]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/mdr.py
2 from __future__
import unicode_literals
4 from .common
import InfoExtractor
5 from ..compat
import compat_urlparse
17 class MDRIE(InfoExtractor
):
18 IE_DESC
= 'MDR.DE and KiKA'
19 _VALID_URL
= r
'https?://(?:www\.)?(?:mdr|kika)\.de/(?:.*)/[a-z-]+-?(?P<id>\d+)(?:_.+?)?\.html'
21 _GEO_COUNTRIES
= ['DE']
24 # MDR regularly deletes its videos
25 'url': 'http://www.mdr.de/fakt/video189002.html',
26 'only_matching': True,
29 'url': 'http://www.mdr.de/kultur/audio1312272_zc-15948bad_zs-86171fdd.html',
30 'md5': '64c4ee50f0a791deb9479cd7bbe9d2fa',
34 'title': 'Feuilleton vom 30. Oktober 2015',
36 'uploader': 'MITTELDEUTSCHER RUNDFUNK',
38 'skip': '404 not found',
40 'url': 'http://www.kika.de/baumhaus/videos/video19636.html',
41 'md5': '4930515e36b06c111213e80d1e4aad0e',
45 'title': 'Baumhaus vom 30. Oktober 2015',
49 'skip': '404 not found',
51 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/videos/video8182.html',
52 'md5': '5fe9c4dd7d71e3b238f04b8fdd588357',
56 'title': 'Beutolomäus und der geheime Weihnachtswunsch',
57 'description': 'md5:b69d32d7b2c55cbe86945ab309d39bbd',
58 'timestamp': 1482541200,
59 'upload_date': '20161224',
64 # audio with alternative playerURL pattern
65 'url': 'http://www.mdr.de/kultur/videos-und-audios/audio-radio/operation-mindfuck-robert-wilson100.html',
69 'title': 'Feature: Operation Mindfuck - Robert Anton Wilson',
71 'uploader': 'MITTELDEUTSCHER RUNDFUNK',
74 # empty bitrateVideo and bitrateAudio
75 'url': 'https://www.kika.de/filme/sendung128372_zc-572e3f45_zs-1d9fb70e.html',
79 'title': 'Der kleine Wichtel kehrt zurück',
80 'description': 'md5:f77fafdff90f7aa1e9dca14f662c052a',
82 'timestamp': 1607823300,
83 'upload_date': '20201213',
87 'skip_download': True,
90 'url': 'http://www.kika.de/baumhaus/sendungen/video19636_zc-fea7f8a0_zs-4bf89c60.html',
91 'only_matching': True,
93 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/einzelsendung2534.html',
94 'only_matching': True,
96 'url': 'http://www.mdr.de/mediathek/mdr-videos/a/video-1334.html',
97 'only_matching': True,
100 def _real_extract(self
, url
):
101 video_id
= self
._match
_id
(url
)
103 webpage
= self
._download
_webpage
(url
, video_id
)
105 data_url
= self
._search
_regex
(
106 r
'(?:dataURL|playerXml(?:["\'])?
)\s
*:\s
*(["\'])(?P<url>.+?-avCustom\.xml)\1',
107 webpage, 'data url', group='url').replace(r'\/', '/')
109 doc = self._download_xml(
110 compat_urlparse.urljoin(url, data_url), video_id)
112 title = xpath_text(doc, ['./title', './broadcast/broadcastName'], 'title', fatal=True)
114 type_ = xpath_text(doc, './type', default=None)
118 for asset in doc.findall('./assets/asset'):
121 'progressiveDownload',
122 'dynamicHttpStreamingRedirector',
123 'adaptiveHttpStreamingRedirector'):
124 url_el = asset.find('./%sUrl' % source)
128 video_url = url_or_none(url_el.text)
129 if not video_url or video_url in processed_urls:
132 processed_urls.append(video_url)
134 ext = determine_ext(video_url)
136 formats.extend(self._extract_m3u8_formats(
137 video_url, video_id, 'mp4', entry_protocol='m3u8_native',
138 quality=1, m3u8_id='HLS', fatal=False))
140 formats.extend(self._extract_f4m_formats(
141 video_url + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id,
142 quality=1, f4m_id='HDS', fatal=False))
144 media_type = xpath_text(asset, './mediaType', 'media type', default='MP4')
145 vbr = int_or_none(xpath_text(asset, './bitrateVideo', 'vbr'), 1000)
146 abr = int_or_none(xpath_text(asset, './bitrateAudio', 'abr'), 1000)
147 filesize = int_or_none(xpath_text(asset, './fileSize', 'file size'))
151 'format_id': join_nonempty(media_type, vbr or abr),
152 'filesize': filesize,
159 'width': int_or_none(xpath_text(asset, './frameWidth', 'width')),
160 'height': int_or_none(xpath_text(asset, './frameHeight', 'height')),
168 self._sort_formats(formats)
170 description = xpath_text(doc, './broadcast/broadcastDescription', 'description')
171 timestamp = parse_iso8601(
174 './broadcast/broadcastDate',
175 './broadcast/broadcastStartDate',
176 './broadcast/broadcastEndDate'],
177 'timestamp', default=None))
178 duration = parse_duration(xpath_text(doc, './duration', 'duration'))
179 uploader = xpath_text(doc, './rights', 'uploader')
184 'description': description,
185 'timestamp': timestamp,
186 'duration': duration,
187 'uploader': uploader,