w3m

Unnamed repository; edit this file to name it for gitweb.
git clone https://logand.com/git/w3m.git/
Log | Files | Refs | README

mk_gbk_ucs_map.pl (2366B)

      1 
      2 @NAME = ();
      3 while(<DATA>) {
      4   chop;
      5   s/\s*$//;
      6   (($n, $m, $c) = split(" ", $_, 3)) >= 3 || next;
      7   push(@NAME, $n);
      8   $MAP{$n} = $m;
      9   $CODE{$n} = $c;
     10 }
     11 
     12 foreach $name (@NAME) {
     13 
     14 $code = $CODE{$name};
     15 $map = $MAP{$name};
     16 
     17 print "$name\t$map\t$code\n";
     18 
     19 %to_ucs = ();
     20 %from_ucs = ();
     21 open(MAP, "< $map");
     22 open(OUT, "> ${name}_ucs.map");
     23 while(<MAP>) {
     24   /^#/ && next;
     25   s/#.*//;
     26   (($i, $u) = split(" ")) || next;
     27   $i = hex($i);
     28   $u = hex($u);
     29   $a = $i >> 8;
     30   $b = $i & 0x00FF;
     31   if ($a >= 0xA1 && $a <= 0xFE && $b >= 0xA1 && $b <= 0xFE) {
     32     ($i >= 0xA2A1 && $i <= 0xA2AA) ||
     33     ($i >= 0xA6E0 && $i <= 0xA6F5) ||
     34     ($i >= 0xA8BB && $i <= 0xA8BB) ||
     35     ($i >= 0xA8BD && $i <= 0xA8BD) ||
     36     ($i >= 0xA8BE && $i <= 0xA8BE) ||
     37     ($i >= 0xA8C0 && $i <= 0xA8C0) || next;
     38   }
     39   ($u < 0x80) && next;
     40   $to_ucs{$i} = $u;
     41   $from_ucs{$u} = $i;
     42 }
     43 
     44 # print OUT <<EOF;
     45 # /*
     46 #   These conversion tables between $code and
     47 #   Unicode were made from
     48 # 
     49 #     ftp://ftp.unicode.org/Public/MAPPINGS/$map.
     50 # */
     51 print OUT <<EOF;
     52 /* $code */
     53 
     54 #define N_ucs_gbk_80_map 1
     55 
     56 static wc_map ucs_gbk_80_map[ N_ucs_gbk_80_map ] = {
     57   { 0x20AC, 0x0080 },
     58 };
     59 
     60 static wc_uint16 ${name}_ucs_map[ 0x7E * 0xBE - 0x5E * 0x5E + 0x0A + 0x16 + 0x06 ] = {
     61 EOF
     62 
     63 for $ub (0x81 .. 0xA0) {
     64   for $lb (0x40 .. 0x7E, 0x80 .. 0xFE) {
     65     $_ = ($ub << 8) + $lb;
     66     printf OUT "  0x%.4X,\t/* 0x%.4X */\n", $to_ucs{$_}, $_;
     67   }
     68 }
     69 for $ub (0xA1 .. 0xFE) {
     70   for $lb (0x40 .. 0x7E, 0x80 .. 0xA0) {
     71     $_ = ($ub << 8) + $lb;
     72     printf OUT "  0x%.4X,\t/* 0x%.4X */\n", $to_ucs{$_}, $_;
     73   }
     74   if ($ub == 0xA2) {
     75   for $lb (0xA1 .. 0xAA) {	# 0x0A
     76     $_ = ($ub << 8) + $lb;
     77     printf OUT "  0x%.4X,\t/* 0x%.4X */\n", $to_ucs{$_}, $_;
     78   }
     79   } elsif ($ub == 0xA6) {
     80   for $lb (0xE0 .. 0xF5) {	# 0x16
     81     $_ = ($ub << 8) + $lb;
     82     printf OUT "  0x%.4X,\t/* 0x%.4X */\n", $to_ucs{$_}, $_;
     83   }
     84   } elsif ($ub == 0xA8) {
     85   for $lb (0xBB .. 0xC0) {	# 0x06
     86     $_ = ($ub << 8) + $lb;
     87     printf OUT "  0x%.4X,\t/* 0x%.4X */\n", $to_ucs{$_}, $_;
     88   }
     89   }
     90 }
     91 
     92 @ucs = sort { $a <=> $b } keys %from_ucs;
     93 $nucs = @ucs + 0;
     94 
     95 print OUT <<EOF;
     96 };
     97 
     98 #define N_ucs_${name}_map $nucs
     99 
    100 static wc_map ucs_${name}_map[ N_ucs_${name}_map ] = {
    101 EOF
    102 for(@ucs) {
    103   printf OUT "  { 0x%.4X, 0x%.4X },\n", $_, $from_ucs{$_};
    104 }
    105 
    106 print OUT <<EOF;
    107 };
    108 EOF
    109 
    110 close(MAP);
    111 }
    112 
    113 __END__
    114 gbk		VENDORS/MICSFT/WINDOWS/CP936.TXT	GBK/CP936 (Chinese)