1 from __future__ import unicode_literals
5 from .common import InfoExtractor
10 from ..compat import (
16 class DiscoveryIE(InfoExtractor):
17 _VALID_URL = r'''(?x)https?://(?:www\.)?(?:
19 investigationdiscovery|
27 )\.com/(?:[^/]+/)*(?P<id>[^./?#]+)'''
29 'url': 'http://www.discovery.com/tv-shows/mythbusters/videos/mission-impossible-outtakes.htm',
33 'title': 'Mission Impossible Outtakes',
34 'description': ('Watch Jamie Hyneman and Adam Savage practice being'
35 ' each other -- to the point of confusing Jamie\'s dog -- and '
36 'don\'t miss Adam moon-walking as Jamie ... behind Jamie\'s'
39 'timestamp': 1302032462,
40 'upload_date': '20110405',
43 'skip_download': True, # requires ffmpeg
46 'url': 'http://www.discovery.com/tv-shows/mythbusters/videos/mythbusters-the-simpsons',
48 'id': 'mythbusters-the-simpsons',
49 'title': 'MythBusters: The Simpsons',
51 'playlist_mincount': 10,
53 'url': 'http://www.animalplanet.com/longfin-eels-maneaters/',
57 'title': 'Longfin Eels: Maneaters?',
58 'description': 'Jeremy Wade tests whether or not New Zealand\'s longfin eels are man-eaters by covering himself in fish guts and getting in the water with them.',
59 'upload_date': '20140725',
60 'timestamp': 1406246400,
65 def _real_extract(self, url):
66 display_id = self._match_id(url)
67 info = self._download_json(url + '?flat=1', display_id)
69 video_title = info.get('playlist_title') or info.get('video_title')
73 for idx, video_info in enumerate(info['playlist']):
74 m3u8_url = video_info['src']
75 formats = m3u8_formats = self._extract_m3u8_formats(
76 m3u8_url, display_id, 'mp4', 'm3u8_native', m3u8_id='hls',
77 note='Download m3u8 information for video %d' % (idx + 1))
78 qualities_basename = self._search_regex(
79 '/([^/]+)\.csmil/', m3u8_url, 'qualities basename', default=None)
80 if qualities_basename:
81 m3u8_path = compat_urlparse.urlparse(m3u8_url).path
82 QUALITIES_RE = r'((,\d+k)+,?)'
83 qualities = self._search_regex(
84 QUALITIES_RE, qualities_basename,
85 'qualities', default=None)
87 qualities = list(map(lambda q: int(q[:-1]), qualities.strip(',').split(',')))
89 http_path = m3u8_path[1:].split('/', 1)[1]
90 http_template = re.sub(QUALITIES_RE, r'%dk', http_path)
91 http_template = http_template.replace('.csmil/master.m3u8', '')
92 http_template = compat_urlparse.urljoin(
93 'http://discsmil.edgesuite.net/', http_template)
95 self._sort_formats(m3u8_formats)
96 m3u8_formats = list(filter(
97 lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
99 if len(qualities) == len(m3u8_formats):
100 for q, m3u8_format in zip(qualities, m3u8_formats):
101 f = m3u8_format.copy()
103 'url': http_template % q,
104 'format_id': f['format_id'].replace('hls', 'http'),
111 'url': http_template % q,
113 'format_id': 'http-%d' % q,
116 self._sort_formats(formats)
118 'id': compat_str(video_info['id']),
120 'title': video_info['title'],
121 'description': video_info.get('description'),
122 'duration': parse_duration(video_info.get('video_length')),
123 'webpage_url': video_info.get('href') or video_info.get('url'),
124 'thumbnail': video_info.get('thumbnailURL'),
125 'alt_title': video_info.get('secondary_title'),
126 'timestamp': parse_iso8601(video_info.get('publishedDate')),
129 return self.playlist_result(entries, display_id, video_title)