share/format.pl

   1 #!/usr/bin/env perl
   2 #
   3 #  Format the dictionaries according to a standard scheme.
   4 #
   5 #  Usage: cat dictionary | ./format.pl > new
   6 #
   7 #  We don't over-write the dictionaries in place, so that the process
   8 #  can be double-checked by hand.
   9 #
  10 #  This is a bit of a hack.
  11 #
  12 #  FIXME: get lengths from variables, rather than hard-coding.
  13 #
  14 #  $Id$
  15 #
  16
  17 $begin_vendor = 0;
  18 $blank = 0;
  19
  20 while (@ARGV) {
  21     $filename = shift;
  22
  23     open FILE, "<$filename" or die "Failed to open $filename: $!\n";
  24
  25     @output = ();
  26
  27     while (<FILE>) {
  28         #
  29         #  Clear out trailing whitespace
  30         #
  31         s/[ \t]+$//;
  32
  33         #
  34         #  And CR's
  35         #
  36         s/\r//g;
  37
  38         #
  39         #  Suppress multiple blank lines
  40         #
  41         if (/^\s+$/) {
  42             next if ($blank == 1);
  43             $blank = 1;
  44             push @output, "\n";
  45             next;
  46         }
  47         $blank = 0;
  48
  49         #
  50         #  Remember the vendor
  51         #
  52         if (/^VENDOR\s+([\w-]+)\s+(\w+)(.*)/) {
  53             $name=$1;
  54             $len = length $name;
  55             if ($len < 32) {
  56                 $lenx = 32 - $len;
  57                 $lenx += 7;             # round up
  58                 $lenx /= 8;
  59                 $lenx = int $lenx;
  60                 $tabs = "\t" x $lenx;
  61             } else {
  62                 $tabs = " ";
  63             }
  64             push @output, "VENDOR\t\t$name$tabs$2$3\n";
  65             $vendor = $name;
  66             next;
  67         }
  68
  69         #
  70         #  Remember if we did begin-vendor.
  71         #
  72         if (/^BEGIN-VENDOR\s+([\w-]+)/) {
  73             $begin_vendor = 1;
  74             if (!defined $vendor) {
  75                 $vendor = $1;
  76             } elsif ($vendor ne $1) {
  77                 # do something smart
  78             }
  79
  80             push @output, "BEGIN-VENDOR\t$vendor\n";
  81             next;
  82         }
  83
  84         #
  85         #  Get attribute.
  86         #
  87         if (/^ATTRIBUTE\s+([\w-]+)\s+(\w+)\s+(\w+)(.*)/) {
  88             $name=$1;
  89             $len = length $name;
  90             if ($len < 40) {
  91                 $lenx = 40 - $len;
  92                 $lenx += 7;             # round up
  93                 $lenx /= 8;
  94                 $lenx = int $lenx;
  95                 $tabs = "\t" x $lenx;
  96                 if ($tabs eq "") {
  97                     $tabs = " ";
  98                 }
  99             } else {
 100                 $tabs = " ";
 101             }
 102
 103             $value = $2;
 104             $type = $3;
 105             $stuff = $4;
 106
 107             #
 108             #  See if it's old format, with the vendor at the end of
 109             #  the line.  If so, make it the new format.
 110             #
 111             if ($stuff =~ /$vendor/) {
 112                 if ($begin_vendor == 0) {
 113                     push @output, "BEGIN-VENDOR\t$vendor\n\n";
 114                     $begin_vendor = 1;
 115                 }
 116                 $stuff =~ s/$vendor//;
 117                 $stuff =~ s/\s+$//;
 118             }
 119
 120             push @output, "ATTRIBUTE\t$name$tabs$value\t$type$stuff\n";
 121             next;
 122         }
 123
 124         #
 125         #  Values.
 126         #
 127         if (/^VALUE\s+([\w-]+)\s+([\w-\/,.]+)\s+(\w+)(.*)/) {
 128             $attr=$1;
 129             $len = length $attr;
 130             if ($len < 32) {
 131                 $lenx = 32 - $len;
 132                 $lenx += 7;             # round up
 133                 $lenx /= 8;
 134                 $lenx = int $lenx;
 135                 $tabsa = "\t" x $lenx;
 136                 if ($tabsa eq "") {
 137                     $tabsa = " ";
 138                     $len += 1;
 139                 } else {
 140                     $len -= $len % 8;
 141                     $len += 8 * length $tabsa;
 142                 }
 143             } else {
 144                 $tabsa = " ";
 145                 $len += 1;
 146             }
 147
 148             #
 149             #  For the code below, we assume that the attribute lengths
 150             #
 151             if ($len < 32) {
 152                 $lena = 0;
 153             } else {
 154                 $lena = $len - 32;
 155             }
 156
 157             $name = $2;
 158             $len = length $name;
 159             if ($len < 24) {
 160                 $lenx = 24 - $lena - $len;
 161                 $lenx += 7;             # round up
 162                 $lenx /= 8;
 163                 $lenx = int $lenx;
 164                 $tabsn = "\t" x $lenx;
 165                 if ($tabsn eq "") {
 166                     $tabsn = " ";
 167                 }
 168             } else {
 169                 $tabsn = " ";
 170             }
 171
 172             push @output, "VALUE\t$attr$tabsa$name$tabsn$3$4\n";
 173             next;
 174         }
 175
 176         #
 177         #  Remember if we did this.
 178         #
 179         if (/^END-VENDOR/) {
 180             $begin_vendor = 0;
 181         }
 182
 183         #
 184         #  Everything else gets dumped out as-is.
 185         #
 186         push @output, $_;
 187     }
 188
 189 #
 190 #  If we changed the format, print the end vendor, too.
 191 #
 192     if ($begin_vendor) {
 193         push @output, "\nEND-VENDOR\t$vendor\n";
 194     }
 195
 196     close FILE;
 197
 198     open FILE, ">$filename" or die "Failed to open $filename: $!\n";
 199
 200     print FILE @output;
 201
 202     close FILE;
 203 }