Fix exporting large playlists to spotify

We cannot add more than 100 tracks to a playlist, therefore break down
submitting tracks into chunks. Earlier, we used to replace all tracks
in one go. But replace endpoint doesn't take a position parameter to
tell it which tracks should be removed. Therefore, we can't use it
directly. The new workflow is to call replace items with an empty uris
array to reset the playlist tracks. Then, call add items on each chunk
of tracks to insert those.

Author: amCap1712

requirements.txt

@@ -6,3 +6,4 @@ countryinfo>=0.1.2
 pylistenbrainz@git+https://github.com/metabrainz/[email protected]
 python-dateutil>=2.8.2
 spotipy==2.22.1
+more_itertools

troi/playlist.py

@@ -1,10 +1,10 @@
 from collections import defaultdict
 import json
-from typing import Dict, Tuple
 
 
 import requests
 import spotipy
+from more_itertools import chunked
 from spotipy import SpotifyException
 
 from troi import Recording, Playlist, PipelineError, Element, Artist, Release
@@ -277,8 +277,14 @@ def submit_to_spotify(self,
                                                            description=playlist.description)
                 playlist_id = spotify_playlist["id"]
                 playlist_url = spotify_playlist["external_urls"]["spotify"]
+            else:
+                # existing playlist, clear it
+                sp.playlist_replace_items(playlist_id, [])
+
+            # spotify API allows a max of 100 tracks in 1 request
+            for chunk in chunked(spotify_track_ids, 100):
+                sp.playlist_add_items(playlist_id, chunk)
 
-            result = sp.playlist_replace_items(playlist_id, spotify_track_ids)
             fixup_spotify_playlist(sp, playlist_id, mbid_spotify_index, spotify_mbid_index)
             submitted.append((playlist_url, playlist_id))
 

troi/tests/test_playlist.py

@@ -93,7 +93,9 @@ def test_submit_to_spotify(self, mock_requests):
             }
         })
 
-        mock_requests.put(f"https://api.spotify.com/v1/playlists/{playlist_id}/tracks", json={"snapshot_id": "foo"})
+        mock_requests.put(f"https://api.spotify.com/v1/playlists/{playlist_id}/tracks", json={"snapshot_id": "baz"})
+
+        mock_requests.post(f"https://api.spotify.com/v1/playlists/{playlist_id}/tracks", json={"snapshot_id": "foo"})
 
         mock_requests.get(f"https://api.spotify.com/v1/playlists/{playlist_id}/tracks", json={
             "items": [
@@ -172,27 +174,24 @@ def test_submit_to_spotify(self, mock_requests):
         # we had 4 tracks with mbids, out of those no matching spotify id was found for the last track id.
         # so we submitted the first spotify track id for the remaining tracks preserving the order in which
         # occur in the original playlist
-        self.assertEqual(history[2].json(), {
-            "uris": [
-                "spotify:track:47BBI51FKFwOMlIiX6m8ya",
-                "spotify:track:7y9bltr6hV3CsbqXWgwVZv",
-                "spotify:track:4hyVrAsoKKjxAvQjPRt0ai"
-            ]
-        })
+        self.assertEqual(history[2].json(), [
+            "spotify:track:47BBI51FKFwOMlIiX6m8ya",
+            "spotify:track:7y9bltr6hV3CsbqXWgwVZv",
+            "spotify:track:4hyVrAsoKKjxAvQjPRt0ai"
+        ])
 
         # history[3] is the request to retrieve tracks for checking whether the playlist has unplayable tracks
         # history[4] is the request to retrieve info for alternative spotify ids of unplayable tracks, only thing
         # to check about this is the query strings which the request matcher in the beginning of this method does
+        # history[5] is the request to reset the playlist tracks
 
         # we had found one track to be unplayable and replaced it with an alternative playable track. now updating the
         # playlist again so check the all the correct tracks and the new replaced tracks are sent and also in the
         # original order in which the tracks occur in the playlist
-        self.assertEqual(history[5].json(), {
-            "uris": [
-                "spotify:track:47BBI51FKFwOMlIiX6m8ya",
-                "spotify:track:4LWQfAhwP1Tf1wbzmT6NwW",
-                "spotify:track:4hyVrAsoKKjxAvQjPRt0ai"
-            ]
-        })
+        self.assertEqual(history[6].json(), [
+            "spotify:track:47BBI51FKFwOMlIiX6m8ya",
+            "spotify:track:4LWQfAhwP1Tf1wbzmT6NwW",
+            "spotify:track:4hyVrAsoKKjxAvQjPRt0ai"
+        ])
 
         self.assertEqual(playlist.playlists[0].additional_metadata["external_urls"]["spotify"], playlist_url)

troi/tools/spotify_lookup.py

@@ -2,6 +2,7 @@
 
 import requests
 import spotipy
+from more_itertools import chunked
 
 SPOTIFY_IDS_LOOKUP_URL = "https://labs.api.listenbrainz.org/spotify-id-from-mbid/json"
 
@@ -106,4 +107,9 @@ def fixup_spotify_playlist(sp: spotipy.Spotify, playlist_id: str, mbid_spotify_i
     all_items.sort(key=lambda x: x[0])
     # update all track ids the spotify playlist
     finalized_ids = [x[1] for x in all_items]
-    result = sp.playlist_replace_items(playlist_id, finalized_ids)
+
+    # clear existing playlist
+    sp.playlist_replace_items(playlist_id, [])
+    # spotify API allows a max of 100 tracks in 1 request
+    for chunk in chunked(finalized_ids, 100):
+        sp.playlist_add_items(playlist_id, chunk)