Make certain fields optional to speed downloads that don't need them

[pavelkomarov]

We finally solved #32, but in doing so we have to grab albums, which can only be gotten 20 at a time. Even using a Set to deduplicate these queries, that 20 limit slows down the app quite a bit.

We could get rid of queries to the artists endpoint if genre information were populated in the album responses, but the genres fields come back as empty lists! Unless/until Spotify fixes this, we're stuck querying both endpoints.

We could alternatively make genres, record label, and audio features optional in the output by adding a selector to the UI and switching in the code.

[pavelkomarov]

They've decided to officially deprecate the genres field at the albums endpoint, which means we're stuck querying artists too. So the only way to decrease traffic would be to get sophisticated and make certain fields optional, doable but a bit more lift.

[Starttoaster]

FYI it looks like Spotify deprecated the multi-albums endpoint as well, though it still appears usable for now. I imagine the singular albums API endpoint would slow down exportify a fair bit more.

Not super familiar with Jupyter, but I think Jupyter has the ability to make requests concurrently via asyncio and aiohttp? Eventually it's likely that pattern would end up getting 429 rate limited by Spotify, but it looks like they're kind enough to give a Retry-After header to set a sleep for, then retry the request(s).

[pavelkomarov]

Yes, we had a recent discussion about this. See #123.

As for rate limits, I've implemented retries, but if we have to make individual calls per artist or album, it will overwhelm the rate limit very quickly and end up waiting an unreasonably long time to complete downloads.