Preprocess lyrics for RNN

filter_genre.py

import csv
import requests
import sys
import json


LASTM_FM_KEY = ''
genre_cache = {}


def get_genres(artist_name):
    artist_name = artist_name.strip().lower()
    if artist_name not in genre_cache:
        try:
            print('Downloading genres for: %s' % artist_name)
            data = requests.get('http://ws.audioscrobbler.com/2.0/?method=artist.getinfo&artist=%s&format=json&api_key=%s' % (artist_name, LASTM_FM_KEY))
            genres = [n['name'] for n in data.json()['artist']['tags']['tag']]
        except Exception as e:
            print('Error while downloading genres for %s' % artist_name)
            genres = []
        finally:
            genre_cache[artist_name] = genres

    return genre_cache[artist_name]


def filter_row(row, target_genre):
    # artist,song,link,text
    genres = get_genres(row[0])
    return target_genre in genres

input_file = sys.argv[1]
output_file = sys.argv[2]
target_genre = sys.argv[3]
genres_file = sys.argv[4]

with open(genres_file) as f:
    genre_cache = json.load(f)

with open(input_file) as f:
    reader = csv.reader(f, delimiter=',', quotechar='"')
    rows = list(reader)

filtered_rows = [r for r in rows if filter_row(r, target_genre)]

with open(output_file, 'w') as f:
    for filtered_row in filtered_rows:
        f.write(filtered_row[3])
        f.write('\n')

with open(genres_file, 'w') as f:
    json.dump(genre_cache, f)