tools/unimap.pl

   1 #!/usr/bin/perl
   2 #
   3 # Reads the Unicode 2.0 "unidata2.txt" file and selects encodings
   4 # for case pairs, then builds an include file "casemap.h" for the
   5 # case conversion routines.
   6
   7 use integer
   8
   9 $INFILE = "unidata2.txt";
  10 $OUT    = ">casemap.h";
  11 $TEST   = ">allcodes";
  12
  13 # Open the data file ...
  14 open INFILE     or die "Can't open input file $INFILE!\n";
  15 open OUT        or die "Can't open output file $OUT!\n";
  16 open TEST       or die "Can't open output file $OUT!\n";
  17
  18 #Initialize the upper and lower hashes
  19 %lwrtable = ();
  20 %uprtable = ();
  21 @low = ("0") x 256;
  22 @upr = ("0") x 256;
  23
  24 while ($line = <INFILE> )
  25 {
  26         # Decode the fields ...
  27         ($code, $name, $cat, $comb, $bidi,
  28          $decomp, $dec, $dig, $num, $mirror,
  29          $oldname, $comment, $upper, $lower, $title) = split /;/, $line;
  30
  31         #Get the high byte of the code
  32         $high = substr $code, 0, 2;
  33         if ($lower ne "") {
  34                 $low[hex $high] = "lblk" . $high;
  35                 $lwrtable{$code} = $lower;
  36         }
  37         if ($upper ne "") {
  38                 $upr[hex $high] = "ublk" . $high;
  39                 $uprtable{$code} = $upper;
  40         }
  41         #Write everything to the test file
  42         printf TEST "%s %s %s\n", $code,
  43                 $upper ne "" ? $upper : "0000",
  44                 $lower ne "" ? $lower : "0000";
  45
  46 }
  47 close(FILE);
  48 close TEST;
  49
  50 #Generate the header file
  51 print OUT "/*\n";
  52 print OUT " * Automatically generated file -- do not edit!\n";
  53 print OUT " * (Use tools/unimap.pl for generation)\n";
  54 print OUT " *\n";
  55 print OUT " * Mapping tables for Unicode case conversion\n";
  56 print OUT " */\n\n";
  57
  58 #Write out the non-trivial mappings
  59 for ($high = 0; $high < 256; $high++) {
  60         #Check whether the table is needed
  61         if (length $low[$high] < 6) {
  62                 next;
  63         }
  64         printf OUT "/* Lowercase mappings %02X00 - %02XFF */\n",
  65                 $high, $high;
  66         printf OUT "static const WCHAR lblk%02X[256] = {\n", $high;
  67         for ($low = 0; $low < 256; $low += 8) {
  68                 @patch = ();
  69                 for ($i = 0; $i < 8; $i++) {
  70                         $code = sprintf "%02X%02X", $high, $low + $i;
  71                         $map = $lwrtable{$code};
  72                         if ($map eq "") {
  73                                 $map = $code;
  74                         }
  75                         $patch[$i] = "0x" . $map;
  76                 }
  77                 printf OUT "\t%s, %s, %s, %s, %s, %s, %s, %s,\n",
  78                         @patch;
  79         }
  80         print OUT "};\n\n";
  81 }
  82 print OUT "static const WCHAR * const lwrtable[256] = {\n";
  83 for ($i = 0; $i < 256; $i += 8) {
  84         @patch = @low[$i+0 .. $i+7];
  85         printf OUT "\t%06s, %06s, %06s, %06s, %06s, %06s, %06s, %06s,\n",
  86                 @patch;
  87 }
  88 print OUT "};\n\n";
  89
  90 for ($high = 0; $high < 256; $high++) {
  91         #Check whether the table is needed
  92         if (length $upr[$high] < 6) {
  93                 next;
  94         }
  95         printf OUT "/* Uppercase mappings %02X00 - %02XFF */\n",
  96                 $high, $high;
  97         printf OUT "static const WCHAR ublk%02X[256] = {\n", $high;
  98         for ($low = 0; $low < 256; $low += 8) {
  99                 @patch = ();
 100                 for ($i = 0; $i < 8; $i++) {
 101                         $code = sprintf "%02X%02X", $high, $low + $i;
 102                         $map = $uprtable{$code};
 103                         if ($map eq "") {
 104                                 $map = $code;
 105                         }
 106                         $patch[$i] = "0x" . $map;
 107                 }
 108                 printf OUT "\t%s, %s, %s, %s, %s, %s, %s, %s,\n",
 109                         @patch;
 110         }
 111         print OUT "};\n\n";
 112 }
 113 print OUT "static const WCHAR * const uprtable[256] = {\n";
 114 for ($i = 0; $i < 256; $i += 8) {
 115         @patch = @upr[$i+0 .. $i+7];
 116         printf OUT "\t%06s, %06s, %06s, %06s, %06s, %06s, %06s, %06s,\n",
 117                 @patch;
 118 }
 119 print OUT "};\n\n";
 120
 121 close(OUT);