#!/usr/bin/perl use warnings; use strict; # Load current template #@`wget s7s8.com/SL/template -O template.html`; #@open my $in, '< template.html'; #@my $template= join '', <$in>; #@close $in; #@ # Load candidates for wrapping my @files= $ARGV[0] || <*.html>; my $limit = $ARGV[1] || 10; # Local list of keywords my %keywords= ( #"kerberos.html" => [ qw/MIT 1 2 3 4/], ); for( @files) { open my $in, "< $_"; my $content= join '', <$in>; close $in; # Remove certainly unneeded stuff # $content=~ s###g; # $content=~ s###s; # $content=~ s#.*##s; # # Final data that'll be output ( my $outname= $_)=~ s#.*/##; # my $data= $template; # # # Extract and insert description # if( $content=~ m##s) { # my $desc= $1; # $data=~ s/\{DESCRIPTION\}/$desc/s; # } # Extract keywords my $tmp= $content; $tmp=~ s#<.*?>##gs; $tmp=~ s/\&\w+;//gs; # Remove HTML entities $tmp=~ s/[^a-zA-Z0-9_-]/ /gs; my @words= split / +/, $tmp; my %ladder; for( @words) { $_= lc $_; next if ( length($_)< 4 and $_!~ /^(afs)$/i) or # <- USE LIST $_=~ /^(will|that|sudo|with|this|root|file|mirko|have|your|which|files|from|there|they|would|just|monarch|also|more|then|them|some|first|other|called|8212|found|local|using|network|group)$/; # <- DISCARD LIST $ladder{$_}++; } my @top= sort { $ladder{$b} <=> $ladder{$a} } keys %ladder; $#top= $limit; # Insert keywords my @local_keywords= $keywords{$outname}? @{ $keywords{$outname}}: (); my $keyword_list= join(', ', @local_keywords, @top[0..( $limit- scalar( @local_keywords)- 1)]); # $data=~ s/\{KEYWORDS\}/$keyword_list/s; # # # Remove whole head after its been parsed # $content=~ s#(.*)##s; # # # Insert content # $data=~ s/\{BODY\}/$content/s; # # print $outname, "\n"; # open my $out, "> $outname"; # print $out $data; # close $out; if( @files> 1) { print "$_: "; } print $keyword_list, "\n"; }