1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
6 from .common import InfoExtractor
19 class WDRIE(InfoExtractor):
20 _CURRENT_MAUS_URL = r'https?://www.wdrmaus.de/aktuelle-sendung/(wdr|index).php5'
21 _PAGE_REGEX = r'/mediathek/(?P<media_type>[^/]+)/(?P<type>[^/]+)/(?P<display_id>.+)\.html'
22 _VALID_URL = r'(?P<page_url>https?://(?:www\d\.)?wdr\d?\.de)' + _PAGE_REGEX + '|' + _CURRENT_MAUS_URL
24 _JS_URL_REGEX = r'(https?://deviceids-medp.wdr.de/ondemand/\d+/\d+\.js)'
28 'url': 'http://www1.wdr.de/mediathek/video/sendungen/doku-am-freitag/video-geheimnis-aachener-dom-100.html',
29 'md5': 'e58c39c3e30077141d258bf588700a7b',
33 'display_id': 'doku-am-freitag/video-geheimnis-aachener-dom-100',
34 'title': 'Geheimnis Aachener Dom',
35 'alt_title': 'Doku am Freitag',
36 'upload_date': '20160304',
37 'description': 'md5:87be8ff14d8dfd7a7ee46f0299b52318',
39 'subtitles': {'de': [{
40 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml'
43 'skip': 'Page Not Found',
46 'url': 'http://www1.wdr.de/mediathek/audio/wdr3/wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100.html',
47 'md5': 'f4c1f96d01cf285240f53ea4309663d8',
51 'display_id': 'wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100',
52 'title': 'Schriftstellerin Juli Zeh',
53 'alt_title': 'WDR 3 Gespräch am Samstag',
54 'upload_date': '20160312',
55 'description': 'md5:e127d320bc2b1f149be697ce044a3dd7',
59 'skip': 'Page Not Found',
62 'url': 'http://www1.wdr.de/mediathek/video/live/index.html',
66 'display_id': 'index',
67 'title': r're:^WDR Fernsehen im Livestream [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
68 'alt_title': 'WDR Fernsehen Live',
70 'description': 'md5:ae2ff888510623bf8d4b115f95a9b7c9',
75 'skip_download': True, # m3u8 download
79 'url': 'http://www1.wdr.de/mediathek/video/sendungen/aktuelle-stunde/aktuelle-stunde-120.html',
80 'playlist_mincount': 10,
82 'id': 'aktuelle-stunde/aktuelle-stunde-120',
86 'url': 'http://www.wdrmaus.de/aktuelle-sendung/index.php5',
90 'upload_date': 're:^[0-9]{8}$',
91 'title': 're:^Die Sendung mit der Maus vom [0-9.]{10}$',
92 'description': '- Die Sendung mit der Maus -',
94 'skip': 'The id changes from week to week because of the new episode'
98 def _real_extract(self, url):
99 mobj = re.match(self._VALID_URL, url)
100 url_type = mobj.group('type')
101 page_url = mobj.group('page_url')
102 display_id = mobj.group('display_id')
103 webpage = self._download_webpage(url, display_id)
105 js_url = self._search_regex(self._JS_URL_REGEX, webpage, 'js_url', default=None)
109 self.url_result(page_url + href[0], 'WDR')
110 for href in re.findall(
111 r'<a href="(%s)"' % self._PAGE_REGEX,
115 if entries: # Playlist page
116 return self.playlist_result(entries, playlist_id=display_id)
118 raise ExtractorError('No downloadable streams found', expected=True)
120 metadata = self._download_json(
121 js_url, 'metadata', transform_source=strip_jsonp)
123 metadata_tracker_data = metadata['trackerData']
124 metadata_media_resource = metadata['mediaResource']
128 # check if the metadata contains a direct URL to a file
129 metadata_media_alt = metadata_media_resource.get('alt')
130 if metadata_media_alt:
131 for tag_name in ['videoURL', 'audioURL']:
132 if tag_name in metadata_media_alt:
133 alt_url = metadata_media_alt[tag_name]
134 if determine_ext(alt_url) == 'm3u8':
135 m3u_fmt = self._extract_m3u8_formats(
136 alt_url, display_id, 'mp4', 'm3u8_native',
138 formats.extend(m3u_fmt)
144 # check if there are flash-streams for this video
145 if 'dflt' in metadata_media_resource and 'videoURL' in metadata_media_resource['dflt']:
146 video_url = metadata_media_resource['dflt']['videoURL']
147 if video_url.endswith('.f4m'):
148 full_video_url = video_url + '?hdcore=3.2.0&plugin=aasp-3.2.0.77.18'
149 formats.extend(self._extract_f4m_formats(full_video_url, display_id, f4m_id='hds', fatal=False))
150 elif video_url.endswith('.smil'):
151 formats.extend(self._extract_smil_formats(video_url, 'stream', fatal=False))
154 caption_url = metadata_media_resource.get('captionURL')
160 title = metadata_tracker_data.get('trackerClipTitle')
161 is_live = url_type == 'live'
164 title = self._live_title(title)
166 elif 'trackerClipAirTime' in metadata_tracker_data:
167 upload_date = metadata_tracker_data['trackerClipAirTime']
169 upload_date = self._html_search_meta('DC.Date', webpage, 'upload date')
172 upload_date = unified_strdate(upload_date)
174 self._sort_formats(formats)
177 'id': metadata_tracker_data.get('trackerClipId', display_id),
178 'display_id': display_id,
180 'alt_title': metadata_tracker_data.get('trackerClipSubcategory'),
182 'upload_date': upload_date,
183 'description': self._html_search_meta('Description', webpage),
185 'subtitles': subtitles,
189 class WDRMobileIE(InfoExtractor):
190 _VALID_URL = r'''(?x)
191 https?://mobile-ondemand\.wdr\.de/
192 .*?/fsk(?P<age_limit>[0-9]+)
194 (?P<id>[0-9]+)_(?P<title>[0-9]+)'''
195 IE_NAME = 'wdr:mobile'
197 'url': 'http://mobile-ondemand.wdr.de/CMS2010/mdb/ondemand/weltweit/fsk0/42/421735/421735_4283021.mp4',
204 'skip': 'Problems with loading data.'
207 def _real_extract(self, url):
208 mobj = re.match(self._VALID_URL, url)
210 'id': mobj.group('id'),
211 'title': mobj.group('title'),
212 'age_limit': int(mobj.group('age_limit')),
215 'User-Agent': 'mobile',
220 class WDRMausIE(InfoExtractor):
221 _VALID_URL = 'https?://(?:www\.)?wdrmaus\.de/(?:[^/]+/){,2}(?P<id>[^/?#]+)((?<!index)\.php5|/(?:$|[?#]))'
222 IE_DESC = 'Sendung mit der Maus'
224 'url': 'http://www.wdrmaus.de/sachgeschichten/sachgeschichten/achterbahn.php5',
225 'md5': '178b432d002162a14ccb3e0876741095',
229 'thumbnail': 're:^http://.+\.jpg',
230 'upload_date': '20131001',
231 'title': '19.09.2013 - Achterbahn',
235 def _real_extract(self, url):
236 video_id = self._match_id(url)
238 webpage = self._download_webpage(url, video_id)
239 param_code = self._html_search_regex(
240 r'<a href="\?startVideo=1&([^"]+)"', webpage, 'parameters')
242 title_date = self._search_regex(
243 r'<div class="sendedatum"><p>Sendedatum:\s*([0-9\.]+)</p>',
245 title_str = self._html_search_regex(
246 r'<h1>(.*?)</h1>', webpage, 'title')
247 title = '%s - %s' % (title_date, title_str)
248 upload_date = unified_strdate(
249 self._html_search_meta('dc.date', webpage))
251 fields = compat_parse_qs(param_code)
252 video_url = fields['firstVideo'][0]
253 thumbnail = compat_urlparse.urljoin(url, fields['startPicture'][0])
260 jscode = self._download_webpage(
261 'http://www.wdrmaus.de/codebase/js/extended-medien.min.js',
262 video_id, fatal=False,
263 note='Downloading URL translation table',
264 errnote='Could not download URL translation table')
266 for m in re.finditer(
267 r"stream:\s*'dslSrc=(?P<stream>[^']+)',\s*download:\s*'(?P<dl>[^']+)'\s*\}",
269 if video_url.startswith(m.group('stream')):
270 http_url = video_url.replace(
271 m.group('stream'), m.group('dl'))
278 self._sort_formats(formats)
284 'thumbnail': thumbnail,
285 'upload_date': upload_date,