removing punctuation from spotify search

analysis/prep/feature.py

@@ -59,7 +59,7 @@ def populated_features(spotnet: SpotNetwork,
         except (KeyError, TypeError):
             for_pulling.append(row)
     
-    # GET SPOTIFY TRACKS
+    # GET SPOTIFY TRACKS FOR CACHE FAILURES
     logger.info('pulling tracks')
     tracks = spotnet.tracks(uris=[i.uri for i in for_pulling])
 

analysis/prep/scrobble.py

@@ -1,5 +1,6 @@
 import datetime
 import logging
+import string
 from csv import DictWriter
 
 from analysis.cache import Cache
@@ -33,26 +34,34 @@ def populated_scrobbles(spotnet: SpotNetwork,
     # get all scrobbles for date range
     scrobbles = fmnet.recent_tracks(limit=limit, from_time=from_date, to_time=to_date, page_limit=200)
 
+    punc_stripper = str.maketrans('', '', string.punctuation)
+
     # populate with uris
     for scrobble in scrobbles:
 
         cache_entry = cache.get_track(name=scrobble.track.name.lower(), artist=scrobble.track.artist.name.lower())
 
-        if cache_entry is not None and cache_entry.get('uri'):
+        try:
             # uri is cached
-            scrobble.uri = cache_entry.get('uri')
-        else:
+            scrobble.uri = cache_entry['uri']
+        except (TypeError, KeyError): 
+            # TypeError for cache_entry == None, KeyError for uri key existing
             # cache missed or doesn't have uri
             logger.info(f'pulling {scrobble.track}')
             spotify_search = spotnet.search(query_types=[Uri.ObjectType.track],
-                                            track=scrobble.track.name,
-                                            artist=scrobble.track.artist.name,
+                                            track=scrobble.track.name.translate(punc_stripper),
+                                            artist=scrobble.track.artist.name.translate(punc_stripper),
                                             response_limit=5).tracks
             if len(spotify_search) > 0:
+                top_result = spotify_search[0]
+
+                if top_result.name.lower() != scrobble.track.name.lower() or top_result.artists[0].name.lower() != scrobble.track.artist.name.lower():
+                    logger.info(f'{scrobble.track.name} - {scrobble.track.artist.name} != {top_result.name} - {top_result.artists[0].name}')
+
                 cache.set_track(name=scrobble.track.name.lower(),
                                 artist=scrobble.track.artist.name.lower(),
-                                uri=str(spotify_search[0].uri))
-                scrobble.uri = str(spotify_search[0].uri)
+                                uri=str(top_result.uri))
+                scrobble.uri = str(top_result.uri)
             else:
                 logger.debug('no search tracks returned')
                 scrobble.uri = None