2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_str
16 class Revision3IE(InfoExtractor):
17 _VALID_URL = r'https?://(?:www\.)?(?P<domain>(?:revision3|testtube|animalist)\.com)/(?P<id>[^/]+(?:/[^/?#]+)?)'
19 'url': 'http://www.revision3.com/technobuffalo/5-google-predictions-for-2016',
20 'md5': 'd94a72d85d0a829766de4deb8daaf7df',
23 'display_id': 'technobuffalo/5-google-predictions-for-2016',
25 'title': '5 Google Predictions for 2016',
26 'description': 'Google had a great 2015, but it\'s already time to look ahead. Here are our five predictions for 2016.',
27 'upload_date': '20151228',
28 'timestamp': 1451325600,
30 'uploader': 'TechnoBuffalo',
31 'uploader_id': 'technobuffalo',
34 'url': 'http://testtube.com/brainstuff',
37 'title': 'BrainStuff',
38 'description': 'Whether the topic is popcorn or particle physics, you can count on the HowStuffWorks team to explore-and explain-the everyday science in the world around us on BrainStuff.',
40 'playlist_mincount': 93,
42 'url': 'https://testtube.com/dnews/5-weird-ways-plants-can-eat-animals?utm_source=FB&utm_medium=DNews&utm_campaign=DNewsSocial',
45 'display_id': 'dnews/5-weird-ways-plants-can-eat-animals',
48 'title': '5 Weird Ways Plants Can Eat Animals',
49 'description': 'Why have some plants evolved to eat meat?',
50 'upload_date': '20150120',
51 'timestamp': 1421763300,
53 'uploader_id': 'dnews',
56 'url': 'http://testtube.com/tt-editors-picks/the-israel-palestine-conflict-explained-in-ten-min',
60 'display_id': 'tt-editors-picks/the-israel-palestine-conflict-explained-in-ten-min',
61 'title': 'The Israel-Palestine Conflict Explained in Ten Minutes',
62 'description': 'If you\'d like to learn about the struggle between Israelis and Palestinians, this video is a great place to start',
63 'uploader': 'Editors\' Picks',
64 'uploader_id': 'tt-editors-picks',
65 'timestamp': 1453309200,
66 'upload_date': '20160120',
68 'add_ie': ['Youtube'],
70 _PAGE_DATA_TEMPLATE = 'http://www.%s/apiProxy/ddn/%s?domain=%s'
71 _API_KEY = 'ba9c741bce1b9d8e3defcc22193f3651b8867e62'
73 def _real_extract(self, url):
74 domain, display_id = re.match(self._VALID_URL, url).groups()
75 page_info = self._download_json(
76 self._PAGE_DATA_TEMPLATE % (domain, display_id, domain), display_id)
78 page_data = page_info['data']
79 page_type = page_data['type']
80 if page_type == 'episode' or page_type == 'embed':
81 show_data = page_data['show']['data']
82 video_id = compat_str(page_data['video']['data']['id'])
84 preference = qualities(['mini', 'small', 'medium', 'large'])
88 'preference': preference(image_id)
89 } for image_id, image_url in page_data.get('images', {}).items()]
93 'display_id': display_id,
94 'title': unescapeHTML(page_data['name']),
95 'description': unescapeHTML(page_data.get('summary')),
96 'timestamp': parse_iso8601(page_data.get('publishTime'), ' '),
97 'author': page_data.get('author'),
98 'uploader': show_data.get('name'),
99 'uploader_id': show_data.get('slug'),
100 'thumbnails': thumbnails,
103 if page_type == 'embed':
105 '_type': 'url_transparent',
106 'url': page_data['video']['data']['embed'],
110 video_data = self._download_json(
111 'http://revision3.com/api/getPlaylist.json?api_key=%s&codecs=h264,vp8,theora&video_id=%s' % (self._API_KEY, video_id),
112 video_id)['items'][0]
115 for vcodec, media in video_data['media'].items():
116 for quality_id, quality in media.items():
117 if quality_id == 'hls':
118 formats.extend(self._extract_m3u8_formats(
119 quality['url'], video_id, 'mp4',
120 'm3u8_native', m3u8_id='hls', fatal=False))
123 'url': quality['url'],
124 'format_id': '%s-%s' % (vcodec, quality_id),
125 'tbr': int_or_none(quality.get('bitrate')),
128 self._sort_formats(formats)
131 'title': unescapeHTML(video_data['title']),
132 'description': unescapeHTML(video_data.get('summary')),
133 'uploader': video_data.get('show', {}).get('name'),
134 'uploader_id': video_data.get('show', {}).get('slug'),
135 'duration': int_or_none(video_data.get('duration')),
140 show_data = page_info['show']['data']
141 episodes_data = page_info['episodes']['data']
142 num_episodes = page_info['meta']['totalEpisodes']
143 processed_episodes = 0
147 entries.extend([self.url_result(
148 'http://%s/%s/%s' % (domain, display_id, episode['slug'])) for episode in episodes_data])
149 processed_episodes += len(episodes_data)
150 if processed_episodes == num_episodes:
153 episodes_data = self._download_json(self._PAGE_DATA_TEMPLATE % (
154 domain, display_id + '/' + compat_str(page_num), domain),
155 display_id)['episodes']['data']
157 return self.playlist_result(
158 entries, compat_str(show_data['id']),
159 show_data.get('name'), show_data.get('summary'))