1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
6 from .common import InfoExtractor
20 class WDRIE(InfoExtractor):
21 _CURRENT_MAUS_URL = r'https?://www.wdrmaus.de/aktuelle-sendung/(wdr|index).php5'
22 _PAGE_REGEX = r'/mediathek/(?P<media_type>[^/]+)/(?P<type>[^/]+)/(?P<display_id>.+)\.html'
23 _VALID_URL = r'(?P<page_url>https?://(?:www\d\.)?wdr\d?\.de)' + _PAGE_REGEX + '|' + _CURRENT_MAUS_URL
27 'url': 'http://www1.wdr.de/mediathek/video/sendungen/doku-am-freitag/video-geheimnis-aachener-dom-100.html',
28 'md5': 'e58c39c3e30077141d258bf588700a7b',
32 'display_id': 'doku-am-freitag/video-geheimnis-aachener-dom-100',
33 'title': 'Geheimnis Aachener Dom',
34 'alt_title': 'Doku am Freitag',
35 'upload_date': '20160304',
36 'description': 'md5:87be8ff14d8dfd7a7ee46f0299b52318',
38 'subtitles': {'de': [{
39 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml'
42 'skip': 'Page Not Found',
45 'url': 'http://www1.wdr.de/mediathek/audio/wdr3/wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100.html',
46 'md5': 'f4c1f96d01cf285240f53ea4309663d8',
50 'display_id': 'wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100',
51 'title': 'Schriftstellerin Juli Zeh',
52 'alt_title': 'WDR 3 Gespräch am Samstag',
53 'upload_date': '20160312',
54 'description': 'md5:e127d320bc2b1f149be697ce044a3dd7',
58 'skip': 'Page Not Found',
61 'url': 'http://www1.wdr.de/mediathek/video/live/index.html',
65 'display_id': 'index',
66 'title': r're:^WDR Fernsehen im Livestream [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
67 'alt_title': 'WDR Fernsehen Live',
69 'description': 'md5:ae2ff888510623bf8d4b115f95a9b7c9',
74 'skip_download': True, # m3u8 download
78 'url': 'http://www1.wdr.de/mediathek/video/sendungen/aktuelle-stunde/aktuelle-stunde-120.html',
79 'playlist_mincount': 10,
81 'id': 'aktuelle-stunde/aktuelle-stunde-120',
85 'url': 'http://www.wdrmaus.de/aktuelle-sendung/index.php5',
89 'upload_date': 're:^[0-9]{8}$',
90 'title': 're:^Die Sendung mit der Maus vom [0-9.]{10}$',
91 'description': '- Die Sendung mit der Maus -',
93 'skip': 'The id changes from week to week because of the new episode'
97 def _real_extract(self, url):
98 mobj = re.match(self._VALID_URL, url)
99 url_type = mobj.group('type')
100 page_url = mobj.group('page_url')
101 display_id = mobj.group('display_id')
102 webpage = self._download_webpage(url, display_id)
104 # for wdr.de the data-extension is in a tag with the class "mediaLink"
105 # for wdrmaus its in a link to the page in a multiline "videoLink"-tag
106 json_metadata = self._html_search_regex(
107 r'class=(?:"mediaLink\b[^"]*"[^>]+|"videoLink\b[^"]*"[\s]*>\n[^\n]*)data-extension="([^"]+)"',
108 webpage, 'media link', default=None, flags=re.MULTILINE)
110 if not json_metadata:
112 self.url_result(page_url + href[0], 'WDR')
113 for href in re.findall(
114 r'<a href="(%s)"' % self._PAGE_REGEX,
118 if entries: # Playlist page
119 return self.playlist_result(entries, playlist_id=display_id)
121 raise ExtractorError('No downloadable streams found', expected=True)
123 media_link_obj = self._parse_json(json_metadata, display_id,
124 transform_source=js_to_json)
125 jsonp_url = media_link_obj['mediaObj']['url']
127 metadata = self._download_json(
128 jsonp_url, 'metadata', transform_source=strip_jsonp)
130 metadata_tracker_data = metadata['trackerData']
131 metadata_media_resource = metadata['mediaResource']
135 # check if the metadata contains a direct URL to a file
136 metadata_media_alt = metadata_media_resource.get('alt')
137 if metadata_media_alt:
138 for tag_name in ['videoURL', 'audioURL']:
139 if tag_name in metadata_media_alt:
140 alt_url = metadata_media_alt[tag_name]
141 if determine_ext(alt_url) == 'm3u8':
142 m3u_fmt = self._extract_m3u8_formats(
143 alt_url, display_id, 'mp4', 'm3u8_native',
145 formats.extend(m3u_fmt)
151 # check if there are flash-streams for this video
152 if 'dflt' in metadata_media_resource and 'videoURL' in metadata_media_resource['dflt']:
153 video_url = metadata_media_resource['dflt']['videoURL']
154 if video_url.endswith('.f4m'):
155 full_video_url = video_url + '?hdcore=3.2.0&plugin=aasp-3.2.0.77.18'
156 formats.extend(self._extract_f4m_formats(full_video_url, display_id, f4m_id='hds', fatal=False))
157 elif video_url.endswith('.smil'):
158 formats.extend(self._extract_smil_formats(video_url, 'stream', fatal=False))
161 caption_url = metadata_media_resource.get('captionURL')
167 title = metadata_tracker_data.get('trackerClipTitle')
168 is_live = url_type == 'live'
171 title = self._live_title(title)
173 elif 'trackerClipAirTime' in metadata_tracker_data:
174 upload_date = metadata_tracker_data['trackerClipAirTime']
176 upload_date = self._html_search_meta('DC.Date', webpage, 'upload date')
179 upload_date = unified_strdate(upload_date)
181 self._sort_formats(formats)
184 'id': metadata_tracker_data.get('trackerClipId', display_id),
185 'display_id': display_id,
187 'alt_title': metadata_tracker_data.get('trackerClipSubcategory'),
189 'upload_date': upload_date,
190 'description': self._html_search_meta('Description', webpage),
192 'subtitles': subtitles,
196 class WDRMobileIE(InfoExtractor):
197 _VALID_URL = r'''(?x)
198 https?://mobile-ondemand\.wdr\.de/
199 .*?/fsk(?P<age_limit>[0-9]+)
201 (?P<id>[0-9]+)_(?P<title>[0-9]+)'''
202 IE_NAME = 'wdr:mobile'
204 'url': 'http://mobile-ondemand.wdr.de/CMS2010/mdb/ondemand/weltweit/fsk0/42/421735/421735_4283021.mp4',
211 'skip': 'Problems with loading data.'
214 def _real_extract(self, url):
215 mobj = re.match(self._VALID_URL, url)
217 'id': mobj.group('id'),
218 'title': mobj.group('title'),
219 'age_limit': int(mobj.group('age_limit')),
222 'User-Agent': 'mobile',
227 class WDRMausIE(InfoExtractor):
228 _VALID_URL = 'https?://(?:www\.)?wdrmaus\.de/(?:[^/]+/){,2}(?P<id>[^/?#]+)((?<!index)\.php5|/(?:$|[?#]))'
229 IE_DESC = 'Sendung mit der Maus'
231 'url': 'http://www.wdrmaus.de/sachgeschichten/sachgeschichten/achterbahn.php5',
232 'md5': '178b432d002162a14ccb3e0876741095',
236 'thumbnail': 're:^http://.+\.jpg',
237 'upload_date': '20131001',
238 'title': '19.09.2013 - Achterbahn',
242 def _real_extract(self, url):
243 video_id = self._match_id(url)
245 webpage = self._download_webpage(url, video_id)
246 param_code = self._html_search_regex(
247 r'<a href="\?startVideo=1&([^"]+)"', webpage, 'parameters')
249 title_date = self._search_regex(
250 r'<div class="sendedatum"><p>Sendedatum:\s*([0-9\.]+)</p>',
252 title_str = self._html_search_regex(
253 r'<h1>(.*?)</h1>', webpage, 'title')
254 title = '%s - %s' % (title_date, title_str)
255 upload_date = unified_strdate(
256 self._html_search_meta('dc.date', webpage))
258 fields = compat_parse_qs(param_code)
259 video_url = fields['firstVideo'][0]
260 thumbnail = compat_urlparse.urljoin(url, fields['startPicture'][0])
267 jscode = self._download_webpage(
268 'http://www.wdrmaus.de/codebase/js/extended-medien.min.js',
269 video_id, fatal=False,
270 note='Downloading URL translation table',
271 errnote='Could not download URL translation table')
273 for m in re.finditer(
274 r"stream:\s*'dslSrc=(?P<stream>[^']+)',\s*download:\s*'(?P<dl>[^']+)'\s*\}",
276 if video_url.startswith(m.group('stream')):
277 http_url = video_url.replace(
278 m.group('stream'), m.group('dl'))
285 self._sort_formats(formats)
291 'thumbnail': thumbnail,
292 'upload_date': upload_date,