4 from .common import InfoExtractor
12 class TechTVMITIE(InfoExtractor):
13 IE_NAME = u'techtv.mit.edu'
14 _VALID_URL = r'https?://techtv\.mit\.edu/(videos|embeds)/(?P<id>\d+)'
17 u'url': u'http://techtv.mit.edu/videos/25418-mit-dna-learning-center-set',
18 u'file': u'25418.mp4',
19 u'md5': u'1f8cb3e170d41fd74add04d3c9330e5f',
21 u'title': u'MIT DNA Learning Center Set',
22 u'description': u'md5:82313335e8a8a3f243351ba55bc1b474',
26 def _real_extract(self, url):
27 mobj = re.match(self._VALID_URL, url)
28 video_id = mobj.group('id')
29 raw_page = self._download_webpage(
30 'http://techtv.mit.edu/videos/%s' % video_id, video_id)
31 clean_page = re.compile(u'<!--.*?-->', re.S).sub(u'', raw_page)
33 base_url = self._search_regex(r'ipadUrl: \'(.+?cloudfront.net/)',
34 raw_page, u'base url')
35 formats_json = self._search_regex(r'bitrates: (\[.+?\])', raw_page,
37 formats_mit = json.loads(formats_json)
40 'format_id': f['label'],
41 'url': base_url + f['url'].partition(':')[2],
42 'ext': f['url'].partition(':')[0],
50 title = get_element_by_id('edit-title', clean_page)
51 description = clean_html(get_element_by_id('edit-description', clean_page))
52 thumbnail = self._search_regex(r'playlist:.*?url: \'(.+?)\'',
53 raw_page, u'thumbnail', flags=re.DOTALL)
55 return {'id': video_id,
58 'description': description,
59 'thumbnail': thumbnail,
63 class MITIE(TechTVMITIE):
64 IE_NAME = u'video.mit.edu'
65 _VALID_URL = r'https?://video\.mit\.edu/watch/(?P<title>[^/]+)'
68 u'url': u'http://video.mit.edu/watch/the-government-is-profiling-you-13222/',
69 u'file': u'21783.mp4',
70 u'md5': u'7db01d5ccc1895fc5010e9c9e13648da',
72 u'title': u'The Government is Profiling You',
73 u'description': u'md5:ad5795fe1e1623b73620dbfd47df9afd',
77 def _real_extract(self, url):
78 mobj = re.match(self._VALID_URL, url)
79 page_title = mobj.group('title')
80 webpage = self._download_webpage(url, page_title)
81 self.to_screen('%s: Extracting %s url' % (page_title, TechTVMITIE.IE_NAME))
82 embed_url = self._search_regex(r'<iframe .*?src="(.+?)"', webpage,
84 return self.url_result(embed_url, ie='TechTVMIT')
86 class OCWMITIE(InfoExtractor):
87 IE_NAME = u'ocw.mit.edu'
88 _VALID_URL = r'^http://ocw\.mit\.edu/courses/(?P<topic>[a-z0-9\-]+)'
89 _BASE_URL = u'http://ocw.mit.edu/'
93 u'url': u'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/',
94 u'md5': u'348bef727b573c0bd9ad8a7c08c89ebd',
96 u'title': u'7. Discrete Random Variables III',
97 u'description': u'In this lecture, the professor discussed multiple random variables, expectations, and binomial distribution.',
98 u'subtitles': u'http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-041-probabilistic-systems-analysis-and-applied-probability-fall-2010/video-lectures/lecture-7-multiple-variables-expectations-independence/MIT6_041F11_lec07_300k.mp4.srt'
102 u'url': u'http://ocw.mit.edu/courses/mathematics/18-01sc-single-variable-calculus-fall-2010/1.-differentiation/part-a-definition-and-basic-rules/session-1-introduction-to-derivatives/',
103 u'md5': u'f4a434f08f15e581eb67cec0b57bcf6f',
105 u'title': u'Lec 1 _ MIT 18.01 Single Variable Calculus, Fall 2007',
106 u'subtitles': u'http://ocw.mit.edu//courses/mathematics/18-01sc-single-variable-calculus-fall-2010/ocw-18.01-f07-lec01_300k.SRT'
111 def _real_extract(self, url):
112 webpage = self._download_webpage(url, self.IE_NAME)
113 title = self._html_search_meta(u'WT.cg_s', webpage)
114 description = self._html_search_meta(u'Description', webpage)
116 # search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, start, stop, captions_file)
117 embed_chapter_media = re.search(r'ocw_embed_chapter_media\((.+?)\)', webpage)
118 if embed_chapter_media:
119 metadata = re.sub(r'[\'"]', u'', embed_chapter_media.group(1))
120 metadata = re.split(r', ?', metadata)
122 subs = compat_urlparse.urljoin(self._BASE_URL, metadata[7])
124 # search for call to ocw_embed_chapter_media(container_id, media_url, provider, page_url, image_url, captions_file)
125 embed_media = re.search(r'ocw_embed_media\((.+?)\)', webpage)
127 metadata = re.sub(r'[\'"]', u'', embed_media.group(1))
128 metadata = re.split(r', ?', metadata)
130 subs = compat_urlparse.urljoin(self._BASE_URL, metadata[5])
132 raise ExtractorError('Unable to find embedded YouTube video.')
134 data = self.url_result(yt, 'Youtube')
135 data['subtitles'] = subs