contrib/groff/tmac/hyphenex.pl

   1 #! /usr/bin/perl
   2 #
   3 #
   4 # hyphenex.pl
   5 #
   6 # This small filter converts a hyphenation exception log article for
   7 # TUGBoat to a real \hyphenation block.
   8 #
   9 # Written by Werner Lemberg <wl@gnu.org>.
  10 #
  11 # Version 1.1 (2003/04/17)
  12 #
  13 # Public domain.
  14 #
  15 #
  16 # Usage:
  17 #
  18 #   [perl] hyphenex.pl < tugboat-article > hyphenation-exceptions
  19
  20 # print header
  21 print "% Hyphenation exceptions for US English,\n";
  22 print "% based on the hyphenation exception log article in TUGBoat.\n";
  23 print "%\n";
  24 print "% This is an automatically generated file.  Do not edit!\n";
  25 print "%\n";
  26 print "% Please contact Barbara Beeton <bnb\@ams.org>\n";
  27 print "% for corrections and omissions.\n";
  28 print "\n";
  29 print "\\hyphenation{\n";
  30
  31 while (<>) {
  32   # retain only lines starting with \1 ... \6 or \tabalign
  33   next if not (m/^\\[123456]/ || m/^\\tabalign/);
  34   # remove final newline
  35   chop;
  36   # remove all TeX commands except \1 ... \6
  37   s/\\[^123456\s{]+//g;
  38   # remove all paired { ... }
  39   1 while s/{(.*?)}/\1/g;
  40   # skip lines which now have only whitespace before `&'
  41   next if m/^\s*&/;
  42   # remove comments
  43   s/%.*//;
  44   # remove trailing whitespace
  45   s/\s*$//;
  46   # remove trailing `*' (used as a marker in the document)
  47   s/\*$//;
  48   # split at whitespace
  49   @field = split(' ');
  50   if ($field[0] eq "\\1" || $field[0] eq "\\4") {
  51     print "  $field[2]\n";
  52   }
  53   elsif ($field[0] eq "\\2" || $field[0] eq "\\5") {
  54     print "  $field[2]\n";
  55     # handle multiple suffixes separated by commata
  56     @suffix_list = split(/,/, "$field[3]");
  57     foreach $suffix (@suffix_list) {
  58       print "  $field[2]$suffix\n";
  59     }
  60   }
  61   elsif ($field[0] eq "\\3" || $field[0] eq "\\6") {
  62     # handle multiple suffixes separated by commata
  63     @suffix_list = split(/,/, "$field[3],$field[4]");
  64     foreach $suffix (@suffix_list) {
  65       print "  $field[2]$suffix\n";
  66     }
  67   }
  68   else {
  69     # for `&', split at `&' with trailing whitespace
  70     @field = split(/&\s*/);
  71     print "  $field[1]\n";
  72   }
  73 }
  74
  75 # print trailer
  76 print "}\n";
  77 print "\n";
  78 print "% EOF\n";