mk_cns_ucs_map.pl (1944B)
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_ucs1 = (); 20 @to_ucs2 = (); 21 %from_ucs = (); 22 open(MAP, "< $map"); 23 open(OUT, "> ${name}_ucs.map"); 24 while(<MAP>) { 25 /^#/ && next; 26 s/#.*//; 27 (($i, $u) = split(" ")) || next; 28 $i =~ s/0x(.)/0x/; 29 $p = $1; 30 $i = hex($i); 31 $u = hex($u); 32 if ($p == 1) { 33 $to_ucs1[$i] = $u; 34 $from_ucs{$u} = $i; 35 } elsif ($p == 2) { 36 $to_ucs2[$i] = $u; 37 $from_ucs{$u} = $i + 0x8000; 38 } 39 } 40 41 # print OUT <<EOF; 42 # /* 43 # These conversion tables between $code and 44 # Unicode were made from 45 # 46 # ftp://ftp.unicode.org/Public/MAPPINGS/$map. 47 # */ 48 print OUT <<EOF; 49 /* $code */ 50 51 static wc_uint16 ${name}1_ucs_map[ 0x5E * 0x5E ] = { 52 EOF 53 54 for $i (0x21 .. 0x7E) { 55 for $j (0x21 .. 0x7E) { 56 $_ = $i * 0x100 + $j; 57 $u = $to_ucs1[$_]; 58 if ($u) { 59 printf OUT " 0x%.4X,", $u; 60 } else { 61 print OUT " 0,\t"; 62 } 63 printf OUT "\t/* 0x%.4X */\n", $_; 64 } 65 } 66 67 print OUT <<EOF; 68 }; 69 70 static wc_uint16 ${name}2_ucs_map[ 0x5E * 0x5E ] = { 71 EOF 72 73 for $i (0x21 .. 0x7E) { 74 for $j (0x21 .. 0x7E) { 75 $_ = $i * 0x100 + $j; 76 $u = $to_ucs2[$_]; 77 if ($u) { 78 printf OUT " 0x%.4X,", $u; 79 } else { 80 print OUT " 0,\t"; 81 } 82 printf OUT "\t/* 0x%.4X */\n", $_; 83 } 84 } 85 86 @ucs = sort { $a <=> $b } keys %from_ucs; 87 $nucs = @ucs + 0; 88 89 print OUT <<EOF; 90 }; 91 92 #define N_ucs_${name}_map $nucs 93 /* 94 UCS-2 CNS 11643-1 95 UCS-2 CNS 11643-2 | 0x8000 96 */ 97 static wc_map ucs_${name}_map[ N_ucs_${name}_map ] = { 98 EOF 99 for(@ucs) { 100 printf OUT " { 0x%.4X, 0x%.4X },", $_, $from_ucs{$_}; 101 if ($from_ucs{$_} & 0x8000) { 102 print OUT "\t/* CNS 11643-2 */\n"; 103 } else { 104 print OUT "\t/* CNS 11643-1 */\n"; 105 } 106 } 107 108 print OUT <<EOF; 109 }; 110 EOF 111 112 close(MAP); 113 } 114 115 __END__ 116 cns11643 EASTASIA/OTHER/CNS11643.TXT CNS 11643 (Chinese Taiwan)