def get_audio_features(sp: spotipy.client.Spotify, track_uri: str):
track_uri = sp._get_id("track", track_uri)
os.makedirs("audio_features", exist_ok=True)
filename = f"audio_features/{track_uri}"
try:
with open(filename, "r") as f:
contents = f.read()
data = json.loads(contents)
except FileNotFoundError:
features = sp.audio_features([track_uri])
data = features[0]
with open(filename, "w") as f:
contents = json.dumps(data, indent=4)
f.write(contents)
return data
def get_audio_features(spotify: spotipy.client.Spotify, track_ids: list) -> list:
"""
This function gets the following features of a spotify song, hundred songs at a time:
["acousticness", "danceability", "durationms", "energy", "instrumentalness", "key", "liveness", "loudness", "mode", "speechiness"\
, "tempo", "timesignature", "valence"]
Getting more than 100 songs will result in a bad request error
Parameters Required: Authenticated Spotify Client, and list of song IDs
Return Data: List of dictionary containing song features
"""
# Getting features
featurelst = []
count = 0
while count <= len(track_ids):
# Get 100 songs' features at a time. Getting any more will result in bad result error
featurelst.extend(spotify.audio_features(track_ids[count : count + 100]))
count = count + 100
return featurelst
def get_features_for_playlist(sp: spotipy.client.Spotify,
uri: str) -> pd.DataFrame:
"""
Extract features from each track in a playlist
:param sp: spotify client object
:param uri: playlist uri
:return: playlist with audio features of each song
"""
# Get all track metadata from given playlist
playlist_name, names, artists, uris = get_playlist_info(sp, uri)
# Set column names and empty dataframe
column_names = [
'name', 'artist', 'track_URI', 'acousticness', 'danceability',
'energy', 'instrumentalness', 'liveness', 'loudness', 'speechiness',
'tempo', 'valence', 'playlist'
]
df = pd.DataFrame(columns=column_names)
# Iterate through each track to get audio features and save data into dataframe
for name, artist, track_uri in zip(names, artists, uris):
# Access audio features for given track URI via spotipy
audio_features = sp.audio_features(track_uri)
# Get relevant audio features
feature_subset = [
audio_features[0][col] for col in column_names
if col not in ["name", "artist", "track_URI", "playlist"]
]
# Compose a row of the dataframe by flattening the list of audio features
row = [name, artist, track_uri, *feature_subset, playlist_name]
df.loc[len(df.index)] = row
return df