music-tools/spotify_search.py

import logging
import time
import spotipy
from spotipy.oauth2 import SpotifyClientCredentials

logging.basicConfig(
    level=logging.INFO,
    format="{asctime} - {levelname} - [{funcName}:{lineno}] - {message}",
    style="{",
    datefmt="%Y-%m-%d %H:%M",
)

def init_spotify_client():
    """
    Initialize and return a Spotify client.
    Requires SPOTIPY_CLIENT_ID and SPOTIPY_CLIENT_SECRET environment variables.
    """
    return spotipy.Spotify(client_credentials_manager=SpotifyClientCredentials())


def search_track(spotify, artist, title):
    """
    Search for a track on Spotify.
    
    Args:
        spotify: Spotify client instance
        artist: Artist name (string)
        title: Track title (string)
    
    Returns:
        dict with track data if found, None otherwise. Structure:
        {
            'found': True,
            'artist': str,
            'album': str,
            'album_id': str,
            'release_date': str,
            'release_date_precision': str,
            'total_tracks': int,
            'track_number': int,
            'popularity': int,
            'image_url': str,
            'genres': list,
            'label': str,
            'versions_count': int
        }
    """
    try:
        querystring = "artist:{0} track:{1}".format(artist.split("\00")[0], title)
        logging.info("Searching Spotify for track with query: " + querystring)
        
        results = spotify.search(q=querystring, type='track')
        
        if len(results['tracks']['items']) > 0:
            logging.info("Track found on Spotify!")
            track = results['tracks']['items'][0]
            album = track["album"]
            
            # Get artist genres
            artist_search = spotify.artist(track['artists'][0]['external_urls']['spotify'])
            genres = artist_search.get('genres', [])
            
            # Build response
            data = {
                'found': True,
                'artist': album["artists"][0]["name"],
                'album': album["name"],
                'album_id': album["id"],
                'release_date': album["release_date"],
                'release_date_precision': album.get("release_date_precision", "day"),
                'total_tracks': album["total_tracks"],
                'track_number': track["track_number"],
                'popularity': track["popularity"],
                'image_url': album["images"][0]["url"] if album["images"] else None,
                'genres': genres,
                'label': album.get("label", ""),
                'versions_count': len(results["tracks"]["items"])
            }
            
            logging.info(f"Found: {data['artist']} - {data['album']}")
            return data
        else:
            logging.info("No track found on Spotify")
            return None
            
    except Exception as err:
        logging.error(f"Error searching for track on Spotify: {err}")
        return None


def search_album(spotify, artist, album_name):
    """
    Search for an album on Spotify.
    
    Args:
        spotify: Spotify client instance
        artist: Artist name (string)
        album_name: Album name (string)
    
    Returns:
        dict with album data if found, None otherwise. Structure:
        {
            'found': True,
            'artist': str,
            'album': str,
            'album_id': str,
            'release_date': str,
            'release_date_precision': str,
            'total_tracks': int,
            'image_url': str,
            'genres': list,
            'versions_count': int
        }
    """
    try:
        querystring = "artist:{0} album:{1}".format(artist, album_name)
        logging.info("Searching Spotify for album with query: " + querystring)
        
        tries = 0
        found = False
        results = None
        
        while tries < 5 and not found:
            try:
                results = spotify.search(q=querystring, type='album')
                found = True
            except Exception as err:
                logging.error(f"Could not search on Spotify: {err}")
                logging.info("Waiting 30 seconds before trying again")
                time.sleep(30)
                tries += 1
        
        if not found or not results:
            logging.error("Could not search on Spotify after 5 tries")
            return None
        
        if len(results["albums"]["items"]) > 0:
            logging.info("Album found on Spotify!")
            album = results["albums"]["items"][0]
            
            # Get artist genres
            artist_search = spotify.artist(album['artists'][0]['external_urls']['spotify'])
            genres = artist_search.get('genres', [])
            
            # Build response
            data = {
                'found': True,
                'artist': album["artists"][0]["name"],
                'album': album["name"],
                'album_id': album["id"],
                'release_date': album["release_date"],
                'release_date_precision': album.get("release_date_precision", "day"),
                'total_tracks': album["total_tracks"],
                'image_url': album["images"][0]["url"] if album["images"] else None,
                'genres': genres,
                'versions_count': len(results["albums"]["items"])
            }
            
            logging.info(f"Found: {data['artist']} - {data['album']}")
            return data
        else:
            logging.info("No album found on Spotify")
            return None
            
    except Exception as err:
        logging.error(f"Error searching for album on Spotify: {err}")
        return None


def format_genres(genres):
    """
    Format a list of genres into a comma-separated string.
    
    Args:
        genres: list of genre strings
    
    Returns:
        Comma-separated string of genres, or empty string if no genres
    """
    if not genres or len(genres) == 0:
        return ""
    elif len(genres) == 1:
        return str(genres[0])
    else:
        return ",".join(genres)