Cover Retrieval Errors

[Seb Ruiz]

Markey and I were trying to debug problems fetching a cover for the
following artist/album:

Jean-Michel Jarre - Téo & Téa

After removing the &, this gets encoded to:
Jean-Michel%20Jarre%20-%20T%C3%A9o%20%20T%C3%A9a

Which amazon returns no results for:
http://ecs.amazonaws.com/onca/xml?Service=AWSECommerceService&Version=2007-10-29&Operation=ItemSearch&AssociateTag=webservices-20&AWSAccessKeyId=11ZKJS8X1ETSTJ6MT802&Keywords=Jean-Michel%20Jarre%20-%20T%C3%A9o%20%20T%C3%A9a&SearchIndex=Music&ResponseGroup=Small,Images

The CoverFetcher proceeds to try another fetch without the album,
which returns garbage.

---------------
However, if we do a request from Amazon with: Jean-Michel Jarre - Teo
& Tea (no umlauts), then Amazon returns the correct images:
http://ecs.amazonaws.com/onca/xml?Service=AWSECommerceService&Version=2007-10-29&Operation=ItemSearch&AssociateTag=webservices-20&AWSAccessKeyId=11ZKJS8X1ETSTJ6MT802&Keywords=Jean-Michel%20Jarre-Teo%20Tea&SearchIndex=Music&ResponseGroup=Small,Images


Can anybody confirm any other albums/artists which have a better image
retrieval behaviour when umlauts are removed from characters? This
might give us a much better hit rate on cover fetching.

-- 
Seb Ruiz

http://www.sebruiz.net/