1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
6 from .common import InfoExtractor
17 class WDRIE(InfoExtractor):
18 _CURRENT_MAUS_URL = r'https?://(?:www\.)wdrmaus.de/(?:[^/]+/){1,2}[^/?#]+\.php5'
19 _PAGE_REGEX = r'/(?:mediathek/)?(?P<media_type>[^/]+)/(?P<type>[^/]+)/(?P<display_id>.+)\.html'
20 _VALID_URL = r'(?P<page_url>https?://(?:www\d\.)?wdr\d?\.de)' + _PAGE_REGEX + '|' + _CURRENT_MAUS_URL
24 'url': 'http://www1.wdr.de/mediathek/video/sendungen/doku-am-freitag/video-geheimnis-aachener-dom-100.html',
25 # HDS download, MD5 is unstable
29 'display_id': 'doku-am-freitag/video-geheimnis-aachener-dom-100',
30 'title': 'Geheimnis Aachener Dom',
31 'alt_title': 'Doku am Freitag',
32 'upload_date': '20160304',
33 'description': 'md5:87be8ff14d8dfd7a7ee46f0299b52318',
35 'subtitles': {'de': [{
36 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml'
41 'url': 'http://www1.wdr.de/mediathek/audio/wdr3/wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100.html',
42 'md5': 'f4c1f96d01cf285240f53ea4309663d8',
46 'display_id': 'wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100',
47 'title': 'Schriftstellerin Juli Zeh',
48 'alt_title': 'WDR 3 Gespräch am Samstag',
49 'upload_date': '20160312',
50 'description': 'md5:e127d320bc2b1f149be697ce044a3dd7',
56 'url': 'http://www1.wdr.de/mediathek/video/live/index.html',
60 'display_id': 'index',
61 'title': r're:^WDR Fernsehen im Livestream [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
62 'alt_title': 'WDR Fernsehen Live',
64 'description': 'md5:ae2ff888510623bf8d4b115f95a9b7c9',
69 'skip_download': True, # m3u8 download
73 'url': 'http://www1.wdr.de/mediathek/video/sendungen/aktuelle-stunde/aktuelle-stunde-120.html',
74 'playlist_mincount': 8,
76 'id': 'aktuelle-stunde/aktuelle-stunde-120',
80 'url': 'http://www.wdrmaus.de/aktuelle-sendung/index.php5',
84 'upload_date': 're:^[0-9]{8}$',
85 'title': 're:^Die Sendung mit der Maus vom [0-9.]{10}$',
86 'description': '- Die Sendung mit der Maus -',
88 'skip': 'The id changes from week to week because of the new episode'
91 'url': 'http://www.wdrmaus.de/sachgeschichten/sachgeschichten/achterbahn.php5',
92 # HDS download, MD5 is unstable
96 'upload_date': '20130919',
97 'title': 'Sachgeschichte - Achterbahn ',
98 'description': '- Die Sendung mit der Maus -',
102 'url': 'http://www1.wdr.de/radio/player/radioplayer116~_layout-popupVersion.html',
106 'title': 'Funkhaus Europa Livestream',
107 'description': 'md5:2309992a6716c347891c045be50992e4',
108 'upload_date': '20160101',
113 def _real_extract(self, url):
114 mobj = re.match(self._VALID_URL, url)
115 url_type = mobj.group('type')
116 page_url = mobj.group('page_url')
117 display_id = mobj.group('display_id')
118 webpage = self._download_webpage(url, display_id)
120 # for wdr.de the data-extension is in a tag with the class "mediaLink"
121 # for wdr.de radio players, in a tag with the class "wdrrPlayerPlayBtn"
122 # for wdrmaus its in a link to the page in a multiline "videoLink"-tag
123 json_metadata = self._html_search_regex(
124 r'class=(?:"(?:mediaLink|wdrrPlayerPlayBtn)\b[^"]*"[^>]+|"videoLink\b[^"]*"[\s]*>\n[^\n]*)data-extension="([^"]+)"',
125 webpage, 'media link', default=None, flags=re.MULTILINE)
127 if not json_metadata:
129 self.url_result(page_url + href[0], 'WDR')
130 for href in re.findall(
131 r'<a href="(%s)"[^>]+data-extension=' % self._PAGE_REGEX,
135 if entries: # Playlist page
136 return self.playlist_result(entries, playlist_id=display_id)
138 raise ExtractorError('No downloadable streams found', expected=True)
140 media_link_obj = self._parse_json(json_metadata, display_id,
141 transform_source=js_to_json)
142 jsonp_url = media_link_obj['mediaObj']['url']
144 metadata = self._download_json(
145 jsonp_url, 'metadata', transform_source=strip_jsonp)
147 metadata_tracker_data = metadata['trackerData']
148 metadata_media_resource = metadata['mediaResource']
152 # check if the metadata contains a direct URL to a file
153 metadata_media_alt = metadata_media_resource.get('alt')
154 if metadata_media_alt:
155 for tag_name in ['videoURL', 'audioURL']:
156 if tag_name in metadata_media_alt:
157 alt_url = metadata_media_alt[tag_name]
158 ext = determine_ext(alt_url)
160 m3u_fmt = self._extract_m3u8_formats(
161 alt_url, display_id, 'mp4', 'm3u8_native',
163 formats.extend(m3u_fmt)
168 if ext == 'unknown_video':
169 urlh = self._request_webpage(
170 alt_url, display_id, note='Determining extension')
171 ext = urlhandle_detect_ext(urlh)
172 a_format['ext'] = ext
173 formats.append(a_format)
175 # check if there are flash-streams for this video
176 if 'dflt' in metadata_media_resource and 'videoURL' in metadata_media_resource['dflt']:
177 video_url = metadata_media_resource['dflt']['videoURL']
178 if video_url.endswith('.f4m'):
179 full_video_url = video_url + '?hdcore=3.2.0&plugin=aasp-3.2.0.77.18'
180 formats.extend(self._extract_f4m_formats(full_video_url, display_id, f4m_id='hds', fatal=False))
181 elif video_url.endswith('.smil'):
182 formats.extend(self._extract_smil_formats(video_url, 'stream', fatal=False))
185 caption_url = metadata_media_resource.get('captionURL')
191 title = metadata_tracker_data.get('trackerClipTitle')
192 is_live = url_type == 'live'
195 title = self._live_title(title)
197 elif 'trackerClipAirTime' in metadata_tracker_data:
198 upload_date = metadata_tracker_data['trackerClipAirTime']
200 upload_date = self._html_search_meta('DC.Date', webpage, 'upload date')
203 upload_date = unified_strdate(upload_date)
205 self._sort_formats(formats)
208 'id': metadata_tracker_data.get('trackerClipId', display_id),
209 'display_id': display_id,
211 'alt_title': metadata_tracker_data.get('trackerClipSubcategory'),
213 'upload_date': upload_date,
214 'description': self._html_search_meta('Description', webpage),
216 'subtitles': subtitles,
220 class WDRMobileIE(InfoExtractor):
221 _VALID_URL = r'''(?x)
222 https?://mobile-ondemand\.wdr\.de/
223 .*?/fsk(?P<age_limit>[0-9]+)
225 (?P<id>[0-9]+)_(?P<title>[0-9]+)'''
226 IE_NAME = 'wdr:mobile'
228 'url': 'http://mobile-ondemand.wdr.de/CMS2010/mdb/ondemand/weltweit/fsk0/42/421735/421735_4283021.mp4',
235 'skip': 'Problems with loading data.'
238 def _real_extract(self, url):
239 mobj = re.match(self._VALID_URL, url)
241 'id': mobj.group('id'),
242 'title': mobj.group('title'),
243 'age_limit': int(mobj.group('age_limit')),
246 'User-Agent': 'mobile',