0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-14 01:05:06 +00:00
yt-dlp/yt_dlp/extractor/drooble.py

114 lines
4.1 KiB
Python
Raw Normal View History

import json
from .common import InfoExtractor
from ..utils import (
ExtractorError,
int_or_none,
try_get,
)
class DroobleIE(InfoExtractor):
_VALID_URL = r'''(?x)https?://drooble\.com/(?:
(?:(?P<user>[^/]+)/)?(?P<kind>song|videos|music/albums)/(?P<id>\d+)|
(?P<user_2>[^/]+)/(?P<kind_2>videos|music))
'''
_TESTS = [{
'url': 'https://drooble.com/song/2858030',
'md5': '5ffda90f61c7c318dc0c3df4179eb064',
'info_dict': {
'id': '2858030',
'ext': 'mp3',
'title': 'Skankocillin',
'upload_date': '20200801',
'timestamp': 1596241390,
'uploader_id': '95894',
'uploader': 'Bluebeat Shelter',
},
}, {
'url': 'https://drooble.com/karl340758/videos/2859183',
'info_dict': {
'id': 'J6QCQY_I5Tk',
'ext': 'mp4',
'title': 'Skankocillin',
'uploader_id': 'UCrSRoI5vVyeYihtWEYua7rg',
'description': 'md5:ffc0bd8ba383db5341a86a6cd7d9bcca',
'upload_date': '20200731',
'uploader': 'Bluebeat Shelter',
},
}, {
'url': 'https://drooble.com/karl340758/music/albums/2858031',
'info_dict': {
'id': '2858031',
},
'playlist_mincount': 8,
}, {
'url': 'https://drooble.com/karl340758/music',
'info_dict': {
'id': 'karl340758',
},
'playlist_mincount': 8,
}, {
'url': 'https://drooble.com/karl340758/videos',
'info_dict': {
'id': 'karl340758',
},
'playlist_mincount': 8,
}]
def _call_api(self, method, video_id, data=None):
response = self._download_json(
f'https://drooble.com/api/dt/{method}', video_id, data=json.dumps(data).encode())
if not response[0]:
raise ExtractorError('Unable to download JSON metadata')
return response[1]
def _real_extract(self, url):
mobj = self._match_valid_url(url)
user = mobj.group('user') or mobj.group('user_2')
kind = mobj.group('kind') or mobj.group('kind_2')
display_id = mobj.group('id') or user
if mobj.group('kind_2') == 'videos':
data = {'from_user': display_id, 'album': -1, 'limit': 18, 'offset': 0, 'order': 'new2old', 'type': 'video'}
elif kind in ('music/albums', 'music'):
data = {'user': user, 'public_only': True, 'individual_limit': {'singles': 1, 'albums': 1, 'playlists': 1}}
else:
data = {'url_slug': display_id, 'children': 10, 'order': 'old2new'}
method = 'getMusicOverview' if kind in ('music/albums', 'music') else 'getElements'
json_data = self._call_api(method, display_id, data=data)
if kind in ('music/albums', 'music'):
json_data = json_data['singles']['list']
entites = []
for media in json_data:
url = media.get('external_media_url') or media.get('link')
if url.startswith('https://www.youtube.com'):
entites.append({
'_type': 'url',
'url': url,
'ie_key': 'Youtube',
})
continue
is_audio = (media.get('type') or '').lower() == 'audio'
entites.append({
'url': url,
'id': media['id'],
'title': media['title'],
'duration': int_or_none(media.get('duration')),
'timestamp': int_or_none(media.get('timestamp')),
'album': try_get(media, lambda x: x['album']['title']),
'uploader': try_get(media, lambda x: x['creator']['display_name']),
'uploader_id': try_get(media, lambda x: x['creator']['id']),
'thumbnail': media.get('image_comment'),
'like_count': int_or_none(media.get('likes')),
'vcodec': 'none' if is_audio else None,
'ext': 'mp3' if is_audio else None,
})
if len(entites) > 1:
return self.playlist_result(entites, display_id)
return entites[0]