locale/gen-translit.pl

   1 #!/usr/bin/perl -w
   2 open F, "cat C-translit.h.in | gcc -E - |" || die "Cannot preprocess input file";
   3
   4
   5 sub cstrlen {
   6   my($str) = @_;
   7   my($len) = length($str);
   8   my($cnt);
   9   my($res) = 0;
  10
  11   for ($cnt = 0; $cnt < $len; ++$cnt) {
  12     if (substr($str, $cnt, 1) eq '\\') {
  13       # Recognize the escape sequence.
  14       if (substr($str, $cnt + 1, 1) eq 'x') {
  15         my($inner);
  16         for ($inner = $cnt + 2; $inner < $len && $inner < $cnt + 10; ++$inner) {
  17           my($ch) = substr($str, $inner, 1);
  18           next if (($ch ge '0' && $ch le '9')
  19                    || ($ch ge 'a' && $ch le 'f')
  20                    || ($ch ge 'A' && $ch le 'F'));
  21           last;
  22         }
  23         $cnt = $inner;
  24         ++$res;
  25       } else {
  26         die "invalid input" if ($cnt + 1 >= $len);
  27         ++$res;
  28         ++$cnt;
  29       }
  30     } else {
  31       ++$res;
  32     }
  33   }
  34
  35   return $res;
  36 }
  37
  38 while (<F>) {
  39   next if (/^#/);
  40   next if (/^[  ]*$/);
  41   chop;
  42
  43   if (/"([^\"]*)"[      ]*"(.*)"/) {
  44     my($from) = $1;
  45     my($to) = $2;
  46     my($fromlen) = cstrlen($from);
  47     my($tolen) = cstrlen($to);
  48
  49     push(@froms, $from);
  50     push(@fromlens, $fromlen);
  51     push(@tos, $to);
  52     push(@tolens, $tolen);
  53   }
  54 }
  55
  56 printf "#include <stdint.h>\n";
  57
  58 printf "#define NTRANSLIT %d\n", $#froms + 1;
  59
  60 printf "static const uint32_t translit_from_idx[] =\n{\n  ";
  61 $col = 2;
  62 $total = 0;
  63 for ($cnt = 0; $cnt <= $#fromlens; ++$cnt) {
  64   if ($cnt != 0) {
  65     if ($col + 7 >= 79) {
  66       printf(",\n  ");
  67       $col = 2;
  68     } else {
  69       printf(", ");
  70       $col += 2;
  71     }
  72   }
  73   printf("%4d", $total);
  74   $total += $fromlens[$cnt] + 1;
  75   $col += 4;
  76 }
  77 printf("\n};\n");
  78
  79 printf "static const wchar_t translit_from_tbl[] =\n ";
  80 $col = 1;
  81 for ($cnt = 0; $cnt <= $#froms; ++$cnt) {
  82   if ($cnt != 0) {
  83     if ($col + 6 >= 79) {
  84       printf("\n ");
  85       $col = 1;
  86     }
  87     printf(" L\"\\0\"");
  88     $col += 6;
  89   }
  90   if ($col > 2 && $col + length($froms[$cnt]) + 4 >= 79) {
  91     printf("\n  ");
  92     $col = 2;
  93   } else {
  94     printf(" ");
  95     ++$col;
  96   }
  97   printf("L\"$froms[$cnt]\"");
  98   $col += length($froms[$cnt]) + 3;
  99 }
 100 printf(";\n");
 101
 102 printf "static const uint32_t translit_to_idx[] =\n{\n  ";
 103 $col = 2;
 104 $total = 0;
 105 for ($cnt = 0; $cnt <= $#tolens; ++$cnt) {
 106   if ($cnt != 0) {
 107     if ($col + 7 >= 79) {
 108       printf(",\n  ");
 109       $col = 2;
 110     } else {
 111       printf(", ");
 112       $col += 2;
 113     }
 114   }
 115   printf("%4d", $total);
 116   $total += $tolens[$cnt] + 2;
 117   $col += 4;
 118 }
 119 printf("\n};\n");
 120
 121 printf "static const wchar_t translit_to_tbl[] =\n ";
 122 $col = 1;
 123 for ($cnt = 0; $cnt <= $#tos; ++$cnt) {
 124   if ($cnt != 0) {
 125     if ($col + 6 >= 79) {
 126       printf("\n ");
 127       $col = 1;
 128     }
 129     printf(" L\"\\0\"");
 130     $col += 6;
 131   }
 132   if ($col > 2 && $col + length($tos[$cnt]) + 6 >= 79) {
 133     printf("\n  ");
 134     $col = 2;
 135   } else {
 136     printf(" ");
 137     ++$col;
 138   }
 139   printf("%s", "L\"$tos[$cnt]\\0\"");
 140   $col += length($tos[$cnt]) + 5;
 141 }
 142 printf(";\n");
 143
 144 exit 0;