1 from __future__ import unicode_literals
5 from .common import InfoExtractor
8 class SpiegelIE(InfoExtractor):
9 _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<videoID>[0-9]+)(?:\.html)?(?:#.*)?$'
11 'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html',
12 'file': '1259285.mp4',
13 'md5': '2c2754212136f35fb4b19767d242f66e',
15 'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv',
19 'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html',
20 'file': '1309159.mp4',
21 'md5': 'f2cdf638d7aa47654e251e1aee360af1',
23 'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers',
27 def _real_extract(self, url):
28 m = re.match(self._VALID_URL, url)
29 video_id = m.group('videoID')
31 webpage = self._download_webpage(url, video_id)
33 video_title = self._html_search_regex(
34 r'<div class="module-title">(.*?)</div>', webpage, 'title')
36 base_url = self._search_regex(
37 r'var\s+server\s+=\s+\"(http://video\d*\.spiegel\.de/flash/\d+/\d+/)\";',
42 xml_url = base_url + video_id + '.xml'
43 idoc = self._download_xml(
45 note='Downloading XML', errnote='Failed to download XML from "{0}"'.format(xml_url))
49 'format_id': n.tag.rpartition('type')[2],
50 'url': 'http://video2.spiegel.de/flash/' + n.find('./filename').text,
51 'width': int(n.find('./width').text),
52 'height': int(n.find('./height').text),
53 'abr': int(n.find('./audiobitrate').text),
54 'vbr': int(n.find('./videobitrate').text),
55 'vcodec': n.find('./codec').text,
59 # Blacklist type 6, it's extremely LQ and not available on the same server
60 if n.tag.startswith('type') and n.tag != 'type6'
62 duration = float(idoc[0].findall('./duration')[0].text)
64 self._sort_formats(formats)