2utf.c

一个UTF的源代码,可以提供参考

  enum
  {
    UCS2, UCS4, UCSwchar_t, UTF8f
  }
  output_format = UTF8f;
  int index, create_aliases = FALSE, file_arg = 0, html_doc = FALSE, list_charmaps = FALSE, spit = FALSE, swap, switch_to_unicode = FALSE;
  register int c;
  long int l;
  union
  {
    u_int8_t byte[4];
    u_int16_t b16[2];
    u_int32_t b32;
  }
  u_int;
  wchar_t range, spit_min = 0xF000, spit_max = 0xF1FF, wchar;

  UTF8_charset.pipe = USASCII_charset.pipe = unknown_charset.pipe = stdout;

  /* standard UCS byte order is big endian */
#if (__BYTE_ORDER != __BIG_ENDIAN)
  swap = TRUE;
#else
  swap = FALSE;
#endif

  if (argc <= 1)
  {
  }
  else
  {
    /*while ((c = getopt_long (argc, argv, optstring, long_options, NULL)) != -1)*/
    Long2short_options (argc, argv, long2short_options);
    while ((c = getopt (argc, argv, optstring)) != -1)
    {
      if (verbose)
	fprintf (stderr, " -%c", c);
      switch (c)
      {
	case '?':
	case ':':
	  Short_help(3);
	case '2':
	  output_format = UCS2;
	  break;
	case '4':
	case 'w':
	  output_format = UCS4;
	  break;
	case '8':
	  output_format = UTF8f;
	  break;
	case 'C':
	  create_aliases = 1;
	  break;
	case 'V':
	  fprintf (stderr, "%s%s", version, blurb);
	  exit (0);
	case 'c':
	  charmap_filename = xstrdup (optarg);
	  if (verbose)
	    fprintf (stderr, "'%s'", charmap_filename);
	  break;
	case 'd':
	  Debug = 1;
	  if (optarg)
	    Debug = atoi (optarg);
	  if (verbose)
	    fprintf (stderr, "%i", Debug);
	  verbose = TRUE;
	  break;
	case 'e':
	  encode = TRUE;
	  break;
	case 'f':
	  if (optarg)
	  {
	    charmap_format[DEFAULT] = xstrdup (optarg);
	  }
	  else
	    charmap_format[DEFAULT] = default_charmap_format;
	  if (verbose)
	    fprintf (stderr, "'%s'", charmap_format[DEFAULT]);
	  break;
	case 'H':
	  html_doc = TRUE;
	  break;
	case 'h':
	  Help (default_unknown_char);
	  exit (0);
	case 'i':
	  iconv_only = TRUE;
	  if (optarg) {
	    if (Strcasecmp(optarg, "only") == 0)
	      iconv_only = TRUE;
	    else if (Strcasecmp(optarg, "first") == 0) {
	      iconv_only = FALSE;
	      iconv_first = TRUE;
	    } else if (Strcasecmp(optarg, "last") == 0) {
	      iconv_only = FALSE;
	      iconv_first = FALSE;
	    } else
	      Short_help(5);
	  }
	  break;
	case 'l':
	  list_charmaps = TRUE;
	  break;
	case 'o':
	  reverse = FALSE;
	  break;
	case 'p':
	  Read_config_file ();
	  Print_paths ();
	  exit (0);
	case 'r':
	  reverse = TRUE;
	  break;
	case 'S':
	  spit = 1;
	  if (optarg)
	  {
	    if (strcspn (optarg + strspn (optarg, "\t\n "), "-,;") == 0)
	    {
	      spit_min = 0;
	      if ((next_token = strtok (optarg, "\t\n -,;")) == NULL)
		spit = 2;
	    }
	    else
	    {
	      if (strpbrk (optarg + strspn (optarg, "\t\n -,;"), "-,;"))
		spit = 2;
	      if ((next_token = strtok (optarg, "\t\n -,;")) == NULL)
		Short_help(5);
	      errno = 0;
	      range = strtoul (next_token, (char **) &stop_char_ptr, 16);
	      if (errno != ERANGE && *stop_char_ptr == '\0' &&range >= 0 && range <= 0x7FFFFFFF)
		spit_min = range;
	      else
		Short_help(5);
	      next_token = strtok (NULL, "\t\n -,;");
	    }
	    if (next_token)
	    {
	      errno = 0;
	      range = strtoul (next_token, (char **) &stop_char_ptr, 16);
	      if (errno != ERANGE && *stop_char_ptr == '\0' &&range >= 0 && range <= 0x7FFFFFFF)
		spit_max = range;
	      else
		Short_help(5);
	    }
	    else if (spit == 2)
	      spit_max = (spit_min > 0x7FFFFFFF - 511) ? 0x7FFFFFFF : spit_min / 32 * 32 + 511;
	    else
	      spit_max = spit_min / 32 * 32 + 31;
	    if (spit_max < spit_min)
	    {
	      range = spit_min;
	      spit_min = spit_max;
	      spit_max = range;
	    }
	    if (strtok (NULL, "\t\n -,;"))
	      Short_help(5);
	  }
	  /*fprintf(stderr,"min = %lx\n",spit_min); */
	  if (verbose)
	    fprintf (stderr, "0x%.4lX-0x%.4lX\n", (long unsigned int) spit_min, (long unsigned int) spit_max);
	  break;
	case 's':
	  switch_to_unicode = TRUE;
	  break;
	case 'u':
	  if (optarg)
	    if (strlen (optarg) <= 1)
	      unknown_char = optarg[0];
	    else
	    {
	      errno = 0;
	      l = strtol (optarg, (char **) &stop_char_ptr, 0);
	      if (errno != ERANGE && *stop_char_ptr == '\0' && l <= 0xFF)
		unknown_char = l;
	      else
		Short_help(5);
	    }
	  else
	    unknown_char = default_unknown_char;
	  if (verbose)
	    fprintf (stderr, "0x%.2X", unknown_char);
	  break;
	case 'v':
	  verbose = TRUE;
	  fprintf (stderr, " -v");
	  break;
	case 'W':
	  show_charmap = TRUE;
	  break;
	  /*          case 'w':
		      output_format = UCSwchar_t;
		      break; */
	default:
	  fprintf (stderr, "getopt returned character code 0x%X ??\n", c);
      }
    }
    if (verbose)
      putc ('\n', stderr);
  }
  if (encode && !iconv_only)
    Error (add_iconv_only);
  if (!(ptr = strrchr (argv[0], '/')))
    ptr = argv[0];
  else
    ptr++;
  if (Strcasecmp (ptr, "fromUTF") == 0)
    reverse = !reverse;
  if (Debug > 7)
    fprintf (stderr, "2UTF: argv[0]='%s', filename='%s'. \n", argv[0], ptr);
  if (Debug)
    if (reverse)
      fprintf (stderr, "2UTF: from UTF \n");
    else
      fprintf (stderr, "2UTF: to UTF \n");
  if (switch_to_unicode)
    fprintf (stderr, "\033%%G");
  if (iconv_only) {
    paths_number = 0;
  } else {
    Read_config_file ();
  }
  if (atexit (&Close_pipe) != 0)
    Error (internal_err);
  if (create_aliases)
    if (!(c = Create_aliases ()))
    {
      fprintf (stderr, "2UTF: %s '%s' \n", can_not_create, aliases_pathname);
      exit (7);
    }
  if (list_charmaps)
    exit (List_charmaps ()? 0 : 8);
  if (spit)
  {
    Spit (spit_min, spit_max, NULL, unknown_char);
    exit (0);
  }
  if (argc > 1 && optind < argc) {
    charmap_filename = argv[file_arg = argc - 1];
  }
  else if (charmap_filename == NULL)
    if (create_aliases)
      exit (0);
    else
    {
      /* Assuming mail message */
      if (output_format == UCS2 || output_format == UCS4 || reverse)
	Error (unimplemented);
      Mail = TRUE;
      c = parse_message (0);
      if (charset_p->type == BUF_PIPE || charset_p->type == NON_BUF_PIPE)
	pclose (charset_p->pipe);
      Validate_charset ("'unknown'", 0);
      /*charset_p = &unknown_charset;*/
      Pipe_to_UTF8 (stdin);
      if (c != END_BOUNDARY)
      {
	if (Debug)
	  fprintf (stderr, "parse of message failed: %d\n", c);
	exit (1);
      }
      else if (Debug)
	fprintf (stderr, "parse of message complete\n");
      exit (0);
    }
  /* For the output of interractive programms */
  setvbuf (stdout, NULL, _IONBF, 0);
  if (show_charmap)
  {
    reverse = FALSE;
    unknown_wchar = (wchar_t) '?';
  }
  if (!Validate_charset (charmap_filename, 0))
  {				/* (charset_p->type == UNKNOWN) */
    if (strchr (charmap_filename, '/') == NULL)
    {
      fprintf (stderr, "%s '%s' \n", can_not_find_alias, charmap_filename);
      for (index = 0; index < paths_number; index++)
	fprintf (stderr, " %s*%s* \n", paths[index], charmap_filename);
    }
    else
      fprintf (stderr, " %s \n", charmap_filename);
    Error (can_not_open_any);
  }
  else if (charset_p->type == UTF8)
  {
    if (verbose)
      fprintf (stderr, "2UTF: %s '%s' \n", using, "Unicode UTF-8");
  }
  else if (charset_p->type == USASCII)
  {
    if (verbose)
      fprintf (stderr, "2UTF: %s '%s' \n", using, "US-ASCII");
    charset_p->charmap = xmalloc (256 * sizeof (wchar_t));
    charset_p->type = KNOWN;
    charset_p->pipe = stdout;
    charset_p->USASCII_is_subset = IS;
    for (wchar = 0; wchar <= 0x7F; wchar++)
      charset_p->charmap[wchar] = wchar;
    for (wchar = 0x7F; wchar <= 0xFF; wchar++)
      charset_p->charmap[wchar] = unknown_wchar;
  }

  if (charmap_filename != argv[file_arg])
    free (charmap_filename);
  if (show_charmap)
  {
    unknown_wchar = (wchar_t) '?';
    if (charset_p->type != UTF8)
      Spit (0, 255, charset_p->charmap, unknown_char);
    else
      Spit (0, 0x7FFFFFFF, NULL, unknown_char);
    exit (0);
  }
  if (charset_p->type != KNOWN && (output_format == UCS2 || output_format == UCS4))
    Error (unimplemented);
  if (reverse)
  {
    if (output_format == UCS2)
    {
      while (fread (u_int.byte, 1, 2, stdin) == 2)
      {
	/* check Byte Order Mark */
	if (0xFFFE == u_int.b16[0])
	  swap = TRUE;
	else if (0xFEFF == u_int.b16[0])
	  swap = FALSE;
	else
	{
	  if (swap)
	  {
	    c = u_int.byte[0];
	    u_int.byte[0] = u_int.byte[1];
	    u_int.byte[1] = c;
	  }
	  wchar = u_int.b16[0];
	  Print_approx_Macro (1, wchar, unknown_char)
	}
      }
    }
    else if (output_format == UCS4)
    {
      while (fread (u_int.byte, 1, 4, stdin) == 4)
      {
#if (__BYTE_ORDER == __PDP_ENDIAN)
#error
	c = u_int.b16[0];
	u_int.b16[0] = u_int.b16[1];
	u_int.b16[1] = c;
#endif
	/* check Byte Order Mark */
	if (0xFFFE0000 == u_int.b32)
	  swap = TRUE;
	else if (0x0000FEFF == u_int.b32)
	  swap = FALSE;
	else
	{
	  if (swap)
	  {
	    c = u_int.byte[0];
	    u_int.byte[0] = u_int.byte[3];
	    u_int.byte[3] = c;
	    c = u_int.byte[1];
	    u_int.byte[1] = u_int.byte[2];
	    u_int.byte[2] = c;
	  }
	  wchar = u_int.b32;
	  Print_approx_Macro (1, wchar, unknown_char)
	}
      }
    }
    else if (output_format == UTF8f)
    {
      if (charset_p->type == ICONV) {
	while ((c = getc (stdin)) != EOF)
	  Put_to_iconv_buf (c, charset_p);
	Put_to_iconv_buf (-2, charset_p);
      } else if (charset_p->type == KNOWN) {
	while ((c = Freadmb (&wchar, stdin)) != EOF)
	  Print_approx_Macro (c, wchar, unknown_char)
      } else if (charset_p->type == UTF8 || charset_p->type == BUF_PIPE || charset_p->type == NON_BUF_PIPE || charset_p->type == USASCII) {
	while ((c = getc (stdin)) != EOF) {
	  if (putc (c, charset_p->pipe) == EOF)
	    Error (IO_err);
	}
      } else
	Error (internal_err);
      if (ferror (stdin))
	Error (IO_err);
    }
    else
      Error (internal_err);

  }
  else
    /* not reverse */
  {
    if (output_format == UCS2)
    {
      putchar (0xFE);
      putchar (0xFF);
      while ((c = getchar ()) != EOF)
      {
	u_int.b16[0] = (u_int16_t) charset_p->charmap[c];
#if (__BYTE_ORDER != __BIG_ENDIAN)
	c = u_int.byte[0];
	u_int.byte[0] = u_int.byte[1];
	u_int.byte[1] = c;
#endif
	if (fwrite (u_int.byte, 1, 2, stdout) != 2)
	  Error (IO_err);
      }
    }
    else if (output_format == UCS4)
    {
      putchar (0);
      putchar (0);
      putchar (0xFE);
      putchar (0xFF);
      while ((c = getchar ()) != EOF)
      {
	u_int.b32 = (u_int32_t) charset_p->charmap[c];
#if (__BYTE_ORDER == __LITTLE_ENDIAN)
	c = u_int.byte[0];
	u_int.byte[0] = u_int.byte[3];
	u_int.byte[3] = c;
	c = u_int.byte[1];
	u_int.byte[1] = u_int.byte[2];
	u_int.byte[2] = c;
#elif (__BYTE_ORDER == __PDP_ENDIAN)
	c = u_int.byte[0];
	u_int.byte[0] = u_int.byte[1];
	u_int.byte[1] = c;
	c = u_int.byte[2];
	u_int.byte[2] = u_int.byte[3];
	u_int.byte[3] = c;
#elif (__BYTE_ORDER == __BIG_ENDIAN)
#else
#error Unknown byte order.
#endif
	if (fwrite (u_int.byte, 1, 4, stdout) != 4)
	  Error (IO_err);
      }
    }
    /*      else if (output_format == UCSwchar_t)
	    {
	    while ((c = getchar ()) != EOF)
	    if (fwrite (&charset_p->charmap[c], sizeof (wchar_t), 1, stdout) != 1)
	    Error (IO_err);
	    }
     */
	    else if (output_format == UTF8f)
	      Pipe_to_UTF8 (stdin);
	    else
	      Error (internal_err);
  }
  if (feof (stdin))
    exit (0);
  else
    Error (IO_err);
  exit (101);
}