def mark_verses(song_list_path):
song_list = read_list_from_file(song_list_path)
artist_list = read_list_from_file('final_artist_list.txt')
process_song = get_process_song(artist_list)
loop_and_process(song_list, process_song, "Song", lambda x: x,
marked_verses_dir)
remove_duplicates_from_list_file(marked_verses_dir, skipped_artists_file)
def raw_songs_to_verse_split_songs(song_list_path, dir_prefix):
with open(song_list_path) as listfile:
song_list = listfile.readlines()
song_list = [song.strip() for song in song_list]
def process_song(song, bar):
with open("{}/{}.json".format(cleaned_songs_dir,
name_to_file_name(song))) as song_file:
song = json.load(song_file)
verses = []
lyrics = song['lyrics']
i = 0
verse_lyrics = ''
verse_metadata = ''
def write_verse(v_metadata, v_lyrics, verses):
if len(v_metadata.strip()) > 0:
verses.append({'metadata': v_metadata, 'lyrics': v_lyrics})
return verses
while i < len(lyrics):
# Parse the songs into each verse
if lyrics[i] == '[':
# we reached a new verse
# append the previous verse if there was one
verses = write_verse(verse_metadata, verse_lyrics, verses)
# reset for the new verse
verse_lyrics = ''
verse_metadata = ''
# start processing the new verse
while i < len(lyrics) and lyrics[i] != ']':
verse_metadata = verse_metadata + lyrics[i]
i = i + 1
if i < len(lyrics):
verse_metadata = verse_metadata + lyrics[i]
else:
verse_lyrics = verse_lyrics + lyrics[i]
i = i + 1
verses = write_verse(verse_metadata, verse_lyrics, verses)
return {
'title': song['title'],
'verses': verses,
'artist': song['artist'],
'featured_artists': song['featured_artists']
}
def get_song_name(song):
return song
loop_and_process(song_list, process_song, "Song", get_song_name,
dir_prefix)
def process_artist(name, bar):
artist = genius.search_artist(name)
songs = artist.songs
def process_song(song, bar):
return {
'title': song.title,
'artist': song.artist,
'lyrics': song.lyrics,
'featured_artists': [a['name'] for a in song.featured_artists]
}
def get_song_name(song):
return song.artist + artist_song_split_token + song.title
loop_and_process(songs, process_song, "Song", get_song_name, raw_songs_dir)
return None
def run_bpe_on_songs(codes_file, song_list_path, out_dir):
bpe = get_bpe_object(codes_file)
song_list = read_list_from_file(song_list_path)
def process_song(song, bar):
with open("{}/{}.json".format(verses_with_tokens,
name_to_file_name(song))) as song_file:
song = json.load(song_file)
for verse in song['verses']:
if verse['valid']:
lyrics = verse['lyrics']
lyrics = apply_bpe_to_string(lyrics, bpe)
verse['lyrics'] = lyrics
return song
loop_and_process(song_list, process_song, "Song", lambda x: x, out_dir)
def get_songs(name=None, csv=None):
artists = pd.DataFrame([], columns=['Artist'])
if csv is not None:
print("\n Getting lyrics for all artists in {}".format(csv))
with open(csv) as openfile:
artists = openfile.readlines()
artists = [artist.strip() for artist in artists]
elif name is not None:
print("\n Getting lyrics for {}".format(name))
artists = pd.DataFrame([name], columns=['Artist'])
else:
print("No Input Artists")
while len(artists) > 0:
try:
genius = instantiate_genius()
# functions
def process_artist(name, bar):
artist = genius.search_artist(name)
songs = artist.songs
def process_song(song, bar):
return {
'title': song.title,
'artist': song.artist,
'lyrics': song.lyrics,
'featured_artists': [a['name'] for a in song.featured_artists]
}
def get_song_name(song):
return song.artist + artist_song_split_token + song.title
loop_and_process(songs, process_song, "Song", get_song_name, raw_songs_dir)
return None
def get_artist_name(name):
return name
loop_and_process(
artists,
process_artist,
"Artist",
get_artist_name,
artist_lyric_dir,
)
except:
e = sys.exc_info()[0]
print(e)
finally:
completed_artists = read_list_from_file("{}/{}".format(artist_lyric_dir, "_LIST"))
for artist in completed_artists:
if artist in artists:
artists.remove(artist)
def artist_to_raw_song_files(artists_file):
with open(artists_file) as openfile:
artists = openfile.readlines()
artists = [artist.strip() for artist in artists]
for artist_name in tqdm(artists):
with open("{}/{}".format(artist_lyric_dir,
name_to_file_name(artist_name))) as jsonfile:
artist = json.load(jsonfile)
songs = artist["songs"]
def process_song(song):
return {
'title': song['title'],
'artist': song['primary_artist']['name'],
'lyrics': song['lyrics']
}
def get_song_name(song):
return song['title']
loop_and_process(songs, process_song, "Song", get_song_name,
raw_songs_dir)
def clean_song(song_list_path, out_dir):
song_list = read_list_from_file(song_list_path)
loop_and_process(song_list, process_song, "Song", lambda x: x, out_dir)
def verse_songs_extract_artists(song_list_path, dir_prefix):
song_list = read_list_from_file(song_list_path)
loop_and_process(song_list, process_song, "Song", get_song_name,
dir_prefix)
remove_duplicates_from_list_file(verse_artists_dir, artists_list_file)
remove_duplicates_from_list_file(verse_artists_dir, raw_artists_list_file)
def fix_tokens_for_verses(song_list_path, out_dir):
song_list = read_list_from_file(song_list_path)
loop_and_process(song_list, process_song, "Song", lambda x:x, out_dir)
