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 $outheader = "sbcsdat.h";
  11 $outheader = shift @ARGV if defined $ARGV[0];
  12
  13 open FOO, $infile;
  14 open BAR, ">$outfile";
  15 select BAR;
  16
  17 print "/*\n";
  18 print " * sbcsdat.c - data definitions for single-byte character sets.\n";
  19 print " *\n";
  20 print " * Generated by sbcsgen.pl from sbcs.dat.\n";
  21 print " * You should edit those files rather than editing this one.\n";
  22 print " */\n";
  23 print "\n";
  24 print "#ifndef ENUM_CHARSETS\n";
  25 print "\n";
  26 print "#include \"charset.h\"\n";
  27 print "#include \"internal.h\"\n";
  28 print "\n";
  29
  30 my $charsetname = undef;
  31 my @vals = ();
  32
  33 my @charsetnames = ();
  34 my @sortpriority = ();
  35
  36 while (<FOO>) {
  37     chomp;
  38     if (/^charset (.*)$/) {
  39         $charsetname = $1;
  40         @vals = ();
  41         @sortpriority = map { 0 } 0..255;
  42     } elsif (/^sortpriority ([^-]*)-([^-]*) (.*)$/) {
  43         for ($i = hex $1; $i <= hex $2; $i++) {
  44             $sortpriority[$i] += $3;
  45         }
  46     } elsif (/^[0-9a-fA-FX]/) {
  47         push @vals, map { $_ eq "XXXX" ? -1 : hex $_ } split / +/, $_;
  48         if (scalar @vals > 256) {
  49             die "$infile:$.: charset $charsetname has more than 256 values\n";
  50         } elsif (scalar @vals == 256) {
  51             &outcharset($charsetname, \@vals, \@sortpriority);
  52             push @charsetnames, $charsetname;
  53             $charsetname = undef;
  54             @vals = ();
  55             @sortpriority = map { 0 } 0..255;
  56         }
  57     }
  58 }
  59
  60 print "#else /* ENUM_CHARSETS */\n";
  61 print "\n";
  62
  63 foreach $i (@charsetnames) {
  64     print "ENUM_CHARSET($i)\n";
  65 }
  66
  67 print "\n";
  68 print "#endif /* ENUM_CHARSETS */\n";
  69
  70 close BAR;
  71
  72 open BAR, ">$outheader";
  73 select BAR;
  74
  75 print "/*\n";
  76 print " * sbcsdat.h - header file for SBCS data structures.\n";
  77 print " *\n";
  78 print " * Generated by sbcsgen.pl from sbcs.dat.\n";
  79 print " * You should edit those files rather than editing this one.\n";
  80 print " */\n";
  81 print "\n";
  82 print "#ifndef charset_sbcsdat_h\n";
  83 print "#define charset_sbcsdat_h\n";
  84 print "\n";
  85 print "#include \"charset.h\"\n";
  86 print "#include \"internal.h\"\n";
  87 print "\n";
  88 foreach $i (@charsetnames) {
  89     print "extern const sbcs_data sbcsdata_$i;\n";
  90 }
  91 print "\n";
  92 print "#endif /* charset_sbcsdat_h */\n";
  93
  94 close BAR;
  95
  96 sub outcharset($$$) {
  97     my ($name, $vals, $sortpriority) = @_;
  98     my ($prefix, $i, @sorted);
  99
 100     print "const sbcs_data sbcsdata_$name = {\n";
 101     print "    {\n";
 102     $prefix = "    ";
 103     @sorted = ();
 104     for ($i = 0; $i < 256; $i++) {
 105         if ($vals->[$i] < 0) {
 106             printf "%sERROR ", $prefix;
 107         } else {
 108             printf "%s0x%04x", $prefix, $vals->[$i];
 109             die "ooh? $i\n" unless defined $sortpriority->[$i];
 110             push @sorted, [$i, $vals->[$i], 0+$sortpriority->[$i]];
 111         }
 112         if ($i % 8 == 7) {
 113             $prefix = ",\n    ";
 114         } else {
 115             $prefix = ", ";
 116         }
 117     }
 118     print "\n    },\n    {\n";
 119     @sorted = sort { $a->[1] == $b->[1] ?
 120                      $b->[2] <=> $a->[2] :
 121                      $a->[1] <=> $b->[1] } @sorted;
 122     $prefix = "    ";
 123     $uval = -1;
 124     for ($i = $j = 0; $i < scalar @sorted; $i++) {
 125         next if ($uval == $sorted[$i]->[1]); # low-priority alternative
 126         $uval = $sorted[$i]->[1];
 127         printf "%s0x%02x", $prefix, $sorted[$i]->[0];
 128         if ($j % 8 == 7) {
 129             $prefix = ",\n    ";
 130         } else {
 131             $prefix = ", ";
 132         }
 133         $j++;
 134     }
 135     printf "\n    },\n    %d\n", $j;
 136     print "};\n";
 137     print "const charset_spec charset_$name = {\n" .
 138           "    $name, read_sbcs, write_sbcs, &sbcsdata_$name\n};\n\n";
 139 }