2 from __future__ import unicode_literals
6 from .common import InfoExtractor
11 compat_urllib_parse_urlparse,
20 class ARDMediathekIE(InfoExtractor):
21 IE_NAME = 'ARD:mediathek'
22 _VALID_URL = r'^https?://(?:(?:www\.)?ardmediathek\.de|mediathek\.daserste\.de)/(?:.*/)(?P<video_id>[0-9]+|[^0-9][^/\?]+)[^/\?]*(?:\?.*)?'
25 'url': 'http://mediathek.daserste.de/sendungen_a-z/328454_anne-will/22429276_vertrauen-ist-gut-spionieren-ist-besser-geht',
26 'file': '22429276.mp4',
27 'md5': '469751912f1de0816a9fc9df8336476c',
29 'title': 'Vertrauen ist gut, Spionieren ist besser - Geht so deutsch-amerikanische Freundschaft?',
30 'description': 'Das Erste Mediathek [ARD]: Vertrauen ist gut, Spionieren ist besser - Geht so deutsch-amerikanische Freundschaft?, Anne Will, Über die Spionage-Affäre diskutieren Clemens Binninger, Katrin Göring-Eckardt, Georg Mascolo, Andrew B. Denison und Constanze Kurz.. Das Video zur Sendung Anne Will am Mittwoch, 16.07.2014',
32 'skip': 'Blocked outside of Germany',
34 'url': 'http://www.ardmediathek.de/tv/Tatort/Das-Wunder-von-Wolbeck-Video-tgl-ab-20/Das-Erste/Video?documentId=22490580&bcastId=602916',
38 'title': 'Das Wunder von Wolbeck (Video tgl. ab 20 Uhr)',
39 'description': 'Auf einem restaurierten Hof bei Wolbeck wird der Heilpraktiker Raffael Lembeck eines morgens von seiner Frau Stella tot aufgefunden. Das Opfer war offensichtlich in seiner Praxis zu Fall gekommen und ist dann verblutet, erklärt Prof. Boerne am Tatort.',
41 'skip': 'Blocked outside of Germany',
44 def _real_extract(self, url):
45 # determine video id from url
46 m = re.match(self._VALID_URL, url)
48 numid = re.search(r'documentId=([0-9]+)', url)
50 video_id = numid.group(1)
52 video_id = m.group('video_id')
54 webpage = self._download_webpage(url, video_id)
56 if '>Der gewünschte Beitrag ist nicht mehr verfügbar.<' in webpage:
57 raise ExtractorError('Video %s is no longer available' % video_id, expected=True)
59 title = self._html_search_regex(
60 [r'<h1(?:\s+class="boxTopHeadline")?>(.*?)</h1>',
61 r'<meta name="dcterms.title" content="(.*?)"/>',
62 r'<h4 class="headline">(.*?)</h4>'],
64 description = self._html_search_meta(
65 'dcterms.abstract', webpage, 'description', default=None)
66 if description is None:
67 description = self._html_search_meta(
68 'description', webpage, 'meta description')
70 # Thumbnail is sometimes not present.
71 # It is in the mobile version, but that seems to use a different URL
72 # structure altogether.
73 thumbnail = self._og_search_thumbnail(webpage, default=None)
75 media_streams = re.findall(r'''(?x)
76 mediaCollection\.addMediaStream\([0-9]+,\s*[0-9]+,\s*"[^"]*",\s*
77 "([^"]+)"''', webpage)
80 QUALITIES = qualities(['lo', 'hi', 'hq'])
82 for furl in set(media_streams):
83 if furl.endswith('.f4m'):
86 fid_m = re.match(r'.*\.([^.]+)\.[^.]+$', furl)
87 fid = fid_m.group(1) if fid_m else None
89 'quality': QUALITIES(fid),
93 else: # request JSON file
94 media_info = self._download_json(
95 'http://www.ardmediathek.de/play/media/%s' % video_id, video_id)
96 # The second element of the _mediaArray contains the standard http urls
97 streams = media_info['_mediaArray'][1]['_mediaStreamArray']
99 if '"fsk"' in webpage:
100 raise ExtractorError('This video is only available after 20:00')
104 if type(s['_stream']) == list:
105 for index, url in enumerate(s['_stream'][::-1]):
106 quality = s['_quality'] + index
110 'format_id': '%s-%s' % (determine_ext(url), quality)
115 'quality': s['_quality'],
119 format['format_id'] = '%s-%s' % (
120 determine_ext(format['url']), format['quality'])
122 formats.append(format)
124 self._sort_formats(formats)
129 'description': description,
131 'thumbnail': thumbnail,
135 class ARDIE(InfoExtractor):
136 _VALID_URL = '(?P<mainurl>https?://(www\.)?daserste\.de/[^?#]+/videos/(?P<display_id>[^/?#]+)-(?P<id>[0-9]+))\.html'
138 'url': 'http://www.daserste.de/information/reportage-dokumentation/dokus/videos/die-story-im-ersten-mission-unter-falscher-flagge-100.html',
139 'md5': 'd216c3a86493f9322545e045ddc3eb35',
141 'display_id': 'die-story-im-ersten-mission-unter-falscher-flagge',
145 'title': 'Die Story im Ersten: Mission unter falscher Flagge',
146 'upload_date': '20140804',
147 'thumbnail': 're:^https?://.*\.jpg$',
151 def _real_extract(self, url):
152 mobj = re.match(self._VALID_URL, url)
153 display_id = mobj.group('display_id')
155 player_url = mobj.group('mainurl') + '~playerXml.xml'
156 doc = self._download_xml(player_url, display_id)
157 video_node = doc.find('./video')
158 upload_date = unified_strdate(xpath_text(
159 video_node, './broadcastDate'))
160 thumbnail = xpath_text(video_node, './/teaserImage//variant/url')
163 for a in video_node.findall('.//asset'):
165 'format_id': a.attrib['type'],
166 'width': int_or_none(a.find('./frameWidth').text),
167 'height': int_or_none(a.find('./frameHeight').text),
168 'vbr': int_or_none(a.find('./bitrateVideo').text),
169 'abr': int_or_none(a.find('./bitrateAudio').text),
170 'vcodec': a.find('./codecVideo').text,
171 'tbr': int_or_none(a.find('./totalBitrate').text),
173 if a.find('./serverPrefix').text:
174 f['url'] = a.find('./serverPrefix').text
175 f['playpath'] = a.find('./fileName').text
177 f['url'] = a.find('./fileName').text
179 self._sort_formats(formats)
182 'id': mobj.group('id'),
184 'display_id': display_id,
185 'title': video_node.find('./title').text,
186 'duration': parse_duration(video_node.find('./duration').text),
187 'upload_date': upload_date,
188 'thumbnail': thumbnail,