admin/charsets/gb180302.awk

   1 BEGIN {
   2   tohex["A"] = 10;
   3   tohex["B"] = 11;
   4   tohex["C"] = 12;
   5   tohex["D"] = 13;
   6   tohex["E"] = 14;
   7   tohex["F"] = 15;
   8   tohex["a"] = 10;
   9   tohex["b"] = 11;
  10   tohex["c"] = 12;
  11   tohex["d"] = 13;
  12   tohex["e"] = 14;
  13   tohex["f"] = 15;
  14   from_gb = 0;
  15   to_gb = -1;
  16   to_unicode = 0;
  17   from_unicode = 0;
  18 }
  19
  20 function decode_hex(str) {
  21   n = 0;
  22   len = length(str);
  23   for (i = 1; i <= len; i++)
  24     {
  25       c = substr (str, i, 1);
  26       if (c >= "0" && c <= "9")
  27         n = n * 16 + (c - "0");
  28       else
  29         n = n * 16 + tohex[c];
  30     }
  31   return n;
  32 }
  33
  34 function gb_to_index(gb) {
  35   b0 = int(gb / 256);
  36   b1 = gb % 256;
  37   idx = (((b0 - 129)) * 191 + b1 - 64);
  38 #  if (b1 >= 128)
  39 #    idx--;
  40   return idx
  41 }
  42
  43 function index_to_gb(idx) {
  44   b0 = int(idx / 191) + 129;
  45   b1 = (idx % 191) + 64;
  46 #  if (b1 >= 127)
  47 #    b1++;
  48   return (b0 * 256 + b1);
  49 }
  50
  51 /^\#/ {
  52   print;
  53   next;
  54 }
  55
  56 {
  57   gb = gb_to_index(decode_hex(substr($1, 3, 4)));
  58   unicode = decode_hex(substr($2, 3, 4));
  59   if ((gb == to_gb + 1) && (unicode == to_unicode + 1))
  60     {
  61       to_gb++;
  62       to_unicode++;
  63     }
  64   else
  65     {
  66       if (from_gb == to_gb)
  67         printf "0x%04X 0x%04X\n", index_to_gb(from_gb), from_unicode;
  68       else if (from_gb < to_gb)
  69         printf "0x%04X-0x%04X 0x%04X\n",
  70           index_to_gb(from_gb), index_to_gb(to_gb), from_unicode;
  71       from_gb = to_gb = gb;
  72       from_unicode = to_unicode = unicode;
  73     }
  74 }
  75
  76 END {
  77   if (from_gb <= to_gb)
  78     printf "0x%04X-0x%04X 0x%04X\n",
  79       index_to_gb(from_gb), index_to_gb(to_gb), from_unicode;
  80 }
  81
  82 # arch-tag: d7dbad89-a512-41a4-8ee0-ba1a4505b8c1