fix: add handling for empty response for audio features API endpoint

audio features endpoint is now deprecated and used by AI compalies to crawl data from spotify => always returns fixed 401.This commit adds a fallback to empty data if this situation is encountered while allowing the feature to be used when Spotify allows data fetching again.
mashb1t · Dec 4, 2024 · cd65ff2 · cd65ff2
1 parent 3acb4b4
commit cd65ff2
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 25 deletions.
diff --git a/php/src/app/Crawler/SpotifyCrawler.php b/php/src/app/Crawler/SpotifyCrawler.php
@@ -81,9 +81,9 @@ public function crawlAll(string $username): void
         $session = $spotifySession->getUnderlyingObject();
         $spotifyWebApi = $this->factory->getSpotifyWebAPI($session);
 
-        logs('crawler')->debug("starting spotify crawler for username $username");
+        logs('crawler')->info("starting spotify crawler for username $username");
         $this->crawlTrackHistoryAndAudioFeatures($username, $spotifyWebApi);
-        logs('crawler')->debug("finished spotify crawler for username $username");
+        logs('crawler')->info("finished spotify crawler for username $username");
 
         $this->sessionHandler->saveSession($spotifySession, $username);
     }
@@ -94,10 +94,10 @@ public function crawlAll(string $username): void
     protected function crawlTrackHistoryAndAudioFeatures(string $username, SpotifyWebAPI $spotifyWebApi): void
     {
         // TODO add "after" instead of limit if last crawl was last hour
-        $recentTracks = $spotifyWebApi->getMyRecentTracks([
+	    $recentTracks = $spotifyWebApi->getMyRecentTracks([
             'limit' => config('services.spotify.crawl_bulk_limit'),
         ])->items;
-        logs('crawler')->debug("retrieved recent logs for user $username");
+        logs('crawler')->info("retrieved recent logs for user $username");
 
         $recentTracksIds = [];
         foreach ($recentTracks as $recentTrack) {
@@ -112,12 +112,13 @@ protected function crawlTrackHistoryAndAudioFeatures(string $username, SpotifyWe
                 $artistIds[$artist->id] = $artist->id;
             }
         }
-
+        logs('crawler')->debug("getting artists for user $username");
         $artistsById = $this->getArtistsById($spotifyWebApi, $artistIds);
+        logs('crawler')->debug("getting audio features for user $username");
         $audioFeatures = $this->getAudioFeatures($spotifyWebApi, $recentTracksIds);
-
+        logs('crawler')->debug("writing history for user $username");
         foreach ($recentTracks as $recentTrack) {
-            $this->writeTrackHistoryPoint($audioFeatures[$recentTrack->track->id], $recentTrack, $username);
+            $this->writeTrackHistoryPoint($audioFeatures[$recentTrack->track->id] ?? null, $recentTrack, $username);
             $this->writeGenreHistoryPoints($recentTrack, $artistsById, $username);
         }
 
@@ -132,13 +133,17 @@ protected function crawlTrackHistoryAndAudioFeatures(string $username, SpotifyWe
      */
     protected function getArtistsById(SpotifyWebAPI $spotifyWebApi, array $artistIds): array
     {
+        logs('crawler')->debug("getting artists for user $username");
         $artistsFromAPI = $this->getCachedArtistsAndCleanupIds($artistIds);
+
         $cache = Cache::tags([
             ServiceEnum::Spotify->value,
             ServiceEnum::Spotify->value . CacheKeyEnum::CACHE_KEY_SEPARATOR .  CacheKeyEnum::Artist->value
         ]);
 
-        if (count($artistIds) > 0) {
+        $newArtistIdCount = count($artistIds);
+        if (newArtistIdCount > 0) {
+            logs('crawler')->debug("retrieving uncached data for $newArtistIdCount artists");
             // artistIds count could be more than crawl_bulk_limit
             $artistIdsChunks = array_chunk($artistIds, config('services.spotify.crawl_bulk_limit'));
 
@@ -211,13 +216,16 @@ protected function getAudioFeatures(SpotifyWebAPI $spotifyWebApi, array $trackId
                 logs('crawler')->debug("found audio feature for $trackId in cache");
             }
         }
-
         if (count($trackIds) > 0) {
-            $response = $spotifyWebApi->getMultipleAudioFeatures(array_values($trackIds));
-            foreach ($response->audio_features as $audioFeature) {
-                $cache->put($audioFeature->id, $audioFeature, config('services.spotify.cache_ttl'));
-                $audioFeatures[$audioFeature->id] = $audioFeature;
-                logs('crawler')->debug("set audio feature for $audioFeature->id to cache");
+            try {
+                $response = $spotifyWebApi->getMultipleAudioFeatures(array_values($trackIds));
+                foreach ($response->audio_features as $audioFeature) {
+                    $cache->put($audioFeature->id, $audioFeature, config('services.spotify.cache_ttl'));
+                    $audioFeatures[$audioFeature->id] = $audioFeature;
+                    logs('crawler')->debug("set audio feature for $audioFeature->id to cache");
+                }
+            } catch (Exception $e) {
+                logs('crawler')->warning('error while fetching audio feature, skipping. Exception: ' . $e->getMessage());
             }
         }
 
@@ -227,7 +235,7 @@ protected function getAudioFeatures(SpotifyWebAPI $spotifyWebApi, array $trackId
     /**
      * @throws Exception
      */
-    protected function writeTrackHistoryPoint(stdClass $audioFeature, stdClass $track, string $username): void
+    protected function writeTrackHistoryPoint(?stdClass $audioFeature, stdClass $track, string $username): void
     {
         $point = $this->factory->getTrackHistoryPoint(
             $username, ServiceEnum::Spotify->value, $audioFeature, $track

diff --git a/php/src/app/Factory.php b/php/src/app/Factory.php
@@ -60,7 +60,7 @@ public function getSpotifyWebAPI(Session $session): SpotifyWebAPI
     public function getTrackHistoryPoint(
         string $username,
         string $service,
-        stdClass $audioFeature,
+        ?stdClass $audioFeature,
         stdClass $track
     ): Point {
         $artists = [];
@@ -81,15 +81,15 @@ public function getTrackHistoryPoint(
             ->addField('track', $track->track->name)
             ->addField('track_id', $track->track->id)
             ->addField('duration_ms', (int)$track->track->duration_ms)
-            ->addField('danceability', (float)$audioFeature->danceability)
-            ->addField('energy', (float)$audioFeature->energy)
-            ->addField('key', (int)$audioFeature->key)
-            ->addField('speechiness', (float)$audioFeature->speechiness)
-            ->addField('acousticness', (float)$audioFeature->acousticness)
-            ->addField('instrumentalness', (float)$audioFeature->instrumentalness)
-            ->addField('liveness', (float)$audioFeature->liveness)
-            ->addField('valence', (float)$audioFeature->valence)
-            ->addField('tempo', round((float)$audioFeature->tempo))
+            #->addField('danceability', (float)$audioFeature->danceability)
+            #->addField('energy', (float)$audioFeature->energy)
+            #->addField('key', (int)$audioFeature->key)
+            #->addField('speechiness', (float)$audioFeature->speechiness)
+            #->addField('acousticness', (float)$audioFeature->acousticness)
+            #->addField('instrumentalness', (float)$audioFeature->instrumentalness)
+            #->addField('liveness', (float)$audioFeature->liveness)
+            #->addField('valence', (float)$audioFeature->valence)
+            #->addField('tempo', round((float)$audioFeature->tempo))
             ->time($playedAtDateTime);
     }