regenc.c

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

  int c, i, len;
  OnigCodePoint n;

  len = enc_len(enc, p);
  n = (OnigCodePoint )(*p++);
  if (len == 1) return n;

  for (i = 1; i < len; i++) {
    if (p >= end) break;
    c = *p++;
    n <<= 8;  n += c;
  }
  return n;
}

extern int
onigenc_mbn_mbc_to_normalize(OnigEncoding enc, OnigAmbigType flag,
                             const UChar** pp, const UChar* end, UChar* lower)
{
  int len;
  const UChar *p = *pp;

  if (ONIGENC_IS_MBC_ASCII(p)) {
    if ((flag & ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) != 0) {
      *lower = ONIGENC_ASCII_CODE_TO_LOWER_CASE(*p);
    }
    else {
      *lower = *p;
    }
    (*pp)++;
    return 1;
  }
  else {
    len = enc_len(enc, p);
    if (lower != p) {
      int i;
      for (i = 0; i < len; i++) {
	*lower++ = *p++;
      }
    }
    (*pp) += len;
    return len; /* return byte length of converted to lower char */
  }
}

extern int
onigenc_mbn_is_mbc_ambiguous(OnigEncoding enc, OnigAmbigType flag,
                             const UChar** pp, const UChar* end)
{
  const UChar* p = *pp;

  if (ONIGENC_IS_MBC_ASCII(p)) {
    (*pp)++;
    if ((flag & ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) != 0) {
      return ONIGENC_IS_ASCII_CODE_CASE_AMBIG(*p);
    }
    else {
      return FALSE;
    }
  }

  (*pp) += enc_len(enc, p);
  return FALSE;
}

extern int
onigenc_mb2_code_to_mbclen(OnigCodePoint code)
{
  if ((code & 0xff00) != 0) return 2;
  else return 1;
}

extern int
onigenc_mb4_code_to_mbclen(OnigCodePoint code)
{
       if ((code & 0xff000000) != 0) return 4;
  else if ((code & 0xff0000) != 0) return 3;
  else if ((code & 0xff00) != 0) return 2;
  else return 1;
}

extern int
onigenc_mb2_code_to_mbc_first(OnigCodePoint code)
{
  int first;

  if ((code & 0xff00) != 0) {
    first = (code >> 8) & 0xff;
  }
  else {
    return (int )code;
  }
  return first;
}

extern int
onigenc_mb4_code_to_mbc_first(OnigCodePoint code)
{
  int first;

  if ((code & 0xff000000) != 0) {
    first = (code >> 24) & 0xff;
  }
  else if ((code & 0xff0000) != 0) {
    first = (code >> 16) & 0xff;
  }
  else if ((code & 0xff00) != 0) {
    first = (code >>  8) & 0xff;
  }
  else {
    return (int )code;
  }
  return first;
}

extern int
onigenc_mb2_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
{
  UChar *p = buf;

  if ((code & 0xff00) != 0) {
    *p++ = (UChar )((code >>  8) & 0xff);
  }
  *p++ = (UChar )(code & 0xff);

#if 1
  if (enc_len(enc, buf) != (p - buf))
    return ONIGENCERR_INVALID_WIDE_CHAR_VALUE;
#endif
  return p - buf;
}

extern int
onigenc_mb4_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
{
  UChar *p = buf;

  if ((code & 0xff000000) != 0) {
    *p++ = (UChar )((code >> 24) & 0xff);
  }
  if ((code & 0xff0000) != 0) {
    *p++ = (UChar )((code >> 16) & 0xff);
  }
  if ((code & 0xff00) != 0) {
    *p++ = (UChar )((code >> 8) & 0xff);
  }
  *p++ = (UChar )(code & 0xff);

#if 1
  if (enc_len(enc, buf) != (p - buf))
    return ONIGENCERR_INVALID_WIDE_CHAR_VALUE;
#endif
  return p - buf;
}

extern int
onigenc_mb2_is_code_ctype(OnigEncoding enc, OnigCodePoint code,
			  unsigned int ctype)
{
  if ((ctype & ONIGENC_CTYPE_WORD) != 0) {
    if (code < 128)
      return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
    else
      return (ONIGENC_CODE_TO_MBCLEN(enc, code) > 1 ? TRUE : FALSE);

    ctype &= ~ONIGENC_CTYPE_WORD;
    if (ctype == 0) return FALSE;
  }

  if (code < 128)
    return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
  else
    return FALSE;
}

extern int
onigenc_mb4_is_code_ctype(OnigEncoding enc, OnigCodePoint code,
			  unsigned int ctype)
{
  if ((ctype & ONIGENC_CTYPE_WORD) != 0) {
    if (code < 128)
      return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
    else
      return (ONIGENC_CODE_TO_MBCLEN(enc, code) > 1 ? TRUE : FALSE);

    ctype &= ~ONIGENC_CTYPE_WORD;
    if (ctype == 0) return FALSE;
  }

  if (code < 128)
    return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
  else
    return FALSE;
}

extern int
onigenc_with_ascii_strncmp(OnigEncoding enc, const UChar* p, const UChar* end,
                           const UChar* sascii /* ascii */, int n)
{
  int x, c;

  while (n-- > 0) {
    if (p >= end) return (int )(*sascii);

    c = (int )ONIGENC_MBC_TO_CODE(enc, p, end);
    x = *sascii - c;
    if (x) return x;

    sascii++;
    p += enc_len(enc, p);
  }
  return 0;
}

#else /* ONIG_RUBY_M17N */

extern int
onigenc_is_code_ctype(OnigEncoding enc, OnigCodePoint code, int ctype)
{
  switch (ctype) {
  case ONIGENC_CTYPE_NEWLINE:
    if (code == 0x0a) return 1;
    break;

  case ONIGENC_CTYPE_ALPHA:
    return m17n_isalpha(enc, code);
    break;
  case ONIGENC_CTYPE_BLANK:
    return ONIGENC_IS_CODE_BLANK(enc, (int )(code));
    break;
  case ONIGENC_CTYPE_CNTRL:
    return m17n_iscntrl(enc, code);
    break;
  case ONIGENC_CTYPE_DIGIT:
    return m17n_isdigit(enc, code);
    break;
  case ONIGENC_CTYPE_GRAPH:
    return ONIGENC_IS_CODE_GRAPH(enc, (int )(code));
    break;
  case ONIGENC_CTYPE_LOWER:
    return m17n_islower(enc, code);
    break;
  case ONIGENC_CTYPE_PRINT:
    return m17n_isprint(enc, code);
    break;
  case ONIGENC_CTYPE_PUNCT:
    return m17n_ispunct(enc, code);
    break;
  case ONIGENC_CTYPE_SPACE:
    return m17n_isspace(enc, code);
    break;
  case ONIGENC_CTYPE_UPPER:
    return m17n_isupper(enc, code);
    break;
  case ONIGENC_CTYPE_XDIGIT:
    return m17n_isxdigit(enc, code);
    break;
  case ONIGENC_CTYPE_WORD:
    return m17n_iswchar(enc, code);
    break;
  case ONIGENC_CTYPE_ASCII:
    return (code < 128 ? TRUE : FALSE);
    break;
  case ONIGENC_CTYPE_ALNUM:
    return m17n_isalnum(enc, code);
    break;
  default:
    break;
  }

  return 0;
}

extern int
onigenc_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
{
  int c, len;

  m17n_mbcput(enc, code, buf);
  c = m17n_firstbyte(enc, code);
  len = enc_len(enc, c);
  return len;
}

extern int
onigenc_mbc_to_lower(OnigEncoding enc, UChar* p, UChar* buf)
{
  unsigned int c, low;

  c   = m17n_codepoint(enc, p, p + enc_len(enc, *p));
  low = m17n_tolower(enc, c);
  m17n_mbcput(enc, low, buf);

  return m17n_codelen(enc, low);
}

extern int
onigenc_is_mbc_ambiguous(OnigEncoding enc, OnigAmbigType flag,
                         UChar** pp, UChar* end)
{
  int len;
  unsigned int c;
  UChar* p = *pp;

  len = enc_len(enc, *p);
  (*pp) += len;
  c = m17n_codepoint(enc, p, p + len);

  if ((flag & ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE) != 0) {
    if (m17n_isupper(enc, c) || m17n_islower(enc, c))
      return TRUE;
  }

  return FALSE;
}

extern UChar*
onigenc_get_left_adjust_char_head(OnigEncoding enc, UChar* start, UChar* s)
{
  UChar *p;
  int len;

  if (s <= start) return s;
  p = s;

  while (!m17n_islead(enc, *p) && p > start) p--;
  while (p + (len = enc_len(enc, *p)) < s) {
    p += len;
  }
  if (p + len == s) return s;
  return p;
}

extern int
onigenc_is_allowed_reverse_match(OnigEncoding enc,
				 const UChar* s, const UChar* end)
{
  return ONIGENC_IS_SINGLEBYTE(enc);
}

extern void
onigenc_set_default_caseconv_table(UChar* table) { }

#endif /* ONIG_RUBY_M17N */