2 from __future__
import unicode_literals
6 from .common
import InfoExtractor
9 get_elements_text_and_html_by_attribute
,
15 class TVOpenGrBaseIE(InfoExtractor
):
16 def _return_canonical_url(self
, url
, video_id
):
17 webpage
= self
._download
_webpage
(url
, video_id
)
18 canonical_url
= self
._og
_search
_url
(webpage
)
19 title
= self
._og
_search
_title
(webpage
)
20 return self
.url_result(canonical_url
, ie
=TVOpenGrWatchIE
.ie_key(), video_id
=video_id
, video_title
=title
)
23 class TVOpenGrWatchIE(TVOpenGrBaseIE
):
24 IE_NAME
= 'tvopengr:watch'
25 IE_DESC
= 'tvopen.gr (and ethnos.gr) videos'
26 _VALID_URL
= r
'https?://(?P<netloc>(?:www\.)?(?:tvopen|ethnos)\.gr)/watch/(?P<id>\d+)/(?P<slug>[^/]+)'
27 _API_ENDPOINT
= 'https://www.tvopen.gr/templates/data/player'
30 'url': 'https://www.ethnos.gr/watch/101009/nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
31 'md5': '8728570e3a72e0f8d9475ba94859fdc1',
34 'title': 'md5:51f68773dcb6c70498cd326f45fefdf0',
35 'display_id': 'nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
36 'description': 'md5:78fff49f18fb3effe41b070e5c7685d6',
37 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/d573ba71-ec5f-43c6-b4cb-d181f327d3a8.jpg',
39 'upload_date': '20220109',
40 'timestamp': 1641686400,
43 'url': 'https://www.tvopen.gr/watch/100979/se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
44 'md5': '38f98a1be0c577db4ea2d1b1c0770c48',
47 'title': 'md5:e021f3001e16088ee40fa79b20df305b',
48 'display_id': 'se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
49 'description': 'md5:ba17db53954134eb8d625d199e2919fb',
50 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/9bb71cf1-21da-43a9-9d65-367950fde4e3.jpg',
52 'upload_date': '20220108',
53 'timestamp': 1641600000,
57 def _extract_formats_and_subs(self
, response
, video_id
):
58 formats
, subs
= [], {}
59 for format_id
, format_url
in response
.items():
60 if format_id
not in ('stream', 'httpstream', 'mpegdash'):
62 ext
= determine_ext(format_url
)
64 formats_
, subs_
= self
._extract
_m
3u8_formats
_and
_subtitles
(
65 format_url
, video_id
, 'mp4', m3u8_id
=format_id
,
68 formats_
, subs_
= self
._extract
_mpd
_formats
_and
_subtitles
(
69 format_url
, video_id
, 'mp4', fatal
=False)
73 'format_id': format_id
,
76 formats
.extend(formats_
)
77 self
._merge
_subtitles
(subs_
, target
=subs
)
78 self
._sort
_formats
(formats
)
82 def _scale_thumbnails_to_max_width(formats
, thumbnails
, url_width_re
):
83 _keys
= ('width', 'height')
85 [tuple(format
.get(k
) or 0 for k
in _keys
) for format
in formats
],
87 if not max_dimensions
[0]:
91 {'url': re.sub(url_width_re, str(max_dimensions[0]), thumbnail['url'])}
,
92 dict(zip(_keys
, max_dimensions
)), thumbnail
)
93 for thumbnail
in thumbnails
96 def _real_extract(self
, url
):
97 netloc
, video_id
, display_id
= self
._match
_valid
_url
(url
).group('netloc', 'id', 'slug')
98 if netloc
.find('tvopen.gr') == -1:
99 return self
._return
_canonical
_url
(url
, video_id
)
100 webpage
= self
._download
_webpage
(url
, video_id
)
101 info
= self
._search
_json
_ld
(webpage
, video_id
, expected_type
='VideoObject')
102 info
['formats'], info
['subtitles'] = self
._extract
_formats
_and
_subs
(
103 self
._download
_json
(self
._API
_ENDPOINT
, video_id
, query
={'cid': video_id}
),
105 info
['thumbnails'] = self
._scale
_thumbnails
_to
_max
_width
(
106 info
['formats'], info
['thumbnails'], r
'(?<=/imgHandler/)\d+')
107 description
, _html
= next(get_elements_text_and_html_by_attribute('class', 'description', webpage
))
108 if description
and _html
.startswith('<span '):
109 info
['description'] = description
110 info
['id'] = video_id
111 info
['display_id'] = display_id
115 class TVOpenGrEmbedIE(TVOpenGrBaseIE
):
116 IE_NAME
= 'tvopengr:embed'
117 IE_DESC
= 'tvopen.gr embedded videos'
118 _VALID_URL
= r
'(?:https?:)?//(?:www\.|cdn\.|)(?:tvopen|ethnos).gr/embed/(?P<id>\d+)'
119 _EMBED_RE
= re
.compile(rf
'''<iframe[^>]+?src=(?P<_q1>["'])(?P<url>{_VALID_URL})(?P=_q1)''')
122 'url': 'https://cdn.ethnos.gr/embed/100963',
123 'md5': '2da147881f45571d81662d94d086628b',
126 'display_id': 'koronoiosapotoysdieythyntestonsxoleionselftestgiaosoysdenbrhkan',
127 'title': 'md5:2c71876fadf0cda6043da0da5fca2936',
128 'description': 'md5:17482b4432e5ed30eccd93b05d6ea509',
129 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/5804e07f-799a-4247-a696-33842c94ca37.jpg',
131 'upload_date': '20220108',
132 'timestamp': 1641600000,
137 def _extract_urls(cls
, webpage
):
138 for mobj
in cls
._EMBED
_RE
.finditer(webpage
):
139 yield unescapeHTML(mobj
.group('url'))
141 def _real_extract(self
, url
):
142 video_id
= self
._match
_id
(url
)
143 return self
._return
_canonical
_url
(url
, video_id
)