debian/lynx-dump-postprocess

   1 #!/usr/bin/perl
   2
   3 open IN, '<&STDIN';
   4 #open IN, 'dump';
   5 {
   6   local $/;
   7   $content=<IN>;
   8 }
   9 close IN;
  10
  11 ($title,$body,$links) =
  12   ($content =~
  13    /^(.*)\n\s+_+\n\n
  14     (\s+Table\ of\ Contents.*)
  15     (References\n\n\s+1\..*)/sx);
  16
  17 die unless ($title);
  18
  19 print "$title\n".'-' x length($title)."\n\n";
  20
  21 # Sort out local links.
  22 # The regex might not be entirely accurate.
  23 foreach (split /\n/, $links) {
  24   ($index, $url) = /^\s*(\d+)\. (.+)$/;
  25   if ($url !~ /file:\/\/.*#.*$/) {
  26     $links[$index] = $url;
  27   }
  28 }
  29
  30 $linkno=0;
  31 # Split paragraphs
  32 foreach (split /\n(?:\s+_+\n)?\n/, $body) {
  33   my $footnote = '';
  34   my $rest = $_;
  35   while ( $rest =~ /^(.*?)\[(\d+)\](.*)$/s ) {
  36     print $1;
  37     if (defined $links[$2]) {
  38       $linkno++;
  39       print "[$linkno]";
  40       $footnote.="     $linkno. $links[$2]\n";
  41     }
  42     $rest = $3;
  43   }
  44   print $rest;
  45   print "\n\n";
  46   if ($footnote ne '') {
  47     print "$footnote\n";
  48   }
  49 }