1 from __future__ import unicode_literals
5 from .common import InfoExtractor
15 class DigitalSpeakingIE(InfoExtractor):
16 _VALID_URL = r'http://evt.dispeak.com/([^/]+/)+xml/(?P<id>[^.]+).xml'
19 # From http://evt.dispeak.com/ubm/gdc/sf16/xml/840376_BQRC.xml
20 'url': 'http://evt.dispeak.com/ubm/gdc/sf16/xml/840376_BQRC.xml',
21 'md5': 'a8efb6c31ed06ca8739294960b2dbabd',
25 'title': 'Tenacious Design and The Interface of \'Destiny\'',
29 def _parse_mp4(self, metadata):
33 mp4_video = xpath_text(metadata, './mp4video', default=None)
34 if mp4_video is not None:
35 mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video)
36 video_root = mobj.group('root')
37 if video_root is None:
38 http_host = xpath_text(metadata, 'httpHost', default=None)
40 video_root = 'http://%s/' % http_host
41 if video_root is None:
42 # Hard-coded in http://evt.dispeak.com/ubm/gdc/sf16/custom/player2.js
43 # Works for GPUTechConf, too
44 video_root = 'http://s3-2u.digitallyspeaking.com/'
46 formats = metadata.findall('./MBRVideos/MBRVideo')
49 for a_format in formats:
50 stream_name = xpath_text(a_format, 'streamName', fatal=True)
51 video_path = re.match(r'mp4\:(?P<path>.*)', stream_name).group('path')
52 url = video_root + video_path
53 vbr = xpath_text(a_format, 'bitrate')
54 video_formats.append({
56 'vbr': int_or_none(vbr),
60 def _parse_flv(self, metadata):
62 akamai_url = xpath_text(metadata, './akamaiHost', fatal=True)
63 audios = metadata.find('./audios')
64 if audios is not None:
67 'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
68 'play_path': remove_end(audio.get('url'), '.flv'),
71 'format_id': audio.get('code'),
73 slide_video_path = xpath_text(metadata, './slideVideo', fatal=True)
75 'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
76 'play_path': remove_end(slide_video_path, '.flv'),
78 'format_note': 'slide deck video',
81 'format_id': 'slides',
83 speaker_video_path = xpath_text(metadata, './speakerVideo', fatal=True)
85 'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
86 'play_path': remove_end(speaker_video_path, '.flv'),
88 'format_note': 'speaker video',
91 'format_id': 'speaker',
95 def _real_extract(self, url):
96 video_id = self._match_id(url)
98 xml_description = self._download_xml(url, video_id)
99 metadata = xpath_element(xml_description, 'metadata')
101 video_formats = self._parse_mp4(metadata)
102 if video_formats is None:
103 video_formats = self._parse_flv(metadata)
107 'formats': video_formats,
108 'title': xpath_text(metadata, 'title', fatal=True),
109 'duration': parse_duration(xpath_text(metadata, 'endTime')),
110 'creator': xpath_text(metadata, 'speaker'),