3 from .common
import InfoExtractor
19 class OnetBaseIE(InfoExtractor
):
20 _URL_BASE_RE
= r
'https?://(?:(?:www\.)?onet\.tv|onet100\.vod\.pl)/[a-z]/'
22 def _search_mvp_id(self
, webpage
):
23 return self
._search
_regex
(
24 r
'id=(["\'])mvp
:(?P
<id>.+?
)\
1', webpage, 'mvp
id', group='id')
26 def _extract_from_id(self, video_id, webpage=None):
27 response = self._download_json(
28 'http
://qi
.ckm
.onetapi
.pl
/', video_id,
31 'body
[jsonrpc
]': '2.0',
32 'body
[method
]': 'get_asset_detail
',
33 'body
[params
][ID_Publikacji
]': video_id,
34 'body
[params
][Service
]': 'www
.onet
.pl
',
35 'content
-type': 'application
/jsonp
',
36 'x
-onet
-app
': 'player
.front
.onetapi
.pl
',
39 error = response.get('error
')
42 '%s said
: %s' % (self.IE_NAME, error['message
']), expected=True)
44 video = response['result
'].get('0')
47 for format_type, formats_dict in video['formats
'].items():
48 if not isinstance(formats_dict, dict):
50 for format_id, format_list in formats_dict.items():
51 if not isinstance(format_list, list):
54 video_url = f.get('url
')
57 ext = determine_ext(video_url)
58 if format_id.startswith('ism
'):
59 formats.extend(self._extract_ism_formats(
60 video_url, video_id, 'mss
', fatal=False))
62 formats.extend(self._extract_mpd_formats(
63 video_url, video_id, mpd_id='dash
', fatal=False))
64 elif format_id.startswith('hls
'):
65 formats.extend(self._extract_m3u8_formats(
66 video_url, video_id, 'mp4
', 'm3u8_native
',
67 m3u8_id='hls
', fatal=False))
71 'format_id
': format_id,
72 'abr
': float_or_none(f.get('audio_bitrate
')),
74 if format_type == 'audio
':
75 http_f['vcodec
'] = 'none
'
78 'height
': int_or_none(f.get('vertical_resolution
')),
79 'width
': int_or_none(f.get('horizontal_resolution
')),
80 'vbr
': float_or_none(f.get('video_bitrate
')),
82 formats.append(http_f)
83 self._sort_formats(formats)
85 meta = video.get('meta
', {})
87 title = (self._og_search_title(
88 webpage, default=None) if webpage else None) or meta['title
']
89 description = (self._og_search_description(
90 webpage, default=None) if webpage else None) or meta.get('description
')
91 duration = meta.get('length
') or meta.get('lenght
')
92 timestamp = parse_iso8601(meta.get('addDate
'), ' ')
97 'description
': description,
99 'timestamp
': timestamp,
104 class OnetMVPIE(OnetBaseIE):
105 _VALID_URL = r'onetmvp
:(?P
<id>\d
+\
.\d
+)'
108 'url
': 'onetmvp
:381027.1509591944',
109 'only_matching
': True,
112 def _real_extract(self, url):
113 return self._extract_from_id(self._match_id(url))
116 class OnetIE(OnetBaseIE):
117 _VALID_URL = OnetBaseIE._URL_BASE_RE + r'[a
-z
]+/(?P
<display_id
>[0-9a
-z
-]+)/(?P
<id>[0-9a
-z
]+)'
121 'url
': 'http
://onet
.tv
/k
/openerfestival
/open-er
-festival
-2016-najdziwniejsze
-wymagania
-gwiazd
/qbpyqc
',
122 'md5
': '436102770fb095c75b8bb0392d3da9ff
',
125 'display_id
': 'open-er
-festival
-2016-najdziwniejsze
-wymagania
-gwiazd
',
127 'title
': 'Open
\'er Festival
2016: najdziwniejsze wymagania gwiazd
',
128 'description
': 'Trzy samochody
, których nigdy nie użyto
, prywatne spa
, hotel dekorowany czarnym suknem czy nielegalne używki
. Organizatorzy koncertów i festiwali muszą stawać przed nie lada wyzwaniem zapraszając gwia
...',
129 'upload_date
': '20160705',
130 'timestamp
': 1467721580,
133 'url
': 'https
://onet100
.vod
.pl
/k
/openerfestival
/open-er
-festival
-2016-najdziwniejsze
-wymagania
-gwiazd
/qbpyqc
',
134 'only_matching
': True,
137 def _real_extract(self, url):
138 mobj = self._match_valid_url(url)
139 display_id, video_id = mobj.group('display_id
', 'id')
141 webpage = self._download_webpage(url, display_id)
143 mvp_id = self._search_mvp_id(webpage)
145 info_dict = self._extract_from_id(mvp_id, webpage)
148 'display_id
': display_id,
154 class OnetChannelIE(OnetBaseIE):
155 _VALID_URL = OnetBaseIE._URL_BASE_RE + r'(?P
<id>[a
-z
]+)(?
:[?
#]|$)'
156 IE_NAME
= 'onet.tv:channel'
159 'url': 'http://onet.tv/k/openerfestival',
161 'id': 'openerfestival',
162 'title': "Open'er Festival",
163 'description': "Tak było na Open'er Festival 2016! Oglądaj nasze reportaże i wywiady z artystami.",
165 'playlist_mincount': 35,
167 'url': 'https://onet100.vod.pl/k/openerfestival',
168 'only_matching': True,
171 def _real_extract(self
, url
):
172 channel_id
= self
._match
_id
(url
)
174 webpage
= self
._download
_webpage
(url
, channel_id
)
176 current_clip_info
= self
._parse
_json
(self
._search
_regex
(
177 r
'var\s+currentClip\s*=\s*({[^}]+})', webpage
, 'video info'), channel_id
,
178 transform_source
=lambda s
: js_to_json(re
.sub(r
'\'\s
*\
+\s
*\'', '', s)))
179 video_id = remove_start(current_clip_info['ckmId
'], 'mvp
:')
180 video_name = url_basename(current_clip_info['url
'])
182 if not self._yes_playlist(channel_id, video_name, playlist_label='channel
'):
183 return self._extract_from_id(video_id, webpage)
185 matches = re.findall(
186 r'<a
[^
>]+href
=[\'"](%s[a-z]+/[0-9a-z-]+/[0-9a-z]+)' % self._URL_BASE_RE,
189 self.url_result(video_link, OnetIE.ie_key())
190 for video_link in matches]
192 channel_title = strip_or_none(get_element_by_class('o_channelName', webpage))
193 channel_description = strip_or_none(get_element_by_class('o_channelDesc', webpage))
194 return self.playlist_result(entries, channel_id, channel_title, channel_description)
197 class OnetPlIE(InfoExtractor):
198 _VALID_URL = r'https?://(?:[^/]+\.)?(?:onet|businessinsider\.com|plejada)\.pl/(?:[^/]+/)+(?P<id>[0-9a-z]+)'
202 'url': 'http://eurosport.onet.pl/zimowe/skoki-narciarskie/ziobro-wygral-kwalifikacje-w-pjongczangu/9ckrly',
203 'md5': 'b94021eb56214c3969380388b6e73cb0',
205 'id': '1561707.1685479',
207 'title': 'Ziobro wygrał kwalifikacje w Pjongczangu',
208 'description': 'md5:61fb0740084d2d702ea96512a03585b4',
209 'upload_date': '20170214',
210 'timestamp': 1487078046,
213 # embedded via pulsembed
214 'url': 'http://film.onet.pl/pensjonat-nad-rozlewiskiem-relacja-z-planu-serialu/y428n0',
216 'id': '501235.965429946',
218 'title': '"Pensjonat nad rozlewiskiem
": relacja z planu serialu',
219 'upload_date': '20170622',
220 'timestamp': 1498159955,
223 'skip_download': True,
226 'url': 'http://film.onet.pl/zwiastuny/ghost-in-the-shell-drugi-zwiastun-pl/5q6yl3',
227 'only_matching': True,
229 'url': 'http://moto.onet.pl/jak-wybierane-sa-miejsca-na-fotoradary/6rs04e',
230 'only_matching': True,
232 'url': 'http://businessinsider.com.pl/wideo/scenariusz-na-koniec-swiata-wedlug-nasa/dwnqptk',
233 'only_matching': True,
235 'url': 'http://plejada.pl/weronika-rosati-o-swoim-domniemanym-slubie/n2bq89',
236 'only_matching': True,
239 def _search_mvp_id(self, webpage, default=NO_DEFAULT):
240 return self._search_regex(
241 r'data-(?:params-)?mvp=["\'](\d
+\
.\d
+)', webpage, 'mvp
id',
244 def _real_extract(self, url):
245 video_id = self._match_id(url)
247 webpage = self._download_webpage(url, video_id)
249 mvp_id = self._search_mvp_id(webpage, default=None)
252 pulsembed_url = self._search_regex(
253 r'data
-src
=(["\'])(?P<url>(?:https?:)?//pulsembed\.eu/.+?)\1',
254 webpage, 'pulsembed url', group='url')
255 webpage = self._download_webpage(
256 pulsembed_url, video_id, 'Downloading pulsembed webpage')
257 mvp_id = self._search_mvp_id(webpage)
259 return self.url_result(
260 'onetmvp:%s' % mvp_id, OnetMVPIE.ie_key(), video_id=mvp_id)