regcomp.c

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

      add_opcode(reg, OP_CCLASS);

    r = add_bitset(reg, cc->bs);
  }
  else {
    if (ONIGENC_MBC_MINLEN(reg->enc) > 1 || bitset_is_empty(cc->bs)) {
      if (IS_CCLASS_NOT(cc))
        add_opcode(reg, OP_CCLASS_MB_NOT);
      else
        add_opcode(reg, OP_CCLASS_MB);

      r = add_multi_byte_cclass(cc->mbuf, reg);
    }
    else {
      if (IS_CCLASS_NOT(cc))
        add_opcode(reg, OP_CCLASS_MIX_NOT);
      else
        add_opcode(reg, OP_CCLASS_MIX);

      r = add_bitset(reg, cc->bs);
      if (r) return r;
      r = add_multi_byte_cclass(cc->mbuf, reg);
    }
  }

  return r;
}

static int
entry_repeat_range(regex_t* reg, int id, int lower, int upper)
{
#define REPEAT_RANGE_ALLOC  4

  OnigRepeatRange* p;

  if (reg->repeat_range_alloc == 0) {
    p = (OnigRepeatRange* )xmalloc(sizeof(OnigRepeatRange) * REPEAT_RANGE_ALLOC);
    CHECK_NULL_RETURN_VAL(p, ONIGERR_MEMORY);
    reg->repeat_range = p;
    reg->repeat_range_alloc = REPEAT_RANGE_ALLOC;
  }
  else if (reg->repeat_range_alloc <= id) {
    int n;
    n = reg->repeat_range_alloc + REPEAT_RANGE_ALLOC;
    p = (OnigRepeatRange* )xrealloc(reg->repeat_range,
                                    sizeof(OnigRepeatRange) * n);
    CHECK_NULL_RETURN_VAL(p, ONIGERR_MEMORY);
    reg->repeat_range = p;
    reg->repeat_range_alloc = n;
  }
  else {
    p = reg->repeat_range;
  }

  p[id].lower = lower;
  p[id].upper = upper;
  return 0;
}

static int
compile_range_repeat_node(QualifierNode* qn, int target_len, int empty_info,
                          regex_t* reg)
{
  int r;
  int num_repeat = reg->num_repeat;

  r = add_opcode(reg, qn->greedy ? OP_REPEAT : OP_REPEAT_NG);
  if (r) return r;
  r = add_mem_num(reg, num_repeat); /* OP_REPEAT ID */
  reg->num_repeat++;
  if (r) return r;
  r = add_rel_addr(reg, target_len + SIZE_OP_REPEAT_INC);
  if (r) return r;

  r = entry_repeat_range(reg, num_repeat, qn->lower, qn->upper);
  if (r) return r;

  r = compile_tree_empty_check(qn->target, reg, empty_info);
  if (r) return r;

  if (
#ifdef USE_SUBEXP_CALL
      reg->num_call > 0 ||
#endif
      IS_QUALIFIER_IN_REPEAT(qn)) {
    r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC_SG : OP_REPEAT_INC_NG_SG);
  }
  else {
    r = add_opcode(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG);
  }
  if (r) return r;
  r = add_mem_num(reg, num_repeat); /* OP_REPEAT ID */
  return r;
}

#define QUALIFIER_EXPAND_LIMIT_SIZE   50

static int
compile_length_qualifier_node(QualifierNode* qn, regex_t* reg)
{
  int len, mod_tlen;
  int infinite = IS_REPEAT_INFINITE(qn->upper);
  int empty_info = qn->target_empty_info;
  int tlen = compile_length_tree(qn->target, reg);

  if (tlen < 0) return tlen;

  /* anychar repeat */
  if (NTYPE(qn->target) == N_ANYCHAR) {
    if (qn->greedy && infinite) {
      if (IS_NOT_NULL(qn->next_head_exact))
        return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower;
      else
        return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower;
    }
  }

  if (empty_info != 0)
    mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);
  else
    mod_tlen = tlen;

  if (infinite &&
      (qn->lower <= 1 || tlen * qn->lower <= QUALIFIER_EXPAND_LIMIT_SIZE)) {
    if (qn->lower == 1 && tlen > QUALIFIER_EXPAND_LIMIT_SIZE) {
      len = SIZE_OP_JUMP;
    }
    else {
      len = tlen * qn->lower;
    }

    if (qn->greedy) {
      if (IS_NOT_NULL(qn->head_exact))
	len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP;
      else if (IS_NOT_NULL(qn->next_head_exact))
	len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP;
      else
	len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP;
    }
    else
      len += SIZE_OP_JUMP + mod_tlen + SIZE_OP_PUSH;
  }
  else if (qn->upper == 0 && qn->is_refered != 0) { /* /(?<n>..){0}/ */
    len = SIZE_OP_JUMP + tlen;
  }
  else if (!infinite && qn->greedy &&
           (qn->upper == 1 || (tlen + SIZE_OP_PUSH) * qn->upper
                                      <= QUALIFIER_EXPAND_LIMIT_SIZE)) {
    len = tlen * qn->lower;
    len += (SIZE_OP_PUSH + tlen) * (qn->upper - qn->lower);
  }
  else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */
    len = SIZE_OP_PUSH + SIZE_OP_JUMP + tlen;
  }
  else {
    len = SIZE_OP_REPEAT_INC
        + mod_tlen + SIZE_OPCODE + SIZE_RELADDR + SIZE_MEMNUM;
  }

  return len;
}

static int
is_anychar_star_qualifier(QualifierNode* qn)
{
  if (qn->greedy && IS_REPEAT_INFINITE(qn->upper) &&
      NTYPE(qn->target) == N_ANYCHAR)
    return 1;
  else
    return 0;
}

static int
compile_qualifier_node(QualifierNode* qn, regex_t* reg)
{
  int i, r, mod_tlen;
  int infinite = IS_REPEAT_INFINITE(qn->upper);
  int empty_info = qn->target_empty_info;
  int tlen = compile_length_tree(qn->target, reg);

  if (tlen < 0) return tlen;

  if (is_anychar_star_qualifier(qn)) {
    r = compile_tree_n_times(qn->target, qn->lower, reg);
    if (r) return r;
    if (IS_NOT_NULL(qn->next_head_exact)) {
      if (IS_MULTILINE(reg->options))
	r = add_opcode(reg, OP_ANYCHAR_ML_STAR_PEEK_NEXT);
      else
	r = add_opcode(reg, OP_ANYCHAR_STAR_PEEK_NEXT);
      if (r) return r;
      return add_bytes(reg, NSTRING(qn->next_head_exact).s, 1);
    }
    else {
      if (IS_MULTILINE(reg->options))
	return add_opcode(reg, OP_ANYCHAR_ML_STAR);
      else
	return add_opcode(reg, OP_ANYCHAR_STAR);
    }
  }

  if (empty_info != 0)
    mod_tlen = tlen + (SIZE_OP_NULL_CHECK_START + SIZE_OP_NULL_CHECK_END);
  else
    mod_tlen = tlen;

  if (infinite &&
      (qn->lower <= 1 || tlen * qn->lower <= QUALIFIER_EXPAND_LIMIT_SIZE)) {
    if (qn->lower == 1 && tlen > QUALIFIER_EXPAND_LIMIT_SIZE) {
      if (qn->greedy) {
	if (IS_NOT_NULL(qn->head_exact))
	  r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_OR_JUMP_EXACT1);
	else if (IS_NOT_NULL(qn->next_head_exact))
	  r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH_IF_PEEK_NEXT);
	else
	  r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_PUSH);
      }
      else {
	r = add_opcode_rel_addr(reg, OP_JUMP, SIZE_OP_JUMP);
      }
      if (r) return r;
    }
    else {
      r = compile_tree_n_times(qn->target, qn->lower, reg);
      if (r) return r;
    }

    if (qn->greedy) {
      if (IS_NOT_NULL(qn->head_exact)) {
	r = add_opcode_rel_addr(reg, OP_PUSH_OR_JUMP_EXACT1,
			     mod_tlen + SIZE_OP_JUMP);
	if (r) return r;
	add_bytes(reg, NSTRING(qn->head_exact).s, 1);
	r = compile_tree_empty_check(qn->target, reg, empty_info);
	if (r) return r;
	r = add_opcode_rel_addr(reg, OP_JUMP,
	-(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1));
      }
      else if (IS_NOT_NULL(qn->next_head_exact)) {
	r = add_opcode_rel_addr(reg, OP_PUSH_IF_PEEK_NEXT,
				mod_tlen + SIZE_OP_JUMP);
	if (r) return r;
	add_bytes(reg, NSTRING(qn->next_head_exact).s, 1);
	r = compile_tree_empty_check(qn->target, reg, empty_info);
	if (r) return r;
	r = add_opcode_rel_addr(reg, OP_JUMP,
          -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH_IF_PEEK_NEXT));
      }
      else {
	r = add_opcode_rel_addr(reg, OP_PUSH, mod_tlen + SIZE_OP_JUMP);
	if (r) return r;
	r = compile_tree_empty_check(qn->target, reg, empty_info);
	if (r) return r;
	r = add_opcode_rel_addr(reg, OP_JUMP,
		     -(mod_tlen + (int )SIZE_OP_JUMP + (int )SIZE_OP_PUSH));
      }
    }
    else {
      r = add_opcode_rel_addr(reg, OP_JUMP, mod_tlen);
      if (r) return r;
      r = compile_tree_empty_check(qn->target, reg, empty_info);
      if (r) return r;
      r = add_opcode_rel_addr(reg, OP_PUSH, -(mod_tlen + (int )SIZE_OP_PUSH));
    }
  }
  else if (qn->upper == 0 && qn->is_refered != 0) { /* /(?<n>..){0}/ */
    r = add_opcode_rel_addr(reg, OP_JUMP, tlen);
    if (r) return r;
    r = compile_tree(qn->target, reg);
  }
  else if (!infinite && qn->greedy &&
           (qn->upper == 1 || (tlen + SIZE_OP_PUSH) * qn->upper
                                  <= QUALIFIER_EXPAND_LIMIT_SIZE)) {
    int n = qn->upper - qn->lower;

    r = compile_tree_n_times(qn->target, qn->lower, reg);
    if (r) return r;

    for (i = 0; i < n; i++) {
      r = add_opcode_rel_addr(reg, OP_PUSH,
			   (n - i) * tlen + (n - i - 1) * SIZE_OP_PUSH);
      if (r) return r;
      r = compile_tree(qn->target, reg);
      if (r) return r;
    }
  }
  else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */
    r = add_opcode_rel_addr(reg, OP_PUSH, SIZE_OP_JUMP);
    if (r) return r;
    r = add_opcode_rel_addr(reg, OP_JUMP, tlen);
    if (r) return r;
    r = compile_tree(qn->target, reg);
  }
  else {
    r = compile_range_repeat_node(qn, mod_tlen, empty_info, reg);
  }
  return r;
}

static int
compile_length_option_node(EffectNode* node, regex_t* reg)
{
  int tlen;
  OnigOptionType prev = reg->options;

  reg->options = node->option;
  tlen = compile_length_tree(node->target, reg);
  reg->options = prev;

  if (tlen < 0) return tlen;

  if (IS_DYNAMIC_OPTION(prev ^ node->option)) {
    return SIZE_OP_SET_OPTION_PUSH + SIZE_OP_SET_OPTION + SIZE_OP_FAIL
           + tlen + SIZE_OP_SET_OPTION;
  }
  else
    return tlen;
}

static int
compile_option_node(EffectNode* node, regex_t* reg)
{
  int r;
  OnigOptionType prev = reg->options;

  if (IS_DYNAMIC_OPTION(prev ^ node->option)) {
    r = add_opcode_option(reg, OP_SET_OPTION_PUSH, node->option);
    if (r) return r;
    r = add_opcode_option(reg, OP_SET_OPTION, prev);
    if (r) return r;
    r = add_opcode(reg, OP_FAIL);
    if (r) return r;
  }

  reg->options = node->option;
  r = compile_tree(node->target, reg);
  reg->options = prev;

  if (IS_DYNAMIC_OPTION(prev ^ node->option)) {
    if (r) return r;
    r = add_opcode_option(reg, OP_SET_OPTION, prev);
  }
  return r;
}

static int
compile_length_effect_node(EffectNode* node, regex_t* reg)
{
  int len;
  int tlen;

  if (node->type == EFFECT_OPTION)
    return compile_length_option_node(node, reg);

  if (node->target) {
    tlen = compile_length_tree(node->target, reg);
    if (tlen < 0) return tlen;
  }
  else
    tlen = 0;

  switch (node->type) {
  case EFFECT_MEMORY:
#ifdef USE_SUBEXP_CALL
    if (IS_EFFECT_CALLED(node)) {
      len = SIZE_OP_MEMORY_START_PUSH + tlen
	  + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;
      if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))
	len += (IS_EFFECT_RECURSION(node)
		? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);
      else
	len += (IS_EFFECT_RECURSION(node)
		? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);
    }
    else
#endif
    {
      if (BIT_STATUS_AT(reg->bt_mem_start, node->regnum))
	len = SIZE_OP_MEMORY_START_PUSH;
      else
	len = SIZE_OP_MEMORY_START;

      len += tlen + (BIT_STATUS_AT(reg->bt_mem_end, node->regnum)
		     ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END);
    }
    break;

  case EFFECT_STOP_BACKTRACK:
    if (IS_EFFECT_STOP_BT_SIMPLE_REPEAT(node)) {
      QualifierNode* qn = &NQUALIFIER(node->target);
      tlen = compile_length_tree(qn->target, reg);
      if (tlen < 0) return tlen;

      len = tlen * qn->lower
	  + SIZE_OP_PUSH + tlen + SIZE_OP_POP + SIZE_OP_JUMP;
    }
    else {
      len = SIZE_OP_PUSH_STOP_BT + tlen + SIZE_OP_POP_STOP_BT;
    }
    break;

  default:
    return ONIGERR_TYPE_BUG;
    break;
  }

  return len;
}

static int get_char_length_tree(Node* node, regex_t* reg, int* len);

static int
compile_effect_node(EffectNode* node, regex_t* reg)
{
  int r, len;

  if (node->type == EFFECT_OPTION)
    return compile_option_node(node, reg);

  switch (node->type) {
  case EFFECT_MEMORY:
#ifdef USE_SUBEXP_CALL
    if (IS_EFFECT_CALLED(node)) {
      r = add_opcode(reg, OP_CALL);
      if (r) return r;
      node->call_addr = BBUF_GET_OFFSET_POS(reg) + SIZE_ABSADDR + SIZE_OP_JUMP;
      node->state |= NST_ADDR_FIXED;
      r = add_abs_addr(reg, (int )node->call_addr);
      if (r) return r;
      len = compile_length_tree(node->target, reg);
      len += (SIZE_OP_MEMORY_START_PUSH + SIZE_OP_RETURN);
      if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))
	len += (IS_EFFECT_RECURSION(node)
		? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);
      else
	len += (IS_EFFECT_RECURSION(node)
		? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);

      r = add_opcode_rel_addr(reg, OP_JUMP, len);
      if (r) return r;
    }
#endif
    if (BIT_STATUS_AT(reg->bt_mem_start, node->regnum))
      r = add_opcode(reg, OP_MEMORY_START_PUSH);
    else
      r = add_opcode(reg, OP_MEMORY_START);
    if (r) return r;
    r = add_mem_num(reg, node->regnum);
    if (r) return r;
    r = compile_tree(node->target, reg);
    if (r) return r;
#ifdef USE_SUBEXP_CALL
    if (IS_EFFECT_CALLED(node)) {
      if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))
	r = add_opcode(reg, (IS_EFFECT_RECURSION(node)
			     ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH));
      else
	r = add_opcode(reg, (IS_EFFECT_RECURSION(node)
			     ? OP_MEMORY_END_REC : OP_MEMORY_END));

      if (r) return r;
      r = add_mem_num(reg, node->regnum);
      if (r) return r;
      r = add_opcode(reg, OP_RETURN);
    }
    else
#endif
    {
      if (BIT_STATUS_AT(reg->bt_mem_end, node->regnum))
	r = add_opcode(reg, OP_MEMORY_END_PUSH);
      else
	r = add_opcode(reg, OP_MEMORY_END);
      if (r) return r;
      r = add_mem_num(reg, node->regnum);
    }
    break;

  case EFFECT_STOP_BACKTRACK:
    if (IS_EFFECT_STOP_BT_SIMPLE_REPEAT(node)) {
      QualifierNode* qn = &NQUALIFIER(node->target);
      r = compile_tree_n_times(qn->target, qn->lower, reg);
      if (r) return r;

      len = compile_length_tree(qn->target, reg);
      if (len < 0) return len;

      r = add_opcode_rel_addr(reg, OP_PUSH, len + SIZE_OP_POP + SIZE_OP_JUMP);
      if (r) return r;
      r = compile_tree(qn->target, reg);
      if (r) return r;
      r = add_opcode(reg, OP_POP);
      if (r) return r;
      r = add_opcode_rel_addr(reg, OP_JUMP,
	 -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP + (int )SIZE_OP_JUMP));
    }
    else {
      r = add_opcode(reg, OP_PUSH_STOP_BT);
      if (r) return r;
      r = compile_tree(node->target, reg);
      if (r) return r;
      r = add_opcode(reg, OP_POP_STOP_BT);
    }
    break;

  default:
    return ONIGERR_TYPE_BUG;
    break;
  }

  return r;
}

static int
compile_length_anchor_node(AnchorNode* node, regex_t* reg)
{
  int len;
  int tlen = 0;

  if (node->target) {
    tlen = compile_length_tree(node->target, reg);
    if (tlen < 0) return tlen;
  }

  switch (node->type) {
  case ANCHOR_PREC_READ:
    len = SIZE_OP_PUSH_POS + tlen + SIZE_OP_POP_POS;
    break;
  case ANCHOR_PREC_READ_NOT:
    len = SIZE_OP_PUSH_POS_NOT + tlen + SIZE_OP_FAIL_POS;
    break;
  case ANCHOR_LOOK_BEHIND:
    len = SIZE_OP_LOOK_BEHIND + tlen;
    break;
  case ANCHOR_LOOK_BEHIND_NOT:
    len = SIZE_OP_PUSH_LOOK_BEHIND_NOT + tlen + SIZE_OP_FAIL_LOOK_BEHIND_NOT;
    break;

  default:
    len = SIZE_OPCODE;
    break;
  }

  return len;
}

static int
compile_anchor_node(AnchorNode* node, regex_t* reg)
{
  int r, len;

  switch (node->type) {
  case ANCHOR_BEGIN_BUF:      r = add_opcode(reg, OP_BEGIN_BUF);      break;
  case ANCHOR_END_BUF:        r = add_opcode(reg, OP_END_BUF);        break;
  case ANCHOR_BEGIN_LINE:     r = add_opcode(reg, OP_BEGIN_LINE);     break;
  case ANCHOR_END_LINE:       r = add_opcode(reg, OP_END_LINE);       break;
  case ANCHOR_SEMI_END_BUF:   r = add_opcode(reg, OP_SEMI_END_BUF);   break;
  case ANCHOR_BEGIN_POSITION: r = add_opcode(reg, OP_BEGIN_POSITION); break;

  case ANCHOR_WORD_BOUND:     r = add_opcode(reg, OP_WORD_BOUND);     break;
  case ANCHOR_NOT_WORD_BOUND: r = add_opcode(reg, OP_NOT_WORD_BOUND); break;
#ifdef USE_WORD_BEGIN_END
  case ANCHOR_WORD_BEGIN:     r = add_opcode(reg, OP_WORD_BEGIN);     break;
  case ANCHOR_WORD_END:       r = add_opcode(reg, OP_WORD_END);       break;
#endif

  case ANCHOR_PREC_READ:
    r = add_opcode(reg, OP_PUSH_POS);
    if (r) return r;
    r = compile_tree(node->target, reg);
    if (r) return r;
    r = add_opcode(reg, OP_POP_POS);
    break;

  case ANCHOR_PREC_READ_NOT:
    len = compile_length_tree(node->target, reg);
    if (len < 0) return len;
    r = add_opcode_rel_addr(reg, OP_PUSH_POS_NOT, len + SIZE_OP_FAIL_POS);
    if (r) return r;
    r = compile_tree(node->target, reg);
    if (r) return r;
    r = add_opcode(reg, OP_FAIL_POS);
    break;

  case ANCHOR_LOOK_BEHIND:
    {
      int n;
      r = add_opcode(reg, OP_LOOK_BEHIND);
      if (r) return r;
      if (node->char_len < 0) {
	r = get_char_length_tree(node->target, reg, &n);
	if (r) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;