Committed, thanks.

what about a commit ?

attached patch changes lib/Jonah.php to look for encoding value in the 
xml declaration and set charset if there is no charset specified in 
Content-Type.

I love pattern-matching  -  I will look for it at evening.

Why not matching the complete <?xml tag on the complete feed content? 
Something like (untested):

/<\?xml[^>]+encoding=["']?([^"'\s?]+)[^?]*?>/i

today at evening I will try something like this in lib/Jonah.php :



looking in the first 80 characters of the source-file (should contain 
the XML declaration ordered : version - encoding - standalone)

for the string after  encoding=  (should be the charset).





         if (preg_match('/.*;\s?charset="?([^"]*)/', $content_type, $match)) {

             $result['charset'] = $match[1];

+        } else {

+            $t_start = strpos(substr($result['body'], 1, 60), 
'encoding=') + 10;

+            if ($t_start) {

+                $t_stop  = strpos(substr($result['body'], $t_start, 
20), '"', $t_start);

+                $result['charset'] = 
strtolower(trim(substr($result['body'], $t_start, $t_stop - $t_start)));

+            }

         }



not very inventive (I do not know php), but it should extract the right thing.

yes - there is a problem with $t_stop if the encoding value is 
included in single quotes (I will look after it).

I am not sure about  preg_match  , but the following should also work :



         if (preg_match('/.*;\s?charset="?([^"]*)/', $content_type, $match)) {

             $result['charset'] = $match[1];

+        } elsif (preg_match('/.*\s?encoding="?([^"]*)/', 
substr($result['body'], 1, 80), $match)) {

+            $result['charset'] = $match[1];

         }

No, this doesn't help much unfortunately because conversion is not the 
problem, but detecting the correct charset in the first place.

Remembered this link when thinking about this: 
http://weierophinney.net/matthew/archives/111-mbstring-comes-to-the-rescue.html



might be relevant but might not, too.

thank you - I will try to read the first line of the xml-file in 
lib/Jonah.php if no charset is given in the Content-Type Header 
($result['charset'] is NULL) and set the charset to the "encoding=" 
value from within the xml-file.

I am not familiar with php, so it will last a little time.

The problem is that PHP's xml parser is not able to properly detect 
the feed's charset from the "encoding" parameter. *sigh*

Thus we have to rely on the charset being sent by the feed's web 
server in the Content-Type HTTP header. If no charset is sent, we fall 
back to UTF-8 which is happening here.

Ideas for a better solution are welcome.

with the current HEAD two of my news-channels 
(http://www.wdr.de/xml/newsticker.rdf and 
http://www.frag-mutti.de/newsfeed/rss-de.xml) are not showed anymore 
("No stories are currently available.").

The Problem seems to be the ISO-8859-1 encoding : if I remove any 
special character, everything is fine.

This worked in HEAD to at least June (did not look very nice but the 
channels have been available).

UTF-8 encoded files like http://rss.orf.at/oesterreich.xml are 
displayed great.