From: Caesar <redacted>
Date: Sat, 17 May 2014 10:49:00 +0000 (+0200)
Subject: (irssi-feed) Fix issues with character sets (hackily)
X-Git-Url: http://git.99rst.org/?a=commitdiff_plain;h=bd4b842c94805a82b8e1e53a0ba0bd0aeb3c5c4a;p=irssi-scripts.irssi.org.git

(irssi-feed) Fix issues with character sets (hackily)
---

diff --git a/scripts/irssi-feed.pl b/scripts/irssi-feed.pl
index 2753ba8..7bcafbd 100644
--- a/scripts/irssi-feed.pl
+++ b/scripts/irssi-feed.pl
@@ -290,7 +290,7 @@ sub feed_io_event_write {
 		my $req = "GET " . $query . " HTTP/1.0\r\n" .
 				"Host: " . $self->{uri}->host . "\r\n" .
 				"User-Agent: Irssi feed reader " . $VERSION . "\r\n" .
-				"Accept-Encoding: gzip\r\n" .
+				"Accept-Encoding: " . HTTP::Message->decodable() . "\r\n" .
 				"Connection: close\r\n\r\n";
 		$self->{io}->{conn}->send($req);
 		Irssi::input_remove($self->{io}->{writetag}) if $self->{io}->{writetag};
@@ -337,7 +337,12 @@ sub feed_parse_buffer {
 		}
 	}
 	return if not $http->is_success;
-	my $httpcontent = $http->decoded_content;
+	my $httpcontent = $http->decoded_content();
+	my $contenttype = $http->header('Content-Type');
+	if($contenttype and $contenttype =~ /^.*charset\s*=\s*(?<charset>[^;]*)(;|$)/ and $+{charset} !~ /UTF-8/i) {
+		# http specifies a non-utf8 content, but HTTP::Response will convert that to utf-8 for us.
+		$httpcontent =~ s/encoding\s*=\s*("[^"]*"|[^" ]*)\s(?<!\?>)//;
+	}
 	my $data = eval { $feed->{io}->{xml} = XML::Feed->parse(\$httpcontent) };
 	if($data) {
 		$feed->{name} = $data->title;