mdw@git.distorted.org.uk Git - sgt/charset/blob - sbcsgen.pl

   1 #!/usr/bin/env perl -w
   2
   3 # This script generates sbcsdat.c (the data for all the SBCSes) from its
   4 # source form sbcs.dat.
   5
   6 $infile = "sbcs.dat";
   7 $infile = shift @ARGV if defined $ARGV[0];
   8 $outfile = "sbcsdat.c";
   9 $outfile = shift @ARGV if defined $ARGV[0];
  10
  11 open FOO, $infile;
  12 open BAR, ">$outfile";
  13 select BAR;
  14
  15 print "/*\n";
  16 print " * sbcsdat.c - data definitions for single-byte character sets.\n";
  17 print " *\n";
  18 print " * Generated by sbcsgen.pl from sbcs.dat.\n";
  19 print " * You should edit those files rather than editing this one.\n";
  20 print " */\n";
  21 print "\n";
  22 print "#ifndef ENUM_CHARSETS\n";
  23 print "\n";
  24 print "#include \"charset.h\"\n";
  25 print "#include \"internal.h\"\n";
  26 print "\n";
  27
  28 my $charsetname = undef;
  29 my @vals = ();
  30
  31 my @charsetnames = ();
  32 my @sortpriority = ();
  33
  34 while (<FOO>) {
  35     chomp;
  36     if (/^charset (.*)$/) {
  37         $charsetname = $1;
  38         @vals = ();
  39         @sortpriority = map { 0 } 0..255;
  40     } elsif (/^sortpriority ([^-]*)-([^-]*) (.*)$/) {
  41         for ($i = hex $1; $i <= hex $2; $i++) {
  42             $sortpriority[$i] += $3;
  43         }
  44     } elsif (/^[0-9a-fA-FX]/) {
  45         push @vals, map { $_ eq "XXXX" ? -1 : hex $_ } split / +/, $_;
  46         if (scalar @vals > 256) {
  47             die "$infile:$.: charset $charsetname has more than 256 values\n";
  48         } elsif (scalar @vals == 256) {
  49             &outcharset($charsetname, \@vals, \@sortpriority);
  50             push @charsetnames, $charsetname;
  51             $charsetname = undef;
  52             @vals = ();
  53             @sortpriority = map { 0 } 0..255;
  54         }
  55     }
  56 }
  57
  58 print "#else /* ENUM_CHARSETS */\n";
  59 print "\n";
  60
  61 foreach $i (@charsetnames) {
  62     print "ENUM_CHARSET($i)\n";
  63 }
  64
  65 print "\n";
  66 print "#endif /* ENUM_CHARSETS */\n";
  67
  68 sub outcharset($$$) {
  69     my ($name, $vals, $sortpriority) = @_;
  70     my ($prefix, $i, @sorted);
  71
  72     print "static const sbcs_data data_$name = {\n";
  73     print "    {\n";
  74     $prefix = "    ";
  75     @sorted = ();
  76     for ($i = 0; $i < 256; $i++) {
  77         if ($vals->[$i] < 0) {
  78             printf "%sERROR ", $prefix;
  79         } else {
  80             printf "%s0x%04x", $prefix, $vals->[$i];
  81             die "ooh? $i\n" unless defined $sortpriority->[$i];
  82             push @sorted, [$i, $vals->[$i], 0+$sortpriority->[$i]];
  83         }
  84         if ($i % 8 == 7) {
  85             $prefix = ",\n    ";
  86         } else {
  87             $prefix = ", ";
  88         }
  89     }
  90     print "\n    },\n    {\n";
  91     @sorted = sort { $a->[1] == $b->[1] ?
  92                      $b->[2] <=> $a->[2] :
  93                      $a->[1] <=> $b->[1] } @sorted;
  94     $prefix = "    ";
  95     $uval = -1;
  96     for ($i = $j = 0; $i < scalar @sorted; $i++) {
  97         next if ($uval == $sorted[$i]->[1]); # low-priority alternative
  98         $uval = $sorted[$i]->[1];
  99         printf "%s0x%02x", $prefix, $sorted[$i]->[0];
 100         if ($j % 8 == 7) {
 101             $prefix = ",\n    ";
 102         } else {
 103             $prefix = ", ";
 104         }
 105         $j++;
 106     }
 107     printf "\n    },\n    %d\n", $j;
 108     print "};\n";
 109     print "const charset_spec charset_$name = {\n" .
 110           "    $name, read_sbcs, write_sbcs, &data_$name\n};\n\n";
 111 }