]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/americastestkitchen.py
Update upstream source from tag 'upstream/2019.06.08'
[youtubedl] / youtube_dl / extractor / americastestkitchen.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6 clean_html,
7 int_or_none,
8 try_get,
9 unified_strdate,
10 )
11
12
13 class AmericasTestKitchenIE(InfoExtractor):
14 _VALID_URL = r'https?://(?:www\.)?americastestkitchen\.com/(?:episode|videos)/(?P<id>\d+)'
15 _TESTS = [{
16 'url': 'https://www.americastestkitchen.com/episode/548-summer-dinner-party',
17 'md5': 'b861c3e365ac38ad319cfd509c30577f',
18 'info_dict': {
19 'id': '1_5g5zua6e',
20 'title': 'Summer Dinner Party',
21 'ext': 'mp4',
22 'description': 'md5:858d986e73a4826979b6a5d9f8f6a1ec',
23 'thumbnail': r're:^https?://.*\.jpg',
24 'timestamp': 1497285541,
25 'upload_date': '20170612',
26 'uploader_id': 'roger.metcalf@americastestkitchen.com',
27 'release_date': '20170617',
28 'series': "America's Test Kitchen",
29 'season_number': 17,
30 'episode': 'Summer Dinner Party',
31 'episode_number': 24,
32 },
33 'params': {
34 'skip_download': True,
35 },
36 }, {
37 'url': 'https://www.americastestkitchen.com/videos/3420-pan-seared-salmon',
38 'only_matching': True,
39 }]
40
41 def _real_extract(self, url):
42 video_id = self._match_id(url)
43
44 webpage = self._download_webpage(url, video_id)
45
46 video_data = self._parse_json(
47 self._search_regex(
48 r'window\.__INITIAL_STATE__\s*=\s*({.+?})\s*;\s*</script>',
49 webpage, 'initial context'),
50 video_id)
51
52 ep_data = try_get(
53 video_data,
54 (lambda x: x['episodeDetail']['content']['data'],
55 lambda x: x['videoDetail']['content']['data']), dict)
56 ep_meta = ep_data.get('full_video', {})
57
58 zype_id = ep_meta.get('zype_id')
59 if zype_id:
60 embed_url = 'https://player.zype.com/embed/%s.js?api_key=jZ9GUhRmxcPvX7M3SlfejB6Hle9jyHTdk2jVxG7wOHPLODgncEKVdPYBhuz9iWXQ' % zype_id
61 ie_key = 'Zype'
62 else:
63 partner_id = self._search_regex(
64 r'src=["\'](?:https?:)?//(?:[^/]+\.)kaltura\.com/(?:[^/]+/)*(?:p|partner_id)/(\d+)',
65 webpage, 'kaltura partner id')
66 external_id = ep_data.get('external_id') or ep_meta['external_id']
67 embed_url = 'kaltura:%s:%s' % (partner_id, external_id)
68 ie_key = 'Kaltura'
69
70 title = ep_data.get('title') or ep_meta.get('title')
71 description = clean_html(ep_meta.get('episode_description') or ep_data.get(
72 'description') or ep_meta.get('description'))
73 thumbnail = try_get(ep_meta, lambda x: x['photo']['image_url'])
74 release_date = unified_strdate(ep_data.get('aired_at'))
75
76 season_number = int_or_none(ep_meta.get('season_number'))
77 episode = ep_meta.get('title')
78 episode_number = int_or_none(ep_meta.get('episode_number'))
79
80 return {
81 '_type': 'url_transparent',
82 'url': embed_url,
83 'ie_key': ie_key,
84 'title': title,
85 'description': description,
86 'thumbnail': thumbnail,
87 'release_date': release_date,
88 'series': "America's Test Kitchen",
89 'season_number': season_number,
90 'episode': episode,
91 'episode_number': episode_number,
92 }