/[pcre]/code/trunk/pcre_jit_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_jit_compile.c

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

revision 1402 by zherczeg, Fri Nov 15 12:04:55 2013 UTC revision 1423 by zherczeg, Tue Dec 31 07:57:56 2013 UTC
# Line 306  typedef struct then_trap_backtrack { Line 306  typedef struct then_trap_backtrack {
306    int framesize;    int framesize;
307  } then_trap_backtrack;  } then_trap_backtrack;
308    
309  #define MAX_RANGE_SIZE 6  #define MAX_RANGE_SIZE 4
310    
311  typedef struct compiler_common {  typedef struct compiler_common {
312    /* The sljit ceneric compiler. */    /* The sljit ceneric compiler. */
# Line 363  typedef struct compiler_common { Line 363  typedef struct compiler_common {
363    BOOL positive_assert;    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;
372    /* Tables. */    /* Tables. */
373    sljit_sw ctypes;    sljit_sw ctypes;
   int digits[2 + MAX_RANGE_SIZE];  
374    /* Named capturing brackets. */    /* Named capturing brackets. */
375    pcre_uchar *name_table;    pcre_uchar *name_table;
376    sljit_sw name_count;    sljit_sw name_count;
# Line 404  typedef struct compiler_common { Line 405  typedef struct compiler_common {
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 524  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 533  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   check_opcode_types   check_opcode_types
# Line 2457  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 2467  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);    JUMPHERE(jump);
2485    }    }
2486  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  #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));  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);
2544      }
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 SLJIT_INLINE void read_char(compiler_common *common)
2632    {
2633    read_char_max(common, READ_CHAR_ANY, TRUE);
2634  }  }
2635    
2636  static void read_char8_type(compiler_common *common)  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. */  /* 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 2595  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 2626  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 2697  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 2818  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 2894  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 2949  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 2960  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;      break;
3135    
3136      default:  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3137      must_stop = 2;      case OP_XCLASS:
3138        any = TRUE;
3139        cc += GET(cc, 1);
3140      break;      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;
3154    
3155    #ifdef SUPPORT_UCP
3156        case OP_NOTPROP:
3157        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 (must_stop == 2)    if (any)
3173        break;      {
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        repeat = 1;
3200        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);
3211        if (caseless == 0)
3212          return consumed;
3213    #ifdef COMPILE_PCRE8
3214        caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));
3215    #else
3216        if ((caseless & 0x100) != 0)
3217          caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));
3218        else
3219          caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));
3220    #endif
3221        }
3222      else
3223        caseless = 0;
3224    
3225      len_save = len;
3226      cc_save = cc;
3227      while (TRUE)
3228        {
3229        do
3230          {
3231          chr = *cc;
3232    #ifdef COMPILE_PCRE32
3233          if (SLJIT_UNLIKELY(chr == NOTACHAR))
3234            return consumed;
3235    #endif
3236          mask = 0;
3237          if (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        if (--repeat == 0)
3266          break;
3267    
3268        len = len_save;
3269        cc = cc_save;
3270        }
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      caseless = char_get_othercase_bit(common, cc);      ones[i] += ones_in_half_byte[mask & 0xf];
3310      if (caseless == 0)      mask >>= 4;
       return FALSE;  
 #ifdef COMPILE_PCRE8  
     caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));  
 #else  
     if ((caseless & 0x100) != 0)  
       caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));  
     else  
       caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));  
 #endif  
3311      }      }
3312    else    }
     caseless = 0;  
3313    
3314    while (len > 0 && location < MAX_N_CHARS * 2)  offsets[0] = -1;
3315      {  /* Scan forward. */
3316      c = *cc;  for (i = 0; i < max; i++)
3317      bit = 0;    if (ones[i] <= 2) {
3318      if (len == (caseless & 0xff))      offsets[0] = i;
3319        break;
3320      }
3321    
3322    if (offsets[0] == -1)
3323      return FALSE;
3324    
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        bit = caseless >> 8;        offsets[2] = i;
3341        c |= bit;        break;
3342        }        }
3343    
3344      chars[location] = c;    if (offsets[2] == -1)
3345      chars[location + 1] = bit;      {
3346        for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--)
3347      len--;        if (ones[i] <= 2)
3348      location += 2;          {
3349      cc++;          offsets[2] = i;
3350            break;
3351            }
3352      }      }
   
   if (location >= MAX_N_CHARS * 2 || must_stop != 0)  
     break;  
3353    }    }
3354    
3355  /* At least two characters are required. */  SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1]));
3356  if (location < 2 * 2)  SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2]));
3357      return FALSE;  
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 3057  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 3177  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 3206  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 3239  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 3248  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 3497  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE Line 3847  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSE
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 3615  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 3628  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 4021  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;  pcre_int32 c, charoffset;
 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;
# Line 4032  int compares, invertcmp, numberofcmps; Line 4391  int compares, invertcmp, numberofcmps;
4391  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4392  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4393  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4394    const pcre_uint32 *other_cases;
4395  pcre_int32 typeoffset;  pcre_int32 typeoffset;
4396  #endif  #endif
4397    
# Line 4040  pcre_int32 typeoffset; Line 4400  pcre_int32 typeoffset;
4400  detect_partial_match(common, backtracks);  detect_partial_match(common, backtracks);
4401  read_char(common);  read_char(common);
4402    
4403  if ((*cc++ & XCL_MAP) != 0)  cc++;
4404    if ((cc[-1] & XCL_HASPROP) == 0)
4405    {    {
4406    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    if ((cc[-1] & XCL_MAP) != 0)
4407  #ifndef COMPILE_PCRE8      {
4408    jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4409  #elif defined SUPPORT_UTF  #ifdef SUPPORT_UCP
4410    if (common->utf)      charsaved = TRUE;
     jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
4411  #endif  #endif
4412        if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, FALSE, backtracks))
4413          {
4414          jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4415    
4416          OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4417          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4418          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4419          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4420          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4421          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4422          add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4423    
4424          JUMPHERE(jump);
4425          }
4426        else
4427          add_jump(compiler, &found, CMP(SLJIT_C_LESS_EQUAL, TMP3, 0, SLJIT_IMM, 0xff));
4428    
4429    if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list))      OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4430        cc += 32 / sizeof(pcre_uchar);
4431        }
4432      else
4433        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0xff));
4434      }
4435    else if ((cc[-1] & XCL_MAP) != 0)
4436      {
4437      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4438    #ifdef SUPPORT_UCP
4439      charsaved = TRUE;
4440    #endif
4441      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4442      {      {
4443    #ifdef COMPILE_PCRE8
4444        SLJIT_ASSERT(common->utf);
4445    #endif
4446        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4447    
4448      OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);      OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4449      OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);      OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4450      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4451      OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);      OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4452      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);
4453      add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));      add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
     }  
4454    
 #ifndef COMPILE_PCRE8  
   JUMPHERE(jump);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
4455      JUMPHERE(jump);      JUMPHERE(jump);
4456  #endif      }
4457    
4458    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);    OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
 #ifdef SUPPORT_UCP  
   charsaved = TRUE;  
 #endif  
4459    cc += 32 / sizeof(pcre_uchar);    cc += 32 / sizeof(pcre_uchar);
4460    }    }
4461    
# Line 4431  while (*cc != XCL_END) Line 4817  while (*cc != XCL_END)
4817        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4818        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);
4819    
4820        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4821        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4822    
4823        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
4824        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4825    
4826        JUMPHERE(jump);        JUMPHERE(jump);
# Line 4447  while (*cc != XCL_END) Line 4833  while (*cc != XCL_END)
4833        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
4834        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
4835    
4836        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
4837        OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);        OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
4838    
4839        jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);        jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
# Line 4457  while (*cc != XCL_END) Line 4843  while (*cc != XCL_END)
4843        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
4844        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);
4845    
4846        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
4847        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4848    
4849        JUMPHERE(jump);        JUMPHERE(jump);
# Line 4509  struct sljit_label *label; Line 4895  struct sljit_label *label;
4895  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4896  pcre_uchar propdata[5];  pcre_uchar propdata[5];
4897  #endif  #endif
4898  #endif  #endif /* SUPPORT_UTF */
4899    
4900  switch(type)  switch(type)
4901    {    {
# Line 4534  switch(type) Line 4920  switch(type)
4920    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
4921    case OP_DIGIT:    case OP_DIGIT:
4922    /* 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);  
4923    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4924    /* Flip the starting bit in the negative case. */  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4925    if (type == OP_NOT_DIGIT)    if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_digit, FALSE))
4926      common->digits[1] ^= 1;      read_char7_type(common, type == OP_NOT_DIGIT);
4927    if (!check_ranges(common, common->digits, backtracks, TRUE))    else
4928      {  #endif
4929      read_char8_type(common);      read_char8_type(common, type == OP_NOT_DIGIT);
4930      OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);      /* Flip the starting bit in the negative case. */
4931      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);
4932      }    add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
   if (type == OP_NOT_DIGIT)  
     common->digits[1] ^= 1;  
4933    return cc;    return cc;
4934    
4935    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
4936    case OP_WHITESPACE:    case OP_WHITESPACE:
4937    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4938    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4939      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_space, FALSE))
4940        read_char7_type(common, type == OP_NOT_WHITESPACE);
4941      else
4942    #endif
4943        read_char8_type(common, type == OP_NOT_WHITESPACE);
4944    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);
4945    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));
4946    return cc;    return cc;
# Line 4561  switch(type) Line 4948  switch(type)
4948    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
4949    case OP_WORDCHAR:    case OP_WORDCHAR:
4950    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4951    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
4952      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_word, FALSE))
4953        read_char7_type(common, type == OP_NOT_WORDCHAR);
4954      else
4955    #endif
4956        read_char8_type(common, type == OP_NOT_WORDCHAR);
4957    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);
4958    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));
4959    return cc;    return cc;
4960    
4961    case OP_ANY:    case OP_ANY:
4962    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
4963    read_char(common);    read_char_max(common, common->nlmax, TRUE);
4964    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
4965      {      {
4966      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);
# Line 4624  switch(type) Line 5016  switch(type)
5016  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5017    case OP_NOTPROP:    case OP_NOTPROP:
5018    case OP_PROP:    case OP_PROP:
5019    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
5020    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
5021    propdata[2] = cc[0];    propdata[2] = cc[0];
5022    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4636  switch(type) Line 5028  switch(type)
5028    
5029    case OP_ANYNL:    case OP_ANYNL:
5030    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5031    read_char(common);    read_char_max(common, common->bsr_nlmax, FALSE);
5032    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);
5033    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
5034    end_list = NULL;    end_list = NULL;
# Line 4658  switch(type) Line 5050  switch(type)
5050    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
5051    case OP_HSPACE:    case OP_HSPACE:
5052    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5053    read_char(common);    read_char_max(common, 0x3000, type == OP_NOT_HSPACE);
5054    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));
5055    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));
5056    return cc;    return cc;
# Line 4666  switch(type) Line 5058  switch(type)
5058    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
5059    case OP_VSPACE:    case OP_VSPACE:
5060    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5061    read_char(common);    read_char_max(common, 0x2029, type == OP_NOT_VSPACE);
5062    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));
5063    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));
5064    return cc;    return cc;
# Line 4765  switch(type) Line 5157  switch(type)
5157      else      else
5158        {        {
5159        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);
5160        read_char(common);        read_char_max(common, common->nlmax, TRUE);
5161        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));
5162        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
5163        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4813  switch(type) Line 5205  switch(type)
5205    else    else
5206      {      {
5207      skip_char_back(common);      skip_char_back(common);
5208      read_char(common);      read_char_max(common, common->nlmax, TRUE);
5209      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5210      }      }
5211    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4888  switch(type) Line 5280  switch(type)
5280  #endif  #endif
5281      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);
5282      }      }
5283    
5284    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
   read_char(common);  
5285  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5286    if (common->utf)    if (common->utf)
5287      {      {
# Line 4898  switch(type) Line 5290  switch(type)
5290    else    else
5291  #endif  #endif
5292      c = *cc;      c = *cc;
5293    
5294    if (type == OP_CHAR || !char_has_othercase(common, cc))    if (type == OP_CHAR || !char_has_othercase(common, cc))
5295      {      {
5296        read_char_max(common, c, FALSE);
5297      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));
5298      return cc + length;      return cc + length;
5299      }      }
5300    oc = char_othercase(common, c);    oc = char_othercase(common, c);
5301      read_char_max(common, c > oc ? c : oc, FALSE);
5302    bit = c ^ oc;    bit = c ^ oc;
5303    if (is_powerof2(bit))    if (is_powerof2(bit))
5304      {      {
# Line 4911  switch(type) Line 5306  switch(type)
5306      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));
5307      return cc + length;      return cc + length;
5308      }      }
5309    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);
5310    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));
5311    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));  
5312    return cc + length;    return cc + length;
5313    
5314    case OP_NOT:    case OP_NOT:
# Line 4950  switch(type) Line 5343  switch(type)
5343  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
5344        {        {
5345        GETCHARLEN(c, cc, length);        GETCHARLEN(c, cc, length);
       read_char(common);  
5346        }        }
5347      }      }
5348    else    else
5349  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
     {  
     read_char(common);  
5350      c = *cc;      c = *cc;
     }  
5351    
5352    if (type == OP_NOT || !char_has_othercase(common, cc))    if (type == OP_NOT || !char_has_othercase(common, cc))
5353        {
5354        read_char_max(common, c, TRUE);
5355      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));
5356        }
5357    else    else
5358      {      {
5359      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5360        read_char_max(common, c > oc ? c : oc, TRUE);
5361      bit = c ^ oc;      bit = c ^ oc;
5362      if (is_powerof2(bit))      if (is_powerof2(bit))
5363        {        {
# Line 4982  switch(type) Line 5375  switch(type)
5375    case OP_CLASS:    case OP_CLASS:
5376    case OP_NCLASS:    case OP_NCLASS:
5377    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5378    read_char(common);  
5379    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5380      bit = (common->utf && is_char7_bitset((const pcre_uint8 *)cc, type == OP_NCLASS)) ? 127 : 255;
5381      read_char_max(common, bit, type == OP_NCLASS);
5382    #else
5383      read_char_max(common, 255, type == OP_NCLASS);
5384    #endif
5385    
5386      if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5387      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5388    
5389  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5390    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. */  
5391    if (common->utf)    if (common->utf)
 #endif /* COMPILE_PCRE8 */  
5392      {      {
5393      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, bit);
5394      if (type == OP_CLASS)      if (type == OP_CLASS)
5395        {        {
5396        add_jump(compiler, backtracks, jump[0]);        add_jump(compiler, backtracks, jump[0]);
5397        jump[0] = NULL;        jump[0] = NULL;
5398        }        }
5399      }      }
5400  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #elif !defined COMPILE_PCRE8
5401      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
5402      if (type == OP_CLASS)
5403        {
5404        add_jump(compiler, backtracks, jump[0]);
5405        jump[0] = NULL;
5406        }
5407    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
5408    
5409    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5410    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5411    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5412    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5413    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);
5414    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
5415    
5416  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5417    if (jump[0] != NULL)    if (jump[0] != NULL)
5418      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5419  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif
5420    
5421    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5422    
5423  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 9030  switch(re->options & PCRE_NEWLINE_BITS) Line 9436  switch(re->options & PCRE_NEWLINE_BITS)
9436    case PCRE_NEWLINE_ANYCRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break;    case PCRE_NEWLINE_ANYCRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break;
9437    default: return;    default: return;
9438    }    }
9439    common->nlmax = READ_CHAR_ANY;
9440  if ((re->options & PCRE_BSR_ANYCRLF) != 0)  if ((re->options & PCRE_BSR_ANYCRLF) != 0)
9441    common->bsr_nltype = NLTYPE_ANYCRLF;    common->bsr_nltype = NLTYPE_ANYCRLF;
9442  else if ((re->options & PCRE_BSR_UNICODE) != 0)  else if ((re->options & PCRE_BSR_UNICODE) != 0)
# Line 9042  else Line 9449  else
9449    common->bsr_nltype = NLTYPE_ANY;    common->bsr_nltype = NLTYPE_ANY;
9450  #endif  #endif
9451    }    }
9452    common->bsr_nlmax = READ_CHAR_ANY;
9453  common->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;  common->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0;
9454  common->ctypes = (sljit_sw)(tables + ctypes_offset);  common->ctypes = (sljit_sw)(tables + ctypes_offset);
 common->digits[0] = -2;  
9455  common->name_table = ((pcre_uchar *)re) + re->name_table_offset;  common->name_table = ((pcre_uchar *)re) + re->name_table_offset;
9456  common->name_count = re->name_count;  common->name_count = re->name_count;
9457  common->name_entry_size = re->name_entry_size;  common->name_entry_size = re->name_entry_size;
# Line 9055  common->utf = (re->options & PCRE_UTF8) Line 9462  common->utf = (re->options & PCRE_UTF8)
9462  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
9463  common->use_ucp = (re->options & PCRE_UCP) != 0;  common->use_ucp = (re->options & PCRE_UCP) != 0;
9464  #endif  #endif
9465    if (common->utf)
9466      {
9467      if (common->nltype == NLTYPE_ANY)
9468        common->nlmax = 0x2029;
9469      else if (common->nltype == NLTYPE_ANYCRLF)
9470        common->nlmax = (CHAR_CR > CHAR_NL) ? CHAR_CR : CHAR_NL;
9471      else
9472        {
9473        /* We only care about the first newline character. */
9474        common->nlmax = common->newline & 0xff;
9475        }
9476    
9477      if (common->bsr_nltype == NLTYPE_ANY)
9478        common->bsr_nlmax = 0x2029;
9479      else
9480        common->bsr_nlmax = (CHAR_CR > CHAR_NL) ? CHAR_CR : CHAR_NL;
9481      }
9482  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
9483  ccend = bracketend(rootbacktrack.cc);  ccend = bracketend(rootbacktrack.cc);
9484    
# Line 9218  if ((re->options & PCRE_ANCHORED) == 0) Line 9642  if ((re->options & PCRE_ANCHORED) == 0)
9642      else if ((re->flags & PCRE_STARTLINE) != 0)      else if ((re->flags & PCRE_STARTLINE) != 0)
9643        fast_forward_newline(common, (re->options & PCRE_FIRSTLINE) != 0);        fast_forward_newline(common, (re->options & PCRE_FIRSTLINE) != 0);
9644      else if ((re->flags & PCRE_STARTLINE) == 0 && study != NULL && (study->flags & PCRE_STUDY_MAPPED) != 0)      else if ((re->flags & PCRE_STARTLINE) == 0 && study != NULL && (study->flags & PCRE_STUDY_MAPPED) != 0)
9645        fast_forward_start_bits(common, (sljit_uw)study->start_bits, (re->options & PCRE_FIRSTLINE) != 0);        fast_forward_start_bits(common, study->start_bits, (re->options & PCRE_FIRSTLINE) != 0);
9646      }      }
9647    }    }
9648  else  else
# Line 9456  if (common->reset_match != NULL) Line 9880  if (common->reset_match != NULL)
9880    JUMPTO(SLJIT_JUMP, reset_match_label);    JUMPTO(SLJIT_JUMP, reset_match_label);
9881    }    }
9882  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
9883  #ifndef COMPILE_PCRE32  #ifdef COMPILE_PCRE8
9884  if (common->utfreadchar != NULL)  if (common->utfreadchar != NULL)
9885    {    {
9886    set_jumps(common->utfreadchar, LABEL());    set_jumps(common->utfreadchar, LABEL());
9887    do_utfreadchar(common);    do_utfreadchar(common);
9888    }    }
9889  #endif /* !COMPILE_PCRE32 */  if (common->utfreadchar16 != NULL)
9890  #ifdef COMPILE_PCRE8    {
9891      set_jumps(common->utfreadchar16, LABEL());
9892      do_utfreadchar16(common);
9893      }
9894  if (common->utfreadtype8 != NULL)  if (common->utfreadtype8 != NULL)
9895    {    {
9896    set_jumps(common->utfreadtype8, LABEL());    set_jumps(common->utfreadtype8, LABEL());

Legend:
Removed from v.1402  
changed lines
  Added in v.1423

  ViewVC Help
Powered by ViewVC 1.1.5