get_title: UTF8 decode title

This commit is contained in:
Pragmatic Software 2019-08-25 10:46:52 -07:00
parent 9c803b2d14
commit 198a7d6c5b
1 changed files with 3 additions and 1 deletions

View File

@ -24,6 +24,7 @@ $arguments =~ s/\W$//;
exit if $arguments =~ m{https?://git}i;
exit if $arguments =~ m{https://.*swissborg.com}i;
exit if $arguments =~ m{https://streamable.com}i;
exit if $arguments =~ m{https://matrix.org}i;
exit if $arguments =~ m{https://freenode.net/news/spam-shake}i;
exit if $arguments =~ m{https://twitter.com/ISCdotORG}i;
@ -96,7 +97,7 @@ if (not $response->is_success)
die "Couldn't get link: $arguments";
}
my $text = $response->content;
my $text = $response->decoded_content;
if ($text =~ m/<title>(.*?)<\/title>/msi)
{
@ -157,6 +158,7 @@ if ($distance / $length < 0.75) {
exit if $t !~ m/\s/; # exit if title is only one word -- this isn't usually interesting
exit if $t =~ m{christel}i;
exit if $t =~ m{streamable}i;
exit if $t =~ m{freenode}i;
exit if $t =~ m{ico scam}i;
exit if $t =~ m{Freenode head of infrastructure}i;