[F] v2.38 Unicode bugs with Amazon/Discogs

There exist multiple bugs in v2.38 when attempting to look up albums in Amazon/Discogs. I'll spell them out here:

1) HTTP requests are not being URL encoded. The text entered in the GUI is being put directly into the HTTP request, which fails because the server expects proper URL encoding. An example is this artist:

Sébastien Léger

If you type that name into the Discogs Artist dialog box, mp3tag makes a request to the server: HTTP GET /artist/S\351bastien+L\351ger HTTP/1.1

This is clearly wrong. If, instead, you type "S%C3%A9bastien+L%C3%A9ger" into the dialog box, it correctly locates the artist.

2) Amazon/Discogs responses are not being correctly handled as UTF-8. If you make a request to fetch an album which has non-ASCII UTF-8 encoded characters in it, the dialog shows you garbage in place of the real characters because they are not being translated from UTF-8 back into UTF-16 before being used in Windows and the mp3 tags.

An example album is:

If you type in the Discogs search "Isol%C3%A9e" to get around the first bug, and then select the "Rest (CD)" album, you will see garbage in the track listing where it should display "Démon"

i encountered the same problem, starting with v2.38, as well as 2.39

basically the search doesnt work with some unicode characters.


Discogs search for Trentemøller = works in 2.37, does not work in 2.38/39

Discogs search for Pär Grindvik = works in 2.37, does not work in 2.38/39

also, the second problem that andrew described happens in the new versions, too.

i switched back to 2.37 now, although it tells me to upgrade everytime i run the program ("the message will not be shown again"..yeah right :wink:

is there any chance of getting this fixed, because basically the program is useless for any non-english searches (french accentuation, german umlauts etc.) now.


i must say, however, that accentuations do not work in 2.37 either

example: Sébastien Léger = does not work in any version.

For discogs, it seems to work if you add [Encoding]=utf-8 to the script.
I've changed my scripts, please try it:

you're right, that works for the search, but I still get wrong characters displayed when i try to fetch a release containing those characters

Yes there's something wrong with the decoding detection.
I hope Florian can find the problem.

This should be fixed with the latest Development Build.

Thanks for reporting!

Kind regards,

seems to be working...thanks!

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.