tmac/hyphenex.pl

   1 #!/usr/bin/env perl
   2 #@ hyphenex.pl
   3 #@ This small filter converts a hyphenation exception log article for
   4 #@ TUGBoat to a real \hyphenation block.
   5 #@ Version 1.2 (2007/11/16)
   6 #@ Synopsis:
   7 #@   [perl] hyphenex.pl < tugboat-article > hyphenation-exceptions
   8 #
   9 # Written by Werner Lemberg <wl@gnu.org>.
  10 # Public domain.
  11 #
  12 # Adapted to S-roff by Steffen (Daode) Nurpmeso <steffen@sdaoden.eu>.
  13
  14 # print header
  15 print "% Hyphenation exceptions for US English,\n";
  16 print "% based on hyphenation exception log articles in TUGboat.\n";
  17 print "%\n";
  18 print "% Copyright 2007 TeX Users Group.\n";
  19 print "% You may freely use, modify and/or distribute this file.\n";
  20 print "%\n";
  21 print "% This is an automatically generated file.  Do not edit!\n";
  22 print "%\n";
  23 print "% Please contact the TUGboat editorial staff <tugboat\@tug.org>\n";
  24 print "% for corrections and omissions.\n";
  25 print "\n";
  26 print "\\hyphenation{\n";
  27
  28 while (<>) {
  29   # retain only lines starting with \1 ... \6 or \tabalign
  30   next if not (m/^\\[123456]/ || m/^\\tabalign/);
  31   # remove final newline
  32   chop;
  33   # remove all TeX commands except \1 ... \6
  34   s/\\[^123456\s{]+//g;
  35   # remove all paired { ... }
  36   1 while s/{(.*?)}/\1/g;
  37   # skip lines which now have only whitespace before `&'
  38   next if m/^\s*&/;
  39   # remove comments
  40   s/%.*//;
  41   # remove trailing whitespace
  42   s/\s*$//;
  43   # remove trailing `*' (used as a marker in the document)
  44   s/\*$//;
  45   # split at whitespace
  46   @field = split(' ');
  47   if ($field[0] eq "\\1" || $field[0] eq "\\4") {
  48     print "  $field[2]\n";
  49   }
  50   elsif ($field[0] eq "\\2" || $field[0] eq "\\5") {
  51     print "  $field[2]\n";
  52     # handle multiple suffixes separated by commata
  53     @suffix_list = split(/,/, "$field[3]");
  54     foreach $suffix (@suffix_list) {
  55       print "  $field[2]$suffix\n";
  56     }
  57   }
  58   elsif ($field[0] eq "\\3" || $field[0] eq "\\6") {
  59     # handle multiple suffixes separated by commata
  60     @suffix_list = split(/,/, "$field[3],$field[4]");
  61     foreach $suffix (@suffix_list) {
  62       print "  $field[2]$suffix\n";
  63     }
  64   }
  65   else {
  66     # for `&', split at `&' with trailing whitespace
  67     @field = split(/&\s*/);
  68     print "  $field[1]\n";
  69   }
  70 }
  71
  72 # print trailer
  73 print "}\n";
  74 print "\n";
  75 print "% EOF\n";
  76 # s-it2-mode