mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-01-07 17:16:08 +00:00
parent
0f897e0929
commit
45d20488f1
|
@ -1143,6 +1143,7 @@ from .udemy import (
|
|||
from .udn import UDNEmbedIE
|
||||
from .uktvplay import UKTVPlayIE
|
||||
from .digiteka import DigitekaIE
|
||||
from .umg import UMGDeIE
|
||||
from .unistra import UnistraIE
|
||||
from .unity import UnityIE
|
||||
from .uol import UOLIE
|
||||
|
|
103
youtube_dl/extractor/umg.py
Normal file
103
youtube_dl/extractor/umg.py
Normal file
|
@ -0,0 +1,103 @@
|
|||
# coding: utf-8
|
||||
from __future__ import unicode_literals
|
||||
|
||||
from .common import InfoExtractor
|
||||
from ..utils import (
|
||||
int_or_none,
|
||||
parse_filesize,
|
||||
parse_iso8601,
|
||||
)
|
||||
|
||||
|
||||
class UMGDeIE(InfoExtractor):
|
||||
IE_NAME = 'umg:de'
|
||||
IE_DESC = 'Universal Music Deutschland'
|
||||
_VALID_URL = r'https?://(?:www\.)?universal-music\.de/[^/]+/videos/[^/?#]+-(?P<id>\d+)'
|
||||
_TEST = {
|
||||
'url': 'https://www.universal-music.de/sido/videos/jedes-wort-ist-gold-wert-457803',
|
||||
'md5': 'ebd90f48c80dcc82f77251eb1902634f',
|
||||
'info_dict': {
|
||||
'id': '457803',
|
||||
'ext': 'mp4',
|
||||
'title': 'Jedes Wort ist Gold wert',
|
||||
'timestamp': 1513591800,
|
||||
'upload_date': '20171218',
|
||||
}
|
||||
}
|
||||
|
||||
def _real_extract(self, url):
|
||||
video_id = self._match_id(url)
|
||||
video_data = self._download_json(
|
||||
'https://api.universal-music.de/graphql',
|
||||
video_id, query={
|
||||
'query': '''{
|
||||
universalMusic(channel:16) {
|
||||
video(id:%s) {
|
||||
headline
|
||||
formats {
|
||||
formatId
|
||||
url
|
||||
type
|
||||
width
|
||||
height
|
||||
mimeType
|
||||
fileSize
|
||||
}
|
||||
duration
|
||||
createdDate
|
||||
}
|
||||
}
|
||||
}''' % video_id})['data']['universalMusic']['video']
|
||||
|
||||
title = video_data['headline']
|
||||
hls_url_template = 'http://mediadelivery.universal-music-services.de/vod/mp4:autofill/storage/' + '/'.join(list(video_id)) + '/content/%s/file/playlist.m3u8'
|
||||
|
||||
thumbnails = []
|
||||
formats = []
|
||||
|
||||
def add_m3u8_format(format_id):
|
||||
m3u8_formats = self._extract_m3u8_formats(
|
||||
hls_url_template % format_id, video_id, 'mp4',
|
||||
'm3u8_native', m3u8_id='hls', fatal='False')
|
||||
if m3u8_formats and m3u8_formats[0].get('height'):
|
||||
formats.extend(m3u8_formats)
|
||||
|
||||
for f in video_data.get('formats', []):
|
||||
f_url = f.get('url')
|
||||
mime_type = f.get('mimeType')
|
||||
if not f_url or mime_type == 'application/mxf':
|
||||
continue
|
||||
fmt = {
|
||||
'url': f_url,
|
||||
'width': int_or_none(f.get('width')),
|
||||
'height': int_or_none(f.get('height')),
|
||||
'filesize': parse_filesize(f.get('fileSize')),
|
||||
}
|
||||
f_type = f.get('type')
|
||||
if f_type == 'Image':
|
||||
thumbnails.append(fmt)
|
||||
elif f_type == 'Video':
|
||||
format_id = f.get('formatId')
|
||||
if format_id:
|
||||
fmt['format_id'] = format_id
|
||||
if mime_type == 'video/mp4':
|
||||
add_m3u8_format(format_id)
|
||||
urlh = self._request_webpage(f_url, video_id, fatal=False)
|
||||
if urlh:
|
||||
first_byte = urlh.read(1)
|
||||
if first_byte not in (b'F', b'\x00'):
|
||||
continue
|
||||
formats.append(fmt)
|
||||
if not formats:
|
||||
for format_id in (867, 836, 940):
|
||||
add_m3u8_format(format_id)
|
||||
self._sort_formats(formats, ('width', 'height', 'filesize', 'tbr'))
|
||||
|
||||
return {
|
||||
'id': video_id,
|
||||
'title': title,
|
||||
'duration': int_or_none(video_data.get('duration')),
|
||||
'timestamp': parse_iso8601(video_data.get('createdDate'), ' '),
|
||||
'thumbnails': thumbnails,
|
||||
'formats': formats,
|
||||
}
|
Loading…
Reference in a new issue