mirror of
https://github.com/pragma-/pbot.git
synced 2024-11-05 03:29:33 +01:00
115 lines
2.6 KiB
Perl
Executable File
115 lines
2.6 KiB
Perl
Executable File
#!/usr/bin/perl -w -I /home/msmud/lib/lib/perl5/site_perl/5.10.0/
|
|
|
|
# Quick and dirty by :pragma
|
|
|
|
use LWP::UserAgent;
|
|
use HTML::Entities;
|
|
use Text::Levenshtein qw(fastdistance);
|
|
|
|
my ($text);
|
|
|
|
if ($#ARGV <= 0)
|
|
{
|
|
print "Usage: title nick URL\n";
|
|
exit;
|
|
}
|
|
|
|
my $nick = shift(@ARGV);
|
|
$arguments = join("%20", @ARGV);
|
|
|
|
exit if($arguments =~ m/stackoverflow.com/i);
|
|
exit if($arguments =~ m/scratch.mit.edu/i);
|
|
exit if($arguments =~ m/imgur.com/i);
|
|
exit if($arguments =~ m/sprunge.us/i);
|
|
exit if($arguments =~ m/hastebin.com/i);
|
|
exit if($arguments =~ m/lmgtfy.com/i);
|
|
exit if($arguments =~ m/gyazo/i);
|
|
exit if($arguments =~ m/imagebin/i);
|
|
exit if($arguments =~ m/\/wiki\//i);
|
|
exit if($arguments =~ m/github.com/i);
|
|
exit if($arguments =~ m/wiki.osdev.org/i);
|
|
exit if($arguments =~ m/wikipedia.org/i);
|
|
exit if($arguments =~ m/everfall.com/i);
|
|
exit if($arguments =~ m/fukung.net/i);
|
|
exit if($arguments =~ m/\/paste\//i);
|
|
exit if($arguments =~ m/paste\./i);
|
|
exit if($arguments =~ m/pastie/i);
|
|
exit if($arguments =~ m/ideone.com/i);
|
|
exit if($arguments =~ m/codepad.org/i);
|
|
exit if($arguments =~ m/^http\:\/\/past(e|ing)\./i);
|
|
exit if($arguments =~ m/paste.*\.(?:com|org|net|ch|ca|de|uk|info)/i);
|
|
exit if($arguments =~ m/pasting.*\.(?:com|org|net|ca|de|uk|info|ch)/i);
|
|
|
|
my $ua = LWP::UserAgent->new;
|
|
$ua->agent("Mozilla/5.0");
|
|
$ua->max_size(200 * 1024);
|
|
|
|
my $response = $ua->get("$arguments");
|
|
|
|
if (not $response->is_success)
|
|
{
|
|
#print "Couldn't get link.\n";
|
|
die "Couldn't get link: $arguments";
|
|
}
|
|
|
|
$text = $response->content;
|
|
|
|
if($text =~ m/<title>(.*?)<\/title>/msi)
|
|
{
|
|
$t = $1;
|
|
} else {
|
|
#print "No title for link.\n";
|
|
exit;
|
|
}
|
|
|
|
my $quote = chr(226) . chr(128) . chr(156);
|
|
my $quote2 = chr(226) . chr(128) . chr(157);
|
|
my $dash = chr(226) . chr(128) . chr(147);
|
|
|
|
$t =~ s/\s+/ /g;
|
|
$t =~ s/^\s+//g;
|
|
$t =~ s/\s+$//g;
|
|
$t =~ s/<[^>]+>//g;
|
|
$t =~ s/<\/[^>]+>//g;
|
|
$t =~ s/$quote/"/g;
|
|
$t =~ s/$quote2/"/g;
|
|
$t =~ s/$dash/-/g;
|
|
$t =~ s/"/"/g;
|
|
$t =~ s/“/"/g;
|
|
$t =~ s/”/"/g;
|
|
$t =~ s/&/&/g;
|
|
$t =~ s/&nsb;/ /g;
|
|
$t =~ s/'/'/g;
|
|
$t =~ s/</</g;
|
|
$t =~ s/>/>/g;
|
|
$t =~ s/«/<</g;
|
|
$t =~ s/»/>>/g;
|
|
$t =~ s/>/>/g;
|
|
$t =~ s/•/-/g;
|
|
$t =~ s/<em>//g;
|
|
$t =~ s/<\/em>//g;
|
|
|
|
if(length $t > 150) {
|
|
$t = substr($t, 0, 150);
|
|
$t = "$t [...]";
|
|
}
|
|
|
|
# $nick =~ s/^(.)(.*)/$1|$2/;
|
|
|
|
$t = decode_entities($t);
|
|
|
|
$t =~ s/^\s+//;
|
|
$t =~ s/\s+$//;
|
|
|
|
my ($file) = $arguments =~ m/.*\/(.*)$/;
|
|
$file =~ s/[_-]/ /g;
|
|
|
|
my $distance = fastdistance(lc $file, lc $t);
|
|
my $length = (length $file > length $t) ? length $file : length $t;
|
|
|
|
if($distance / $length < 0.75) {
|
|
exit;
|
|
}
|
|
|
|
print "Title of $nick\'s link: $t\n" if length $t;
|