use HTML::Tidy;
my $call_dir = "Html3";
my $contents_of_file = 1;
#my $tidy = HTML::Tidy->new();
my $tidy = HTML::Tidy->new({
tidy_mark => 1,
#output_xhtml => 1, # yes
output_html => 1, # yes
add_xml_decl => 1, # no
wrap => 76,
error_file => errs.txt,
char_encoding => utf8,
indent_cdata => 1
});
#my $tidy = HTML::Tidy->new({config_file => 'config3.txt'});
#my $tidy = HTML::Tidy->new( {config_file => 'path/to/config'} );
my @files = glob "$call_dir/*.html";
printf "Got %d files\n", scalar @files;
for my $file (@files) {
open my $in_fh, '<', $file
or die "Could not open $file : $!";
my $contents_of_file = do { local $/;<$in_fh> };
close $in_fh;
$tidy->parse( $file, $contents_of_file );
$tidy->clean( $file);
#parse( $filename, $str [, $str...] )
#or warn "Error parsing $file :$!";
for my $message ( $tidy->messages ) {
#print $message->as_string;
}
}
####
Aberdeen%20Animal%20trait%20analysis , News Search |
Ask.com
####
http://www.ask.com/news?q=Aberdeen%2520Animal%2520trait%2520analysis+&qsrc=167&qo=channelNavigation&o=0&l=dir
Aberdeen%20Animal%20trait%20analysis , News Search | Ask.com