]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/europa.py
1 from .common
import InfoExtractor
13 class EuropaIE(InfoExtractor
):
14 _VALID_URL
= r
'https?://ec\.europa\.eu/avservices/(?:video/player|audio/audioDetails)\.cfm\?.*?\bref=(?P<id>[A-Za-z0-9-]+)'
16 'url': 'http://ec.europa.eu/avservices/video/player.cfm?ref=I107758',
17 'md5': '574f080699ddd1e19a675b0ddf010371',
21 'title': 'TRADE - Wikileaks on TTIP',
22 'description': 'NEW LIVE EC Midday press briefing of 11/08/2015',
23 'thumbnail': r
're:^https?://.*\.jpg$',
24 'upload_date': '20150811',
27 'formats': 'mincount:3',
30 'url': 'http://ec.europa.eu/avservices/video/player.cfm?sitelang=en&ref=I107786',
31 'only_matching': True,
33 'url': 'http://ec.europa.eu/avservices/audio/audioDetails.cfm?ref=I-109295&sitelang=en',
34 'only_matching': True,
37 def _real_extract(self
, url
):
38 video_id
= self
._match
_id
(url
)
40 playlist
= self
._download
_xml
(
41 'http://ec.europa.eu/avservices/video/player/playlist.cfm?ID=%s' % video_id
, video_id
)
43 def get_item(type_
, preference
):
45 for item
in playlist
.findall('./info/%s/item' % type_
):
46 lang
, label
= xpath_text(item
, 'lg', default
=None), xpath_text(item
, 'label', default
=None)
48 items
[lang
] = label
.strip()
54 preferred_lang
= query
.get('sitelang', ('en', ))[0]
56 preferred_langs
= orderedSet((preferred_lang
, 'en', 'int'))
58 title
= get_item('title', preferred_langs
) or video_id
59 description
= get_item('description', preferred_langs
)
60 thumbnail
= xpath_text(playlist
, './info/thumburl', 'thumbnail')
61 upload_date
= unified_strdate(xpath_text(playlist
, './info/date', 'upload date'))
62 duration
= parse_duration(xpath_text(playlist
, './info/duration', 'duration'))
63 view_count
= int_or_none(xpath_text(playlist
, './info/views', 'views'))
65 language_preference
= qualities(preferred_langs
[::-1])
68 for file_
in playlist
.findall('./files/file'):
69 video_url
= xpath_text(file_
, './url')
72 lang
= xpath_text(file_
, './lg')
76 'format_note': xpath_text(file_
, './lglabel'),
77 'language_preference': language_preference(lang
)
79 self
._sort
_formats
(formats
)
84 'description': description
,
85 'thumbnail': thumbnail
,
86 'upload_date': upload_date
,
88 'view_count': view_count
,