Cover Fetching Data Source

[Dan Meltzer]

On Mon, Oct 27, 2008 at 3:01 PM, Casey Link <unnamedrambler at gmail.com> wrote:
> I think everyone would agree that album art auto-fetching being
> enabled by default wouldn't be such a hot-button issue if there as a
> high likelihood of accurate covers being retrieved. So, we should be
> working to improve the reliability of the fetcher before we discuss if
> the auto fetch should be enabled or not.
>
> I sent the following back in June in an attempt to speed up the
> dropping of amazon as a primary data source. Since then last.fm
> released a new API, so I've updated it with the current API methods.
> Lets discuss this, and come to a decision on whether we are going to
> keep amazon, or switch (and what to if that is the case).

Hi,

I've had an implicit (and, to anyone thats been on irc, explicit) plan
all along to replace the amazon stuff with something else.  One of the
main reasons is that we are breaking the amazon TOS as it stands, and
just hoping that they ignore it.  Add in that it's inaccurate and it
seems like a great place for improvement.

The reasons it hasn't been done thus far is because all of our lastfm
code is in the service plugin.  We cannot depend on a plugin, so
theres no easy way to make lastfm availible to the rest of the
application.  At one point I had plans of breaking out our lastfm code
into a separate shared library that the service linked to, but with
Leo's recent work to integrate liblastfm it makes more sense to wait
and use that.

I *do not* think we should use the web api, or implement lastfm code
in any other form.  We should use their code and their api.  The
question is one of whether it's feasible (and sensible) to replace it
this close to what is hopefully going to be a final release.  I
personally have issues with replacing our current lastfm code (which
is known to work) with new lastfm code before 2.0, but that may become
moot depending on when it becomes availible.  If we don't have
liblastfm, we cannot replace amazon cover fetching before 2.0.  If we
do have liblastfm, we probably shouldn't replace amazon cover fetching
before 2.0, but at least it would be possible to do so.

Dan,
>
> Amazon has no way of targeting album searches for a particular artist.
> The problem? Try finding a cover for a "Greatest Hits" album in
> Amarok, you'll almost always get back some artist that wasn't the one
> you wanted. Last.fm provice
>
> With last.fm we can make targeted queries by searching the Artist and
> Album fields. Basically you make a request like this:
>
> http://ws.audioscrobbler.com/2.0/?method=album.getinfo&api_key=b25b959554ed76058ac220b7b2e0a026&artist=<ARTIST
> NAME>&album=<ALBUM NAME>
> and in the returned xml there is:
> <image size="small">http://images.amazon.com/images/P/B000002NYC.01.MZZZZZZZ.jpg</image>
> <image size="medium">http://images.amazon.com/images/P/B000002NYC.01.MZZZZZZZ.jpg</image>
> <image size="large">http://images.amazon.com/images/P/B000002NYC.01.MZZZZZZZ.jpg</image>
> <image size="extralarge">http://images.amazon.com/images/P/B000002NYC.01.MZZZZZZZ.jpg</image>
>
> See: http://www.last.fm/api/show?service=290
>
> A disadvantage is the clause stated in last.fm's TOS:
> "You will not make more than 5 requests per originating IP address per
> second, averaged over a 5 minute period, without prior written
> consent."
>
> For more legal/licensing issues see http://www.last.fm/api/tos
>
> So, what are your thoughts on switching from Amazon to last.fm as a
> data source for cover fetching? Please, don't bring up "Make cover
> fetching a plugin", that is irrelevant to this topic. Amarok will
> offer some default cover fetching mechanism, and I think it should use
> last.fm.
>
> Casey
> _______________________________________________
> Amarok-devel mailing list
> Amarok-devel at kde.org
> https://mail.kde.org/mailman/listinfo/amarok-devel
>