1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-23 02:32:16 +00:00

Scrape actors

modified:   youtube_dl/extractor/xvideos.py
This commit is contained in:
DarkFighterLuke 2022-02-26 14:55:53 +01:00
parent b7340781df
commit 16a2888b1a

View file

@ -36,6 +36,7 @@ class XVideosIE(InfoExtractor):
'age_limit': 18, 'age_limit': 18,
'tags': ['lesbian', 'teen', 'hardcore', 'latina', 'rough', 'squirt', 'big-ass', 'cheater', 'twistys', 'cheat', 'ass-play', 'when-girls-play'], 'tags': ['lesbian', 'teen', 'hardcore', 'latina', 'rough', 'squirt', 'big-ass', 'cheater', 'twistys', 'cheat', 'ass-play', 'when-girls-play'],
'creator': 'Twistys', 'creator': 'Twistys',
'actors': [{'given_name': 'Adriana Chechik', 'url': 'https://www.xvideos.com/pornstars/adriana-chechik'}, {'given_name': 'Abella Danger', 'url': 'https://www.xvideos.com/pornstars/abella-danger'}],
} }
}, { }, {
'url': 'https://flashservice.xvideos.com/embedframe/4588838', 'url': 'https://flashservice.xvideos.com/embedframe/4588838',
@ -144,6 +145,14 @@ class XVideosIE(InfoExtractor):
creator = self._search_regex(r'<span class="name">(?P<creator>.+?)<', webpage, 'creator', group='creator') creator = self._search_regex(r'<span class="name">(?P<creator>.+?)<', webpage, 'creator', group='creator')
actors_data = re.findall(r'href="(?P<actor_url>/pornstars/.+?)" class="btn btn-default label profile hover-name"><span class="name">(?P<actor_name>.+?)</span>', webpage)
actors = []
for actor_tuple in actors_data:
actors.append({
'given_name': actor_tuple[1],
'url': urljoin(url, actor_tuple[0]),
})
return { return {
'id': video_id, 'id': video_id,
'formats': formats, 'formats': formats,
@ -153,4 +162,5 @@ class XVideosIE(InfoExtractor):
'age_limit': 18, 'age_limit': 18,
'tags': tags, 'tags': tags,
'creator': creator, 'creator': creator,
'actors': actors,
} }