dynamicgenetics / spotify-rehydrator Goto Github PK

Rank API search results based on a distance measure.

This will require re-organising the repo a little, and changing the default arguments.

Other things that could be achieved through this option include:

• Choosing whether to get TrackIDs, Features, or something else? (Lyrics)

• Re-organise into package structure

• setup.py

• Register with PyPi

• Change input/output file options to arguments.

About 5% of tracks aren't returned by the API. Quite a few of them seem to have special characters so that could be the reason. It would be good to understand if there's anything we can change about the search queries to increase chances of them returning successfully.

At the moment you lose the information about the files if you have to restart

Both get_track_id and get_track_features are now very long and would benefit from being separated into different functions.

1. Retrieving data to be processed
2. Processing the data
3. Saving it out

Maybe a good time to consider a Data Class?

https://developer.spotify.com/documentation/web-api/#rate-limiting

Error status 429

This returns a 401 error.

Quick google returns some possible solution from Spotipy issues / questions section.

• Dataset doesn't contain artistName or trackName in Tracks

Before running check if the output file exists.
If it does then skip that person and log as info.

Check the joins match as expected etc and any data manipulation is tested.
Will need to mock the API calls.

• get_ids
• read
• unmatched_tracks
• get_URI
• add_URI
• add_feature_cols
• run_pipeline
• rehydrate

Instead of having 'None' as the value in the output csv it would be better to have a null placeholder in each column. Otherwise reading the file back in can cause problems.

https://developer.spotify.com/documentation/web-api/reference/#endpoint-get-audio-features-for-several-tracks

Can get up to 100 at a time.