spotify_lastfm_recs #

Local tools for collecting Spotify playlist metadata and normalizing Spotify + Last.fm listening data into SQLite for recommendation experiments.

What is committed #

• pull_spotify_playlists.py fetches Spotify playlists, playlist detail, and playlist item metadata.
• normalize_music_data.py builds music.db from local Spotify export files, Last.fm JSON exports, and Spotify playlist pulls.
• .gitignore keeps credentials, tokens, exports, and generated databases out of git.

Local inputs #

Expected private local folders:

• spotify_extended_streaming_history/
• spotify_playlists_export/
• lastfm/

Expected private local config:

• .env
• .spotify_token.json

Run:

python3 pull_spotify_playlists.py
python3 normalize_music_data.py
python3 reconcile_tracks.py
python3 qa_reconciliation.py
python3 apply_playlist_weights.py
python3 build_taste_features.py
python3 generate_recommendations.py

The normalized database is written to music.db. Reconciliation and QA add canonical track tables to music.db and local CSV/Markdown reports under reports/. Playlist weighting writes playlist_weights to SQLite and creates an ignored playlist_weight_overrides.csv for manual edits. Taste features write taste_features to SQLite plus first-pass profile and seed reports under reports/. Era boundaries are read from private local taste_eras.json, which is ignored by git. Recommendations are written to ignored local CSV/Markdown files under reports/.