSSESSIVE0      (2 * sizeof(sljit_sw))
476  #define POSSESSIVE1      (3 * sizeof(sljit_sw))  #define POSSESSIVE1      (3 * sizeof(sljit_sw))
477  /* Max limit of recursions. */  /* Max limit of recursions. */
478  #define CALL_LIMIT       (4 * sizeof(sljit_sw))  #define LIMIT_MATCH      (4 * sizeof(sljit_sw))
479  /* The output vector is stored on the stack, and contains pointers  /* The output vector is stored on the stack, and contains pointers
480  to characters. The vector data is divided into two groups: the first  to characters. The vector data is divided into two groups: the first
481  group contains the start / end character pointers, and the second is  group contains the start / end character pointers, and the second is
482  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. */
483  #define OVECTOR_START    (common->ovector_start)  #define OVECTOR_START    (common->ovector_start)
484  #define OVECTOR(i)       (OVECTOR_START + (i) * sizeof(sljit_sw))  #define OVECTOR(i)       (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw))
485  #define OVECTOR_PRIV(i)  (common->cbraptr + (i) * sizeof(sljit_sw))  #define OVECTOR_PRIV(i)  (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw))
486  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])
487    
488  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
# Line 475  the start pointers when the end of the c Line 524  the start pointers when the end of the c
524  #define GET_LOCAL_BASE(dst, dstw, offset) \  #define GET_LOCAL_BASE(dst, dstw, offset) \
525    sljit_get_local_base(compiler, (dst), (dstw), (offset))    sljit_get_local_base(compiler, (dst), (dstw), (offset))
526    
527    #define READ_CHAR_ANY 0x7fffffff
528    
529  static pcre_uchar* bracketend(pcre_uchar* cc)  static pcre_uchar* bracketend(pcre_uchar* cc)
530  {  {
531  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 484  cc += 1 + LINK_SIZE; Line 535  cc += 1 + LINK_SIZE;
535  return cc;  return cc;
536  }  }
537    
538    static int ones_in_half_byte[16] = {
539      /* 0 */ 0, 1, 1, 2, /* 4 */ 1, 2, 2, 3,
540      /* 8 */ 1, 2, 2, 3, /* 12 */ 2, 3, 3, 4
541    };
542    
543  /* Functions whose might need modification for all new supported opcodes:  /* Functions whose might need modification for all new supported opcodes:
544   next_opcode   next_opcode
545   get_private_data_length   check_opcode_types
546   set_private_data_ptrs   set_private_data_ptrs
547   get_framesize   get_framesize
548   init_frame   init_frame
549   get_private_data_length_for_copy   get_private_data_copy_length
550   copy_private_data   copy_private_data
551   compile_matchingpath   compile_matchingpath
552   compile_backtrackingpath   compile_backtrackingpath
# Line 514  switch(*cc) Line 570  switch(*cc)
570    case OP_WORDCHAR:    case OP_WORDCHAR:
571    case OP_ANY:    case OP_ANY:
572    case OP_ALLANY:    case OP_ALLANY:
573      case OP_NOTPROP:
574      case OP_PROP:
575    case OP_ANYNL:    case OP_ANYNL:
576    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
577    case OP_HSPACE:    case OP_HSPACE:
# Line 526  switch(*cc) Line 584  switch(*cc)
584    case OP_CIRCM:    case OP_CIRCM:
585    case OP_DOLL:    case OP_DOLL:
586    case OP_DOLLM:    case OP_DOLLM:
   case OP_TYPESTAR:  
   case OP_TYPEMINSTAR:  
   case OP_TYPEPLUS:  
   case OP_TYPEMINPLUS:  
   case OP_TYPEQUERY:  
   case OP_TYPEMINQUERY:  
   case OP_TYPEPOSSTAR:  
   case OP_TYPEPOSPLUS:  
   case OP_TYPEPOSQUERY:  
587    case OP_CRSTAR:    case OP_CRSTAR:
588    case OP_CRMINSTAR:    case OP_CRMINSTAR:
589    case OP_CRPLUS:    case OP_CRPLUS:
590    case OP_CRMINPLUS:    case OP_CRMINPLUS:
591    case OP_CRQUERY:    case OP_CRQUERY:
592    case OP_CRMINQUERY:    case OP_CRMINQUERY:
593      case OP_CRRANGE:
594      case OP_CRMINRANGE:
595      case OP_CRPOSSTAR:
596      case OP_CRPOSPLUS:
597      case OP_CRPOSQUERY:
598      case OP_CRPOSRANGE:
599      case OP_CLASS:
600      case OP_NCLASS:
601      case OP_REF:
602      case OP_REFI:
603      case OP_DNREF:
604      case OP_DNREFI:
605      case OP_RECURSE:
606      case OP_CALLOUT:
607      case OP_ALT:
608      case OP_KET:
609      case OP_KETRMAX:
610      case OP_KETRMIN:
611      case OP_KETRPOS:
612      case OP_REVERSE:
613      case OP_ASSERT:
614      case OP_ASSERT_NOT:
615      case OP_ASSERTBACK:
616      case OP_ASSERTBACK_NOT:
617      case OP_ONCE:
618      case OP_ONCE_NC:
619      case OP_BRA:
620      case OP_BRAPOS:
621      case OP_CBRA:
622      case OP_CBRAPOS:
623      case OP_COND:
624      case OP_SBRA:
625      case OP_SBRAPOS:
626      case OP_SCBRA:
627      case OP_SCBRAPOS:
628      case OP_SCOND:
629      case OP_CREF:
630      case OP_DNCREF:
631      case OP_RREF:
632      case OP_DNRREF:
633    case OP_DEF:    case OP_DEF:
634    case OP_BRAZERO:    case OP_BRAZERO:
635    case OP_BRAMINZERO:    case OP_BRAMINZERO:
636    case OP_BRAPOSZERO:    case OP_BRAPOSZERO:
637      case OP_PRUNE:
638      case OP_SKIP:
639      case OP_THEN:
640    case OP_COMMIT:    case OP_COMMIT:
641    case OP_FAIL:    case OP_FAIL:
642    case OP_ACCEPT:    case OP_ACCEPT:
643    case OP_ASSERT_ACCEPT:    case OP_ASSERT_ACCEPT:
644      case OP_CLOSE:
645    case OP_SKIPZERO:    case OP_SKIPZERO:
646    return cc + 1;    return cc + PRIV(OP_lengths)[*cc];
   
   case OP_ANYBYTE:  
 #ifdef SUPPORT_UTF  
   if (common->utf) return NULL;  
 #endif  
   return cc + 1;  
647    
648    case OP_CHAR:    case OP_CHAR:
649    case OP_CHARI:    case OP_CHARI:
# Line 568  switch(*cc) Line 655  switch(*cc)
655    case OP_MINPLUS:    case OP_MINPLUS:
656    case OP_QUERY:    case OP_QUERY:
657    case OP_MINQUERY:    case OP_MINQUERY:
658      case OP_UPTO:
659      case OP_MINUPTO:
660      case OP_EXACT:
661    case OP_POSSTAR:    case OP_POSSTAR:
662    case OP_POSPLUS:    case OP_POSPLUS:
663    case OP_POSQUERY:    case OP_POSQUERY:
664      case OP_POSUPTO:
665    case OP_STARI:    case OP_STARI:
666    case OP_MINSTARI:    case OP_MINSTARI:
667    case OP_PLUSI:    case OP_PLUSI:
668    case OP_MINPLUSI:    case OP_MINPLUSI:
669    case OP_QUERYI:    case OP_QUERYI:
670    case OP_MINQUERYI:    case OP_MINQUERYI:
671      case OP_UPTOI:
672      case OP_MINUPTOI:
673      case OP_EXACTI:
674    case OP_POSSTARI:    case OP_POSSTARI:
675    case OP_POSPLUSI:    case OP_POSPLUSI:
676    case OP_POSQUERYI:    case OP_POSQUERYI:
677      case OP_POSUPTOI:
678    case OP_NOTSTAR:    case OP_NOTSTAR:
679    case OP_NOTMINSTAR:    case OP_NOTMINSTAR:
680    case OP_NOTPLUS:    case OP_NOTPLUS:
681    case OP_NOTMINPLUS:    case OP_NOTMINPLUS:
682    case OP_NOTQUERY:    case OP_NOTQUERY:
683    case OP_NOTMINQUERY:    case OP_NOTMINQUERY:
684      case OP_NOTUPTO:
685      case OP_NOTMINUPTO:
686      case OP_NOTEXACT:
687    case OP_NOTPOSSTAR:    case OP_NOTPOSSTAR:
688    case OP_NOTPOSPLUS:    case OP_NOTPOSPLUS:
689    case OP_NOTPOSQUERY:    case OP_NOTPOSQUERY:
690      case OP_NOTPOSUPTO:
691    case OP_NOTSTARI:    case OP_NOTSTARI:
692    case OP_NOTMINSTARI:    case OP_NOTMINSTARI:
693    case OP_NOTPLUSI:    case OP_NOTPLUSI:
694    case OP_NOTMINPLUSI:    case OP_NOTMINPLUSI:
695    case OP_NOTQUERYI:    case OP_NOTQUERYI:
696    case OP_NOTMINQUERYI:    case OP_NOTMINQUERYI:
   case OP_NOTPOSSTARI:  
   case OP_NOTPOSPLUSI:  
   case OP_NOTPOSQUERYI:  
   cc += 2;  
 #ifdef SUPPORT_UTF  
   if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
   return cc;  
   
   case OP_UPTO:  
   case OP_MINUPTO:  
   case OP_EXACT:  
   case OP_POSUPTO:  
   case OP_UPTOI:  
   case OP_MINUPTOI:  
   case OP_EXACTI:  
   case OP_POSUPTOI:  
   case OP_NOTUPTO:  
   case OP_NOTMINUPTO:  
   case OP_NOTEXACT:  
   case OP_NOTPOSUPTO:  
697    case OP_NOTUPTOI:    case OP_NOTUPTOI:
698    case OP_NOTMINUPTOI:    case OP_NOTMINUPTOI:
699    case OP_NOTEXACTI:    case OP_NOTEXACTI:
700      case OP_NOTPOSSTARI:
701      case OP_NOTPOSPLUSI:
702      case OP_NOTPOSQUERYI:
703    case OP_NOTPOSUPTOI:    case OP_NOTPOSUPTOI:
704    cc += 2 + IMM2_SIZE;    cc += PRIV(OP_lengths)[*cc];
705  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
706    if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);    if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
707  #endif  #endif
708    return cc;    return cc;
709    
710    case OP_NOTPROP:    /* Special cases. */
711    case OP_PROP:    case OP_TYPESTAR:
712    return cc + 1 + 2;    case OP_TYPEMINSTAR:
713      case OP_TYPEPLUS:
714      case OP_TYPEMINPLUS:
715      case OP_TYPEQUERY:
716      case OP_TYPEMINQUERY:
717    case OP_TYPEUPTO:    case OP_TYPEUPTO:
718    case OP_TYPEMINUPTO:    case OP_TYPEMINUPTO:
719    case OP_TYPEEXACT:    case OP_TYPEEXACT:
720      case OP_TYPEPOSSTAR:
721      case OP_TYPEPOSPLUS:
722      case OP_TYPEPOSQUERY:
723    case OP_TYPEPOSUPTO:    case OP_TYPEPOSUPTO:
724    case OP_REF:    return cc + PRIV(OP_lengths)[*cc] - 1;
   case OP_REFI:  
   case OP_CREF:  
   case OP_NCREF:  
   case OP_RREF:  
   case OP_NRREF:  
   case OP_CLOSE:  
   cc += 1 + IMM2_SIZE;  
   return cc;  
   
   case OP_CRRANGE:  
   case OP_CRMINRANGE:  
   return cc + 1 + 2 * IMM2_SIZE;  
725    
726    case OP_CLASS:    case OP_ANYBYTE:
727    case OP_NCLASS:  #ifdef SUPPORT_UTF
728    return cc + 1 + 32 / sizeof(pcre_uchar);    if (common->utf) return NULL;
729    #endif
730      return cc + 1;
731    
732  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
733    case OP_XCLASS:    case OP_XCLASS:
734    return cc + GET(cc, 1);    return cc + GET(cc, 1);
735  #endif  #endif
736    
   case OP_RECURSE:  
   case OP_ASSERT:  
   case OP_ASSERT_NOT:  
   case OP_ASSERTBACK:  
   case OP_ASSERTBACK_NOT:  
   case OP_REVERSE:  
   case OP_ONCE:  
   case OP_ONCE_NC:  
   case OP_BRA:  
   case OP_BRAPOS:  
   case OP_COND:  
   case OP_SBRA:  
   case OP_SBRAPOS:  
   case OP_SCOND:  
   case OP_ALT:  
   case OP_KET:  
   case OP_KETRMAX:  
   case OP_KETRMIN:  
   case OP_KETRPOS:  
   return cc + 1 + LINK_SIZE;  
   
   case OP_CBRA:  
   case OP_CBRAPOS:  
   case OP_SCBRA:  
   case OP_SCBRAPOS:  
   return cc + 1 + LINK_SIZE + IMM2_SIZE;  
   
737    case OP_MARK:    case OP_MARK:
738      case OP_PRUNE_ARG:
739      case OP_SKIP_ARG:
740      case OP_THEN_ARG:
741    return cc + 1 + 2 + cc[1];    return cc + 1 + 2 + cc[1];
742    
   case OP_CALLOUT:  
   return cc + 2 + 2 * LINK_SIZE;  
   
743    default:    default:
744      /* All opcodes are supported now! */
745      SLJIT_ASSERT_STOP();
746    return NULL;    return NULL;
747    }    }
748  }  }
749    
750  #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)  
751  {  {
752  int private_data_length = 0;  int count;
753  pcre_uchar *alternative;  pcre_uchar *slot;
 pcre_uchar *name;  
 pcre_uchar *end = NULL;  
 int space, size, i;  
 pcre_uint32 bracketlen;  
754    
755  /* 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. */
756  while (cc < ccend)  while (cc < ccend)
757    {    {
   space = 0;  
   size = 0;  
   bracketlen = 0;  
758    switch(*cc)    switch(*cc)
759      {      {
760      case OP_SET_SOM:      case OP_SET_SOM:
# Line 800  while (cc < ccend) Line 768  while (cc < ccend)
768      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
769      break;      break;
770    
     case OP_ASSERT:  
     case OP_ASSERT_NOT:  
     case OP_ASSERTBACK:  
     case OP_ASSERTBACK_NOT:  
     case OP_ONCE:  
     case OP_ONCE_NC:  
     case OP_BRAPOS:  
     case OP_SBRA:  
     case OP_SBRAPOS:  
     private_data_length += sizeof(sljit_sw);  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
771      case OP_CBRAPOS:      case OP_CBRAPOS:
772      case OP_SCBRAPOS:      case OP_SCBRAPOS:
     private_data_length += sizeof(sljit_sw);  
773      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;
774      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;      cc += 1 + LINK_SIZE + IMM2_SIZE;
775      break;      break;
776    
777      case OP_COND:      case OP_COND:
# Line 825  while (cc < ccend) Line 779  while (cc < ccend)
779      /* Only AUTO_CALLOUT can insert this opcode. We do      /* Only AUTO_CALLOUT can insert this opcode. We do
780         not intend to support this case. */         not intend to support this case. */
781      if (cc[1 + LINK_SIZE] == OP_CALLOUT)      if (cc[1 + LINK_SIZE] == OP_CALLOUT)
782        return -1;        return FALSE;
783        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;  
784      break;      break;
785    
786      case OP_CREF:      case OP_CREF:
787      i = GET2(cc, 1);      common->optimized_cbracket[GET2(cc, 1)] = 0;
     common->optimized_cbracket[i] = 0;  
788      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
789      break;      break;
790    
791      case OP_NCREF:      case OP_DNREF:
792      bracketlen = GET2(cc, 1);      case OP_DNREFI:
793      name = (pcre_uchar *)common->name_table;      case OP_DNCREF:
794      alternative = name;      count = GET2(cc, 1 + IMM2_SIZE);
795      for (i = 0; i < common->name_count; i++)      slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
796        while (count-- > 0)
797        {        {
798        if (GET2(name, 0) == bracketlen) break;        common->optimized_cbracket[GET2(slot, 0)] = 0;
799        name += common->name_entry_size;        slot += common->name_entry_size;
800        }        }
801      SLJIT_ASSERT(i != common->name_count);      cc += 1 + 2 * IMM2_SIZE;
   
     for (i = 0; i < common->name_count; i++)  
       {  
       if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0)  
         common->optimized_cbracket[GET2(alternative, 0)] = 0;  
       alternative += common->name_entry_size;  
       }  
     bracketlen = 0;  
     cc += 1 + IMM2_SIZE;  
802      break;      break;
803    
     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);  
     break;  
   
 #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  
     case OP_XCLASS:  
     size = GET(cc, 1);  
     space = get_class_iterator_size(cc + size);  
     break;  
 #endif  
   
804      case OP_RECURSE:      case OP_RECURSE:
805      /* Set its value only once. */      /* Set its value only once. */
806      if (common->recursive_head_ptr == 0)      if (common->recursive_head_ptr == 0)
# Line 939  while (cc < ccend) Line 820  while (cc < ccend)
820      cc += 2 + 2 * LINK_SIZE;      cc += 2 + 2 * LINK_SIZE;
821      break;      break;
822    
823        case OP_THEN_ARG:
824        common->has_then = TRUE;
825        common->control_head_ptr = 1;
826        /* Fall through. */
827    
828        case OP_PRUNE_ARG:
829        common->needs_start_ptr = TRUE;
830        /* Fall through. */
831    
832      case OP_MARK:      case OP_MARK:
833      if (common->mark_ptr == 0)      if (common->mark_ptr == 0)
834        {        {
# Line 948  while (cc < ccend) Line 838  while (cc < ccend)
838      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
839      break;      break;
840    
841        case OP_THEN:
842        common->has_then = TRUE;
843        common->control_head_ptr = 1;
844        /* Fall through. */
845    
846        case OP_PRUNE:
847        case OP_SKIP:
848        common->needs_start_ptr = TRUE;
849        cc += 1;
850        break;
851    
852        case OP_SKIP_ARG:
853        common->control_head_ptr = 1;
854        common->has_skip_arg = TRUE;
855        cc += 1 + 2 + cc[1];
856        break;
857    
858      default:      default:
859      cc = next_opcode(common, cc);      cc = next_opcode(common, cc);
860      if (cc == NULL)      if (cc == NULL)
861        return -1;        return FALSE;
862      break;      break;
863      }      }
864      }
865    return TRUE;
866    }
867    
868    if (space > 0 && cc >= end)  static int get_class_iterator_size(pcre_uchar *cc)
869      private_data_length += sizeof(sljit_sw) * space;  {
870    switch(*cc)
871      {
872      case OP_CRSTAR:
873      case OP_CRPLUS:
874      return 2;
875    
876    if (size != 0)    case OP_CRMINSTAR:
877      case OP_CRMINPLUS:
878      case OP_CRQUERY:
879      case OP_CRMINQUERY:
880      return 1;
881    
882      case OP_CRRANGE:
883      case OP_CRMINRANGE:
884      if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))
885        return 0;
886      return 2;
887    
888      default:
889      return 0;
890      }
891    }
892    
893    static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin)
894    {
895    pcre_uchar *end = bracketend(begin);
896    pcre_uchar *next;
897    pcre_uchar *next_end;
898    pcre_uchar *max_end;
899    pcre_uchar type;
900    sljit_sw length = end - begin;
901    int min, max, i;
902    
903    /* Detect fixed iterations first. */
904    if (end[-(1 + LINK_SIZE)] != OP_KET)
905      return FALSE;
906    
907    /* Already detected repeat. */
908    if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0)
909      return TRUE;
910    
911    next = end;
912    min = 1;
913    while (1)
914      {
915      if (*next != *begin)
916        break;
917      next_end = bracketend(next);
918      if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0)
919        break;
920      next = next_end;
921      min++;
922      }
923    
924    if (min == 2)
925      return FALSE;
926    
927    max = 0;
928    max_end = next;
929    if (*next == OP_BRAZERO || *next == OP_BRAMINZERO)
930      {
931      type = *next;
932      while (1)
933      {      {
934      if (size < 0)      if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin)
935        {        break;
936        cc += -size;      next_end = bracketend(next + 2 + LINK_SIZE);
937  #ifdef SUPPORT_UTF      if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0)
938        if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);        break;
939  #endif      next = next_end;
940        }      max++;
     else  
       cc += size;  
941      }      }
942    
943    if (bracketlen != 0)    if (next[0] == type && next[1] == *begin && max >= 1)
944      {      {
945      if (cc >= end)      next_end = bracketend(next + 1);
946        if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0)
947        {        {
948        end = bracketend(cc);        for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE)
949        if (end[-1 - LINK_SIZE] == OP_KET)          if (*next_end != OP_KET)
950          end = NULL;            break;
951    
952          if (i == max)
953            {
954            common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end;
955            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO;
956            /* +2 the original and the last. */
957            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2;
958            if (min == 1)
959              return TRUE;
960            min--;
961            max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE);
962            }
963        }        }
     cc += bracketlen;  
964      }      }
965    }    }
966  return private_data_length;  
967    if (min >= 3)
968      {
969      common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end;
970      common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT;
971      common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min;
972      return TRUE;
973      }
974    
975    return FALSE;
976  }  }
977    
978  static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend)  #define CASE_ITERATOR_PRIVATE_DATA_1 \
979        case OP_MINSTAR: \
980        case OP_MINPLUS: \
981        case OP_QUERY: \
982        case OP_MINQUERY: \
983        case OP_MINSTARI: \
984        case OP_MINPLUSI: \
985        case OP_QUERYI: \
986        case OP_MINQUERYI: \
987        case OP_NOTMINSTAR: \
988        case OP_NOTMINPLUS: \
989        case OP_NOTQUERY: \
990        case OP_NOTMINQUERY: \
991        case OP_NOTMINSTARI: \
992        case OP_NOTMINPLUSI: \
993        case OP_NOTQUERYI: \
994        case OP_NOTMINQUERYI:
995    
996    #define CASE_ITERATOR_PRIVATE_DATA_2A \
997        case OP_STAR: \
998        case OP_PLUS: \
999        case OP_STARI: \
1000        case OP_PLUSI: \
1001        case OP_NOTSTAR: \
1002        case OP_NOTPLUS: \
1003        case OP_NOTSTARI: \
1004        case OP_NOTPLUSI:
1005    
1006    #define CASE_ITERATOR_PRIVATE_DATA_2B \
1007        case OP_UPTO: \
1008        case OP_MINUPTO: \
1009        case OP_UPTOI: \
1010        case OP_MINUPTOI: \
1011        case OP_NOTUPTO: \
1012        case OP_NOTMINUPTO: \
1013        case OP_NOTUPTOI: \
1014        case OP_NOTMINUPTOI:
1015    
1016    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \
1017        case OP_TYPEMINSTAR: \
1018        case OP_TYPEMINPLUS: \
1019        case OP_TYPEQUERY: \
1020        case OP_TYPEMINQUERY:
1021    
1022    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \
1023        case OP_TYPESTAR: \
1024        case OP_TYPEPLUS:
1025    
1026    #define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \
1027        case OP_TYPEUPTO: \
1028        case OP_TYPEMINUPTO:
1029    
1030    static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend)
1031  {  {
1032  pcre_uchar *cc = common->start;  pcre_uchar *cc = common->start;
1033  pcre_uchar *alternative;  pcre_uchar *alternative;
1034  pcre_uchar *end = NULL;  pcre_uchar *end = NULL;
1035    int private_data_ptr = *private_data_start;
1036  int space, size, bracketlen;  int space, size, bracketlen;
1037    
1038  while (cc < ccend)  while (cc < ccend)
# Line 997  while (cc < ccend) Line 1040  while (cc < ccend)
1040    space = 0;    space = 0;
1041    size = 0;    size = 0;
1042    bracketlen = 0;    bracketlen = 0;
1043      if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE)
1044        return;
1045    
1046      if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND)
1047        if (detect_repeat(common, cc))
1048          {
1049          /* These brackets are converted to repeats, so no global
1050          based single character repeat is allowed. */
1051          if (cc >= end)
1052            end = bracketend(cc);
1053          }
1054    
1055    switch(*cc)    switch(*cc)
1056      {      {
1057        case OP_KET:
1058        if (common->private_data_ptrs[cc + 1 - common->start] != 0)
1059          {
1060          common->private_data_ptrs[cc - common->start] = private_data_ptr;
1061          private_data_ptr += sizeof(sljit_sw);
1062          cc += common->private_data_ptrs[cc + 1 - common->start];
1063          }
1064        cc += 1 + LINK_SIZE;
1065        break;
1066    
1067      case OP_ASSERT:      case OP_ASSERT:
1068      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1069      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1092  while (cc < ccend) Line 1157  while (cc < ccend)
1157      break;      break;
1158      }      }
1159    
1160      /* Character iterators, which are not inside a repeated bracket,
1161         gets a private slot instead of allocating it on the stack. */
1162    if (space > 0 && cc >= end)    if (space > 0 && cc >= end)
1163      {      {
1164      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
# Line 1122  while (cc < ccend) Line 1189  while (cc < ccend)
1189      cc += bracketlen;      cc += bracketlen;
1190      }      }
1191    }    }
1192    *private_data_start = private_data_ptr;
1193  }  }
1194    
1195  /* Returns with -1 if no need for frame. */  /* Returns with a frame_types (always < 0) if no need for frame. */
1196  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)
1197  {  {
 pcre_uchar *ccend = bracketend(cc);  
1198  int length = 0;  int length = 0;
1199  int possessive = 0;  int possessive = 0;
1200    BOOL stack_restore = FALSE;
1201  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1202  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1203  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
1204  BOOL capture_last_found = FALSE;  BOOL capture_last_found = FALSE;
1205    
1206  if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))  #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
1207    SLJIT_ASSERT(common->control_head_ptr != 0);
1208    *needs_control_head = TRUE;
1209    #else
1210    *needs_control_head = FALSE;
1211    #endif
1212    
1213    if (ccend == NULL)
1214    {    {
1215    possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;    ccend = bracketend(cc) - (1 + LINK_SIZE);
1216    /* This is correct regardless of common->capture_last_ptr. */    if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))
1217    capture_last_found = TRUE;      {
1218        possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;
1219        /* This is correct regardless of common->capture_last_ptr. */
1220        capture_last_found = TRUE;
1221        }
1222      cc = next_opcode(common, cc);
1223    }    }
1224    
 cc = next_opcode(common, cc);  
1225  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1226  while (cc < ccend)  while (cc < ccend)
1227    switch(*cc)    switch(*cc)
1228      {      {
1229      case OP_SET_SOM:      case OP_SET_SOM:
1230      SLJIT_ASSERT(common->has_set_som);      SLJIT_ASSERT(common->has_set_som);
1231        stack_restore = TRUE;
1232      if (!setsom_found)      if (!setsom_found)
1233        {        {
1234        length += 2;        length += 2;
# Line 1158  while (cc < ccend) Line 1238  while (cc < ccend)
1238      break;      break;
1239    
1240      case OP_MARK:      case OP_MARK:
1241        case OP_PRUNE_ARG:
1242        case OP_THEN_ARG:
1243      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1244        stack_restore = TRUE;
1245      if (!setmark_found)      if (!setmark_found)
1246        {        {
1247        length += 2;        length += 2;
1248        setmark_found = TRUE;        setmark_found = TRUE;
1249        }        }
1250        if (common->control_head_ptr != 0)
1251          *needs_control_head = TRUE;
1252      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
1253      break;      break;
1254    
1255      case OP_RECURSE:      case OP_RECURSE:
1256        stack_restore = TRUE;
1257      if (common->has_set_som && !setsom_found)      if (common->has_set_som && !setsom_found)
1258        {        {
1259        length += 2;        length += 2;
# Line 1190  while (cc < ccend) Line 1276  while (cc < ccend)
1276      case OP_CBRAPOS:      case OP_CBRAPOS:
1277      case OP_SCBRA:      case OP_SCBRA:
1278      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1279        stack_restore = TRUE;
1280      if (common->capture_last_ptr != 0 && !capture_last_found)      if (common->capture_last_ptr != 0 && !capture_last_found)
1281        {        {
1282        length += 2;        length += 2;
# Line 1200  while (cc < ccend) Line 1287  while (cc < ccend)
1287      break;      break;
1288    
1289      default:      default:
1290        stack_restore = TRUE;
1291        /* Fall through. */
1292    
1293        case OP_NOT_WORD_BOUNDARY:
1294        case OP_WORD_BOUNDARY:
1295        case OP_NOT_DIGIT:
1296        case OP_DIGIT:
1297        case OP_NOT_WHITESPACE:
1298        case OP_WHITESPACE:
1299        case OP_NOT_WORDCHAR:
1300        case OP_WORDCHAR:
1301        case OP_ANY:
1302        case OP_ALLANY:
1303        case OP_ANYBYTE:
1304        case OP_NOTPROP:
1305        case OP_PROP:
1306        case OP_ANYNL:
1307        case OP_NOT_HSPACE:
1308        case OP_HSPACE:
1309        case OP_NOT_VSPACE:
1310        case OP_VSPACE:
1311        case OP_EXTUNI:
1312        case OP_EODN:
1313        case OP_EOD:
1314        case OP_CIRC:
1315        case OP_CIRCM:
1316        case OP_DOLL:
1317        case OP_DOLLM:
1318        case OP_CHAR:
1319        case OP_CHARI:
1320        case OP_NOT:
1321        case OP_NOTI:
1322    
1323        case OP_EXACT:
1324        case OP_POSSTAR:
1325        case OP_POSPLUS:
1326        case OP_POSQUERY:
1327        case OP_POSUPTO:
1328    
1329        case OP_EXACTI:
1330        case OP_POSSTARI:
1331        case OP_POSPLUSI:
1332        case OP_POSQUERYI:
1333        case OP_POSUPTOI:
1334    
1335        case OP_NOTEXACT:
1336        case OP_NOTPOSSTAR:
1337        case OP_NOTPOSPLUS:
1338        case OP_NOTPOSQUERY:
1339        case OP_NOTPOSUPTO:
1340    
1341        case OP_NOTEXACTI:
1342        case OP_NOTPOSSTARI:
1343        case OP_NOTPOSPLUSI:
1344        case OP_NOTPOSQUERYI:
1345        case OP_NOTPOSUPTOI:
1346    
1347        case OP_TYPEEXACT:
1348        case OP_TYPEPOSSTAR:
1349        case OP_TYPEPOSPLUS:
1350        case OP_TYPEPOSQUERY:
1351        case OP_TYPEPOSUPTO:
1352    
1353        case OP_CLASS:
1354        case OP_NCLASS:
1355        case OP_XCLASS:
1356    
1357      cc = next_opcode(common, cc);      cc = next_opcode(common, cc);
1358      SLJIT_ASSERT(cc != NULL);      SLJIT_ASSERT(cc != NULL);
1359      break;      break;
# Line 1207  while (cc < ccend) Line 1361  while (cc < ccend)
1361    
1362  /* Possessive quantifiers can use a special case. */  /* Possessive quantifiers can use a special case. */
1363  if (SLJIT_UNLIKELY(possessive == length))  if (SLJIT_UNLIKELY(possessive == length))
1364    return -1;    return stack_restore ? no_frame : no_stack;
1365    
1366  if (length > 0)  if (length > 0)
1367    return length + 1;    return length + 1;
1368  return -1;  return stack_restore ? no_frame : no_stack;
1369  }  }
1370    
1371  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)
1372  {  {
1373  DEFINE_COMPILER;  DEFINE_COMPILER;
 pcre_uchar *ccend = bracketend(cc);  
1374  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1375  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1376  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
# Line 1229  SLJIT_UNUSED_ARG(stacktop); Line 1382  SLJIT_UNUSED_ARG(stacktop);
1382  SLJIT_ASSERT(stackpos >= stacktop + 2);  SLJIT_ASSERT(stackpos >= stacktop + 2);
1383    
1384  stackpos = STACK(stackpos);  stackpos = STACK(stackpos);
1385  if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))  if (ccend == NULL)
1386    cc = next_opcode(common, cc);    {
1387      ccend = bracketend(cc) - (1 + LINK_SIZE);
1388      if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))
1389        cc = next_opcode(common, cc);
1390      }
1391    
1392  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1393  while (cc < ccend)  while (cc < ccend)
1394    switch(*cc)    switch(*cc)
# Line 1250  while (cc < ccend) Line 1408  while (cc < ccend)
1408      break;      break;
1409    
1410      case OP_MARK:      case OP_MARK:
1411        case OP_PRUNE_ARG:
1412        case OP_THEN_ARG:
1413      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1414      if (!setmark_found)      if (!setmark_found)
1415        {        {
# Line 1330  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st Line 1490  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st
1490  SLJIT_ASSERT(stackpos == STACK(stacktop));  SLJIT_ASSERT(stackpos == STACK(stacktop));
1491  }  }
1492    
1493  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)
1494  {  {
1495  int private_data_length = 2;  int private_data_length = needs_control_head ? 3 : 2;
1496  int size;  int size;
1497  pcre_uchar *alternative;  pcre_uchar *alternative;
1498  /* Calculate the sum of the private machine words. */  /* Calculate the sum of the private machine words. */
# Line 1341  while (cc < ccend) Line 1501  while (cc < ccend)
1501    size = 0;    size = 0;
1502    switch(*cc)    switch(*cc)
1503      {      {
1504        case OP_KET:
1505        if (PRIVATE_DATA(cc) != 0)
1506          private_data_length++;
1507        cc += 1 + LINK_SIZE;
1508        break;
1509    
1510      case OP_ASSERT:      case OP_ASSERT:
1511      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1512      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1445  return private_data_length; Line 1611  return private_data_length;
1611  }  }
1612    
1613  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,
1614    BOOL save, int stackptr, int stacktop)    BOOL save, int stackptr, int stacktop, BOOL needs_control_head)
1615  {  {
1616  DEFINE_COMPILER;  DEFINE_COMPILER;
1617  int srcw[2];  int srcw[2];
# Line 1466  stacktop = STACK(stacktop - 1); Line 1632  stacktop = STACK(stacktop - 1);
1632    
1633  if (!save)  if (!save)
1634    {    {
1635    stackptr += sizeof(sljit_sw);    stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw);
1636    if (stackptr < stacktop)    if (stackptr < stacktop)
1637      {      {
1638      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);
# Line 1482  if (!save) Line 1648  if (!save)
1648    /* The tmp1next must be TRUE in either way. */    /* The tmp1next must be TRUE in either way. */
1649    }    }
1650    
1651  while (status != end)  do
1652    {    {
1653    count = 0;    count = 0;
1654    switch(status)    switch(status)
# Line 1491  while (status != end) Line 1657  while (status != end)
1657      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);
1658      count = 1;      count = 1;
1659      srcw[0] = common->recursive_head_ptr;      srcw[0] = common->recursive_head_ptr;
1660        if (needs_control_head)
1661          {
1662          SLJIT_ASSERT(common->control_head_ptr != 0);
1663          count = 2;
1664          srcw[1] = common->control_head_ptr;
1665          }
1666      status = loop;      status = loop;
1667      break;      break;
1668    
# Line 1503  while (status != end) Line 1675  while (status != end)
1675    
1676      switch(*cc)      switch(*cc)
1677        {        {
1678          case OP_KET:
1679          if (PRIVATE_DATA(cc) != 0)
1680            {
1681            count = 1;
1682            srcw[0] = PRIVATE_DATA(cc);
1683            }
1684          cc += 1 + LINK_SIZE;
1685          break;
1686    
1687        case OP_ASSERT:        case OP_ASSERT:
1688        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
1689        case OP_ASSERTBACK:        case OP_ASSERTBACK:
# Line 1715  while (status != end) Line 1896  while (status != end)
1896        }        }
1897      }      }
1898    }    }
1899    while (status != end);
1900    
1901  if (save)  if (save)
1902    {    {
# Line 1748  if (save) Line 1930  if (save)
1930  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));
1931  }  }
1932    
1933    static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset)
1934    {
1935    pcre_uchar *end = bracketend(cc);
1936    BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT;
1937    
1938    /* Assert captures then. */
1939    if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT)
1940      current_offset = NULL;
1941    /* Conditional block does not. */
1942    if (*cc == OP_COND || *cc == OP_SCOND)
1943      has_alternatives = FALSE;
1944    
1945    cc = next_opcode(common, cc);
1946    if (has_alternatives)
1947      current_offset = common->then_offsets + (cc - common->start);
1948    
1949    while (cc < end)
1950      {
1951      if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND))
1952        cc = set_then_offsets(common, cc, current_offset);
1953      else
1954        {
1955        if (*cc == OP_ALT && has_alternatives)
1956          current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start);
1957        if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL)
1958          *current_offset = 1;
1959        cc = next_opcode(common, cc);
1960        }
1961      }
1962    
1963    return end;
1964    }
1965    
1966  #undef CASE_ITERATOR_PRIVATE_DATA_1  #undef CASE_ITERATOR_PRIVATE_DATA_1
1967  #undef CASE_ITERATOR_PRIVATE_DATA_2A  #undef CASE_ITERATOR_PRIVATE_DATA_2A
1968  #undef CASE_ITERATOR_PRIVATE_DATA_2B  #undef CASE_ITERATOR_PRIVATE_DATA_2B
# Line 1782  if (list_item) Line 1997  if (list_item)
1997    }    }
1998  }  }
1999    
2000  static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start)  static void add_stub(compiler_common *common, struct sljit_jump *start)
2001  {  {
2002  DEFINE_COMPILER;  DEFINE_COMPILER;
2003  stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list));  stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list));
2004    
2005  if (list_item)  if (list_item)
2006    {    {
   list_item->type = type;  
   list_item->data = data;  
2007    list_item->start = start;    list_item->start = start;
2008    list_item->quit = LABEL();    list_item->quit = LABEL();
2009    list_item->next = common->stubs;    list_item->next = common->stubs;
# Line 1806  stub_list* list_item = common->stubs; Line 2019  stub_list* list_item = common->stubs;
2019  while (list_item)  while (list_item)
2020    {    {
2021    JUMPHERE(list_item->start);    JUMPHERE(list_item->start);
2022    switch(list_item->type)    add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL));
     {  
     case stack_alloc:  
     add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL));  
     break;  
     }  
2023    JUMPTO(SLJIT_JUMP, list_item->quit);    JUMPTO(SLJIT_JUMP, list_item->quit);
2024    list_item = list_item->next;    list_item = list_item->next;
2025    }    }
2026  common->stubs = NULL;  common->stubs = NULL;
2027  }  }
2028    
2029  static SLJIT_INLINE void decrease_call_count(compiler_common *common)  static SLJIT_INLINE void count_match(compiler_common *common)
2030  {  {
2031  DEFINE_COMPILER;  DEFINE_COMPILER;
2032    
2033  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);
2034  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));
2035  }  }
2036    
# Line 1839  OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); Line 2047  OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0);
2047  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0);
2048  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0);
2049  #endif  #endif
2050  add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0));  add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0));
2051  }  }
2052    
2053  static SLJIT_INLINE void free_stack(compiler_common *common, int size)  static SLJIT_INLINE void free_stack(compiler_common *common, int size)
# Line 1853  static SLJIT_INLINE void reset_ovector(c Line 2061  static SLJIT_INLINE void reset_ovector(c
2061  DEFINE_COMPILER;  DEFINE_COMPILER;
2062  struct sljit_label *loop;  struct sljit_label *loop;
2063  int i;  int i;
2064    
2065  /* At this point we can freely use all temporary registers. */  /* At this point we can freely use all temporary registers. */
2066    SLJIT_ASSERT(length > 1);
2067  /* TMP1 returns with begin - 1. */  /* TMP1 returns with begin - 1. */
2068  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));
2069  if (length < 8)  if (length < 8)
2070    {    {
2071    for (i = 0; i < length; i++)    for (i = 1; i < length; i++)
2072      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0);
2073    }    }
2074  else  else
2075    {    {
2076    GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START - sizeof(sljit_sw));    GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START);
2077    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length);    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length - 1);
2078    loop = LABEL();    loop = LABEL();
2079    OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0);    OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0);
2080    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1);
# Line 1872  else Line 2082  else
2082    }    }
2083  }  }
2084    
2085    static SLJIT_INLINE void do_reset_match(compiler_common *common, int length)
2086    {
2087    DEFINE_COMPILER;
2088    struct sljit_label *loop;
2089    int i;
2090    
2091    SLJIT_ASSERT(length > 1);
2092    /* OVECTOR(1) contains the "string begin - 1" constant. */
2093    if (length > 2)
2094      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
2095    if (length < 8)
2096      {
2097      for (i = 2; i < length; i++)
2098        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), TMP1, 0);
2099      }
2100    else
2101      {
2102      GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw));
2103      OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2);
2104      loop = LABEL();
2105      OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
2106      OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1);
2107      JUMPTO(SLJIT_C_NOT_ZERO, loop);
2108      }
2109    
2110    OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);
2111    if (common->mark_ptr != 0)
2112      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0);
2113    if (common->control_head_ptr != 0)
2114      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0);
2115    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));
2116    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr);
2117    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base));
2118    }
2119    
2120    static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg)
2121    {
2122    while (current != NULL)
2123      {
2124      switch (current[-2])
2125        {
2126        case type_then_trap:
2127        break;
2128    
2129        case type_mark:
2130        if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0)
2131          return current[-4];
2132        break;
2133    
2134        default:
2135        SLJIT_ASSERT_STOP();
2136        break;
2137        }
2138      current = (sljit_sw*)current[-1];
2139      }
2140    return -1;
2141    }
2142    
2143  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)
2144  {  {
2145  DEFINE_COMPILER;  DEFINE_COMPILER;
# Line 1925  else Line 2193  else
2193  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)
2194  {  {
2195  DEFINE_COMPILER;  DEFINE_COMPILER;
2196    struct sljit_jump *jump;
2197    
2198  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);
2199  SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));  SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0
2200      && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));
2201    
2202  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0);
2203  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);
2204  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offset_count));  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count));
2205  CMPTO(SLJIT_C_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit);  CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit);
2206    
2207  /* Store match begin and end. */  /* Store match begin and end. */
2208  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin));  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin));
2209  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));
2210    
2211    jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3);
2212    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);
2213    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2214    OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);
2215    #endif
2216    OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0);
2217    JUMPHERE(jump);
2218    
2219  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start);
2220  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);
2221  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 2098  return (bit < 256) ? ((0 << 8) | bit) : Line 2377  return (bit < 256) ? ((0 << 8) | bit) :
2377    
2378  static void check_partial(compiler_common *common, BOOL force)  static void check_partial(compiler_common *common, BOOL force)
2379  {  {
2380  /* Checks whether a partial matching is occured. Does not modify registers. */  /* Checks whether a partial matching is occurred. Does not modify registers. */
2381  DEFINE_COMPILER;  DEFINE_COMPILER;
2382  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
2383    
# Line 2113  else if (common->mode == JIT_PARTIAL_SOF Line 2392  else if (common->mode == JIT_PARTIAL_SOF
2392    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);
2393    
2394  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2395    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2396  else  else
2397    {    {
2398    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
# Line 2126  if (jump != NULL) Line 2405  if (jump != NULL)
2405    JUMPHERE(jump);    JUMPHERE(jump);
2406  }  }
2407    
2408  static struct sljit_jump *check_str_end(compiler_common *common)  static void check_str_end(compiler_common *common, jump_list **end_reached)
2409  {  {
2410  /* Does not affect registers. Usually used in a tight spot. */  /* Does not affect registers. Usually used in a tight spot. */
2411  DEFINE_COMPILER;  DEFINE_COMPILER;
2412  struct sljit_jump *jump;  struct sljit_jump *jump;
 struct sljit_jump *nohit;  
 struct sljit_jump *return_value;  
2413    
2414  if (common->mode == JIT_COMPILE)  if (common->mode == JIT_COMPILE)
2415    return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);    {
2416      add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
2417      return;
2418      }
2419    
2420  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);
2421  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2422    {    {
2423    nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2424    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2425    JUMPHERE(nohit);    add_jump(compiler, end_reached, JUMP(SLJIT_JUMP));
   return_value = JUMP(SLJIT_JUMP);  
2426    }    }
2427  else  else
2428    {    {
2429    return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2430    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
2431      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
2432    else    else
2433      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
2434    }    }
2435  JUMPHERE(jump);  JUMPHERE(jump);
 return return_value;  
2436  }  }
2437    
2438  static void detect_partial_match(compiler_common *common, jump_list **backtracks)  static void detect_partial_match(compiler_common *common, jump_list **backtracks)
# Line 2173  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR Line 2451  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR
2451  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2452  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2453    {    {
2454    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2455    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
2456    }    }
2457  else  else
# Line 2186  else Line 2464  else
2464  JUMPHERE(jump);  JUMPHERE(jump);
2465  }  }
2466    
2467  static void read_char(compiler_common *common)  static void peek_char(compiler_common *common)
2468  {  {
2469  /* Reads the character into TMP1, updates STR_PTR.  /* Reads the character into TMP1, keeps STR_PTR.
2470  Does not check STR_END. TMP2 Destroyed. */  Does not check STR_END. TMP2 Destroyed. */
2471  DEFINE_COMPILER;  DEFINE_COMPILER;
2472  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
# Line 2196  struct sljit_jump *jump; Line 2474  struct sljit_jump *jump;
2474  #endif  #endif
2475    
2476  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2477  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2478  if (common->utf)  if (common->utf)
2479    {    {
 #if defined COMPILE_PCRE8  
2480    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2481  #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] */  
2482    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2483      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2484      JUMPHERE(jump);
2485      }
2486    #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2487    
2488    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2489    if (common->utf)
2490      {
2491      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2492      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2493      /* TMP2 contains the high surrogate. */
2494      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2495      OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2496      OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2497      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2498      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2499      JUMPHERE(jump);
2500      }
2501    #endif
2502    }
2503    
2504    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2505    
2506    static BOOL is_char7_bitset(const pcre_uint8* bitset, BOOL nclass)
2507    {
2508    /* Tells whether the character codes below 128 are enough
2509    to determine a match. */
2510    const pcre_uint8 value = nclass ? 0xff : 0;
2511    const pcre_uint8* end = bitset + 32;
2512    
2513    bitset += 16;
2514    do
2515      {
2516      if (*bitset++ != value)
2517        return FALSE;
2518      }
2519    while (bitset < end);
2520    return TRUE;
2521    }
2522    
2523    static void read_char7_type(compiler_common *common, BOOL full_read)
2524    {
2525    /* Reads the precise character type of a character into TMP1, if the character
2526    is less than 128. Otherwise it returns with zero. Does not check STR_END. The
2527    full_read argument tells whether characters above max are accepted or not. */
2528    DEFINE_COMPILER;
2529    struct sljit_jump *jump;
2530    
2531    SLJIT_ASSERT(common->utf);
2532    
2533    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2534    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2535    
2536    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2537    
2538    if (full_read)
2539      {
2540      jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2541      OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2542      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2543    JUMPHERE(jump);    JUMPHERE(jump);
2544    }    }
 #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  
 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
2545  }  }
2546    
2547  static void peek_char(compiler_common *common)  #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2548    
2549    static void read_char_max(compiler_common *common, pcre_uint32 max, BOOL full_read)
2550  {  {
2551  /* Reads the character into TMP1, keeps STR_PTR.  /* Reads the precise value of a character into TMP1, if the character is
2552  Does not check STR_END. TMP2 Destroyed. */  less than or equal to max. Otherwise it returns with a value greater than max.
2553    Does not check STR_END. The full_read argument tells whether characters above
2554    max are accepted or not. */
2555  DEFINE_COMPILER;  DEFINE_COMPILER;
2556  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2557  struct sljit_jump *jump;  struct sljit_jump *jump;
2558  #endif  #endif
2559    
2560    SLJIT_UNUSED_ARG(full_read);
2561    SLJIT_UNUSED_ARG(max);
2562    
2563  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2564  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2565    
2566    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2567  if (common->utf)  if (common->utf)
2568    {    {
2569  #if defined COMPILE_PCRE8    if (max < 128 && !full_read)
2570        return;
2571    
2572    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2573  #elif defined COMPILE_PCRE16    if (max >= 0x800)
2574    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);      add_jump(compiler, (max < 0x10000) ? &common->utfreadchar16 : &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2575  #endif /* COMPILE_PCRE[8|16] */    else if (max < 128)
2576    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));      {
2577    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);      OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2578        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2579        }
2580      else
2581        {
2582        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2583        if (!full_read)
2584          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2585        else
2586          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2587        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2588        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2589        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2590        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2591        if (full_read)
2592          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2593        }
2594    JUMPHERE(jump);    JUMPHERE(jump);
2595    }    }
2596  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  #endif
2597    
2598    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2599    if (common->utf)
2600      {
2601      if (max >= 0x10000)
2602        {
2603        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2604        jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2605        /* TMP2 contains the high surrogate. */
2606        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2607        OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2608        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2609        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2610        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2611        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2612        JUMPHERE(jump);
2613        return;
2614        }
2615    
2616      if (max < 0xd800 && !full_read)
2617        return;
2618    
2619      /* Skip low surrogate if necessary. */
2620      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2621      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2622      if (full_read)
2623        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2624      if (max >= 0xd800)
2625        OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000);
2626      JUMPHERE(jump);
2627      }
2628    #endif
2629  }  }
2630    
2631  static void read_char8_type(compiler_common *common)  static SLJIT_INLINE void read_char(compiler_common *common)
2632  {  {
2633  /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */  read_char_max(common, READ_CHAR_ANY, TRUE);
2634    }
2635    
2636    static void read_char8_type(compiler_common *common, BOOL full_read)
2637    {
2638    /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END.
2639    The full_read argument tells whether characters above max are accepted or not. */
2640  DEFINE_COMPILER;  DEFINE_COMPILER;
2641  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2642  struct sljit_jump *jump;  struct sljit_jump *jump;
2643  #endif  #endif
2644    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2645    struct sljit_jump *jump2;
2646    #endif
2647    
2648  #ifdef SUPPORT_UTF  SLJIT_UNUSED_ARG(full_read);
2649    
2650    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2651    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2652    
2653    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2654  if (common->utf)  if (common->utf)
2655    {    {
   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  
2656    /* This can be an extra read in some situations, but hopefully    /* This can be an extra read in some situations, but hopefully
2657    it is needed in most cases. */    it is needed in most cases. */
2658    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2659    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2660    add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));    if (!full_read)
2661    JUMPHERE(jump);      {
2662  #elif defined COMPILE_PCRE16      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2663    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2664    jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);      OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2665    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);      OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2666    JUMPHERE(jump);      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2667    /* Skip low surrogate if necessary. */      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
2668    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2669    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800);      jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2670    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2671    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);      JUMPHERE(jump2);
2672    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);      }
2673  #elif defined COMPILE_PCRE32    else
2674    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);      add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));
   jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
   OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
2675    JUMPHERE(jump);    JUMPHERE(jump);
 #endif /* COMPILE_PCRE[8|16|32] */  
2676    return;    return;
2677    }    }
2678  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2679  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
2680  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  #if !defined COMPILE_PCRE8
 #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
2681  /* The ctypes array contains only 256 values. */  /* The ctypes array contains only 256 values. */
2682  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2683  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2684  #endif  #endif
2685  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2686  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if !defined COMPILE_PCRE8
2687  JUMPHERE(jump);  JUMPHERE(jump);
2688  #endif  #endif
2689    
2690    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2691    if (common->utf && full_read)
2692      {
2693      /* Skip low surrogate if necessary. */
2694      OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
2695      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2696      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2697      JUMPHERE(jump);
2698      }
2699    #endif /* SUPPORT_UTF && COMPILE_PCRE16 */
2700  }  }
2701    
2702  static void skip_char_back(compiler_common *common)  static void skip_char_back(compiler_common *common)
# Line 2324  if (common->utf) Line 2734  if (common->utf)
2734  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));
2735  }  }
2736    
2737  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)
2738  {  {
2739  /* 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. */
2740  DEFINE_COMPILER;  DEFINE_COMPILER;
2741    struct sljit_jump *jump;
2742    
2743  if (nltype == NLTYPE_ANY)  if (nltype == NLTYPE_ANY)
2744    {    {
2745    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
2746    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));
2747    }    }
2748  else if (nltype == NLTYPE_ANYCRLF)  else if (nltype == NLTYPE_ANYCRLF)
2749    {    {
2750    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR);    if (jumpifmatch)
2751    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);      {
2752    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));
2753    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));
2754    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));      }
2755      else
2756        {
2757        jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
2758        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
2759        JUMPHERE(jump);
2760        }
2761    }    }
2762  else  else
2763    {    {
2764    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);
2765    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));
2766    }    }
2767  }  }
2768    
# Line 2355  else Line 2772  else
2772  static void do_utfreadchar(compiler_common *common)  static void do_utfreadchar(compiler_common *common)
2773  {  {
2774  /* Fast decoding a UTF-8 character. TMP1 contains the first byte  /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2775  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. */
2776  DEFINE_COMPILER;  DEFINE_COMPILER;
2777  struct sljit_jump *jump;  struct sljit_jump *jump;
2778    
2779  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2780    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2781    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2782    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2783    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2784    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2785    
2786  /* Searching for the first zero. */  /* Searching for the first zero. */
2787  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);
2788  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2789  /* Two byte sequence. */  /* Two byte sequence. */
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
2790  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));
2791  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2));
2792    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2793    
2794    JUMPHERE(jump);
2795    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2796    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2797  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2798  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2799  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);  
2800    
2801  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);
2802  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2803  /* 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));  
2804  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));
2805  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));  
2806  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2807    
2808  /* Four byte sequence. */  /* Four byte sequence. */
2809  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  JUMPHERE(jump);
2810  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07);  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
2811  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18);  OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2812    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2813    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
2814  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);  
2815  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2816  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(4));
2817    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2818    }
2819    
2820    static void do_utfreadchar16(compiler_common *common)
2821    {
2822    /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2823    of the character (>= 0xc0). Return value in TMP1. */
2824    DEFINE_COMPILER;
2825    struct sljit_jump *jump;
2826    
2827    sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2828    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2829    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2830    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2831  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);  
2832  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2833  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3));  
2834  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));  /* Searching for the first zero. */
2835    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800);
2836    jump = JUMP(SLJIT_C_NOT_ZERO);
2837    /* Two byte sequence. */
2838    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2839    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2840    
2841    JUMPHERE(jump);
2842    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400);
2843    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_NOT_ZERO);
2844    /* This code runs only in 8 bit mode. No need to shift the value. */
2845    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2846    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2847    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2848    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2849  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2850  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2851  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));  /* Three byte sequence. */
2852    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2853  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2854  }  }
2855    
# Line 2426  jump = JUMP(SLJIT_C_NOT_ZERO); Line 2869  jump = JUMP(SLJIT_C_NOT_ZERO);
2869  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2870  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));
2871  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);
2872    /* The upper 5 bits are known at this point. */
2873    compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x3);
2874  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2875  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2876  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);  
2877  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2878  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2879    
2880  JUMPHERE(compare);  JUMPHERE(compare);
2881  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2882  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2883    
2884  /* 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);  
   
2885  JUMPHERE(jump);  JUMPHERE(jump);
2886  /* Combine two 16 bit characters. */  OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2887  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2888  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);  
2889  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2890  }  }
2891    
2892  #endif /* COMPILE_PCRE[8|16] */  #endif /* COMPILE_PCRE8 */
2893    
2894  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
2895    
# Line 2547  if (firstline) Line 2964  if (firstline)
2964      mainloop = LABEL();      mainloop = LABEL();
2965      /* Continual stores does not cause data dependency. */      /* Continual stores does not cause data dependency. */
2966      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);
2967      read_char(common);      read_char_max(common, common->nlmax, TRUE);
2968      check_newlinechar(common, common->nltype, &newline, TRUE);      check_newlinechar(common, common->nltype, &newline, TRUE);
2969      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);
2970      JUMPHERE(end);      JUMPHERE(end);
# Line 2623  if (newlinecheck) Line 3040  if (newlinecheck)
3040  return mainloop;  return mainloop;
3041  }  }
3042    
3043  #define MAX_N_CHARS 3  static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, int max_chars)
   
 static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)  
3044  {  {
3045  DEFINE_COMPILER;  /* Recursive function, which scans prefix literals. */
3046  struct sljit_label *start;  int len, repeat, len_save, consumed = 0;
3047  struct sljit_jump *quit;  pcre_uint32 caseless, chr, mask;
3048  pcre_uint32 chars[MAX_N_CHARS * 2];  pcre_uchar *alternative, *cc_save;
3049  pcre_uchar *cc = common->start + 1 + LINK_SIZE;  BOOL last, any;
 int location = 0;  
 pcre_int32 len, c, bit, caseless;  
 int must_stop;  
   
 /* We do not support alternatives now. */  
 if (*(common->start + GET(common->start, 1)) == OP_ALT)  
   return FALSE;  
3050    
3051    repeat = 1;
3052  while (TRUE)  while (TRUE)
3053    {    {
3054      last = TRUE;
3055      any = FALSE;
3056    caseless = 0;    caseless = 0;
3057    must_stop = 1;    switch (*cc)
   switch(*cc)  
3058      {      {
     case OP_CHAR:  
     must_stop = 0;  
     cc++;  
     break;  
   
3059      case OP_CHARI:      case OP_CHARI:
3060      caseless = 1;      caseless = 1;
3061      must_stop = 0;      case OP_CHAR:
3062        last = FALSE;
3063      cc++;      cc++;
3064      break;      break;
3065    
# Line 2678  while (TRUE) Line 3084  while (TRUE)
3084      cc++;      cc++;
3085      break;      break;
3086    
3087        case OP_EXACTI:
3088        caseless = 1;
3089      case OP_EXACT:      case OP_EXACT:
3090        repeat = GET2(cc, 1);
3091        last = FALSE;
3092      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3093      break;      break;
3094    
# Line 2689  while (TRUE) Line 3099  while (TRUE)
3099      cc++;      cc++;
3100      break;      break;
3101    
3102      case OP_EXACTI:      case OP_KET:
3103      caseless = 1;      cc += 1 + LINK_SIZE;
3104      cc += 1 + IMM2_SIZE;      continue;
3105    
3106        case OP_ALT:
3107        cc += GET(cc, 1);
3108        continue;
3109    
3110        case OP_ONCE:
3111        case OP_ONCE_NC:
3112        case OP_BRA:
3113        case OP_BRAPOS:
3114        case OP_CBRA:
3115        case OP_CBRAPOS:
3116        alternative = cc + GET(cc, 1);
3117        while (*alternative == OP_ALT)
3118          {
3119          max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, max_chars);
3120          if (max_chars == 0)
3121            return consumed;
3122          alternative += GET(alternative, 1);
3123          }
3124    
3125        if (*cc == OP_CBRA || *cc == OP_CBRAPOS)
3126          cc += IMM2_SIZE;
3127        cc += 1 + LINK_SIZE;
3128        continue;
3129    
3130        case OP_CLASS:
3131        case OP_NCLASS:
3132        any = TRUE;
3133        cc += 1 + 32 / sizeof(pcre_uchar);
3134        break;
3135    
3136    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3137        case OP_XCLASS:
3138        any = TRUE;
3139        cc += GET(cc, 1);
3140        break;
3141    #endif
3142    
3143        case OP_NOT_DIGIT:
3144        case OP_DIGIT:
3145        case OP_NOT_WHITESPACE:
3146        case OP_WHITESPACE:
3147        case OP_NOT_WORDCHAR:
3148        case OP_WORDCHAR:
3149        case OP_ANY:
3150        case OP_ALLANY:
3151        any = TRUE;
3152        cc++;
3153      break;      break;
3154    
3155      default:  #ifdef SUPPORT_UCP
3156      must_stop = 2;      case OP_NOTPROP:
3157      break;      case OP_PROP:
3158      }      any = TRUE;
3159        cc += 1 + 2;
3160        break;
3161    #endif
3162    
3163        case OP_TYPEEXACT:
3164        repeat = GET2(cc, 1);
3165        cc += 1 + IMM2_SIZE;
3166        continue;
3167    
3168        default:
3169        return consumed;
3170        }
3171    
3172      if (any)
3173        {
3174    #ifdef SUPPORT_UTF
3175        if (common->utf) return consumed;
3176    #endif
3177    #if defined COMPILE_PCRE8
3178        mask = 0xff;
3179    #elif defined COMPILE_PCRE16
3180        mask = 0xffff;
3181    #elif defined COMPILE_PCRE32
3182        mask = 0xffffffff;
3183    #else
3184        SLJIT_ASSERT_STOP();
3185    #endif
3186    
3187        do
3188          {
3189          chars[0] = mask;
3190          chars[1] = mask;
3191    
3192          if (--max_chars == 0)
3193            return consumed;
3194          consumed++;
3195          chars += 2;
3196          }
3197        while (--repeat > 0);
3198    
3199    if (must_stop == 2)      repeat = 1;
3200        break;      continue;
3201        }
3202    
3203    len = 1;    len = 1;
3204  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3205    if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);    if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3206  #endif  #endif
3207    
3208    if (caseless && char_has_othercase(common, cc))    if (caseless != 0 && char_has_othercase(common, cc))
3209      {      {
3210      caseless = char_get_othercase_bit(common, cc);      caseless = char_get_othercase_bit(common, cc);
3211      if (caseless == 0)      if (caseless == 0)
3212        return FALSE;        return consumed;
3213  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
3214      caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));      caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));
3215  #else  #else
# Line 2724  while (TRUE) Line 3222  while (TRUE)
3222    else    else
3223      caseless = 0;      caseless = 0;
3224    
3225    while (len > 0 && location < MAX_N_CHARS * 2)    len_save = len;
3226      {    cc_save = cc;
3227      c = *cc;    while (TRUE)
3228      bit = 0;      {
3229      if (len == (caseless & 0xff))      do
3230        {        {
3231        bit = caseless >> 8;        chr = *cc;
3232        c |= bit;  #ifdef COMPILE_PCRE32
3233          if (SLJIT_UNLIKELY(chr == NOTACHAR))
3234            return consumed;
3235    #endif
3236          mask = 0;
3237          if ((pcre_uint32)len == (caseless & 0xff))
3238            {
3239            mask = caseless >> 8;
3240            chr |= mask;
3241            }
3242    
3243          if (chars[0] == NOTACHAR)
3244            {
3245            chars[0] = chr;
3246            chars[1] = mask;
3247            }
3248          else
3249            {
3250            mask |= chars[0] ^ chr;
3251            chr |= mask;
3252            chars[0] = chr;
3253            chars[1] |= mask;
3254            }
3255    
3256          len--;
3257          if (--max_chars == 0)
3258            return consumed;
3259          consumed++;
3260          chars += 2;
3261          cc++;
3262        }        }
3263        while (len > 0);
3264    
3265      chars[location] = c;      if (--repeat == 0)
3266      chars[location + 1] = bit;        break;
3267    
3268      len--;      len = len_save;
3269      location += 2;      cc = cc_save;
3270      cc++;      }
3271    
3272      repeat = 1;
3273      if (last)
3274        return consumed;
3275      }
3276    }
3277    
3278    #define MAX_N_CHARS 16
3279    
3280    static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)
3281    {
3282    DEFINE_COMPILER;
3283    struct sljit_label *start;
3284    struct sljit_jump *quit;
3285    pcre_uint32 chars[MAX_N_CHARS * 2];
3286    pcre_uint8 ones[MAX_N_CHARS];
3287    pcre_uint32 mask;
3288    int i, max;
3289    int offsets[3];
3290    
3291    for (i = 0; i < MAX_N_CHARS; i++)
3292      {
3293      chars[i << 1] = NOTACHAR;
3294      chars[(i << 1) + 1] = 0;
3295      }
3296    
3297    max = scan_prefix(common, common->start, chars, MAX_N_CHARS);
3298    
3299    if (max <= 1)
3300      return FALSE;
3301    
3302    for (i = 0; i < max; i++)
3303      {
3304      mask = chars[(i << 1) + 1];
3305      ones[i] = ones_in_half_byte[mask & 0xf];
3306      mask >>= 4;
3307      while (mask != 0)
3308        {
3309        ones[i] += ones_in_half_byte[mask & 0xf];
3310        mask >>= 4;
3311      }      }
3312      }
3313    
3314    if (location >= MAX_N_CHARS * 2 || must_stop != 0)  offsets[0] = -1;
3315    /* Scan forward. */
3316    for (i = 0; i < max; i++)
3317      if (ones[i] <= 2) {
3318        offsets[0] = i;
3319      break;      break;
3320    }    }
3321    
3322  /* At least two characters are required. */  if (offsets[0] == -1)
3323  if (location < 2 * 2)    return FALSE;
3324      return FALSE;  
3325    /* Scan backward. */
3326    offsets[1] = -1;
3327    for (i = max - 1; i > offsets[0]; i--)
3328      if (ones[i] <= 2) {
3329        offsets[1] = i;
3330        break;
3331      }
3332    
3333    offsets[2] = -1;
3334    if (offsets[1] >= 0)
3335      {
3336      /* Scan from middle. */
3337      for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++)
3338        if (ones[i] <= 2)
3339          {
3340          offsets[2] = i;
3341          break;
3342          }
3343    
3344      if (offsets[2] == -1)
3345        {
3346        for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--)
3347          if (ones[i] <= 2)
3348            {
3349            offsets[2] = i;
3350            break;
3351            }
3352        }
3353      }
3354    
3355    SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1]));
3356    SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2]));
3357    
3358    chars[0] = chars[offsets[0] << 1];
3359    chars[1] = chars[(offsets[0] << 1) + 1];
3360    if (offsets[2] >= 0)
3361      {
3362      chars[2] = chars[offsets[2] << 1];
3363      chars[3] = chars[(offsets[2] << 1) + 1];
3364      }
3365    if (offsets[1] >= 0)
3366      {
3367      chars[4] = chars[offsets[1] << 1];
3368      chars[5] = chars[(offsets[1] << 1) + 1];
3369      }
3370    
3371    max -= 1;
3372  if (firstline)  if (firstline)
3373    {    {
3374    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
3375    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
3376    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, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS(max));
3377    }    }
3378  else  else
3379    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));
3380    
3381  start = LABEL();  start = LABEL();
3382  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3383    
3384  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0]));
3385  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  if (offsets[1] >= 0)
3386      OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1]));
3387  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));
3388    
3389  if (chars[1] != 0)  if (chars[1] != 0)
3390    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);
3391  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);
3392  if (location > 2 * 2)  if (offsets[2] >= 0)
3393    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1));
3394  if (chars[3] != 0)  
3395    OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]);  if (offsets[1] >= 0)
 CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start);  
 if (location > 2 * 2)  
3396    {    {
3397    if (chars[5] != 0)    if (chars[5] != 0)
3398      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]);      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]);
3399    CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start);    CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start);
3400      }
3401    
3402    if (offsets[2] >= 0)
3403      {
3404      if (chars[3] != 0)
3405        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]);
3406      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start);
3407    }    }
3408  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));
3409    
# Line 2786  JUMPHERE(quit); Line 3412  JUMPHERE(quit);
3412  if (firstline)  if (firstline)
3413    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3414  else  else
3415    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));
3416  return TRUE;  return TRUE;
3417  }  }
3418    
# Line 2906  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ Line 3532  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_
3532  skip_char_back(common);  skip_char_back(common);
3533    
3534  loop = LABEL();  loop = LABEL();
3535  read_char(common);  read_char_max(common, common->nlmax, TRUE);
3536  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3537  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
3538    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
# Line 2935  if (firstline) Line 3561  if (firstline)
3561    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3562  }  }
3563    
3564  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);
3565    
3566  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)
3567  {  {
3568  DEFINE_COMPILER;  DEFINE_COMPILER;
3569  struct sljit_label *start;  struct sljit_label *start;
3570  struct sljit_jump *quit;  struct sljit_jump *quit;
3571  struct sljit_jump *found = NULL;  struct sljit_jump *found = NULL;
3572  jump_list *matches = NULL;  jump_list *matches = NULL;
 pcre_uint8 inverted_start_bits[32];  
 int i;  
3573  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3574  struct sljit_jump *jump;  struct sljit_jump *jump;
3575  #endif  #endif
3576    
 for (i = 0; i < 32; ++i)  
   inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]);  
   
3577  if (firstline)  if (firstline)
3578    {    {
3579    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
# Line 2968  if (common->utf) Line 3589  if (common->utf)
3589    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
3590  #endif  #endif
3591    
3592  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))
3593    {    {
3594  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3595    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);
# Line 2977  if (!check_class_ranges(common, inverted Line 3598  if (!check_class_ranges(common, inverted
3598  #endif  #endif
3599    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
3600    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
3601    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
3602    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
3603    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);
3604    found = JUMP(SLJIT_C_NOT_ZERO);    found = JUMP(SLJIT_C_NOT_ZERO);
# Line 3121  static void check_wordboundary(compiler_ Line 3742  static void check_wordboundary(compiler_
3742  {  {
3743  DEFINE_COMPILER;  DEFINE_COMPILER;
3744  struct sljit_jump *skipread;  struct sljit_jump *skipread;
3745    jump_list *skipread_list = NULL;
3746  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF
3747  struct sljit_jump *jump;  struct sljit_jump *jump;
3748  #endif  #endif
# Line 3178  else Line 3800  else
3800  JUMPHERE(skipread);  JUMPHERE(skipread);
3801    
3802  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
3803  skipread = check_str_end(common);  check_str_end(common, &skipread_list);
3804  peek_char(common);  peek_char(common);
3805    
3806  /* Testing char type. This is a code duplication. */  /* Testing char type. This is a code duplication. */
# Line 3219  else Line 3841  else
3841      JUMPHERE(jump);      JUMPHERE(jump);
3842  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
3843    }    }
3844  JUMPHERE(skipread);  set_jumps(skipread_list, LABEL());
3845    
3846  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1);  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1);
3847  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);
3848  }  }
3849    
3850  /*  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)  
3851  {  {
3852  DEFINE_COMPILER;  DEFINE_COMPILER;
3853  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];  
3854  pcre_uint8 bit, cbit, all;  pcre_uint8 bit, cbit, all;
3855  int i, byte, length = 0;  int i, byte, length = 0;
3856    
3857  bit = bits[0] & 0x1;  bit = bits[0] & 0x1;
3858  ranges[1] = bit;  /* All bits will be zero or one (since bit is zero or one). */
 /* Can be 0 or 255. */  
3859  all = -bit;  all = -bit;
3860    
3861  for (i = 0; i < 256; )  for (i = 0; i < 256; )
# Line 3343  for (i = 0; i < 256; ) Line 3870  for (i = 0; i < 256; )
3870        {        {
3871        if (length >= MAX_RANGE_SIZE)        if (length >= MAX_RANGE_SIZE)
3872          return FALSE;          return FALSE;
3873        ranges[2 + length] = i;        ranges[length] = i;
3874        length++;        length++;
3875        bit = cbit;        bit = cbit;
3876        all = -cbit;        all = -cbit;
# Line 3356  if (((bit == 0) && nclass) || ((bit == 1 Line 3883  if (((bit == 0) && nclass) || ((bit == 1
3883    {    {
3884    if (length >= MAX_RANGE_SIZE)    if (length >= MAX_RANGE_SIZE)
3885      return FALSE;      return FALSE;
3886    ranges[2 + length] = 256;    ranges[length] = 256;
3887    length++;    length++;
3888    }    }
 ranges[0] = length;  
3889    
3890  return check_ranges(common, ranges, backtracks, FALSE);  if (length < 0 || length > 4)
3891      return FALSE;
3892    
3893    bit = bits[0] & 0x1;
3894    if (invert) bit ^= 0x1;
3895    
3896    /* No character is accepted. */
3897    if (length == 0 && bit == 0)
3898      add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
3899    
3900    switch(length)
3901      {
3902      case 0:
3903      /* When bit != 0, all characters are accepted. */
3904      return TRUE;
3905    
3906      case 1:
3907      add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
3908      return TRUE;
3909    
3910      case 2:
3911      if (ranges[0] + 1 != ranges[1])
3912        {
3913        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
3914        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
3915        }
3916      else
3917        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
3918      return TRUE;
3919    
3920      case 3:
3921      if (bit != 0)
3922        {
3923        add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
3924        if (ranges[0] + 1 != ranges[1])
3925          {
3926          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
3927          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
3928          }
3929        else
3930          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
3931        return TRUE;
3932        }
3933    
3934      add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[0]));
3935      if (ranges[1] + 1 != ranges[2])
3936        {
3937        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]);
3938        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
3939        }
3940      else
3941        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1]));
3942      return TRUE;
3943    
3944      case 4:
3945      if ((ranges[1] - ranges[0]) == (ranges[3] - ranges[2])
3946          && (ranges[0] | (ranges[2] - ranges[0])) == ranges[2]
3947          && is_powerof2(ranges[2] - ranges[0]))
3948        {
3949        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[0]);
3950        if (ranges[2] + 1 != ranges[3])
3951          {
3952          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
3953          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
3954          }
3955        else
3956          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
3957        return TRUE;
3958        }
3959    
3960      if (bit != 0)
3961        {
3962        i = 0;
3963        if (ranges[0] + 1 != ranges[1])
3964          {
3965          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
3966          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
3967          i = ranges[0];
3968          }
3969        else
3970          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
3971    
3972        if (ranges[2] + 1 != ranges[3])
3973          {
3974          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - i);
3975          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
3976          }
3977        else
3978          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2] - i));
3979        return TRUE;
3980        }
3981    
3982      OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
3983      add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[0]));
3984      if (ranges[1] + 1 != ranges[2])
3985        {
3986        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]);
3987        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
3988        }
3989      else
3990        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
3991      return TRUE;
3992    
3993      default:
3994      SLJIT_ASSERT_STOP();
3995      return FALSE;
3996      }
3997  }  }
3998    
3999  static void check_anynewline(compiler_common *common)  static void check_anynewline(compiler_common *common)
# Line 3739  return cc; Line 4371  return cc;
4371    if ((value) != charoffset) \    if ((value) != charoffset) \
4372      { \      { \
4373      if ((value) > charoffset) \      if ((value) > charoffset) \
4374        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)((value) - charoffset)); \
4375      else \      else \
4376        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(charoffset - (value))); \
4377      } \      } \
4378    charoffset = (value);    charoffset = (value);
4379    
# Line 3749  static void compile_xclass_matchingpath( Line 4381  static void compile_xclass_matchingpath(
4381  {  {
4382  DEFINE_COMPILER;  DEFINE_COMPILER;
4383  jump_list *found = NULL;  jump_list *found = NULL;
4384  jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks;  jump_list **list = (cc[0] & XCL_NOT) == 0 ? &found : backtracks;
4385  pcre_int32 c, charoffset;  sljit_uw c, charoffset, max = 0;
 const pcre_uint32 *other_cases;  
4386  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
4387  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
4388  int compares, invertcmp, numberofcmps;  int compares, invertcmp, numberofcmps;
4389    #if defined SUPPORT_UTF && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16)
4390    BOOL utf = common->utf;
4391    #endif
4392    
4393  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4394  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4395  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4396  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4397    const pcre_uint32 *other_cases;
4398  pcre_int32 typeoffset;  pcre_int32 typeoffset;
4399  #endif  #endif
4400    
 /* Although SUPPORT_UTF must be defined, we are  
    not necessary in utf mode even in 8 bit mode. */  
 detect_partial_match(common, backtracks);  
 read_char(common);  
   
 if ((*cc++ & XCL_MAP) != 0)  
   {  
   OP1(SLJIT_MOV, TMP3, 0, TMP1, 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  
   cc += 32 / sizeof(pcre_uchar);  
   }  
   
4401  /* Scanning the necessary info. */  /* Scanning the necessary info. */
4402    cc++;
4403  ccbegin = cc;  ccbegin = cc;
4404  compares = 0;  compares = 0;
4405    if (cc[-1] & XCL_MAP) cc += 32 / sizeof(pcre_uchar);
4406    
4407  while (*cc != XCL_END)  while (*cc != XCL_END)
4408    {    {
4409    compares++;    compares++;
4410    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4411      {      {
4412      cc += 2;      cc ++;
4413  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4414      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c > max) max = c;
 #endif  
4415  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4416      needschar = TRUE;      needschar = TRUE;
4417  #endif  #endif
# Line 3822  while (*cc != XCL_END) Line 4422  while (*cc != XCL_END)
4422  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4423      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
4424  #endif  #endif
4425      cc++;      GETCHARINCTEST(c, cc);
4426  #ifdef SUPPORT_UTF      if (c > max) max = c;
     if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
4427  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4428      needschar = TRUE;      needschar = TRUE;
4429  #endif  #endif
# Line 3835  while (*cc != XCL_END) Line 4433  while (*cc != XCL_END)
4433      {      {
4434      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);
4435      cc++;      cc++;
4436        if (*cc == PT_CLIST)
4437          {
4438          other_cases = PRIV(ucd_caseless_sets) + cc[1];
4439          while (*other_cases != NOTACHAR)
4440            {
4441            if (*other_cases > max) max = *other_cases;
4442            other_cases++;
4443            }
4444          }
4445        else
4446          max = READ_CHAR_ANY;
4447    
4448      switch(*cc)      switch(*cc)
4449        {        {
4450        case PT_ANY:        case PT_ANY:
# Line 3854  while (*cc != XCL_END) Line 4464  while (*cc != XCL_END)
4464        case PT_SPACE:        case PT_SPACE:
4465        case PT_PXSPACE:        case PT_PXSPACE:
4466        case PT_WORD:        case PT_WORD:
4467          case PT_PXGRAPH:
4468          case PT_PXPRINT:
4469          case PT_PXPUNCT:
4470        needstype = TRUE;        needstype = TRUE;
4471        needschar = TRUE;        needschar = TRUE;
4472        break;        break;
4473    
4474        case PT_CLIST:        case PT_CLIST:
4475          case PT_UCNC:
4476        needschar = TRUE;        needschar = TRUE;
4477        break;        break;
4478    
4479        default:        default:
4480        SLJIT_ASSERT_STOP();        SLJIT_ASSERT_STOP();
4481        break;        break;
4482          }
4483        cc += 2;
4484        }
4485    #endif
4486      }
4487    
4488    /* We are not necessary in utf mode even in 8 bit mode. */
4489    cc = ccbegin;
4490    detect_partial_match(common, backtracks);
4491    read_char_max(common, max, (cc[0] & XCL_NOT) != 0);
4492    
4493    if ((cc[-1] & XCL_HASPROP) == 0)
4494      {
4495      if ((cc[-1] & XCL_MAP) != 0)
4496        {
4497        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4498        if (!check_class_ranges(common, (const pcre_uint8 *)cc, (((const pcre_uint8 *)cc)[31] & 0x80) != 0, TRUE, &found))
4499          {
4500          OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4501          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4502          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4503          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4504          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4505          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4506        }        }
4507      cc += 2;  
4508        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4509        JUMPHERE(jump);
4510    
4511        cc += 32 / sizeof(pcre_uchar);
4512      }      }
4513      else
4514        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0xff));
4515      }
4516    else if ((cc[-1] & XCL_MAP) != 0)
4517      {
4518      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4519    #ifdef SUPPORT_UCP
4520      charsaved = TRUE;
4521    #endif
4522      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4523        {
4524    #ifdef COMPILE_PCRE8
4525        SLJIT_ASSERT(common->utf);
4526  #endif  #endif
4527        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4528    
4529        OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4530        OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4531        OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4532        OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4533        OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4534        add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
4535    
4536        JUMPHERE(jump);
4537        }
4538    
4539      OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4540      cc += 32 / sizeof(pcre_uchar);
4541    }    }
4542    
4543  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3912  if (needstype || needsscript) Line 4581  if (needstype || needsscript)
4581  #endif  #endif
4582    
4583  /* Generating code. */  /* Generating code. */
 cc = ccbegin;  
4584  charoffset = 0;  charoffset = 0;
4585  numberofcmps = 0;  numberofcmps = 0;
4586  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3928  while (*cc != XCL_END) Line 4596  while (*cc != XCL_END)
4596    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4597      {      {
4598      cc ++;      cc ++;
4599  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4600    
4601      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4602        {        {
4603        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));
4604        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);
4605        numberofcmps++;        numberofcmps++;
4606        }        }
4607      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4608        {        {
4609        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));
4610        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);
4611        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4612        numberofcmps = 0;        numberofcmps = 0;
4613        }        }
4614      else      else
4615        {        {
4616        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));
4617        numberofcmps = 0;        numberofcmps = 0;
4618        }        }
4619      }      }
4620    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4621      {      {
4622      cc ++;      cc ++;
4623  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4624      SET_CHAR_OFFSET(c);      SET_CHAR_OFFSET(c);
4625  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4626      if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4627      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4628        {        {
4629        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));
4630        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);
4631        numberofcmps++;        numberofcmps++;
4632        }        }
4633      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4634        {        {
4635        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));
4636        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);
4637        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4638        numberofcmps = 0;        numberofcmps = 0;
4639        }        }
4640      else      else
4641        {        {
4642        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));
4643        numberofcmps = 0;        numberofcmps = 0;
4644        }        }
4645      }      }
# Line 4040  while (*cc != XCL_END) Line 4688  while (*cc != XCL_END)
4688    
4689        case PT_SPACE:        case PT_SPACE:
4690        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);  
         }  
4691        SET_CHAR_OFFSET(9);        SET_CHAR_OFFSET(9);
4692        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);
4693        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);
4694        if (*cc == PT_SPACE)  
4695          JUMPHERE(jump);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9);
4696          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4697    
4698          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9);
4699          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4700    
4701        SET_TYPE_OFFSET(ucp_Zl);        SET_TYPE_OFFSET(ucp_Zl);
4702        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 4058  while (*cc != XCL_END) Line 4705  while (*cc != XCL_END)
4705        break;        break;
4706    
4707        case PT_WORD:        case PT_WORD:
4708        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));
4709        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4710        /* ... fall through */        /* Fall through. */
4711    
4712        case PT_ALNUM:        case PT_ALNUM:
4713        SET_TYPE_OFFSET(ucp_Ll);        SET_TYPE_OFFSET(ucp_Ll);
# Line 4106  while (*cc != XCL_END) Line 4753  while (*cc != XCL_END)
4753          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]);
4754          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4755    
4756          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));
4757          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);
4758    
4759          other_cases += 3;          other_cases += 3;
4760          }          }
4761        else        else
4762          {          {
4763          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));
4764          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4765          }          }
4766    
4767        while (*other_cases != NOTACHAR)        while (*other_cases != NOTACHAR)
4768          {          {
4769          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));
4770          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);
4771          }          }
4772        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4773        break;        break;
4774    
4775          case PT_UCNC:
4776          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_DOLLAR_SIGN - charoffset));
4777          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
4778          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_COMMERCIAL_AT - charoffset));
4779          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4780          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_GRAVE_ACCENT - charoffset));
4781          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4782    
4783          SET_CHAR_OFFSET(0xa0);
4784          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(0xd7ff - charoffset));
4785          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4786          SET_CHAR_OFFSET(0);
4787          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0);
4788          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL);
4789          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4790          break;
4791    
4792          case PT_PXGRAPH:
4793          /* C and Z groups are the farthest two groups. */
4794          SET_TYPE_OFFSET(ucp_Ll);
4795          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
4796          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
4797    
4798          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
4799    
4800          /* In case of ucp_Cf, we overwrite the result. */
4801          SET_CHAR_OFFSET(0x2066);
4802          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4803          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4804    
4805          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4806          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4807    
4808          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
4809          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4810    
4811          JUMPHERE(jump);
4812          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
4813          break;
4814    
4815          case PT_PXPRINT:
4816          /* C and Z groups are the farthest two groups. */
4817          SET_TYPE_OFFSET(ucp_Ll);
4818          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
4819          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
4820    
4821          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
4822          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
4823    
4824          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
4825    
4826          /* In case of ucp_Cf, we overwrite the result. */
4827          SET_CHAR_OFFSET(0x2066);
4828          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4829          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4830    
4831          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4832          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4833    
4834          JUMPHERE(jump);
4835          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
4836          break;
4837    
4838          case PT_PXPUNCT:
4839          SET_TYPE_OFFSET(ucp_Sc);
4840          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc);
4841          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
4842    
4843          SET_CHAR_OFFSET(0);
4844          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff);
4845          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4846    
4847          SET_TYPE_OFFSET(ucp_Pc);
4848          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc);
4849          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
4850          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4851          break;
4852        }        }
4853      cc += 2;      cc += 2;
4854      }      }
# Line 4149  int length; Line 4874  int length;
4874  unsigned int c, oc, bit;  unsigned int c, oc, bit;
4875  compare_context context;  compare_context context;
4876  struct sljit_jump *jump[4];  struct sljit_jump *jump[4];
4877    jump_list *end_list;
4878  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4879  struct sljit_label *label;  struct sljit_label *label;
4880  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4881  pcre_uchar propdata[5];  pcre_uchar propdata[5];
4882  #endif  #endif
4883  #endif  #endif /* SUPPORT_UTF */
4884    
4885  switch(type)  switch(type)
4886    {    {
# Line 4179  switch(type) Line 4905  switch(type)
4905    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
4906    case OP_DIGIT:    case OP_DIGIT:
4907    /* 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);  
4908    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4909    /* Flip the starting bit in the negative case. */  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4910    if (type == OP_NOT_DIGIT)    if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_digit, FALSE))
4911      common->digits[1] ^= 1;      read_char7_type(common, type == OP_NOT_DIGIT);
4912    if (!check_ranges(common, common->digits, backtracks, TRUE))    else
4913      {  #endif
4914      read_char8_type(common);      read_char8_type(common, type == OP_NOT_DIGIT);
4915      OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);      /* Flip the starting bit in the negative case. */
4916      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);
4917      }    add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
   if (type == OP_NOT_DIGIT)  
     common->digits[1] ^= 1;  
4918    return cc;    return cc;
4919    
4920    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
4921    case OP_WHITESPACE:    case OP_WHITESPACE:
4922    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4923    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4924      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_space, FALSE))
4925        read_char7_type(common, type == OP_NOT_WHITESPACE);
4926      else
4927    #endif
4928        read_char8_type(common, type == OP_NOT_WHITESPACE);
4929    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);
4930    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));
4931    return cc;    return cc;
# Line 4206  switch(type) Line 4933  switch(type)
4933    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
4934    case OP_WORDCHAR:    case OP_WORDCHAR:
4935    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4936    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4937      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_word, FALSE))
4938        read_char7_type(common, type == OP_NOT_WORDCHAR);
4939      else
4940    #endif
4941        read_char8_type(common, type == OP_NOT_WORDCHAR);
4942    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);
4943    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));
4944    return cc;    return cc;
4945    
4946    case OP_ANY:    case OP_ANY:
4947    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4948    read_char(common);    read_char_max(common, common->nlmax, TRUE);
4949    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
4950      {      {
4951      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);
4952        end_list = NULL;
4953      if (common->mode != JIT_PARTIAL_HARD_COMPILE)      if (common->mode != JIT_PARTIAL_HARD_COMPILE)
4954        jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);        add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
4955      else      else
4956        jump[1] = check_str_end(common);        check_str_end(common, &end_list);
4957    
4958      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
4959      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));
4960      if (jump[1] != NULL)      set_jumps(end_list, LABEL());
       JUMPHERE(jump[1]);  
4961      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
4962      }      }
4963    else    else
# Line 4269  switch(type) Line 5001  switch(type)
5001  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5002    case OP_NOTPROP:    case OP_NOTPROP:
5003    case OP_PROP:    case OP_PROP:
5004    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
5005    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
5006    propdata[2] = cc[0];    propdata[2] = cc[0];
5007    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4281  switch(type) Line 5013  switch(type)
5013    
5014    case OP_ANYNL:    case OP_ANYNL:
5015    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5016    read_char(common);    read_char_max(common, common->bsr_nlmax, FALSE);
5017    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);
5018    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
5019      end_list = NULL;
5020    if (common->mode != JIT_PARTIAL_HARD_COMPILE)    if (common->mode != JIT_PARTIAL_HARD_COMPILE)
5021      jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);      add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
5022    else    else
5023      jump[1] = check_str_end(common);      check_str_end(common, &end_list);
5024    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
5025    jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);    jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);
5026    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));
5027    jump[3] = JUMP(SLJIT_JUMP);    jump[2] = JUMP(SLJIT_JUMP);
5028    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
5029    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);
5030      set_jumps(end_list, LABEL());
5031    JUMPHERE(jump[1]);    JUMPHERE(jump[1]);
5032    JUMPHERE(jump[2]);    JUMPHERE(jump[2]);
   JUMPHERE(jump[3]);  
5033    return cc;    return cc;
5034    
5035    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
5036    case OP_HSPACE:    case OP_HSPACE:
5037    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5038    read_char(common);    read_char_max(common, 0x3000, type == OP_NOT_HSPACE);
5039    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));
5040    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));
5041    return cc;    return cc;
# Line 4310  switch(type) Line 5043  switch(type)
5043    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
5044    case OP_VSPACE:    case OP_VSPACE:
5045    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5046    read_char(common);    read_char_max(common, 0x2029, type == OP_NOT_VSPACE);
5047    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));
5048    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));
5049    return cc;    return cc;
# Line 4409  switch(type) Line 5142  switch(type)
5142      else      else
5143        {        {
5144        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);
5145        read_char(common);        read_char_max(common, common->nlmax, TRUE);
5146        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));
5147        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
5148        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4457  switch(type) Line 5190  switch(type)
5190    else    else
5191      {      {
5192      skip_char_back(common);      skip_char_back(common);
5193      read_char(common);      read_char_max(common, common->nlmax, TRUE);
5194      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5195      }      }
5196    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4532  switch(type) Line 5265  switch(type)
5265  #endif  #endif
5266      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);
5267      }      }
5268    
5269    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
   read_char(common);  
5270  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5271    if (common->utf)    if (common->utf)
5272      {      {
# Line 4542  switch(type) Line 5275  switch(type)
5275    else    else
5276  #endif  #endif
5277      c = *cc;      c = *cc;
5278    
5279    if (type == OP_CHAR || !char_has_othercase(common, cc))    if (type == OP_CHAR || !char_has_othercase(common, cc))
5280      {      {
5281        read_char_max(common, c, FALSE);
5282      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));
5283      return cc + length;      return cc + length;
5284      }      }
5285    oc = char_othercase(common, c);    oc = char_othercase(common, c);
5286      read_char_max(common, c > oc ? c : oc, FALSE);
5287    bit = c ^ oc;    bit = c ^ oc;
5288    if (is_powerof2(bit))    if (is_powerof2(bit))
5289      {      {
# Line 4555  switch(type) Line 5291  switch(type)
5291      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));
5292      return cc + length;      return cc + length;
5293      }      }
5294    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);
5295    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));
5296    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));  
5297    return cc + length;    return cc + length;
5298    
5299    case OP_NOT:    case OP_NOT:
# Line 4594  switch(type) Line 5328  switch(type)
5328  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
5329        {        {
5330        GETCHARLEN(c, cc, length);        GETCHARLEN(c, cc, length);
       read_char(common);  
5331        }        }
5332      }      }
5333    else    else
5334  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
     {  
     read_char(common);  
5335      c = *cc;      c = *cc;
     }  
5336    
5337    if (type == OP_NOT || !char_has_othercase(common, cc))    if (type == OP_NOT || !char_has_othercase(common, cc))
5338        {
5339        read_char_max(common, c, TRUE);
5340      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));
5341        }
5342    else    else
5343      {      {
5344      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5345        read_char_max(common, c > oc ? c : oc, TRUE);
5346      bit = c ^ oc;      bit = c ^ oc;
5347      if (is_powerof2(bit))      if (is_powerof2(bit))
5348        {        {
# Line 4626  switch(type) Line 5360  switch(type)
5360    case OP_CLASS:    case OP_CLASS:
5361    case OP_NCLASS:    case OP_NCLASS:
5362    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5363    read_char(common);  
5364    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5365      bit = (common->utf && is_char7_bitset((const pcre_uint8 *)cc, type == OP_NCLASS)) ? 127 : 255;
5366      read_char_max(common, bit, type == OP_NCLASS);
5367    #else
5368      read_char_max(common, 255, type == OP_NCLASS);
5369    #endif
5370    
5371      if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5372      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5373    
5374  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5375    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. */  
5376    if (common->utf)    if (common->utf)
 #endif /* COMPILE_PCRE8 */  
5377      {      {
5378      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, bit);
5379      if (type == OP_CLASS)      if (type == OP_CLASS)
5380        {        {
5381        add_jump(compiler, backtracks, jump[0]);        add_jump(compiler, backtracks, jump[0]);
5382        jump[0] = NULL;        jump[0] = NULL;
5383        }        }
5384      }      }
5385  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #elif !defined COMPILE_PCRE8
5386      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
5387      if (type == OP_CLASS)
5388        {
5389        add_jump(compiler, backtracks, jump[0]);
5390        jump[0] = NULL;
5391        }
5392    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
5393    
5394    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5395    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5396    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5397    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5398    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);
5399    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
5400    
5401  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5402    if (jump[0] != NULL)    if (jump[0] != NULL)
5403      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5404  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif
5405    
5406    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5407    
5408  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 4760  if (context.length > 0) Line 5507  if (context.length > 0)
5507  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);
5508  }  }
5509    
 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));  
 }  
   
5510  /* Forward definitions. */  /* Forward definitions. */
5511  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);
5512  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);
# Line 4814  static void compile_backtrackingpath(com Line 5539  static void compile_backtrackingpath(com
5539    
5540  #define BACKTRACK_AS(type) ((type *)backtrack)  #define BACKTRACK_AS(type) ((type *)backtrack)
5541    
5542  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)
5543    {
5544    /* The OVECTOR offset goes to TMP2. */
5545    DEFINE_COMPILER;
5546    int count = GET2(cc, 1 + IMM2_SIZE);
5547    pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
5548    unsigned int offset;
5549    jump_list *found = NULL;
5550    
5551    SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
5552    
5553    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
5554    
5555    count--;
5556    while (count-- > 0)
5557      {
5558      offset = GET2(slot, 0) << 1;
5559      GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5560      add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5561      slot += common->name_entry_size;
5562      }
5563    
5564    offset = GET2(slot, 0) << 1;
5565    GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5566    if (backtracks != NULL && !common->jscript_compat)
5567      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5568    
5569    set_jumps(found, LABEL());
5570    }
5571    
5572    static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
5573  {  {
5574  DEFINE_COMPILER;  DEFINE_COMPILER;
5575  int offset = GET2(cc, 1) << 1;  BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5576    int offset = 0;
5577  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
5578  struct sljit_jump *partial;  struct sljit_jump *partial;
5579  struct sljit_jump *nopartial;  struct sljit_jump *nopartial;
5580    
5581  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  if (ref)
5582  /* OVECTOR(1) contains the "string begin - 1" constant. */    {
5583  if (withchecks && !common->jscript_compat)    offset = GET2(cc, 1) << 1;
5584    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));
5585      /* OVECTOR(1) contains the "string begin - 1" constant. */
5586      if (withchecks && !common->jscript_compat)
5587        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5588      }
5589    else
5590      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5591    
5592  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
5593  if (common->utf && *cc == OP_REFI)  if (common->utf && *cc == OP_REFI)
5594    {    {
5595    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);
5596    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));    if (ref)
5597        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5598      else
5599        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5600    
5601    if (withchecks)    if (withchecks)
5602      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);
5603    
# Line 4856  if (common->utf && *cc == OP_REFI) Line 5622  if (common->utf && *cc == OP_REFI)
5622  else  else
5623  #endif /* SUPPORT_UTF && SUPPORT_UCP */  #endif /* SUPPORT_UTF && SUPPORT_UCP */
5624    {    {
5625    OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);    if (ref)
5626        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);
5627      else
5628        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
5629    
5630    if (withchecks)    if (withchecks)
5631      jump = JUMP(SLJIT_C_ZERO);      jump = JUMP(SLJIT_C_ZERO);
5632    
# Line 4893  if (jump != NULL) Line 5663  if (jump != NULL)
5663    else    else
5664      JUMPHERE(jump);      JUMPHERE(jump);
5665    }    }
 return cc + 1 + IMM2_SIZE;  
5666  }  }
5667    
5668  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)
5669  {  {
5670  DEFINE_COMPILER;  DEFINE_COMPILER;
5671    BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5672  backtrack_common *backtrack;  backtrack_common *backtrack;
5673  pcre_uchar type;  pcre_uchar type;
5674    int offset = 0;
5675  struct sljit_label *label;  struct sljit_label *label;
5676  struct sljit_jump *zerolength;  struct sljit_jump *zerolength;
5677  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
# Line 4910  BOOL minimize; Line 5681  BOOL minimize;
5681    
5682  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);
5683    
5684    if (ref)
5685      offset = GET2(cc, 1) << 1;
5686    else
5687      cc += IMM2_SIZE;
5688  type = cc[1 + IMM2_SIZE];  type = cc[1 + IMM2_SIZE];
5689    
5690    SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even);
5691  minimize = (type & 0x1) != 0;  minimize = (type & 0x1) != 0;
5692  switch(type)  switch(type)
5693    {    {
# Line 4948  if (!minimize) Line 5725  if (!minimize)
5725    if (min == 0)    if (min == 0)
5726      {      {
5727      allocate_stack(common, 2);      allocate_stack(common, 2);
5728        if (ref)
5729          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5730      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
5731      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
5732      /* Temporary release of STR_PTR. */      /* Temporary release of STR_PTR. */
5733      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));
5734      zerolength = compile_ref_checks(common, ccbegin, NULL);      /* Handles both invalid and empty cases. Since the minimum repeat,
5735        is zero the invalid case is basically the same as an empty case. */
5736        if (ref)
5737          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5738        else
5739          {
5740          compile_dnref_search(common, ccbegin, NULL);
5741          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5742          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
5743          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5744          }
5745      /* Restore if not zero length. */      /* Restore if not zero length. */
5746      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));
5747      }      }
5748    else    else
5749      {      {
5750      allocate_stack(common, 1);      allocate_stack(common, 1);
5751        if (ref)
5752          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5753      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
5754      zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);      if (ref)
5755          {
5756          add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5757          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5758          }
5759        else
5760          {
5761          compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
5762          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5763          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
5764          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5765          }
5766      }      }
5767    
5768    if (min > 1 || max > 1)    if (min > 1 || max > 1)
5769      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0);
5770    
5771    label = LABEL();    label = LABEL();
5772      if (!ref)
5773        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1);
5774    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);    compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE);
5775    
5776    if (min > 1 || max > 1)    if (min > 1 || max > 1)
# Line 4997  if (!minimize) Line 5801  if (!minimize)
5801    JUMPHERE(zerolength);    JUMPHERE(zerolength);
5802    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();    BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
5803    
5804    decrease_call_count(common);    count_match(common);
5805    return cc;    return cc;
5806    }    }
5807    
5808  allocate_stack(common, 2);  allocate_stack(common, ref ? 2 : 3);
5809    if (ref)
5810      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5811  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
5812  if (type != OP_CRMINSTAR)  if (type != OP_CRMINSTAR)
5813    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
5814    
5815  if (min == 0)  if (min == 0)
5816    {    {
5817    zerolength = compile_ref_checks(common, ccbegin, NULL);    /* Handles both invalid and empty cases. Since the minimum repeat,
5818      is zero the invalid case is basically the same as an empty case. */
5819      if (ref)
5820        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5821      else
5822        {
5823        compile_dnref_search(common, ccbegin, NULL);
5824        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5825        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
5826        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5827        }
5828      /* Length is non-zero, we can match real repeats. */
5829    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);    OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
5830    jump = JUMP(SLJIT_JUMP);    jump = JUMP(SLJIT_JUMP);
5831    }    }
5832  else  else
5833    zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);    {
5834      if (ref)
5835        {
5836        add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5837        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5838        }
5839      else
5840        {
5841        compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);
5842        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5843        OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
5844        zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5845        }
5846      }
5847    
5848  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();  BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL();
5849  if (max > 0)  if (max > 0)
5850    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));
5851    
5852    if (!ref)
5853      OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
5854  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);  compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE);
5855  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
5856    
# Line 5036  if (jump != NULL) Line 5868  if (jump != NULL)
5868    JUMPHERE(jump);    JUMPHERE(jump);
5869  JUMPHERE(zerolength);  JUMPHERE(zerolength);
5870    
5871  decrease_call_count(common);  count_match(common);
5872  return cc;  return cc;
5873  }  }
5874    
# Line 5046  DEFINE_COMPILER; Line 5878  DEFINE_COMPILER;
5878  backtrack_common *backtrack;  backtrack_common *backtrack;
5879  recurse_entry *entry = common->entries;  recurse_entry *entry = common->entries;
5880  recurse_entry *prev = NULL;  recurse_entry *prev = NULL;
5881  int start = GET(cc, 1);  sljit_sw start = GET(cc, 1);
5882    pcre_uchar *start_cc;
5883    BOOL needs_control_head;
5884    
5885  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);
5886    
5887    /* Inlining simple patterns. */
5888    if (get_framesize(common, common->start + start, NULL, TRUE, &needs_control_head) == no_stack)
5889      {
5890      start_cc = common->start + start;
5891      compile_matchingpath(common, next_opcode(common, start_cc), bracketend(start_cc) - (1 + LINK_SIZE), backtrack);
5892      BACKTRACK_AS(recurse_backtrack)->inlined_pattern = TRUE;
5893      return cc + 1 + LINK_SIZE;
5894      }
5895    
5896  while (entry != NULL)  while (entry != NULL)
5897    {    {
5898    if (entry->start == start)    if (entry->start == start)
# Line 5160  allocate_stack(common, CALLOUT_ARG_SIZE Line 6004  allocate_stack(common, CALLOUT_ARG_SIZE
6004  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr);
6005  OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
6006  SLJIT_ASSERT(common->capture_last_ptr != 0);  SLJIT_ASSERT(common->capture_last_ptr != 0);
6007  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_number), SLJIT_IMM, cc[1]);  OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_number), SLJIT_IMM, cc[1]);
6008  OP1(SLJIT_MOV_UI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(capture_last), TMP2, 0);  OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(capture_last), TMP2, 0);
6009    
6010  /* These pointer sized fields temporarly stores internal variables. */  /* These pointer sized fields temporarly stores internal variables. */
6011  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));  OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0));
# Line 5170  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CA Line 6014  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CA
6014    
6015  if (common->mark_ptr != 0)  if (common->mark_ptr != 0)
6016    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr));    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr));
6017  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(pattern_position), SLJIT_IMM, GET(cc, 2));  OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(pattern_position), SLJIT_IMM, GET(cc, 2));
6018  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(next_item_length), SLJIT_IMM, GET(cc, 2 + LINK_SIZE));  OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(next_item_length), SLJIT_IMM, GET(cc, 2 + LINK_SIZE));
6019  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(mark), (common->mark_ptr != 0) ? TMP2 : SLJIT_IMM, 0);  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(mark), (common->mark_ptr != 0) ? TMP2 : SLJIT_IMM, 0);
6020    
6021  /* Needed to save important temporary registers. */  /* Needed to save important temporary registers. */
# Line 5200  static pcre_uchar *compile_assert_matchi Line 6044  static pcre_uchar *compile_assert_matchi
6044  {  {
6045  DEFINE_COMPILER;  DEFINE_COMPILER;
6046  int framesize; &nb