regparse.c

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

  NQUALIFIER(node).is_refered        = 0;
  return node;
}

static Node*
node_new_effect(int type)
{
  Node* node = node_new();
  CHECK_NULL_RETURN(node);
  node->type = N_EFFECT;
  NEFFECT(node).type      = type;
  NEFFECT(node).state     =  0;
  NEFFECT(node).regnum    =  0;
  NEFFECT(node).option    =  0;
  NEFFECT(node).target    = NULL;
  NEFFECT(node).call_addr = -1;
  NEFFECT(node).opt_count =  0;
  return node;
}

extern Node*
onig_node_new_effect(int type)
{
  return node_new_effect(type);
}

static Node*
node_new_effect_memory(OnigOptionType option, int is_named)
{
  Node* node = node_new_effect(EFFECT_MEMORY);
  CHECK_NULL_RETURN(node);
  if (is_named != 0)
    SET_EFFECT_STATUS(node, NST_NAMED_GROUP);

#ifdef USE_SUBEXP_CALL
  NEFFECT(node).option = option;
#endif
  return node;
}

static Node*
node_new_option(OnigOptionType option)
{
  Node* node = node_new_effect(EFFECT_OPTION);
  CHECK_NULL_RETURN(node);
  NEFFECT(node).option = option;
  return node;
}

extern int
onig_node_str_cat(Node* node, const UChar* s, const UChar* end)
{
  int addlen = end - s;

  if (addlen > 0) {
    int len  = NSTRING(node).end - NSTRING(node).s;

    if (NSTRING(node).capa > 0 || (len + addlen > NODE_STR_BUF_SIZE - 1)) {
      UChar* p;
      int capa = len + addlen + NODE_STR_MARGIN;

      if (capa <= NSTRING(node).capa) {
	k_strcpy(NSTRING(node).s + len, s, end);
      }
      else {
	if (NSTRING(node).s == NSTRING(node).buf)
	  p = strcat_capa_from_static(NSTRING(node).s, NSTRING(node).end,
				      s, end, capa);
	else
	  p = k_strcat_capa(NSTRING(node).s, NSTRING(node).end, s, end, capa);

	CHECK_NULL_RETURN_VAL(p, ONIGERR_MEMORY);
	NSTRING(node).s    = p;
	NSTRING(node).capa = capa;
      }
    }
    else {
      k_strcpy(NSTRING(node).s + len, s, end);
    }
    NSTRING(node).end = NSTRING(node).s + len + addlen;
  }

  return 0;
}

static int
node_str_cat_char(Node* node, UChar c)
{
  UChar s[1];

  s[0] = c;
  return onig_node_str_cat(node, s, s + 1);
}

extern void
onig_node_conv_to_str_node(Node* node, int flag)
{
  node->type = N_STRING;

  NSTRING(node).flag = flag;
  NSTRING(node).capa = 0;
  NSTRING(node).s    = NSTRING(node).buf;
  NSTRING(node).end  = NSTRING(node).buf;
}

extern void
onig_node_str_clear(Node* node)
{
  if (NSTRING(node).capa != 0 &&
      IS_NOT_NULL(NSTRING(node).s) && NSTRING(node).s != NSTRING(node).buf) {
    xfree(NSTRING(node).s);
  }

  NSTRING(node).capa = 0;
  NSTRING(node).flag = 0;
  NSTRING(node).s    = NSTRING(node).buf;
  NSTRING(node).end  = NSTRING(node).buf;
}

static Node*
node_new_str(const UChar* s, const UChar* end)
{
  Node* node = node_new();
  CHECK_NULL_RETURN(node);

  node->type = N_STRING;
  NSTRING(node).capa = 0;
  NSTRING(node).flag = 0;
  NSTRING(node).s    = NSTRING(node).buf;
  NSTRING(node).end  = NSTRING(node).buf;
  if (onig_node_str_cat(node, s, end)) {
    onig_node_free(node);
    return NULL;
  }
  return node;
}

extern Node*
onig_node_new_str(const UChar* s, const UChar* end)
{
  return node_new_str(s, end);
}

static Node*
node_new_str_raw(UChar* s, UChar* end)
{
  Node* node = node_new_str(s, end);
  NSTRING_SET_RAW(node);
  return node;
}

static Node*
node_new_empty()
{
  return node_new_str(NULL, NULL);
}

static Node*
node_new_str_raw_char(UChar c)
{
  UChar p[1];

  p[0] = c;
  return node_new_str_raw(p, p + 1);
}

static Node*
str_node_split_last_char(StrNode* sn, OnigEncoding enc)
{
  const UChar *p;
  Node* n = NULL_NODE;

  if (sn->end > sn->s) {
    p = onigenc_get_prev_char_head(enc, sn->s, sn->end);
    if (p && p > sn->s) { /* can be splitted. */
      n = node_new_str(p, sn->end);
      if ((sn->flag & NSTR_RAW) != 0)
	NSTRING_SET_RAW(n);
      sn->end = (UChar* )p;
    }
  }
  return n;
}

static int
str_node_can_be_split(StrNode* sn, OnigEncoding enc)
{
  if (sn->end > sn->s) {
    return ((enc_len(enc, sn->s) < sn->end - sn->s)  ?  1 : 0);
  }
  return 0;
}

extern int
onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc)
{
  unsigned int num, val;
  OnigCodePoint c;
  UChar* p = *src;
  PFETCH_READY;

  num = 0;
  while (!PEND) {
    PFETCH(c);
    if (ONIGENC_IS_CODE_DIGIT(enc, c)) {
      val = (unsigned int )DIGITVAL(c);
      if ((INT_MAX_LIMIT - val) / 10UL < num)
	return -1;  /* overflow */

      num = num * 10 + val;
    }
    else {
      PUNFETCH;
      break;
    }
  }
  *src = p;
  return num;
}

static int
scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int maxlen,
				 OnigEncoding enc)
{
  OnigCodePoint c;
  unsigned int num, val;
  UChar* p = *src;
  PFETCH_READY;

  num = 0;
  while (!PEND && maxlen-- != 0) {
    PFETCH(c);
    if (ONIGENC_IS_CODE_XDIGIT(enc, c)) {
      val = (unsigned int )XDIGITVAL(enc,c);
      if ((INT_MAX_LIMIT - val) / 16UL < num)
	return -1;  /* overflow */

      num = (num << 4) + XDIGITVAL(enc,c);
    }
    else {
      PUNFETCH;
      break;
    }
  }
  *src = p;
  return num;
}

static int
scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen,
			   OnigEncoding enc)
{
  OnigCodePoint c;
  unsigned int num, val;
  UChar* p = *src;
  PFETCH_READY;

  num = 0;
  while (!PEND && maxlen-- != 0) {
    PFETCH(c);
    if (ONIGENC_IS_CODE_DIGIT(enc, c) && c < '8') {
      val = ODIGITVAL(c);
      if ((INT_MAX_LIMIT - val) / 8UL < num)
	return -1;  /* overflow */

      num = (num << 3) + val;
    }
    else {
      PUNFETCH;
      break;
    }
  }
  *src = p;
  return num;
}


#define BBUF_WRITE_CODE_POINT(bbuf,pos,code) \
    BBUF_WRITE(bbuf, pos, &(code), SIZE_CODE_POINT)

/* data format:
     [n][from-1][to-1][from-2][to-2] ... [from-n][to-n]
     (all data size is OnigCodePoint)
 */
static int
new_code_range(BBuf** pbuf)
{
#define INIT_MULTI_BYTE_RANGE_SIZE  (SIZE_CODE_POINT * 5)
  int r;
  OnigCodePoint n;
  BBuf* bbuf;

  bbuf = *pbuf = (BBuf* )xmalloc(sizeof(BBuf));
  CHECK_NULL_RETURN_VAL(*pbuf, ONIGERR_MEMORY);
  r = BBUF_INIT(*pbuf, INIT_MULTI_BYTE_RANGE_SIZE);
  if (r) return r;

  n = 0;
  BBUF_WRITE_CODE_POINT(bbuf, 0, n);
  return 0;
}

static int
add_code_range_to_buf(BBuf** pbuf, OnigCodePoint from, OnigCodePoint to)
{
  int r, inc_n, pos;
  int low, high, bound, x;
  OnigCodePoint n, *data;
  BBuf* bbuf;

  if (from > to) {
    n = from; from = to; to = n;
  }

  if (IS_NULL(*pbuf)) {
    r = new_code_range(pbuf);
    if (r) return r;
    bbuf = *pbuf;
    n = 0;
  }
  else {
    bbuf = *pbuf;
    GET_CODE_POINT(n, bbuf->p);
  }
  data = (OnigCodePoint* )(bbuf->p);
  data++;

  for (low = 0, bound = n; low < bound; ) {
    x = (low + bound) >> 1;
    if (from > data[x*2 + 1])
      low = x + 1;
    else
      bound = x;
  }

  for (high = low, bound = n; high < bound; ) {
    x = (high + bound) >> 1;
    if (to >= data[x*2] - 1)
      high = x + 1;
    else
      bound = x;
  }

  inc_n = low + 1 - high;
  if (n + inc_n > ONIG_MAX_MULTI_BYTE_RANGES_NUM)
    return ONIGERR_TOO_MANY_MULTI_BYTE_RANGES;

  if (inc_n != 1) {
    if (from > data[low*2])
      from = data[low*2];
    if (to < data[(high - 1)*2 + 1])
      to = data[(high - 1)*2 + 1];
  }

  if (inc_n != 0 && (OnigCodePoint )high < n) {
    int from_pos = SIZE_CODE_POINT * (1 + high * 2);
    int to_pos   = SIZE_CODE_POINT * (1 + (low + 1) * 2);
    int size = (n - high) * 2 * SIZE_CODE_POINT;

    if (inc_n > 0) {
      BBUF_MOVE_RIGHT(bbuf, from_pos, to_pos, size);
    }
    else {
      BBUF_MOVE_LEFT_REDUCE(bbuf, from_pos, to_pos);
    }
  }

  pos = SIZE_CODE_POINT * (1 + low * 2);
  BBUF_ENSURE_SIZE(bbuf, pos + SIZE_CODE_POINT * 2);
  BBUF_WRITE_CODE_POINT(bbuf, pos, from);
  BBUF_WRITE_CODE_POINT(bbuf, pos + SIZE_CODE_POINT, to);
  n += inc_n;
  BBUF_WRITE_CODE_POINT(bbuf, 0, n);

  return 0;
}

static int
add_code_range(BBuf** pbuf, ScanEnv* env, OnigCodePoint from, OnigCodePoint to)
{
  if (from > to) {
    if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC))
      return 0;
    else
      return ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS;
  }

  return add_code_range_to_buf(pbuf, from, to);
}

static int
not_code_range_buf(OnigEncoding enc, BBuf* bbuf, BBuf** pbuf)
{
  int r, i, n;
  OnigCodePoint pre, from, *data, to = 0;

  *pbuf = (BBuf* )NULL;
  if (IS_NULL(bbuf)) {
  set_all:
    return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
  }

  data = (OnigCodePoint* )(bbuf->p);
  GET_CODE_POINT(n, data);
  data++;
  if (n <= 0) goto set_all;

  r = 0;
  pre = MBCODE_START_POS(enc);
  for (i = 0; i < n; i++) {
    from = data[i*2];
    to   = data[i*2+1];
    if (pre <= from - 1) {
      r = add_code_range_to_buf(pbuf, pre, from - 1);
      if (r != 0) return r;
    }
    if (to == ~((OnigCodePoint )0)) break;
    pre = to + 1;
  }
  if (to < ~((OnigCodePoint )0)) {
    r = add_code_range_to_buf(pbuf, to + 1, ~((OnigCodePoint )0));
  }
  return r;
}

#define SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2) do {\
  BBuf *tbuf; \
  int  tnot; \
  tnot = not1;  not1  = not2;  not2  = tnot; \
  tbuf = bbuf1; bbuf1 = bbuf2; bbuf2 = tbuf; \
} while (0)

static int
or_code_range_buf(OnigEncoding enc, BBuf* bbuf1, int not1,
                  BBuf* bbuf2, int not2, BBuf** pbuf)
{
  int r;
  OnigCodePoint i, n1, *data1;
  OnigCodePoint from, to;

  *pbuf = (BBuf* )NULL;
  if (IS_NULL(bbuf1) && IS_NULL(bbuf2)) {
    if (not1 != 0 || not2 != 0)
      return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
    return 0;
  }

  r = 0;
  if (IS_NULL(bbuf2))
    SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);

  if (IS_NULL(bbuf1)) {
    if (not1 != 0) {
      return SET_ALL_MULTI_BYTE_RANGE(enc, pbuf);
    }
    else {
      if (not2 == 0) {
	return bbuf_clone(pbuf, bbuf2);
      }
      else {
	return not_code_range_buf(enc, bbuf2, pbuf);
      }
    }
  }

  if (not1 != 0)
    SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);

  data1 = (OnigCodePoint* )(bbuf1->p);
  GET_CODE_POINT(n1, data1);
  data1++;

  if (not2 == 0 && not1 == 0) { /* 1 OR 2 */
    r = bbuf_clone(pbuf, bbuf2);
  }
  else if (not1 == 0) { /* 1 OR (not 2) */
    r = not_code_range_buf(enc, bbuf2, pbuf);
  }
  if (r != 0) return r;

  for (i = 0; i < n1; i++) {
    from = data1[i*2];
    to   = data1[i*2+1];
    r = add_code_range_to_buf(pbuf, from, to);
    if (r != 0) return r;
  }
  return 0;
}

static int
and_code_range1(BBuf** pbuf, OnigCodePoint from1, OnigCodePoint to1,
	        OnigCodePoint* data, int n)
{
  int i, r;
  OnigCodePoint from2, to2;

  for (i = 0; i < n; i++) {
    from2 = data[i*2];
    to2   = data[i*2+1];
    if (from2 < from1) {
      if (to2 < from1) continue;
      else {
	from1 = to2 + 1;
      }
    }
    else if (from2 <= to1) {
      if (to2 < to1) {
	if (from1 <= from2 - 1) {
	  r = add_code_range_to_buf(pbuf, from1, from2-1);
	  if (r != 0) return r;
	}
	from1 = to2 + 1;
      }
      else {
	to1 = from2 - 1;
      }
    }
    else {
      from1 = from2;
    }
    if (from1 > to1) break;
  }
  if (from1 <= to1) {
    r = add_code_range_to_buf(pbuf, from1, to1);
    if (r != 0) return r;
  }
  return 0;
}

static int
and_code_range_buf(BBuf* bbuf1, int not1, BBuf* bbuf2, int not2, BBuf** pbuf)
{
  int r;
  OnigCodePoint i, j, n1, n2, *data1, *data2;
  OnigCodePoint from, to, from1, to1, from2, to2;

  *pbuf = (BBuf* )NULL;
  if (IS_NULL(bbuf1)) {
    if (not1 != 0 && IS_NOT_NULL(bbuf2)) /* not1 != 0 -> not2 == 0 */
      return bbuf_clone(pbuf, bbuf2);
    return 0;
  }
  else if (IS_NULL(bbuf2)) {
    if (not2 != 0)
      return bbuf_clone(pbuf, bbuf1);
    return 0;
  }

  if (not1 != 0)
    SWAP_BBUF_NOT(bbuf1, not1, bbuf2, not2);

  data1 = (OnigCodePoint* )(bbuf1->p);
  data2 = (OnigCodePoint* )(bbuf2->p);
  GET_CODE_POINT(n1, data1);
  GET_CODE_POINT(n2, data2);
  data1++;
  data2++;

  if (not2 == 0 && not1 == 0) { /* 1 AND 2 */
    for (i = 0; i < n1; i++) {
      from1 = data1[i*2];
      to1   = data1[i*2+1];
      for (j = 0; j < n2; j++) {
	from2 = data2[j*2];
	to2   = data2[j*2+1];
	if (from2 > to1) break;
	if (to2 < from1) continue;
	from = MAX(from1, from2);
	to   = MIN(to1, to2);
	r = add_code_range_to_buf(pbuf, from, to);
	if (r != 0) return r;
      }
    }
  }
  else if (not1 == 0) { /* 1 AND (not 2) */
    for (i = 0; i < n1; i++) {
      from1 = data1[i*2];
      to1   = data1[i*2+1];
      r = and_code_range1(pbuf, from1, to1, data2, n2);
      if (r != 0) return r;
    }
  }

  return 0;
}

static int
clear_not_flag_cclass(CClassNode* cc, OnigEncoding enc)
{
  BBuf *tbuf;
  int r;

  if (IS_CCLASS_NOT(cc)) {
    bitset_invert(cc->bs);

    if (! ONIGENC_IS_SINGLEBYTE(enc)) {
      r = not_code_range_buf(enc, cc->mbuf, &tbuf);
      if (r != 0) return r;

      bbuf_free(cc->mbuf);