utf8.c

广泛使用的邮件服务器！同时

	for (ten = 0; ten < MAX_JIS0208_TEN; ten++)
	  if ((u = jis0208tab[ku][ten]) != UBOGON) {
	    int sku = ku + BASE_JIS0208_KU;
	    int sten = ten + BASE_JIS0208_TEN;
	    rmap[u] = ((((sku + 1) >> 1) + ((sku < 95) ? 112 : 176)) << 8) +
	      sten + ((sku % 2) ? ((sten > 95) ? 32 : 31) : 126);
	  }
				/* JIS Roman */
      rmap[UCS2_YEN] = JISROMAN_YEN;
      rmap[UCS2_OVERLINE] = JISROMAN_OVERLINE;
				/* JIS hankaku katakana */
      for (u = 0; u < (MAX_KANA_8 - MIN_KANA_8); u++)
	rmap[UCS2_KATAKANA + u] = MIN_KANA_8 + u;
      break;
    }
				/* hack: map NBSP to SP if otherwise no map */
    if (rmap[0x00a0] == NOCHAR) rmap[0x00a0] = rmap[0x0020];
  }
  return rmap;			/* return map */
}

/* Convert UTF-8 sized text to charset using rmap
 * Accepts: source sized text
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 *	    ISO-2022-JP conversion flag
 * Returns T if successful, NIL if failure
 *
 * This routine doesn't try to convert to all possible charsets; in particular
 * it doesn't support other Unicode encodings or any ISO 2022 other than
 * ISO-2022-JP.
 */

long utf8_rmaptext (SIZEDTEXT *text,unsigned short *rmap,SIZEDTEXT *ret,
		    unsigned long errch,long iso2022jp)
{
  unsigned long i,u,c;
				/* get size of buffer */
  if (i = utf8_rmapsize (text,rmap,errch,iso2022jp)) {
    unsigned char *s = text->data;
    unsigned char *t = ret->data = (unsigned char *) fs_get (i);
    ret->size = i - 1;		/* number of octets in destination buffer */
				/* start non-zero ISO-2022-JP state at 1 */
    if (iso2022jp) iso2022jp = 1;
				/* convert string, ignore BOM */
    for (i = text->size; i;) if ((u = utf8_get (&s,&i)) != UCS2_BOM) {
				/* substitute error character for NOCHAR */
      if ((u & U8GM_NONBMP) || ((c = rmap[u]) == NOCHAR)) c = errch;
      switch (iso2022jp) {	/* depends upon ISO 2022 mode */
      case 0:			/* ISO 2022 not in effect */
				/* two-byte character */
	if (c > 0xff) *t++ = (unsigned char) (c >> 8);
				/* single-byte or low-byte of two-byte */
	*t++ = (unsigned char) (c & 0xff);
	break;
      case 1:			/* ISO 2022 Roman */
				/* <ch> */
	if (c < 0x80) *t++ = (unsigned char) c;
	else {			/* JIS character */
	  *t++ = I2C_ESC;	/* ESC $ B <hi> <lo> */
	  *t++ = I2C_MULTI;
	  *t++ = I2CS_94x94_JIS_NEW;
	  *t++ = (unsigned char) (c >> 8) & 0x7f;
	  *t++ = (unsigned char) c & 0x7f;
	  iso2022jp = 2;	/* shift to ISO 2022 JIS */
	}
	break;
      case 2:			/* ISO 2022 JIS */
	if (c > 0x7f) {		/* <hi> <lo> */
	  *t++ = (unsigned char) (c >> 8) & 0x7f;
	  *t++ = (unsigned char) c & 0x7f;
	}
	else {			/* ASCII character */
	  *t++ = I2C_ESC;	/* ESC ( J <ch> */
	  *t++ = I2C_G0_94;
	  *t++ = I2CS_94_JIS_ROMAN;
	  *t++ = (unsigned char) c;
	  iso2022jp = 1;	/* shift to ISO 2022 Roman */
	}
	break;
      }
    }
    if (iso2022jp == 2) {	/* ISO-2022-JP string must end in Roman */
      *t++ = I2C_ESC;		/* ESC ( J */
      *t++ = I2C_G0_94;
      *t++ = I2CS_94_JIS_ROMAN;
    }
    *t++ = NIL;			/* tie off returned data */
    return LONGT;		/* return success */
  }
  ret->data = NIL;
  ret->size = 0;
  return NIL;			/* failure */
}

/* Calculate size of convertsion of UTF-8 sized text to charset using rmap
 * Accepts: source sized text
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 *	    ISO-2022-JP conversion flag
 * Returns size+1 if successful, NIL if failure
 *
 * This routine doesn't try to handle to all possible charsets; in particular
 * it doesn't support other Unicode encodings or any ISO 2022 other than
 * ISO-2022-JP.
 */

unsigned long utf8_rmapsize (SIZEDTEXT *text,unsigned short *rmap,
			     unsigned long errch,long iso2022jp)
{
  unsigned long i,u,c;
  unsigned long ret = 1;	/* terminating NUL */
  unsigned char *s = text->data;
  if (iso2022jp) iso2022jp = 1;	/* start non-zero ISO-2022-JP state at 1 */
  for (i = text->size; i;) if ((u = utf8_get (&s,&i)) != UCS2_BOM) {
    if ((u & U8GM_NONBMP) || (((c = rmap[u]) == NOCHAR) && !(c = errch)))
      return NIL;		/* not in BMP, or NOCHAR and no err char */
    switch (iso2022jp) {	/* depends upon ISO 2022 mode */
    case 0:			/* ISO 2022 not in effect */
      ret += (c > 0xff) ? 2 : 1;
      break;
    case 1:			/* ISO 2022 Roman */
      if (c < 0x80) ret += 1;	/* <ch> */
      else {			/* JIS character */
	ret += 5;		/* ESC $ B <hi> <lo> */
	iso2022jp = 2;		/* shift to ISO 2022 JIS */
      }
      break;
    case 2:			/* ISO 2022 JIS */
      if (c > 0x7f) ret += 2;	/* <hi> <lo> */
      else {			/* ASCII character */
	ret += 4;		/* ESC ( J <ch> */
	iso2022jp = 1;		/* shift to ISO 2022 Roman */
      }
      break;
    }
  }
  if (iso2022jp == 2) {		/* ISO-2022-JP string must end in Roman */
    ret += 3;			/* ESC ( J */
    iso2022jp = 1;		/* reset state to Roman */
  }
  return ret;
}

/* Convert UCS-4 to charset using rmap
 * Accepts: source UCS-4 character(s)
 *	    numver of UCS-4 characters
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns T if successful, NIL if failure
 *
 * Currently only supports BMP characters, and does not support ISO-2022
 */

long ucs4_rmaptext (unsigned long *ucs4,unsigned long len,unsigned short *rmap,
		    SIZEDTEXT *ret,unsigned long errch)
{
  long size = ucs4_rmaplen (ucs4,len,rmap,errch);
  return (size >= 0) ?		/* build in newly-created buffer */
    ucs4_rmapbuf (ret->data = (unsigned char *) fs_get ((ret->size = size) +1),
		  ucs4,len,rmap,errch) : NIL;
}

/* Return size of UCS-4 string converted to other CS via rmap
 * Accepts: source UCS-4 character(s)
 *	    numver of UCS-4 characters
 *	    conversion rmap
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns: length if success, negative if failure (no-convert)
 */

long ucs4_rmaplen (unsigned long *ucs4,unsigned long len,unsigned short *rmap,
		   unsigned long errch)
{
  long ret;
  unsigned long i,u,c;
				/* count non-BOM characters */
  for (ret = 0,i = 0; i < len; ++i) if ((u = ucs4[i]) != UCS2_BOM) {
    if ((u & U8GM_NONBMP) || (((c = rmap[u]) == NOCHAR) && !(c = errch)))
      return -1;		/* not in BMP, or NOCHAR and no err char? */
    ret += (c > 0xff) ? 2 : 1;
  }
  return ret;
}


/* Stuff buffer with UCS-4 string converted to other CS via rmap
 * Accepts: destination buffer
 *	    source UCS-4 character(s)
 *	    number of UCS-4 characters
 *	    conversion rmap
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns: T, always
 */

long ucs4_rmapbuf (unsigned char *t,unsigned long *ucs4,unsigned long len,
		   unsigned short *rmap,unsigned long errch)
{
  unsigned long i,u,c;
				/* convert non-BOM characters */
  for (i = 0; i < len; ++i) if ((u = ucs4[i]) != UCS2_BOM) {
				/* substitute error character for NOCHAR */
    if ((u & U8GM_NONBMP) || ((c = rmap[u]) == NOCHAR)) c = errch;
				/* two-byte character? */
    if (c > 0xff) *t++ = (unsigned char) (c >> 8);
				/* single-byte or low-byte of two-byte */
    *t++ = (unsigned char) (c & 0xff);
  }
  *t++ = NIL;			/* tie off returned data */
  return LONGT;
}

/* Return UCS-4 Unicode character from UTF-8 string
 * Accepts: pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated
 *	    or error code with pointer and count unchanged
 */

unsigned long utf8_get (unsigned char **s,unsigned long *i)
{
  unsigned char *t = *s;
  unsigned long j = *i;
				/* decode raw UTF-8 string */
  unsigned long ret = utf8_get_raw (&t,&j);
  if (ret & U8G_ERROR);		/* invalid raw UTF-8 decoding? */
				/* no, is it surrogate? */
  else if ((ret >= UTF16_SURR) && (ret <= UTF16_MAXSURR)) ret = U8G_SURROGA;
				/* or in non-Unicode ISO 10646 space? */
  else if (ret > UCS4_MAXUNICODE) ret = U8G_NOTUNIC;
  else {
    *s = t;			/* all is well, update pointer */
    *i = j;			/* and counter */
  }
  return ret;			/* return value */
}

/* Return raw (including non-Unicode) UCS-4 character from UTF-8 string
 * Accepts: pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated
 *	    or error code with pointer and count unchanged
 */

unsigned long utf8_get_raw (unsigned char **s,unsigned long *i)
{
  unsigned char c,c1;
  unsigned char *t = *s;
  unsigned long j = *i;
  unsigned long ret = U8G_NOTUTF8;
  int more = 0;
  do {				/* make sure have source octets available */
    if (!j--) return more ? U8G_ENDSTRI : U8G_ENDSTRG;
				/* UTF-8 continuation? */
    else if (((c = *t++) > 0x7f) && (c < 0xc0)) {
				/* continuation when not in progress */
      if (!more) return U8G_BADCONT;
      --more;			/* found a continuation octet */
      ret <<= 6;		/* shift current value by 6 bits */
      ret |= c & 0x3f;		/* merge continuation octet */
    }
				/* incomplete UTF-8 character */
    else if (more) return U8G_INCMPLT;
    else {			/* start of sequence */
      c1 = j ? *t : 0xbf;	/* assume valid continuation if incomplete */
      if (c < 0x80) ret = c;	/* U+0000 - U+007f */
      else if (c < 0xc2);	/* c0 and c1 never valid */
      else if (c < 0xe0) {	/* U+0080 - U+07ff */
	if (c &= 0x1f) more = 1;
      }
      else if (c < 0xf0) {	/* U+0800 - U+ffff */
	if ((c &= 0x0f) || (c1 >= 0xa0)) more = 2;
      }
      else if (c < 0xf8) {	/* U+10000 - U+10ffff (and 110000 - 1fffff) */
	if ((c &= 0x07) || (c1 >= 0x90)) more = 3;
      }
      else if (c < 0xfc) {	/* ISO 10646 200000 - 3ffffff */
	if ((c &= 0x03) || (c1 >= 0x88)) more = 4;
      }
      else if (c < 0xfe) {	/* ISO 10646 4000000 - 7fffffff */
	if ((c &= 0x01) || (c1 >= 0x84)) more = 5;
      }
				/* fe and ff never valid */
      if (more) {		/* multi-octet, make sure more to come */
	if (!j) return U8G_ENDSTRI;
	ret = c;		/* continuation needed, save start bits */
      }
    }
  } while (more);
  if (!(ret & U8G_ERROR)) {	/* success return? */
    *s = t;			/* yes, update pointer */
    *i = j;			/* and counter */
  }
  return ret;			/* return value */
}

/* Return UCS-4 character from named charset string
 * Accepts: charset
 *	    pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated, negative if error
 *
 * Error codes are the same as utf8_get().
 */

unsigned long ucs4_cs_get (CHARSET *cs,unsigned char **s,unsigned long *i)
{
  unsigned char c,c1,ku,ten;
  unsigned long ret,d;
  unsigned char *t = *s;
  unsigned long j = *i;
  struct utf8_eucparam *p1,*p2,*p3;
  if (j--) c = *t++;		/* get first octet */
  else return U8G_ENDSTRG;	/* empty string */
  switch (cs->type) {		/* convert if type known */
  case CT_UTF8:			/* variable UTF-8 encoded Unicode no table */
    return utf8_get (s,i);
  case CT_ASCII:		/* 7-bit ASCII no table */
    if (c >= 0x80) return U8G_NOTUTF8;
  case CT_1BYTE0:		/* 1 byte no table */
    ret = c;			/* identity */
    break;
  case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
    ret = (c > 0x80) ? ((unsigned short *) cs->tab)[c & BITS7] : c;
    break;
  case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
    ret = ((unsigned short *) cs->tab)[c];
    break;

  case CT_EUC:			/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
    if (c & BIT8) {
      p1 = (struct utf8_eucparam *) cs->tab;
      p2 = p1 + 1;
      p3 = p1 + 2;
      if (j--) c1 = *t++;	/* get second octet */
      else return U8G_ENDSTRI;
      if (!(c1 & BIT8)) return U8G_NOTUTF8;
      switch (c) {		/* check 8bit code set */
      case EUC_CS2:		/* CS2 */
	if (p2->base_ku) {	/* CS2 set up? */
	  if (p2->base_ten) {	/* yes, multibyte? */
	    if (j--) c = *t++;	/* get second octet */
	    else return U8G_ENDSTRI;
	    if ((c & BIT8) &&
		((ku = (c1 & BITS7) - p2->base_ku) < p2->max_ku) &&
		((ten = (c & BITS7) - p2->base_ten) < p2->max_ten)) {
	      ret = ((unsigned short *) p2->tab)[(ku*p2->max_ten) + ten];
	      break;
	    }
	  }
	  else if ((c1 >= p2->base_ku) && (c1 < p2->max_ku)) {
	    ret = c1 + ((unsigned long) p2->tab);
	    break;
	  }
	}
	return U8G_NOTUTF8;	/* CS2 not set up or bogus */
      case EUC_CS3:		/* CS3 */
	if (p3->base_ku) {	/* CS3 set up? */
	  if (p3->base_ten) {	/* yes, multibyte? */
	    if (j--) c = *t++;	/* get second octet */
	    else return U8G_ENDSTRI;
	    if ((c & BIT8) &&
		((ku = (c1 & BITS7) - p3->base_ku) < p3->max_ku) &&
		((ten = (c & BITS7) - p3->base_ten) < p3->max_ten)) {
	      ret = ((unsigned short *) p3->tab)[(ku*p3->max_ten) + ten];
	      break;
	    }
	  }
	  else if ((c1 >= p3->base_ku) && (c1 < p3->max_ku)) {
	    ret = c1 + ((unsigned long) p3->tab);
	    break;
	  }
	}
	return U8G_NOTUTF8;	/* CS3 not set up or bogus */