Re: [Rhythmbox-devel] Album art downloading and Amazon

[Jonathan Matthew <jonathan d14n org>]

On Sun, Aug 02, 2009 at 09:41:02AM +1000, Jonathan Matthew wrote:
> On Thu, Jul 30, 2009 at 2:32 AM, Bastien Nocera<hadess hadess net> wrote:
> > On Sat, 2009-07-25 at 19:13 +0300, Erik Houghton wrote:
> >> What are the plans for handling "album art fetching" from Amazon once
> >> the 15th of August passes?
> >
> > What's so special about that date?

> from August 15 on, clients will need to sign API requests using a
> per-account secret key. As far as I can tell, we will no longer be
> able to use Amazon as a cover art provider, since once you've included
> a secret key in some GPL source code, it's not really a secret any
> more.
> 
> Implementing a new cover art provider will probably fit somewhere into
> my pile of things to work on, but given my current lack of hacking
> time and the amount of other stuff that I'm more interested in working
> on, I probably won't get to it soon.

I've just committed a cover art search implementation using the
discogs.com API.  I haven't done a huge amount of testing, but it seems
to work pretty well, at least for my collection.  It sometimes returns
slightly odd results, but I think that's mostly down to the fact that
the images on discogs are user submitted.  I've tried to exclude promo
releases, white labels, and various other things that don't tend to have
good cover art, but there are always more special cases to cover.

I had previously written a cover art search implementation using
last.fm, but their terms of service are a bit difficult to comply with.
The main problems are that last.fm data must only be made available to
registered last.fm users, and that a 100 pixel wide, ~40 pixel high
last.fm logo must be displayed along with any last.fm data.  These
aren't insurmountable, but I decided to focus elsewhere to see if we can
get similar cover art quality without those restrictions.  If anyone is
interested in finding solutions to these problems, I can provide the
basic cover art search code, but I'd rather not have this code in the
git repo until it's in a state where we can actually use it.