[bandcamp:album] Fix title extraction (Closes #10455)

This commit is contained in:
Sergey M․ 2016-08-31 00:29:49 +07:00
parent 245023a861
commit 64fc49aba0
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
1 changed files with 14 additions and 2 deletions

View File

@ -162,6 +162,15 @@ class BandcampAlbumIE(InfoExtractor):
'uploader_id': 'dotscale', 'uploader_id': 'dotscale',
}, },
'playlist_mincount': 7, 'playlist_mincount': 7,
}, {
# with escaped quote in title
'url': 'https://jstrecords.bandcamp.com/album/entropy-ep',
'info_dict': {
'title': '"Entropy" EP',
'uploader_id': 'jstrecords',
'id': 'entropy-ep',
},
'playlist_mincount': 3,
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -176,8 +185,11 @@ class BandcampAlbumIE(InfoExtractor):
entries = [ entries = [
self.url_result(compat_urlparse.urljoin(url, t_path), ie=BandcampIE.ie_key()) self.url_result(compat_urlparse.urljoin(url, t_path), ie=BandcampIE.ie_key())
for t_path in tracks_paths] for t_path in tracks_paths]
title = self._search_regex( title = self._html_search_regex(
r'album_title\s*:\s*"(.*?)"', webpage, 'title', fatal=False) r'album_title\s*:\s*"((?:\\.|[^"\\])+?)"',
webpage, 'title', fatal=False)
if title:
title = title.replace(r'\"', '"')
return { return {
'_type': 'playlist', '_type': 'playlist',
'uploader_id': uploader_id, 'uploader_id': uploader_id,