diff options
Diffstat (limited to 'scrap/parser.py')
| -rw-r--r-- | scrap/parser.py | 37 |
1 files changed, 13 insertions, 24 deletions
diff --git a/scrap/parser.py b/scrap/parser.py index a191fa2..c447a0d 100644 --- a/scrap/parser.py +++ b/scrap/parser.py @@ -4,54 +4,44 @@ import structures import os -def process_json_file(name, artist_id): +def process_json_file(name, album_id, artist_id): link = getLink(name) file_json = open(link, "r") album_json = file_json.read() file_json.close() - return process_json(album_json, artist_id) + return process_json(album_json, album_id, artist_id) def getLink(name): return "Lyrics_"+name.replace(" ", "")+".json" -def process_json(album_json, artist_id): +def process_json(album_json, album_id, artist_id): data = json.loads(album_json) - name = data["name"].lower() + album_name = data["name"].lower() - artist_name_api = data["artist"]["name"].lower() - artist_name = format_artist_name(artist_name_api) off_cover = data["cover_art_thumbnail_url"] - new_cover = get_cover_link(artist_name, name) - download_cover(off_cover, new_cover, artist_name) + new_cover = get_cover_link(artist_id, album_id) + download_cover(off_cover, new_cover, artist_id) songs = [analyze_song(song) for song in data["tracks"]] - return structures.album(name, new_cover, songs, artist_id) + return structures.album(album_name, new_cover, songs, artist_id) -def format_artist_name(name): - name = name.replace(" ", "") - name = re.sub(r'[\(\[].*?[\)\]]', '', name) - return name +def get_cover_link(artist_id, album_id): + cover_link = f"covers/{artist_id}/{album_id}.png" + return cover_link -def get_cover_link(artist_name, album_name): - album_name = album_name.replace(" ", "") - location = f"covers/{artist_name}/{album_name}.png" - - return location - - -def download_cover(off_cover, new_cover, artist_name): +def download_cover(off_cover, new_cover, artist_id): if not os.path.isdir("covers"): os.system("mkdir covers") - if not os.path.isdir(f"covers/{artist_name}"): - os.system(f"mkdir 'covers/{artist_name}'") + if not os.path.isdir(f"covers/{artist_id}"): + os.system(f"mkdir 'covers/{artist_id}'") if not os.path.isfile(new_cover): os.system(f"wget {off_cover} -O '{new_cover}'") @@ -78,7 +68,6 @@ def format_lyrics(lyrics): lyrics = re.sub(r'[\(\[].*?[\)\]]', '', lyrics) lyrics = re.sub("\n{3,}", "\n\n", lyrics) - lyrics = lyrics.replace("\u0435", "e") lyrics = lyrics.replace("\u2005", " ") while lyrics[0] == '\n': |