regexec.c

PHP v6.0 For Linux 运行环境：Win9X/ WinME/ WinNT/ Win2K/ WinXP

  int level = 0;\
  k = stk;\
  while (k > stk_base) {\
    k--;\
    if ((k->type == STK_MEM_END_MARK || k->type == STK_MEM_END) \
      && k->u.mem.num == (mnum)) {\
      level++;\
    }\
    else if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
      if (level == 0) break;\
      level--;\
    }\
  }\
} while (0)

#define STACK_GET_MEM_RANGE(k, mnum, start, end) do {\
  int level = 0;\
  while (k < stk) {\
    if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
      if (level == 0) (start) = k->u.mem.pstr;\
      level++;\
    }\
    else if (k->type == STK_MEM_END && k->u.mem.num == (mnum)) {\
      level--;\
      if (level == 0) {\
        (end) = k->u.mem.pstr;\
        break;\
      }\
    }\
    k++;\
  }\
} while (0)

#define STACK_PUSH_NULL_CHECK_START(cnum, s) do {\
  STACK_ENSURE(1);\
  stk->type = STK_NULL_CHECK_START;\
  stk->u.null_check.num  = (cnum);\
  stk->u.null_check.pstr = (s);\
  STACK_INC;\
} while(0)

#define STACK_PUSH_NULL_CHECK_END(cnum) do {\
  STACK_ENSURE(1);\
  stk->type = STK_NULL_CHECK_END;\
  stk->u.null_check.num  = (cnum);\
  STACK_INC;\
} while(0)

#define STACK_PUSH_CALL_FRAME(pat) do {\
  STACK_ENSURE(1);\
  stk->type = STK_CALL_FRAME;\
  stk->u.call_frame.ret_addr = (pat);\
  STACK_INC;\
} while(0)

#define STACK_PUSH_RETURN do {\
  STACK_ENSURE(1);\
  stk->type = STK_RETURN;\
  STACK_INC;\
} while(0)


#ifdef ONIG_DEBUG
#define STACK_BASE_CHECK(p) \
  if ((p) < stk_base)  goto stack_error;
#else
#define STACK_BASE_CHECK(p)
#endif

#define STACK_POP_ONE do {\
  stk--;\
  STACK_BASE_CHECK(stk); \
} while(0)

#define STACK_POP  do {\
  switch (pop_level) {\
  case STACK_POP_LEVEL_FREE:\
    while (1) {\
      stk--;\
      STACK_BASE_CHECK(stk); \
      if ((stk->type & STK_MASK_POP_USED) != 0)  break;\
    }\
    break;\
  case STACK_POP_LEVEL_MEM_START:\
    while (1) {\
      stk--;\
      STACK_BASE_CHECK(stk); \
      if ((stk->type & STK_MASK_POP_USED) != 0)  break;\
      else if (stk->type == STK_MEM_START) {\
        mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
        mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
      }\
    }\
    break;\
  default:\
    while (1) {\
      stk--;\
      STACK_BASE_CHECK(stk); \
      if ((stk->type & STK_MASK_POP_USED) != 0)  break;\
      else if (stk->type == STK_MEM_START) {\
        mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
        mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
      }\
      else if (stk->type == STK_REPEAT_INC) {\
        STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
      }\
      else if (stk->type == STK_MEM_END) {\
        mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
        mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
      }\
    }\
    break;\
  }\
} while(0)

#define STACK_POP_TIL_POS_NOT  do {\
  while (1) {\
    stk--;\
    STACK_BASE_CHECK(stk); \
    if (stk->type == STK_POS_NOT) break;\
    else if (stk->type == STK_MEM_START) {\
      mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
      mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
    }\
    else if (stk->type == STK_REPEAT_INC) {\
      STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
    }\
    else if (stk->type == STK_MEM_END) {\
      mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
      mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
    }\
  }\
} while(0)

#define STACK_POP_TIL_LOOK_BEHIND_NOT  do {\
  while (1) {\
    stk--;\
    STACK_BASE_CHECK(stk); \
    if (stk->type == STK_LOOK_BEHIND_NOT) break;\
    else if (stk->type == STK_MEM_START) {\
      mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
      mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
    }\
    else if (stk->type == STK_REPEAT_INC) {\
      STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
    }\
    else if (stk->type == STK_MEM_END) {\
      mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
      mem_end_stk[stk->u.mem.num]   = stk->u.mem.end;\
    }\
  }\
} while(0)

#define STACK_POS_END(k) do {\
  k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (IS_TO_VOID_TARGET(k)) {\
      k->type = STK_VOID;\
    }\
    else if (k->type == STK_POS) {\
      k->type = STK_VOID;\
      break;\
    }\
  }\
} while(0)

#define STACK_STOP_BT_END do {\
  StackType *k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (IS_TO_VOID_TARGET(k)) {\
      k->type = STK_VOID;\
    }\
    else if (k->type == STK_STOP_BT) {\
      k->type = STK_VOID;\
      break;\
    }\
  }\
} while(0)

#define STACK_NULL_CHECK(isnull,id,s) do {\
  StackType* k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_NULL_CHECK_START) {\
      if (k->u.null_check.num == (id)) {\
        (isnull) = (k->u.null_check.pstr == (s));\
        break;\
      }\
    }\
  }\
} while(0)

#define STACK_NULL_CHECK_REC(isnull,id,s) do {\
  int level = 0;\
  StackType* k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_NULL_CHECK_START) {\
      if (k->u.null_check.num == (id)) {\
        if (level == 0) {\
          (isnull) = (k->u.null_check.pstr == (s));\
          break;\
        }\
        else level--;\
      }\
    }\
    else if (k->type == STK_NULL_CHECK_END) {\
      level++;\
    }\
  }\
} while(0)

#define STACK_NULL_CHECK_MEMST(isnull,id,s,reg) do {\
  StackType* k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_NULL_CHECK_START) {\
      if (k->u.null_check.num == (id)) {\
        if (k->u.null_check.pstr != (s)) {\
          (isnull) = 0;\
          break;\
        }\
        else {\
          UChar* endp;\
          (isnull) = 1;\
          while (k < stk) {\
            if (k->type == STK_MEM_START) {\
              if (k->u.mem.end == INVALID_STACK_INDEX) {\
                (isnull) = 0; break;\
              }\
              if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
                endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
              else\
                endp = (UChar* )k->u.mem.end;\
              if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
                (isnull) = 0; break;\
              }\
              else if (endp != s) {\
                (isnull) = -1; /* empty, but position changed */ \
              }\
            }\
            k++;\
          }\
  	  break;\
        }\
      }\
    }\
  }\
} while(0)

#define STACK_NULL_CHECK_MEMST_REC(isnull,id,s,reg) do {\
  int level = 0;\
  StackType* k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_NULL_CHECK_START) {\
      if (k->u.null_check.num == (id)) {\
        if (level == 0) {\
          if (k->u.null_check.pstr != (s)) {\
            (isnull) = 0;\
            break;\
          }\
          else {\
            UChar* endp;\
            (isnull) = 1;\
            while (k < stk) {\
              if (k->type == STK_MEM_START) {\
                if (k->u.mem.end == INVALID_STACK_INDEX) {\
                  (isnull) = 0; break;\
                }\
                if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
                  endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
                else\
                  endp = (UChar* )k->u.mem.end;\
                if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
                  (isnull) = 0; break;\
                }\
                else if (endp != s) {\
                  (isnull) = -1; /* empty, but position changed */ \
                }\
              }\
              k++;\
            }\
  	    break;\
          }\
        }\
        else {\
          level--;\
        }\
      }\
    }\
    else if (k->type == STK_NULL_CHECK_END) {\
      if (k->u.null_check.num == (id)) level++;\
    }\
  }\
} while(0)

#define STACK_GET_REPEAT(id, k) do {\
  int level = 0;\
  k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_REPEAT) {\
      if (level == 0) {\
        if (k->u.repeat.num == (id)) {\
          break;\
        }\
      }\
    }\
    else if (k->type == STK_CALL_FRAME) level--;\
    else if (k->type == STK_RETURN)     level++;\
  }\
} while (0)

#define STACK_RETURN(addr)  do {\
  int level = 0;\
  StackType* k = stk;\
  while (1) {\
    k--;\
    STACK_BASE_CHECK(k); \
    if (k->type == STK_CALL_FRAME) {\
      if (level == 0) {\
        (addr) = k->u.call_frame.ret_addr;\
        break;\
      }\
      else level--;\
    }\
    else if (k->type == STK_RETURN)\
      level++;\
  }\
} while(0)


#define STRING_CMP(s1,s2,len) do {\
  while (len-- > 0) {\
    if (*s1++ != *s2++) goto fail;\
  }\
} while(0)

#define STRING_CMP_IC(ambig_flag,s1,ps2,len) do {\
  if (string_cmp_ic(encode, ambig_flag, s1, ps2, len) == 0) \
    goto fail; \
} while(0)

static int string_cmp_ic(OnigEncoding enc, int ambig_flag,
			 UChar* s1, UChar** ps2, int mblen)
{
  UChar buf1[ONIGENC_MBC_NORMALIZE_MAXLEN];
  UChar buf2[ONIGENC_MBC_NORMALIZE_MAXLEN];
  UChar *p1, *p2, *end, *s2, *end2;
  int len1, len2;

  s2   = *ps2;
  end  = s1 + mblen;
  end2 = s2 + mblen;
  while (s1 < end) {
    len1 = ONIGENC_MBC_TO_NORMALIZE(enc, ambig_flag, &s1, end, buf1);
    len2 = ONIGENC_MBC_TO_NORMALIZE(enc, ambig_flag, &s2, end2, buf2);
    if (len1 != len2) return 0;
    p1 = buf1;
    p2 = buf2;
    while (len1-- > 0) {
      if (*p1 != *p2) return 0;
      p1++;
      p2++;
    }
  }

  *ps2 = s2;
  return 1;
}

#define STRING_CMP_VALUE(s1,s2,len,is_fail) do {\
  is_fail = 0;\
  while (len-- > 0) {\
    if (*s1++ != *s2++) {\
      is_fail = 1; break;\
    }\
  }\
} while(0)

#define STRING_CMP_VALUE_IC(ambig_flag,s1,ps2,len,is_fail) do {\
  if (string_cmp_ic(encode, ambig_flag, s1, ps2, len) == 0) \
    is_fail = 1; \
  else \
    is_fail = 0; \
} while(0)

#define ON_STR_BEGIN(s)  ((s) == str)
#define ON_STR_END(s)    ((s) == end)
#define IS_EMPTY_STR     (str == end)

#define DATA_ENSURE(n) \
  if (s + (n) > end) goto fail

#define DATA_ENSURE_CHECK(n)   (s + (n) <= end)

#ifdef USE_CAPTURE_HISTORY
static int
make_capture_history_tree(OnigCaptureTreeNode* node, StackType** kp,
                          StackType* stk_top, UChar* str, regex_t* reg)
{
  int n, r;
  OnigCaptureTreeNode* child;
  StackType* k = *kp;

  while (k < stk_top) {
    if (k->type == STK_MEM_START) {
      n = k->u.mem.num;
      if (n <= ONIG_MAX_CAPTURE_HISTORY_GROUP &&
          BIT_STATUS_AT(reg->capture_history, n) != 0) {
        child = history_node_new();
        CHECK_NULL_RETURN_VAL(child, ONIGERR_MEMORY);
        child->group = n;
        child->beg = (int )(k->u.mem.pstr - str);
        r = history_tree_add_child(node, child);
        if (r != 0) return r;
        *kp = (k + 1);
        r = make_capture_history_tree(child, kp, stk_top, str, reg);
        if (r != 0) return r;

        k = *kp;
        child->end = (int )(k->u.mem.pstr - str);
      }
    }
    else if (k->type == STK_MEM_END) {
      if (k->u.mem.num == node->group) {
        node->end = (int )(k->u.mem.pstr - str);
        *kp = k;
        return 0;
      }
    }
    k++;
  }

  return 1; /* 1: root node ending. */
}
#endif

#ifdef RUBY_PLATFORM

typedef struct {
  int state;
  regex_t*  reg;
  MatchArg* msa;
  StackType* stk_base;
} TrapEnsureArg;

static VALUE
trap_ensure(VALUE arg)
{
  TrapEnsureArg* ta = (TrapEnsureArg* )arg;

  if (ta->state == 0) { /* trap_exec() is not normal return */
    ONIG_STATE_DEC(ta->reg);
    if (! IS_NULL(ta->msa->stack_p) && ta->stk_base != ta->msa->stack_p)
      xfree(ta->stk_base);

    MATCH_ARG_FREE(*(ta->msa));
  }

  return Qnil;
}

static VALUE
trap_exec(VALUE arg)
{
  TrapEnsureArg* ta;

  rb_trap_exec();

  ta = (TrapEnsureArg* )arg;
  ta->state = 1; /* normal return */
  return Qnil;
}

extern void
onig_exec_trap(regex_t* reg, MatchArg* msa, StackType* stk_base)
{
  VALUE arg;
  TrapEnsureArg ta;

  ta.state    = 0;
  ta.reg      = reg;
  ta.msa      = msa;
  ta.stk_base = stk_base;
  arg = (VALUE )(&ta);
  rb_ensure(trap_exec, arg, trap_ensure, arg);
}

#define CHECK_INTERRUPT_IN_MATCH_AT do {\
  if (rb_trap_pending) {\
    if (! rb_prohibit_interrupt) {\
      onig_exec_trap(reg, msa, stk_base);\
    }\
  }\
} while (0)
#else
#define CHECK_INTERRUPT_IN_MATCH_AT
#endif /* RUBY_PLATFORM */

#ifdef ONIG_DEBUG_STATISTICS

#define USE_TIMEOFDAY

#ifdef USE_TIMEOFDAY
#ifdef HAVE_SYS_TIME_H
#include <sys/time.h>
#endif
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
static struct timeval ts, te;
#define GETTIME(t)        gettimeofday(&(t), (struct timezone* )0)
#define TIMEDIFF(te,ts)   (((te).tv_usec - (ts).tv_usec) + \
                           (((te).tv_sec - (ts).tv_sec)*1000000))
#else
#ifdef HAVE_SYS_TIMES_H
#include <sys/times.h>
#endif
static struct tms ts, te;
#define GETTIME(t)         times(&(t))
#define TIMEDIFF(te,ts)   ((te).tms_utime - (ts).tms_utime)
#endif

static int OpCounter[256];
static int OpPrevCounter[256];
static unsigned long OpTime[256];
static int OpCurr = OP_FINISH;
static int OpPrevTarget = OP_FAIL;
static int MaxStackDepth = 0;

#define STAT_OP_IN(opcode) do {\