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 "#define NTRANSLIT %d\n", $#froms + 1;
  57
  58 printf "static const uint32_t translit_from_idx[] =\n{\n  ";
  59 $col = 2;
  60 $total = 0;
  61 for ($cnt = 0; $cnt <= $#fromlens; ++$cnt) {
  62   if ($cnt != 0) {
  63     if ($col + 7 >= 79) {
  64       printf(",\n  ");
  65       $col = 2;
  66     } else {
  67       printf(", ");
  68       $col += 2;
  69     }
  70   }
  71   printf("%4d", $total);
  72   $total += $fromlens[$cnt] + 1;
  73   $col += 4;
  74 }
  75 printf("\n};\n");
  76
  77 printf "static const wchar_t translit_from_tbl[] =\n ";
  78 $col = 1;
  79 for ($cnt = 0; $cnt <= $#froms; ++$cnt) {
  80   if ($cnt != 0) {
  81     if ($col + 6 >= 79) {
  82       printf("\n ");
  83       $col = 1;
  84     }
  85     printf(" L\"\\0\"");
  86     $col += 6;
  87   }
  88   if ($col > 2 && $col + length($froms[$cnt]) + 4 >= 79) {
  89     printf("\n  ");
  90     $col = 2;
  91   } else {
  92     printf(" ");
  93     ++$col;
  94   }
  95   printf("L\"$froms[$cnt]\"");
  96   $col += length($froms[$cnt]) + 3;
  97 }
  98 printf(";\n");
  99
 100 printf "static const uint32_t translit_to_idx[] =\n{\n  ";
 101 $col = 2;
 102 $total = 0;
 103 for ($cnt = 0; $cnt <= $#tolens; ++$cnt) {
 104   if ($cnt != 0) {
 105     if ($col + 7 >= 79) {
 106       printf(",\n  ");
 107       $col = 2;
 108     } else {
 109       printf(", ");
 110       $col += 2;
 111     }
 112   }
 113   printf("%4d", $total);
 114   $total += $tolens[$cnt] + 2;
 115   $col += 4;
 116 }
 117 printf("\n};\n");
 118
 119 printf "static const wchar_t translit_to_tbl[] =\n ";
 120 $col = 1;
 121 for ($cnt = 0; $cnt <= $#tos; ++$cnt) {
 122   if ($cnt != 0) {
 123     if ($col + 6 >= 79) {
 124       printf("\n ");
 125       $col = 1;
 126     }
 127     printf(" L\"\\0\"");
 128     $col += 6;
 129   }
 130   if ($col > 2 && $col + length($tos[$cnt]) + 6 >= 79) {
 131     printf("\n  ");
 132     $col = 2;
 133   } else {
 134     printf(" ");
 135     ++$col;
 136   }
 137   printf("L\"$tos[$cnt]\\0\"");
 138   $col += length($tos[$cnt]) + 5;
 139 }
 140 printf(";\n");
 141
 142 exit 0;