Simplify Musixmatch multi-part lyrics parser

This commit is contained in:
Evgeniy Gurevich 2021-04-02 10:35:53 +08:00 committed by GitHub
parent a8fe8ce9a7
commit afc072801c
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -292,15 +292,11 @@ class MusiXmatch(Backend):
% url)
return
html_parts = html.split('<p class="mxm-lyrics__content')
# Sometimes lyrics come in 2 parts
if len(html_parts) > 1:
html_part1 = html_parts[-2]
lyrics_part1 = extract_text_between(html_part1, '>', '</p>')
else:
lyrics_part1 = ''
html_part2 = html_parts[-1]
lyrics_part2 = extract_text_between(html_part2, '>', '</p>')
lyrics = lyrics_part1 + '\n' + lyrics_part2
# Sometimes lyrics come in 2 or more parts
lyrics_parts = []
for html_part in html_parts:
lyrics_parts.append(extract_text_between(html_part, '>', '</p>'))
lyrics = '\n'.join(lyrics_parts)
lyrics = lyrics.strip(',"').replace('\\n', '\n')
# another odd case: sometimes only that string remains, for
# missing songs. this seems to happen after being blocked
@ -308,7 +304,7 @@ class MusiXmatch(Backend):
if "Instant lyrics for all your music." in lyrics:
return
# sometimes there are non-existent lyrics with some content
if 'Think is wrong? You can always add the lyrics' in lyrics:
if 'Lyrics | Musixmatch' in lyrics:
return
return lyrics