84 lines
3.2 KiB
Python
84 lines
3.2 KiB
Python
from lyricsgenius import Genius
|
|
import json
|
|
import re
|
|
import os
|
|
|
|
class GetLyrics():
|
|
def __init__(self):
|
|
self.GENIUS_TOKEN = "gMnJyj87FvjyP2W093rQ_mjo5ZwwLw1u2r0AmcVqYcJ8kkjjW6ZbObeGnS726SrH"
|
|
self.notes_re = re.compile('((?:\[[0-9a-zA-Z :()&+-.]+\])(?: \+ \([a-zA-Z -.]+)?(?:\\n)?)')
|
|
self.footer_re = re.compile('((?:EmbedShare)[ ]*(?:URLCopyEmbedCopy))')
|
|
self.multiline_re = re.compile(('(\\n){2,}'))
|
|
|
|
def get_songs(self, artists: dict = None):
|
|
session = Genius(self.GENIUS_TOKEN, retries=2, timeout=20, sleep_time=0.3)
|
|
|
|
if artists is not None:
|
|
# get songs
|
|
for artist in artists:
|
|
try:
|
|
songlist = session.search_artist(artist, max_songs=75, sort='title')
|
|
except Exception as e:
|
|
print(e)
|
|
continue
|
|
|
|
songlist.save_lyrics()
|
|
|
|
return songlist
|
|
else:
|
|
print("Please enter an artist")
|
|
|
|
def sanitize_lyrics(self, input):
|
|
sanitized_input = self.notes_re.sub('', input)
|
|
sanitized_input = self.footer_re.sub('', sanitized_input)
|
|
sanitized_input = sanitized_input.replace("[?]", '')
|
|
sanitized_input = self.multiline_re.sub('\n', sanitized_input)
|
|
return sanitized_input
|
|
|
|
def get_lyrics_from_json(self, json_file):
|
|
artist_dict = json.load(json_file)
|
|
ready_lyrics = []
|
|
print(artist_dict.keys())
|
|
for song in artist_dict['songs']:
|
|
sanitized_lyrics = self.sanitize_lyrics(song['lyrics'])
|
|
print(sanitized_lyrics)
|
|
ready_lyrics.append(sanitized_lyrics)
|
|
return ready_lyrics
|
|
|
|
def save_sanitized_lyrics(self):
|
|
sanitized_lyrics_list = []
|
|
for file in os.listdir("./"):
|
|
if file.endswith(".json"):
|
|
with open(file, 'r', encoding="utf-8") as read_file:
|
|
sanitized_lyrics_list.extend(self.get_lyrics_from_json(read_file))
|
|
print(sanitized_lyrics_list)
|
|
with open('./lyrics.txt', 'a+', encoding="utf-8") as lyrics_file:
|
|
for lyrics in sanitized_lyrics_list:
|
|
print(lyrics)
|
|
lyrics_file.write(f"{lyrics}\n")
|
|
|
|
def sanitize_file(self):
|
|
sanitized_lyrics = ''
|
|
with open('./lyrics.txt', 'r', encoding="utf-8") as lyrics_file:
|
|
file_read_lines = lyrics_file.read()
|
|
sanitized_lyrics = self.sanitize_lyrics(file_read_lines)
|
|
|
|
del lyrics_file
|
|
|
|
with open('./lyrics.txt', 'w', encoding="utf-8") as lyrics_file:
|
|
lyrics_file.write(sanitized_lyrics)
|
|
|
|
|
|
lyrics = GetLyrics()
|
|
|
|
#lyrics.get_songs([])
|
|
#"noel miller", "tiny meat gang", "young gravy", "parov stelar", "mac miller", "marilyn manson"
|
|
#"ozzy osborn", "eminem", "2 chainz", "elvis presley", "jim croce", "nwa", "nas", "biggie", "willie nelson", "snoop dog",
|
|
#"lil nas x", "lil wayne", "post malone", "michael jackson", "misterwives", "p!nk", "labrinth", "elton john",
|
|
#"the lovin spoonful", "blake shelton", "avenge sevenfold", "wiz khalifa", "we the kings", "leonard cohen",
|
|
#"john denver", "yes", "pink floyd", "paul simon", "u2"
|
|
|
|
#lyrics.save_sanitized_lyrics()
|
|
|
|
lyrics.sanitize_file()
|