intl/uconv/tools/unihan2cns.pl

   1 #!/usr/local/bin/perl -w
   2 use IO::File;
   3 my(%tagtofilemap);
   4 $tagtofilemap{"kCNS1986-1" } = IO::File->new("|sort> cns1986p1.txt")
   5       or die "cannot open cns1986p1.txt";
   6 $tagtofilemap{"kCNS1986-2" } = IO::File->new("|sort> cns1986p2.txt")
   7       or die "cannot open cns1986p2.txt";
   8 $tagtofilemap{"kCNS1986-E" } = IO::File->new("|sort> cns1986p14.txt")
   9       or die "cannot open cns1986p14.txt";
  10 $tagtofilemap{"kCNS1992-1" } = IO::File->new("|sort> cns1992p1.txt")
  11       or die "cannot open cns1992p1.txt";
  12 $tagtofilemap{"kCNS1992-2" } = IO::File->new("|sort> cns1992p2.txt")
  13       or die "cannot open cns1992p2.txt";
  14 $tagtofilemap{"kCNS1992-3" } = IO::File->new("|sort> cns1992p3.txt")
  15       or die "cannot open cns1992p3.txt";
  16 $tagtofilemap{"kIRG_TSource-1" } = IO::File->new("|sort> cnsIRGTp1.txt")
  17       or die "cannot open cnsIRGTp1.txt";
  18 $tagtofilemap{"kIRG_TSource-2" } = IO::File->new("|sort> cnsIRGTp2.txt")
  19       or die "cannot open cnsIRGTp2.txt";
  20 $tagtofilemap{"kIRG_TSource-3" } = IO::File->new("|sort> cnsIRGTp3.txt")
  21       or die "cannot open cnsIRGTp3.txt";
  22 $tagtofilemap{"kIRG_TSource-4" } = IO::File->new("|sort> cnsIRGTp4.txt")
  23       or die "cannot open cnsIRGTp4.txt";
  24 $tagtofilemap{"kIRG_TSource-5" } = IO::File->new("|sort> cnsIRGTp5.txt")
  25       or die "cannot open cnsIRGTp5.txt";
  26 $tagtofilemap{"kIRG_TSource-6" } = IO::File->new("|sort> cnsIRGTp6.txt")
  27       or die "cannot open cnsIRGTp6.txt";
  28 $tagtofilemap{"kIRG_TSource-7" } = IO::File->new("|sort> cnsIRGTp7.txt")
  29       or die "cannot open cnsIRGTp7.txt";
  30 $tagtofilemap{"kIRG_TSource-F" } = IO::File->new("|sort> cnsIRGTp15.txt")
  31       or die "cannot open cnsIRGTp15.txt";
  32 $tagtofilemap{"kIRG_TSource-3ExtB" } = IO::File->new("|sort> cnsIRGTp3ExtB.txt")
  33       or die "cannot open cnsIRGTp3ExtB.txt";
  34 $tagtofilemap{"kIRG_TSource-4ExtB" } = IO::File->new("|sort> cnsIRGTp4ExtB.txt")
  35       or die "cannot open cnsIRGTp4ExtB.txt";
  36 $tagtofilemap{"kIRG_TSource-5ExtB" } = IO::File->new("|sort> cnsIRGTp5ExtB.txt")
  37       or die "cannot open cnsIRGTp5ExtB.txt";
  38 $tagtofilemap{"kIRG_TSource-6ExtB" } = IO::File->new("|sort> cnsIRGTp6ExtB.txt")
  39       or die "cannot open cnsIRGTp6ExtB.txt";
  40 $tagtofilemap{"kIRG_TSource-7ExtB" } = IO::File->new("|sort> cnsIRGTp7ExtB.txt")
  41       or die "cannot open cnsIRGTp7ExtB.txt";
  42 $tagtofilemap{"kIRG_TSource-FExtB" } = IO::File->new("|sort> cnsIRGTp15ExtB.txt")
  43       or die "cannot open cnsIRGTp15ExtB.txt";
  44
  45 $nonhan =  IO::File->new("< nonhan.txt")
  46       or die "cannot open nonhan.txt";
  47
  48 while(defined($line = $nonhan->getline()))
  49 {
  50    $tagtofilemap{"kCNS1986-1"}->print($line);
  51    $tagtofilemap{"kCNS1992-1"}->print($line);
  52    $tagtofilemap{"kIRG_TSource-1"}->print($line);
  53 }
  54
  55 while(<STDIN>)
  56 {
  57   if(/^U/)
  58   {
  59     chop();
  60     ($u,$tag,$value) = split(/\t/,$_);
  61     if($tag =~ m/(kCNS|kIRG_TSource)/)
  62     {
  63          ($pnum, $cvalue) = split(/-/,$value);
  64          $tagkey = $tag . "-" . $pnum;
  65          if(length($u) > 6) {
  66            $tagkey .= "ExtB";
  67          }
  68          $fd = $tagtofilemap{$tagkey};
  69          if(length($u) > 6) {
  70            $mapping = substr($u,3,4); # trunkcate 0x2 from 0x2abcd
  71          } else {
  72            $mapping = substr($u,2,4); # trunkcate 0x from 0xabcd
  73          }
  74          $fd->print("0x" . $cvalue . "\t0x" . $mapping . "\t# <CJK>\n");
  75     }
  76   }
  77 }
  78