contrib/groff/tmac/hyphenex.pl

   1 #! /usr/bin/env 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 1.2 (2007/11/16)
  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 hyphenation exception log articles in TUGboat.\n";
  23 print "%\n";
  24 print "% Copyright 2007 TeX Users Group.\n";
  25 print "% You may freely use, modify and/or distribute this file.\n";
  26 print "%\n";
  27 print "% This is an automatically generated file.  Do not edit!\n";
  28 print "%\n";
  29 print "% Please contact the TUGboat editorial staff <tugboat\@tug.org>\n";
  30 print "% for corrections and omissions.\n";
  31 print "\n";
  32 print "\\hyphenation{\n";
  33
  34 while (<>) {
  35   # retain only lines starting with \1 ... \6 or \tabalign
  36   next if not (m/^\\[123456]/ || m/^\\tabalign/);
  37   # remove final newline
  38   chop;
  39   # remove all TeX commands except \1 ... \6
  40   s/\\[^123456\s{]+//g;
  41   # remove all paired { ... }
  42   1 while s/{(.*?)}/\1/g;
  43   # skip lines which now have only whitespace before `&'
  44   next if m/^\s*&/;
  45   # remove comments
  46   s/%.*//;
  47   # remove trailing whitespace
  48   s/\s*$//;
  49   # remove trailing `*' (used as a marker in the document)
  50   s/\*$//;
  51   # split at whitespace
  52   @field = split(' ');
  53   if ($field[0] eq "\\1" || $field[0] eq "\\4") {
  54     print "  $field[2]\n";
  55   }
  56   elsif ($field[0] eq "\\2" || $field[0] eq "\\5") {
  57     print "  $field[2]\n";
  58     # handle multiple suffixes separated by commata
  59     @suffix_list = split(/,/, "$field[3]");
  60     foreach $suffix (@suffix_list) {
  61       print "  $field[2]$suffix\n";
  62     }
  63   }
  64   elsif ($field[0] eq "\\3" || $field[0] eq "\\6") {
  65     # handle multiple suffixes separated by commata
  66     @suffix_list = split(/,/, "$field[3],$field[4]");
  67     foreach $suffix (@suffix_list) {
  68       print "  $field[2]$suffix\n";
  69     }
  70   }
  71   else {
  72     # for `&', split at `&' with trailing whitespace
  73     @field = split(/&\s*/);
  74     print "  $field[1]\n";
  75   }
  76 }
  77
  78 # print trailer
  79 print "}\n";
  80 print "\n";
  81 print "% EOF\n";