w3m

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

mk_jis_ucs_map.pl (4251B)


      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 $name0 = "jisx0208x0212x0213";
     13 # print OUT <<EOF;
     14 # /*
     15 #   These conversion tables between $code and
     16 #   Unicode were made from
     17 # 
     18 #     ftp://ftp.unicode.org/Public/MAPPINGS/$MAP{$NAME[0]},
     19 #     ftp://ftp.unicode.org/Public/MAPPINGS/$MAP{$NAME[1]}.
     20 # 
     21 # */
     22 # EOF
     23 # Unicode(CP932)				Unicode(JIS X 0208)
     24 %from_ucs = (
     25   0x00A5, 0x216F, # YEN SIGN
     26   0x00B5, 0x264C, # MICRO SIGN
     27   0xFF3C, 0x2140, # FULLWIDTH REVERSE SOLIDUS	0x005C REVERSE SOLIDUS
     28   0xFF5E, 0x2141, # FULLWIDTH TILDE		0x301C WAVE DASH
     29   0x2225, 0x2142, # PARALLEL TO			0x2016 DOUBLE VERTICAL LINE
     30   0xFF0D, 0x215D, # FULLWIDTH HYPHEN-MINUS	0x2212 MINUS SIGN
     31   0xFFE0, 0x2171, # FULLWIDTH CENT SIGN		0x00A2 CENT SIGN
     32   0xFFE1, 0x2172, # FULLWIDTH POUND SIGN	0x00A3 POUND SIGN
     33   0xFFE2, 0x224C, # FULLWIDTH NOT SIGN		0x00AC NOT SIGN
     34 );
     35 for(keys %from_ucs) {
     36   ($_ == 0x00A5) && next;
     37   ($_ == 0x00B5) && next;
     38   $to_ucs[$from_ucs{$_}] = $_;
     39 }
     40 
     41 open(OUT, "> ${name0}_ucs.map");
     42 foreach $name (@NAME) {
     43 
     44 $code = $CODE{$name};
     45 $map = $MAP{$name};
     46 
     47 print "$name\t$map\t$code\n";
     48 
     49 open(MAP, "< $map");
     50 while(<MAP>) {
     51   /^#/ && next;
     52   s/#.*//;
     53   if ($name =~ /0208/) {
     54     (($s, $i, $u) = split(" ")) || next;
     55     $i = hex($i);
     56     $u = hex($u);
     57     if ($u == 0x5C) {
     58 	$u = 0xFF3C;
     59     }
     60     $to_ucs[$i] || ($to_ucs[$i] = $u);
     61     $to_ucs_jis[$i] = 0;
     62     $from_ucs{$u} = $i;
     63   } elsif ($name =~ /0212/) {
     64     (($i, $u) = split(" ")) || next;
     65     $i = hex($i);
     66     $u = hex($u);
     67     if ($u == 0x7E) {
     68 	$u = 0xFF5E;
     69     }
     70     $to_ucs2[$i] = $u;
     71     $to_ucs2_jis[$i] = 0;
     72     $from_ucs2{$u} = $i;
     73   } else {
     74     /^\d/ || next;
     75     (($p,$i,$e,$s,$u) = split(" ")) || next;
     76     $i =~ s/j-/0x/;
     77     $u =~ s/u-/0x/;
     78     $i = hex($i);
     79     $u = hex($u);
     80     if ($u == 0xffff) {
     81 	$u = 0;
     82     }
     83     if ($u == 0x7E) {
     84 	$u = 0xFF5E;
     85     }
     86     if ($p =~ /^1/) {
     87       $to_ucs[$i] && next;
     88       $to_ucs[$i] = $u;
     89       $to_ucs_jis[$i] = 1;
     90       $from_ucs3{$u} = $i;
     91     } elsif ($p =~ /^2/) {
     92       $to_ucs2[$i] = $u;
     93       $to_ucs2_jis[$i] = 1;
     94       $from_ucs4{$u} = $i;
     95     }
     96   }
     97 }
     98 
     99 }
    100 
    101 print OUT <<EOF;
    102 /* JIS X 0208, JIS X 0212, JIS X 0213 (Japanese) */
    103 
    104 static wc_uint16 jisx0208x02131_ucs_map[ 0x5E * 0x5E ] = {
    105 EOF
    106 
    107 for $i (0x21 .. 0x7E) {
    108 for $j (0x21 .. 0x7E) {
    109   $_ = $i * 0x100 + $j;
    110   $u = $to_ucs[$_];
    111   if ($u) {
    112     printf OUT " 0x%.4X,", $u;
    113   } else {
    114     print OUT " 0,\t";
    115   }
    116   printf OUT "\t/* %s 0x%.4X */\n", $to_ucs_jis[$_] ? "JIS X 0213-1" : "JIS X 0208", $_;
    117 }
    118 }
    119 
    120 print OUT <<EOF;
    121 };
    122 
    123 static wc_uint16 jisx0212x02132_ucs_map[ 0x5E * 0x5E ] = {
    124 EOF
    125 
    126 for $i (0x21 .. 0x7E) {
    127 for $j (0x21 .. 0x7E) {
    128   $_ = $i * 0x100 + $j;
    129   $u = $to_ucs2[$_];
    130   if ($u) {
    131     printf OUT " 0x%.4X,", $u;
    132   } else {
    133     print OUT " 0,\t";
    134   }
    135   printf OUT "\t/* %s 0x%.4X */\n", $to_ucs2_jis[$_] ? "JIS X 0213-2" : "JIS X 0212", $_;
    136 }
    137 }
    138 
    139 print OUT <<EOF;
    140 };
    141 EOF
    142 
    143 @ucs = sort { $a <=> $b } keys %from_ucs;
    144 $nucs = @ucs + 0;
    145 
    146 print OUT <<EOF;
    147 
    148 #define N_ucs_jisx0208_map $nucs
    149 
    150 static wc_map ucs_jisx0208_map[ N_ucs_jisx0208_map ] = {
    151 EOF
    152 for(@ucs) {
    153   $_ || next;
    154   printf OUT "  { 0x%.4X, 0x%.4X },\n", $_, $from_ucs{$_};
    155 }
    156 
    157 @ucs = sort { $a <=> $b } keys %from_ucs2;
    158 $nucs = @ucs + 0;
    159 
    160 print OUT <<EOF;
    161 };
    162 
    163 #define N_ucs_jisx0212_map $nucs
    164 
    165 static wc_map ucs_jisx0212_map[ N_ucs_jisx0212_map ] = {
    166 EOF
    167 for(@ucs) {
    168   $_ || next;
    169   printf OUT "  { 0x%.4X, 0x%.4X },\n", $_, $from_ucs2{$_};
    170 }
    171 
    172 @ucs = sort { $a <=> $b } keys %from_ucs3;
    173 $nucs = @ucs + 0;
    174 
    175 print OUT <<EOF;
    176 };
    177 
    178 #define N_ucs_jisx02131_map $nucs
    179 
    180 static wc_map ucs_jisx02131_map[ N_ucs_jisx02131_map ] = {
    181 EOF
    182 for(@ucs) {
    183   $_ || next;
    184   printf OUT "  { 0x%.4X, 0x%.4X },\n", $_, $from_ucs3{$_};
    185 }
    186 
    187 @ucs = sort { $a <=> $b } keys %from_ucs4;
    188 $nucs = @ucs + 0;
    189 
    190 print OUT <<EOF;
    191 };
    192 
    193 #define N_ucs_jisx02132_map $nucs
    194 
    195 static wc_map ucs_jisx02132_map[ N_ucs_jisx02132_map ] = {
    196 EOF
    197 for(@ucs) {
    198   $_ || next;
    199   printf OUT "  { 0x%.4X, 0x%.4X },\n", $_, $from_ucs4{$_};
    200 }
    201 
    202 print OUT <<EOF;
    203 };
    204 EOF
    205 
    206 close(MAP);
    207 
    208 __END__
    209 jisx0208	EASTASIA/JIS/JIS0208.TXT	JIS X 0208 (Japanese)
    210 jisx0212	EASTASIA/JIS/JIS0212.TXT	JIS X 0212 (Japanese)
    211 jisx0213	jisx0213code.txt		JIS X 0213 (Japanese)
    212