2 from __future__ import unicode_literals
6 from .common import InfoExtractor
14 class BRIE(InfoExtractor):
15 IE_DESC = 'Bayerischer Rundfunk Mediathek'
16 _VALID_URL = r'https?://(?:www\.)?br\.de/(?:[a-z0-9\-_]+/)+(?P<id>[a-z0-9\-_]+)\.html'
17 _BASE_URL = 'http://www.br.de'
21 'url': 'http://www.br.de/mediathek/video/sendungen/heimatsound/heimatsound-festival-2014-trailer-100.html',
22 'md5': '93556dd2bcb2948d9259f8670c516d59',
24 'id': '25e279aa-1ffd-40fd-9955-5325bd48a53a',
26 'title': 'Wenn das Traditions-Theater wackelt',
27 'description': 'Heimatsound-Festival 2014: Wenn das Traditions-Theater wackelt',
32 'url': 'http://www.br.de/nachrichten/schaeuble-haushaltsentwurf-bundestag-100.html',
33 'md5': '3db0df1a9a9cd9fa0c70e6ea8aa8e820',
35 'id': 'c6aae3de-2cf9-43f2-957f-f17fef9afaab',
37 'title': '"Keine neuen Schulden im nächsten Jahr"',
38 'description': 'Haushaltsentwurf: "Keine neuen Schulden im nächsten Jahr"',
43 'url': 'http://www.br.de/radio/bayern1/service/team/videos/team-video-erdelt100.html',
44 'md5': 'dbab0aef2e047060ea7a21fc1ce1078a',
46 'id': '6ba73750-d405-45d3-861d-1ce8c524e059',
48 'title': 'Umweltbewusster Häuslebauer',
49 'description': 'Uwe Erdelt: Umweltbewusster Häuslebauer',
54 'url': 'http://www.br.de/fernsehen/br-alpha/sendungen/kant-fuer-anfaenger/kritik-der-reinen-vernunft/kant-kritik-01-metaphysik100.html',
55 'md5': '23bca295f1650d698f94fc570977dae3',
57 'id': 'd982c9ce-8648-4753-b358-98abb8aec43d',
59 'title': 'Folge 1 - Metaphysik',
60 'description': 'Kant für Anfänger: Folge 1 - Metaphysik',
62 'uploader': 'Eva Maria Steimle',
63 'upload_date': '20140117',
68 def _real_extract(self, url):
69 mobj = re.match(self._VALID_URL, url)
70 display_id = mobj.group('id')
71 page = self._download_webpage(url, display_id)
72 xml_url = self._search_regex(
73 r"return BRavFramework\.register\(BRavFramework\('avPlayer_(?:[a-f0-9-]{36})'\)\.setup\({dataURL:'(/(?:[a-z0-9\-]+/)+[a-z0-9/~_.-]+)'}\)\);", page, 'XMLURL')
74 xml = self._download_xml(self._BASE_URL + xml_url, None)
78 for xml_media in xml.findall('video') + xml.findall('audio'):
80 'id': xml_media.get('externalId'),
81 'title': xml_media.find('title').text,
82 'duration': parse_duration(xml_media.find('duration').text),
83 'formats': self._extract_formats(xml_media.find('assets')),
84 'thumbnails': self._extract_thumbnails(xml_media.find('teaserImage/variants')),
85 'description': ' '.join(xml_media.find('shareTitle').text.splitlines()),
86 'webpage_url': xml_media.find('permalink').text
88 if xml_media.find('author').text:
89 media['uploader'] = xml_media.find('author').text
90 if xml_media.find('broadcastDate').text:
91 media['upload_date'] = ''.join(reversed(xml_media.find('broadcastDate').text.split('.')))
95 self._downloader.report_warning(
96 'found multiple medias; please '
97 'report this with the video URL to http://yt-dl.org/bug')
99 raise ExtractorError('No media entries found')
102 def _extract_formats(self, assets):
104 def text_or_none(asset, tag):
105 elem = asset.find(tag)
106 return None if elem is None else elem.text
109 'url': text_or_none(asset, 'downloadUrl'),
110 'ext': text_or_none(asset, 'mediaType'),
111 'format_id': asset.get('type'),
112 'width': int_or_none(text_or_none(asset, 'frameWidth')),
113 'height': int_or_none(text_or_none(asset, 'frameHeight')),
114 'tbr': int_or_none(text_or_none(asset, 'bitrateVideo')),
115 'abr': int_or_none(text_or_none(asset, 'bitrateAudio')),
116 'vcodec': text_or_none(asset, 'codecVideo'),
117 'acodec': text_or_none(asset, 'codecAudio'),
118 'container': text_or_none(asset, 'mediaType'),
119 'filesize': int_or_none(text_or_none(asset, 'size')),
120 } for asset in assets.findall('asset')
121 if asset.find('downloadUrl') is not None]
123 self._sort_formats(formats)
126 def _extract_thumbnails(self, variants):
128 'url': self._BASE_URL + variant.find('url').text,
129 'width': int_or_none(variant.find('width').text),
130 'height': int_or_none(variant.find('height').text),
131 } for variant in variants.findall('variant')]
132 thumbnails.sort(key=lambda x: x['width'] * x['height'], reverse=True)