async def extract_lyrics(cls, request: Request) -> Lyrics:
request.headers = {
"user-agent":
"Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:59.0) Gecko/20100101 Firefox/59.0"
}
bs = await request.bs
center = bs.body.find("div",
{"class": "col-xs-12 col-lg-8 text-center"})
if not center:
raise NoLyrics
lyrics = center.find("div", {"class": None}).text
lyrics = re.sub(r"<br>", " ", lyrics)
lyrics = re.sub(r"<i?>\W*", "[", lyrics)
lyrics = re.sub(r"\W*</i>", "]", lyrics)
lyrics = re.sub(r""", "\"", lyrics)
lyrics = re.sub(r"</div>", "", lyrics)
title = center.find("h1").text.strip()[1:-8]
artist = bs.select_one("div.lyricsh h2 b").string[:-7]
lyrics = lyrics.strip()
return Lyrics(title, lyrics, artist=artist)
async def test_can_handle(self):
async with ClientSession() as session:
assert await Lyricsmode.can_handle(
Request(
session,
"https://www.lyricsmode.com/lyrics/e/ed_sheeran/a_team_lyrics.html"
)) is True
async def test_translated_extraction(self):
async with ClientSession() as session:
lyrics = await Animelyrics.extract_lyrics(Request(session, "http://www.animelyrics.com/anime/swimminganime/splashfree.htm"))
lyrics_hash = hashlib.sha256(lyrics.lyrics.encode("utf-8")).hexdigest()
assert lyrics_hash == "f8b7b6fc53bdbc738a1b8beea09d086c6ca362c4c4857930488bc10153716dfd"
assert lyrics.title == "SPLASH FREE"
assert lyrics.artist == "STYLE FIVE"
async with ClientSession() as session:
lyrics = await Animelyrics.extract_lyrics(Request(session, "http://www.animelyrics.com/anime/kmb/fnknhnh.htm"))
assert lyrics.title == "Futari no Kimochi no Honto no Himitsu"
assert lyrics.artist == "Yasuna"
assert lyrics.lyrics == lyrics_text
async def test_can_handle(self):
async with ClientSession() as session:
assert await AZLyrics.can_handle(
Request(
session,
"https://www.azlyrics.com/lyrics/edsheeran/theateam.html")
) is True
async def test_can_handle(self):
async with ClientSession() as session:
assert await MusixMatch.can_handle(
Request(
session,
"https://www.musixmatch.com/lyrics/Dua-Lipa/New-Rules")
) is True
async def test_can_handle(self):
async with ClientSession() as session:
assert await LyricalNonsense.can_handle(
Request(
session,
"http://www.lyrical-nonsense.com/lyrics/radwimps/zen-zen-zense/"
)) is True
async def test_untranslated_extraction(self):
async with ClientSession() as session:
lyrics = await Animelyrics.extract_lyrics(Request(session, "https://www.animelyrics.com/anime/accelworld/chasetheworld.htm"))
lyrics_hash = hashlib.sha256(lyrics.lyrics.encode("utf-8")).hexdigest()
assert lyrics_hash == "3cf6a7bc4ef62cccca49e228ef556b58d459c68f4cb0dba7240cc4dffd6b3b20"
assert lyrics.title == "Chase the world"
assert lyrics.artist == "May'n"
async def test_extraction(self):
async with ClientSession() as session:
lyrics = await Genius.extract_lyrics(
Request(session,
"https://genius.com/Ed-sheeran-the-a-team-lyrics"))
assert lyrics.lyrics == lyrics_text
assert lyrics.title == "The A Team"
assert lyrics.artist == "Ed Sheeran"
assert lyrics.release_date == datetime(2011, 6, 12)
async def test_extraction(self):
async with ClientSession() as session:
lyrics = await Lyricsmode.extract_lyrics(
Request(
session,
"https://www.lyricsmode.com/lyrics/e/ed_sheeran/a_team_lyrics.html"
))
assert lyrics.title == "A Team"
assert lyrics.artist == "Ed Sheeran"
assert lyrics.lyrics == lyrics_ed_sheeran_a_team
async def test_extraction(self):
async with ClientSession() as session:
lyrics = await MusixMatch.extract_lyrics(
Request(
session,
"https://www.musixmatch.com/lyrics/Dua-Lipa/New-Rules"))
assert lyrics.title == "New Rules"
assert lyrics.artist == "Dua Lipa"
assert lyrics.release_date == datetime(2017, 6, 2)
assert lyrics.lyrics == lyrics_dua_lipa_new_rules
async def test_extraction(self):
async with ClientSession() as session:
lyrics = await AZLyrics.extract_lyrics(
Request(
session,
"https://www.azlyrics.com/lyrics/edsheeran/theateam.html"))
lyrics_hash = hashlib.sha256(lyrics.lyrics.encode("utf-8")).hexdigest()
assert lyrics_hash == "e77a63fb93b1d0f373b859963532e41a2dbf2d68d290bf3f919b93b174fe26e3"
assert lyrics.title == "The A Team"
assert lyrics.artist == "Ed Sheeran"
async def test_extraction(self):
async with ClientSession() as session:
lyrics = await LyricalNonsense.extract_lyrics(
Request(
session,
"http://www.lyrical-nonsense.com/lyrics/radwimps/zen-zen-zense/"
))
assert lyrics.title == "Zenzenzense"
assert lyrics.artist == "RADWIMPS"
assert lyrics.lyrics == lyrics_radwimps_zenzenzense
async with ClientSession() as session:
lyrics = await LyricalNonsense.extract_lyrics(
Request(
session,
"https://www.lyrical-nonsense.com/lyrics/himouto-umaru-chan-r-theme-songs/umarun-taisou-sisters/"
))
assert lyrics.title == "Umarun Taisou"
assert lyrics.artist == "SisterS"
assert lyrics.lyrics == lyrics_sisters_umarun_taisou
async def extract_lyrics(cls, request: Request) -> Lyrics:
bs = await request.bs
title = next(bs.select_one("div ~ h1").children).string.strip()
artist = bs.find(text=ARTIST_MATCHER)
if artist:
artist = ARTIST_MATCHER.match(artist).group(1)
lyrics_window = bs.find("table", attrs={"cellspacing": "0", "border": "0"})
if lyrics_window: # shit's been translated
log.info("these lyrics have been translated... sighs...")
lines = lyrics_window.find_all("tr")
lyrics = ""
for line in lines:
p = line.td
if p:
p.span.dt.replace_with("")
for br in p.span.find_all("br"):
br.replace_with("\n")
lyrics += p.span.text
lyrics = lyrics.strip()
else:
text_url = re.sub(r"\.html?", ".txt", request.url)
request.url = text_url
request.resp_kwargs["allow_redirects"] = False
text = await request.text
content = text.strip()
match = re.search(r"-{10,}(.+?)-{10,}", content, flags=re.DOTALL)
if match:
lyrics = match.group(1).strip()
else:
raise NoLyrics
lyrics = lyrics.replace("\xa0", " ").replace("\r", "")
return Lyrics(title, lyrics, artist=artist)
async def extract_lyrics(cls, request: Request) -> Lyrics:
request.headers = {
"user-agent":
"Mozilla/5.0 (Windows NT 6.1; WOW64; rv:40.0) Gecko/20100101 Firefox/40.1"
}
try:
(await request.resp).raise_for_status()
except ClientResponseError:
raise exceptions.NotAllowedError
bs = await request.bs
if bs.find_all("div",
attrs={
"class": "mxm-empty-state",
"data-reactid": "87"
}):
raise exceptions.NoLyrics
lyrics_frame = bs.find_all("div", {"class": "mxm-lyrics"})
if not lyrics_frame:
raise exceptions.NoLyrics
lyrics_window = lyrics_frame[0].find_all("div",
{"class": "mxm-lyrics"})
if not lyrics_window:
raise exceptions.NoLyrics
lyrics_window = lyrics_window[0].span
for garbage in bs.find_all("script"):
garbage.replace_with(2 * "\n")
lyrics = lyrics_window.text
title = bs.find("h1", attrs={
"class": "mxm-track-title__track"
}).contents[-1].strip()
artist = bs.select_one("a.mxm-track-title__artist").string
release_date = None
date_str = bs.select_one(
"div.mxm-track-footer__album h3.mui-cell__subtitle")
if date_str:
date_str = ORDINAL_MATCHER.sub(lambda m: m.group(1).zfill(2),
date_str.string)
release_date = datetime.strptime(date_str, "%b %d %Y")
return Lyrics(title, lyrics, artist=artist, release_date=release_date)
async def test_can_handle(self):
async with ClientSession() as session:
assert await Animelyrics.can_handle(Request(session, "http://www.animelyrics.com/anime/swimminganime/splashfree.htm")) is True
async def test_can_handle(self):
async with ClientSession() as session:
assert await Genius.can_handle(
Request(session,
"https://genius.com/Ed-sheeran-the-a-team-lyrics")
) is True