Backup_Repos/beets
1
0
Fork 0
mirror of https://github.com/beetbox/beets.git synced 2026-02-01 21:11:25 +01:00
Code Issues Projects Releases Wiki Activity

Remove unnecessary intermediate web request to genius.com when fetching lyrics.

Browse source 
The search results already include the correct song page url, making it superfluous to do another request via the /song api just to get it.
This commit is contained in:
stlutz 2020-05-16 16:28:17 +02:00
parent ac043f9be0
commit 46143d9762
2 changed files with 8 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
beetsplug/lyrics.py
View file

@ -359,14 +359,9 @@ class Genius(Backend):
'User-Agent': USER_AGENT,
}
def lyrics_from_song_api_path(self, song_api_path):
song_url = self.base_url + song_api_path
response = requests.get(song_url, headers=self.headers)
json = response.json()
path = json["response"]["song"]["path"]
def lyrics_from_song_page(self, page_url):
# Gotta go regular html scraping... come on Genius.
page_url = "https://genius.com" + path
self._log.debug(u'fetching lyrics from: {0}', page_url)
try:
page = requests.get(page_url)
except requests.RequestException as exc:
@ -404,7 +399,6 @@ class Genius(Backend):
self._log.debug(u'Genius API request returned invalid JSON')
return None
song_info = None
for hit in json["response"]["hits"]:
# Genius uses zero-width characters to denote lowercase
# artist names.
@ -412,15 +406,9 @@ class Genius(Backend):
strip(u'\u200b').lower()
if hit_artist == artist.lower():
song_info = hit
break
return self.lyrics_from_song_page(hit["result"]["url"])
if song_info:
self._log.debug(u'fetched: {0}', song_info["result"]["url"])
song_api_path = song_info["result"]["api_path"]
return self.lyrics_from_song_api_path(song_api_path)
else:
self._log.debug(u'genius: no matching artist')
self._log.debug(u'genius: no matching artist')
class LyricsWiki(SymbolsReplaced):

8
test/test_lyrics.py
View file

@ -456,7 +456,7 @@ class LyricsGeniusBaseTest(unittest.TestCase):
self.skipTest("Python's built-in HTML parser is not good enough")
class LyricsGeniusScrapTest(LyricsGeniusBaseTest):
class LyricsGeniusScrapeTest(LyricsGeniusBaseTest):
"""Checks that Genius backend works as intended.
"""
@ -469,12 +469,12 @@ class LyricsGeniusScrapTest(LyricsGeniusBaseTest):
@patch.object(requests, 'get', GeniusMockGet())
def test_no_lyrics_div(self):
"""Ensure that `lyrics_from_song_api_path` doesn't crash when the html
for a Genius page contain <div class="lyrics"></div>
"""Ensure that `lyrics_from_song_page` doesn't crash when the html
for a Genius page doesn't contain <div class="lyrics"></div>
"""
# https://github.com/beetbox/beets/issues/3535
# expected return value None
self.assertEqual(genius.lyrics_from_song_api_path('/nolyric'),
self.assertEqual(genius.lyrics_from_song_page('https://genius.com/sample'),
None)