5 from .common import InfoExtractor
10 class YahooIE(InfoExtractor):
11 """Information extractor for screen.yahoo.com."""
12 _VALID_URL = r'http://screen\.yahoo\.com/.*?-(?P<id>\d*?)\.html'
14 def _real_extract(self, url):
15 mobj = re.match(self._VALID_URL, url)
17 raise ExtractorError(u'Invalid URL: %s' % url)
18 video_id = mobj.group('id')
19 webpage = self._download_webpage(url, video_id)
20 m_id = re.search(r'YUI\.namespace\("Media"\)\.CONTENT_ID = "(?P<new_id>.+?)";', webpage)
23 # TODO: Check which url parameters are required
24 info_url = 'http://cosmos.bcst.yahoo.com/rest/v2/pops;lmsoverride=1;outputformat=mrss;cb=974419660;id=%s;rd=news.yahoo.com;datacontext=mdb;lg=KCa2IihxG3qE60vQ7HtyUy' % video_id
25 webpage = self._download_webpage(info_url, video_id, u'Downloading info webpage')
26 info_re = r'''<title><!\[CDATA\[(?P<title>.*?)\]\]></title>.*
27 <description><!\[CDATA\[(?P<description>.*?)\]\]></description>.*
28 <media:pubStart><!\[CDATA\[(?P<date>.*?)\ .*\]\]></media:pubStart>.*
29 <media:content\ medium="image"\ url="(?P<thumb>.*?)"\ name="LARGETHUMB"
31 self.report_extraction(video_id)
32 m_info = re.search(info_re, webpage, re.VERBOSE|re.DOTALL)
34 raise ExtractorError(u'Unable to extract video info')
35 video_title = m_info.group('title')
36 video_description = m_info.group('description')
37 video_thumb = m_info.group('thumb')
38 video_date = m_info.group('date')
39 video_date = datetime.datetime.strptime(video_date, '%m/%d/%Y').strftime('%Y%m%d')
41 # TODO: Find a way to get mp4 videos
42 rest_url = 'http://cosmos.bcst.yahoo.com/rest/v2/pops;element=stream;outputformat=mrss;id=%s;lmsoverride=1;bw=375;dynamicstream=1;cb=83521105;tech=flv,mp4;rd=news.yahoo.com;datacontext=mdb;lg=KCa2IihxG3qE60vQ7HtyUy' % video_id
43 webpage = self._download_webpage(rest_url, video_id, u'Downloading video url webpage')
44 m_rest = re.search(r'<media:content url="(?P<url>.*?)" path="(?P<path>.*?)"', webpage)
45 video_url = m_rest.group('url')
46 video_path = m_rest.group('path')
48 raise ExtractorError(u'Unable to extract video url')
50 else: # We have to use a different method if another id is defined
51 long_id = m_id.group('new_id')
52 info_url = 'http://video.query.yahoo.com/v1/public/yql?q=SELECT%20*%20FROM%20yahoo.media.video.streams%20WHERE%20id%3D%22' + long_id + '%22%20AND%20format%3D%22mp4%2Cflv%22%20AND%20protocol%3D%22rtmp%2Chttp%22%20AND%20plrs%3D%2286Gj0vCaSzV_Iuf6hNylf2%22%20AND%20acctid%3D%22389%22%20AND%20plidl%3D%22%22%20AND%20pspid%3D%22792700001%22%20AND%20offnetwork%3D%22false%22%20AND%20site%3D%22ivy%22%20AND%20lang%3D%22en-US%22%20AND%20region%3D%22US%22%20AND%20override%3D%22none%22%3B&env=prod&format=json&callback=YUI.Env.JSONP.yui_3_8_1_1_1368368376830_335'
53 webpage = self._download_webpage(info_url, video_id, u'Downloading info json')
54 json_str = re.search(r'YUI.Env.JSONP.yui.*?\((.*?)\);', webpage).group(1)
55 info = json.loads(json_str)
56 res = info[u'query'][u'results'][u'mediaObj'][0]
57 stream = res[u'streams'][0]
58 video_path = stream[u'path']
59 video_url = stream[u'host']
61 video_title = meta[u'title']
62 video_description = meta[u'description']
63 video_thumb = meta[u'thumbnail']
64 video_date = None # I can't find it
69 'play_path': video_path,
71 'description': video_description,
72 'thumbnail': video_thumb,
73 'upload_date': video_date,