libjava/scripts/encodings.pl

   1 # encodings.pl - Download IANA text and compute alias list.
   2 # Assumes you are running this program from gnu/gcj/convert/.
   3 # Output suitable for direct inclusion in IOConverter.java.
   4
   5 # Map IANA canonical names onto our canonical names.
   6 %map = (
   7         'ANSI_X3.4-1968' => 'ASCII',
   8         'ISO_8859-1:1987' => '8859_1',
   9         'UTF-8' => 'UTF8',
  10         'Shift_JIS' => 'SJIS',
  11         'Extended_UNIX_Code_Packed_Format_for_Japanese' => 'EUCJIS'
  12         );
  13
  14 if ($ARGV[0] eq '')
  15 {
  16     $file = 'character-sets';
  17     if (! -f $file)
  18     {
  19         # Too painful to figure out how to get Perl to do it.
  20         system 'wget -o .wget-log http://www.iana.org/assignments/character-sets';
  21     }
  22 }
  23 else
  24 {
  25     $file = $ARGV[0];
  26 }
  27
  28 open (INPUT, "< $file") || die "couldn't open $file: $!";
  29
  30 $body = 0;
  31 $current = '';
  32 while (<INPUT>)
  33 {
  34     chop;
  35     $body = 1 if /^Name:/;
  36     next unless $body;
  37
  38     if (/^$/)
  39     {
  40         $current = '';
  41         next;
  42     }
  43
  44     ($type, $name) = split (/\s+/);
  45     # Encoding names are case-insensitive.  We do all processing on
  46     # the lower-case form.
  47     my $lower = lc ($name);
  48     if ($type eq 'Name:')
  49     {
  50         $current = $map{$name};
  51         if ($current)
  52         {
  53             print "    hash.put (\"$lower\", \"$current\");\n";
  54         }
  55     }
  56     elsif ($type eq 'Alias:')
  57     {
  58         # The IANA list has some ugliness.
  59         if ($name ne '' && $name ne 'NONE' && $current)
  60         {
  61             print "    hash.put (\"$lower\", \"$current\");\n";
  62         }
  63     }
  64 }
  65
  66 close (INPUT);