Problems encountered implementing AURA

[govynnus]

I've been having a go at implementing the AURA spec as a beets plugin using the existing web plugin as inspiration, however I've got a few problems.

First, AURA says that the artist, genre, composer and albumartist should be arrays, but in beets Item object these attributes are just single strings and there doesn't seem to be a consistent separator. For example in my library &, /, and - are all used to separate multiple artists. At the moment I'm just returning an array with a single string in it (which is whatever beets gives me), but this hardly seems the right thing to do.

Second, I'm wondering what a recording-mbid is compared to a track-mbid because AURA gives both as optional attributes but beets has mb_trackid (which corresponds to track-mbid) and mb_releasetrackid (which I don't think is the same as a recording-mbid?). Also, AURA mentions framerate and framecount which I thought were related to video rather than audio. This issue might be more AURA than beets related, or could just be my ignorance.

Third, is there a way to get a list of all artists in the database like you can tracks or albums? If not I could do something like beet ls -f '$artist' then remove duplicates, or probably more sensibly use a Transaction object to do some SQL directly. On a related note, is there a way to access the beet ls functionality through the API?

Thanks.

[sampsyo]

Awesome!! This seems like a really good idea. Thanks for looking into it!

At the moment I'm just returning an array with a single string in it (which is whatever beets gives me), but this hardly seems the right thing to do.

I think this is the right place to start! We designed AURA with extensibility in mind, to be ready for whenever beets supports multi-valued fields as first-class citizens (c.f. #505). But absent that, I think a single-value list is the right thing to do.

Second, I'm wondering what a recording-mbid is compared to a track-mbid

Good question. Because of history (namely, the evaluation from MB's old database schema to what they call "NGS"), beets names are a little weird. Namely:

• mb_trackid: This is the MusicBrainz recording ID.
• mb_releasetrackid: This is now the ID for the MusicBrainz track.

The framerate and framecount attributes are relevant to audio, though: if you think of "frame" as a synonym for "sample" then maybe they make more sense. 😃

Third, is there a way to get a list of all artists in the database like you can tracks or albums?

There is not, unfortunately. The query you describe is the right thing to do. Tracks and albums are entities in the beets database (distinct from just metadata on other entities), but artists are not.

On a related note, is there a way to access the beet ls functionality through the API?

Absolutely! The items method on the Library class does this. You can provide a string query, but it is probably nicer to provide a Query object programmatically. If you look around for other uses of this method, you should be able to find examples.

[govynnus]

That helps a lot, thanks!