音響特徴に寄せたアイマス楽曲のプレイリストを登録するスクリプト

imaspotify.py

# 音響特徴に寄せたアイマス楽曲のプレイリストを登録するスクリプト
#
# 前提:
#  以下の環境変数を設定(.envファイルに記載)
#   SPOTIPY_CLIENT_ID: spotify web apiのID
#   SPOTIPY_CLIENT_SECRET: spotify web apiのシークレットキー
#   TARGET_PLAYLIST_ID: 更新するプレイリストのID
#  以下の依存ライブラリをインストール
#   pip install spotipy
#   pip install python-dotenv
#   pip install numpy
#   pip install pandas
#
# 実行方法:
#  python imaspotify.py [ブランド名] [音響特徴]
#  例) シャニマスのenergyに寄せた楽曲をプレイリストに登録する
#   python imaspotify.py ShinyColors energy

import json
import math
import os.path
import sys
import time

import numpy as np
import pandas as pd
import spotipy
from dotenv import load_dotenv
from spotipy.oauth2 import SpotifyClientCredentials, SpotifyOAuth

load_dotenv()


features = [
    'danceability',
    'energy',
    'key',
    'loudness',
    'mode',
    'speechiness',
    'acousticness',
    'instrumentalness',
    'liveness',
    'valence',
    'tempo'
]


def get_spotify_client():
    return spotipy.Spotify(client_credentials_manager=SpotifyClientCredentials())


def get_spotify_oauth():
    scope = "playlist-modify-private"
    os.environ["SPOTIPY_REDIRECT_URI"] = "http://localhost:8080/"
    spotify = spotipy.Spotify(auth_manager=SpotifyOAuth(scope=scope))
    return spotify


def get_features_fromplaylist(spotify, playlist_id):
    filename = f"playlist_features_{playlist_id}.json"

    if os.path.exists(filename):
        with open(filename, "r") as f:
            items = json.loads(f.read())
        return items

    time.sleep(10)
    items = []

    # get tracks from playlist
    for t in range(10):
        offset = 100 * t
        results = spotify.playlist_items(playlist_id, offset=offset)
        items.extend(results['items'])
        if results["offset"] + results["limit"] > results["total"]:
            break

    # get features from tracks
    id_idx_map = {e['track']['id']: i for i, e in enumerate(items)}
    number_of_sublist = math.ceil(len(id_idx_map.keys()) / 30)
    list_set = np.array_split(list(id_idx_map.keys()), number_of_sublist)

    counter = 0
    for t, l in enumerate(list_set):
        res2 = spotify.audio_features(l)
        for e in res2:
            items[counter]["features"] = e
            counter += 1

    with open(filename, "w") as f:
        f.write(json.dumps(items))

    return items


def convert_to_dataframe(items):
    columns = ['id', 'name', 'artist', *[*features]]
    data = [
        [
            e['track']['id'],
            e['track']['name'],
            ', '.join([a.get('name', '') for a in e['track'].get('artists', [])]), # [0].get('name', ''),
            *[*[e['features'][f] for f in features]]
        ]
        for i, e in enumerate(items)
    ]
    df = pd.DataFrame(data, columns=columns)
    return df


def get_feature_dataframe(spotify, brand: str="all") -> pd.DataFrame:
    playlists = []
    if brand == "all" or brand == "765":
        playlists.append('3mWR84C5UUyoJIf6d2CF9R')
    if brand == "all" or brand == "ShinyColors":
        playlists.append('0vKD0u46kcm0PxIzzmbBD9')
    if brand == "all" or brand == "CinderellaGirls":
        playlists.append('1OL47KFL8HGi5OtGdsd0Tt')
    if brand == "all" or brand == "MillionLive":
        playlists.extend([
            '328b2gee1LdKta0gurnDES',  # Unit
            '66JxoB4JzYeITPyBZa2UP2',  # Solo
        ])
    if brand == "all" or brand == "SideM":
        playlists.append('5KAAHcj9bJV2rkuR8x4FUn')
    df_list = list()
    for playlist_id in playlists:
        items = get_features_fromplaylist(spotify, playlist_id)
        df_list.append(convert_to_dataframe(items))
    return pd.concat(df_list, ignore_index=True).drop_duplicates()


def get_feature_topK(spotify, feature, brand="all", k=20):
    df = get_feature_dataframe(spotify, brand)
    df_feature = df.sort_values(feature, ascending=False).head(k)
    return df_feature


def update_playlist(spotify_oauth, playlist_id, track_ids):
    items = [f"spotify:track:{id}" for id in track_ids]
    spotify_oauth.playlist_replace_items(playlist_id, items)


if __name__ == '__main__':
    if len(sys.argv) < 3:
        print(f"usage: python imaspotify.py [all|765|ShinyColors|CinderellaGirls|MillionLive|SideM] [{'|'.join(features)}]")
        exit(1)
    brand = sys.argv[1]
    feature = sys.argv[2]

    # get tracks
    spotify = get_spotify_client()
    df = get_feature_topK(spotify, feature, brand)
    print(df[["name", "artist", feature]])

    # update playlist
    playlist_id = os.environ["TARGET_PLAYLIST_ID"]
    spotify_oauth = get_spotify_oauth()
    update_playlist(spotify_oauth, playlist_id, df["id"].values)