]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/allocine.py
2 from __future__
import unicode_literals
4 from .common
import InfoExtractor
5 from ..compat
import compat_str
17 class AllocineIE(InfoExtractor
):
18 _VALID_URL
= r
'https?://(?:www\.)?allocine\.fr/(?:article|video|film)/(?:fichearticle_gen_carticle=|player_gen_cmedia=|fichefilm_gen_cfilm=|video-)(?P<id>[0-9]+)(?:\.html)?'
21 'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
22 'md5': '0c9fcf59a841f65635fa300ac43d8269',
25 'display_id': '18635087',
27 'title': 'Astérix - Le Domaine des Dieux Teaser VF',
28 'description': 'md5:4a754271d9c6f16c72629a8a993ee884',
29 'thumbnail': r
're:http://.*\.jpg',
31 'timestamp': 1404273600,
32 'upload_date': '20140702',
36 'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
37 'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
40 'display_id': '19540403',
42 'title': 'Planes 2 Bande-annonce VF',
43 'description': 'Regardez la bande annonce du film Planes 2 (Planes 2 Bande-annonce VF). Planes 2, un film de Roberts Gannaway',
44 'thumbnail': r
're:http://.*\.jpg',
46 'timestamp': 1385659800,
47 'upload_date': '20131128',
51 'url': 'http://www.allocine.fr/video/player_gen_cmedia=19544709&cfilm=181290.html',
52 'md5': '101250fb127ef9ca3d73186ff22a47ce',
55 'display_id': '19544709',
57 'title': 'Dragons 2 - Bande annonce finale VF',
58 'description': 'md5:6cdd2d7c2687d4c6aafe80a35e17267a',
59 'thumbnail': r
're:http://.*\.jpg',
61 'timestamp': 1397589900,
62 'upload_date': '20140415',
66 'url': 'http://www.allocine.fr/video/video-19550147/',
67 'md5': '3566c0668c0235e2d224fd8edb389f67',
71 'title': 'Faux Raccord N°123 - Les gaffes de Cliffhanger',
72 'description': 'md5:bc734b83ffa2d8a12188d9eb48bb6354',
73 'thumbnail': r
're:http://.*\.jpg',
77 def _real_extract(self
, url
):
78 display_id
= self
._match
_id
(url
)
80 webpage
= self
._download
_webpage
(url
, display_id
)
83 quality
= qualities(['ld', 'md', 'hd'])
85 model
= self
._html
_search
_regex
(
86 r
'data-model="([^"]+)"', webpage
, 'data model', default
=None)
88 model_data
= self
._parse
_json
(model
, display_id
)
89 video
= model_data
['videos'][0]
90 title
= video
['title']
91 for video_url
in video
['sources'].values():
92 video_id
, format_id
= url_basename(video_url
).split('_')[:2]
94 'format_id': format_id
,
95 'quality': quality(format_id
),
98 duration
= int_or_none(video
.get('duration'))
99 view_count
= int_or_none(video
.get('view_count'))
100 timestamp
= unified_timestamp(try_get(
101 video
, lambda x
: x
['added_at']['date'], compat_str
))
103 video_id
= display_id
104 media_data
= self
._download
_json
(
105 'http://www.allocine.fr/ws/AcVisiondataV5.ashx?media=%s' % video_id
, display_id
)
106 title
= remove_end(strip_or_none(self
._html
_extract
_title
(webpage
), ' - AlloCiné'))
107 for key
, value
in media_data
['video'].items():
108 if not key
.endswith('Path'):
110 format_id
= key
[:-len('Path')]
112 'format_id': format_id
,
113 'quality': quality(format_id
),
116 duration
, view_count
, timestamp
= [None] * 3
118 self
._sort
_formats
(formats
)
122 'display_id': display_id
,
124 'description': self
._og
_search
_description
(webpage
),
125 'thumbnail': self
._og
_search
_thumbnail
(webpage
),
126 'duration': duration
,
127 'timestamp': timestamp
,
128 'view_count': view_count
,