]> git.bitcoin.ninja Git - youtube-dl/blob - youtube_dl/extractor/veoh.py
Implement a different adult sites checking algorithm
[youtube-dl] / youtube_dl / extractor / veoh.py
1 from __future__ import unicode_literals
2
3 import re
4 import json
5
6 from .common import InfoExtractor
7
8
9 class VeohIE(InfoExtractor):
10     _VALID_URL = r'http://(?:www\.)?veoh\.com/(?:watch|iphone/#_Watch)/v(?P<id>\d*)'
11
12     _TEST = {
13         'url': 'http://www.veoh.com/watch/v56314296nk7Zdmz3',
14         'file': '56314296.mp4',
15         'md5': '620e68e6a3cff80086df3348426c9ca3',
16         'info_dict': {
17             'title': 'Straight Backs Are Stronger',
18             'uploader': 'LUMOback',
19             'description': 'At LUMOback, we believe straight backs are stronger.  The LUMOback Posture & Movement Sensor:  It gently vibrates when you slouch, inspiring improved posture and mobility.  Use the app to track your data and improve your posture over time. ',
20         }
21     }
22
23     def _real_extract(self, url):
24         mobj = re.match(self._VALID_URL, url)
25         video_id = mobj.group('id')
26         webpage = self._download_webpage(url, video_id)
27
28         m_youtube = re.search(r'http://www\.youtube\.com/v/(.*?)(\&|")', webpage)
29         if m_youtube is not None:
30             youtube_id = m_youtube.group(1)
31             self.to_screen('%s: detected Youtube video.' % video_id)
32             return self.url_result(youtube_id, 'Youtube')
33
34         self.report_extraction(video_id)
35         info = self._search_regex(r'videoDetailsJSON = \'({.*?})\';', webpage, 'info')
36         info = json.loads(info)
37         video_url = info.get('fullPreviewHashHighPath') or info.get('fullPreviewHashLowPath')
38
39         return {
40             'id': info['videoId'],
41             'title': info['title'],
42             'url': video_url,
43             'uploader': info['username'],
44             'thumbnail': info.get('highResImage') or info.get('medResImage'),
45             'description': info['description'],
46             'view_count': info['views'],
47         }