#!/usr/bin/env python3 from html.parser import HTMLParser from urllib.request import urlopen import json class Bandcamp: def __init__(self): self.ldjson = None def mesg(self, msg, t=None): self.util.mesg(msg, t) def match_urls(self, str): r = [ i for i in str.split() if "https://" in i and ("bandcamp.com/album/" in i or "bandcamp.com/track/" in i) ] r = list(dict.fromkeys(r)) n = 0 for i in r: if not i.startswith("http"): r.pop(n) n += 1 return r class parseprop(HTMLParser): def __init__(self): # print("bandcamp parse init") HTMLParser.__init__(self) self.ldjson = False def handle_starttag(self, tag, attrs): if tag == "script" and ("type", "application/ld+json") in attrs: self.ldjson = True else: self.ldjson = False def handle_endtag(self, tag): self.ldjson = False def handle_data(self, data): if self.ldjson: Bandcamp.ldjson = data return def bandcamp(self, url): url = url.rstrip("\x01") p = self.parseprop() data = urlopen(url).read().decode() p.feed(data) irc_string = "[\x0304BandCamp\x03] \x0307ERROR:\x0308 got no data from server! \x0315(check your URL for typos!)\x03" ansi_string = "[\x1b[31mBandCamp\x1b[0m] \x1b[33;2mERROR:\x1b[33;1m got no data from server! \x1b[37;2m(check your URL for typos!)\x1b[0m" data = json.loads(Bandcamp.ldjson) # print(data) # for i in data: # print(i) try: type = data["@type"] except KeyError: print(ansi_string) return irc_string, True id = data["@id"] name = data["name"] date = data["datePublished"] artists = data["byArtist"]["name"] # artists=artists.removeprefix(f'Listen to {name} on Spotify').removeprefix('.').strip().removeprefix('· ') # if artists.startswith("Song · "): artists=artists.removeprefix("Song · ") # elif artists.startswith("Album · "): # artists=artists.removeprefix("Album · ")[::-1].split(" · ",1)[1][::-1] #removes the "10 songs" part from album # artists=artists.removesuffix(f" · {date[:4]}") #remove the year too irc_string = f"[\x0303BandCamp\x03] \x02{name}\x02 by \x1d{artists}\x1d published on {date}" ansi_string = f"[\x1b[32mBandCamp\x1b[0m] \x1b[1m{name}\x1b[0m by \x1b[03m{artists}\x1b[0m published on {date}" # print(("Song: " if type=="MusicRecording" else "Album: " if type=="MusicAlbum" else f"Unknown type ({type}): ")+'"'+name+'"'+" by "+'"'+artists+'"'+" released on "+date) print(ansi_string) return irc_string, False if __name__ == "__main__": import sys Bandcamp.bandcamp(Bandcamp, sys.argv[1])