6 from .common
import InfoExtractor
7 from ..aes
import aes_cbc_decrypt_bytes
, unpad_pkcs7
8 from ..compat
import compat_urllib_parse_unquote
22 class DRTVIE(InfoExtractor
):
26 (?:www\.)?dr\.dk/(?:tv/se|nyheder|(?:radio|lyd)(?:/ondemand)?)/(?:[^/]+/)*|
27 (?:www\.)?(?:dr\.dk|dr-massive\.com)/drtv/(?:se|episode|program)/
32 _GEO_COUNTRIES
= ['DK']
35 'url': 'https://www.dr.dk/tv/se/boern/ultra/klassen-ultra/klassen-darlig-taber-10',
36 'md5': '25e659cccc9a2ed956110a299fdf5983',
38 'id': 'klassen-darlig-taber-10',
40 'title': 'Klassen - Dårlig taber (10)',
41 'description': 'md5:815fe1b7fa656ed80580f31e8b3c79aa',
42 'timestamp': 1539085800,
43 'upload_date': '20181009',
46 'season': 'Klassen I',
48 'season_id': 'urn:dr:mu:bundle:57d7e8216187a4031cfd6f6b',
49 'episode': 'Episode 10',
53 'expected_warnings': ['Unable to download f4m manifest'],
54 'skip': 'this video has been removed',
57 'url': 'https://www.dr.dk/nyheder/indland/live-christianias-rydning-af-pusher-street-er-i-gang',
59 'id': 'urn:dr:mu:programcard:57c926176187a50a9c6e83c6',
61 'title': 'christiania pusher street ryddes drdkrjpo',
62 'description': 'md5:2a71898b15057e9b97334f61d04e6eb5',
63 'timestamp': 1472800279,
64 'upload_date': '20160902',
68 'skip_download': True,
70 'expected_warnings': ['Unable to download f4m manifest'],
72 # with SignLanguage formats
73 'url': 'https://www.dr.dk/tv/se/historien-om-danmark/-/historien-om-danmark-stenalder',
77 'title': 'Historien om Danmark: Stenalder',
78 'description': 'md5:8c66dcbc1669bbc6f873879880f37f2a',
79 'timestamp': 1546628400,
80 'upload_date': '20190104',
82 'formats': 'mincount:20',
84 'season_id': 'urn:dr:mu:bundle:5afc03ad6187a4065ca5fd35',
86 'season': 'Historien om Danmark',
87 'series': 'Historien om Danmark',
90 'skip_download': True,
93 'url': 'https://www.dr.dk/lyd/p4kbh/regionale-nyheder-kh4/p4-nyheder-2019-06-26-17-30-9',
94 'only_matching': True,
96 'url': 'https://www.dr.dk/drtv/se/bonderoeven_71769',
100 'title': 'Bonderøven 2019 (1:8)',
101 'description': 'md5:b6dcfe9b6f0bea6703e9a0092739a5bd',
102 'timestamp': 1603188600,
103 'upload_date': '20201020',
105 'season': 'Bonderøven 2019',
106 'season_id': 'urn:dr:mu:bundle:5c201667a11fa01ca4528ce5',
107 'release_year': 2019,
108 'season_number': 2019,
109 'series': 'Frank & Kastaniegaarden'
112 'skip_download': True,
115 'url': 'https://www.dr.dk/drtv/episode/bonderoeven_71769',
116 'only_matching': True,
118 'url': 'https://dr-massive.com/drtv/se/bonderoeven_71769',
119 'only_matching': True,
121 'url': 'https://www.dr.dk/drtv/program/jagten_220924',
122 'only_matching': True,
124 'url': 'https://www.dr.dk/lyd/p4aarhus/regionale-nyheder-ar4/regionale-nyheder-2022-05-05-12-30-3',
126 'id': 'urn:dr:mu:programcard:6265cb2571401424d0360113',
127 'title': "Regionale nyheder",
130 'series': 'P4 Østjylland regionale nyheder',
131 'timestamp': 1651746600,
132 'season': 'Regionale nyheder',
134 'season_id': 'urn:dr:mu:bundle:61c26889539f0201586b73c5',
136 'upload_date': '20220505',
139 'skip_download': True,
143 def _real_extract(self
, url
):
144 video_id
= self
._match
_id
(url
)
146 webpage
= self
._download
_webpage
(url
, video_id
)
148 if '>Programmet er ikke længere tilgængeligt' in webpage
:
149 raise ExtractorError(
150 'Video %s is not available' % video_id
, expected
=True)
152 video_id
= self
._search
_regex
(
153 (r
'data-(?:material-identifier|episode-slug)="([^"]+)"',
154 r
'data-resource="[^>"]+mu/programcard/expanded/([^"]+)"'),
155 webpage
, 'video id', default
=None)
158 video_id
= self
._search
_regex
(
159 r
'(urn(?:%3A|:)dr(?:%3A|:)mu(?:%3A|:)programcard(?:%3A|:)[\da-f]+)',
160 webpage
, 'urn', default
=None)
162 video_id
= compat_urllib_parse_unquote(video_id
)
164 _PROGRAMCARD_BASE
= 'https://www.dr.dk/mu-online/api/1.4/programcard'
165 query
= {'expanded': 'true'}
168 programcard_url
= '%s/%s' % (_PROGRAMCARD_BASE
, video_id
)
170 programcard_url
= _PROGRAMCARD_BASE
171 page
= self
._parse
_json
(
173 r
'data\s*=\s*({.+?})\s*(?:;|</script)', webpage
,
174 'data'), '1')['cache']['page']
175 page
= page
[list(page
.keys())[0]]
177 page
, (lambda x
: x
['item'], lambda x
: x
['entries'][0]['item']),
179 video_id
= item
['customId'].split(':')[-1]
180 query
['productionnumber'] = video_id
182 data
= self
._download
_json
(
183 programcard_url
, video_id
, 'Downloading video JSON', query
=query
)
185 title
= str_or_none(data
.get('Title')) or re
.sub(
186 r
'\s*\|\s*(?:TV\s*\|\s*DR|DRTV)$', '',
187 self
._og
_search
_title
(webpage
))
188 description
= self
._og
_search
_description
(
189 webpage
, default
=None) or data
.get('Description')
191 timestamp
= unified_timestamp(
192 data
.get('PrimaryBroadcastStartTime') or data
.get('SortDateTime'))
197 restricted_to_denmark
= False
203 primary_asset
= data
.get('PrimaryAsset')
204 if isinstance(primary_asset
, dict):
205 assets
.append(primary_asset
)
206 secondary_assets
= data
.get('SecondaryAssets')
207 if isinstance(secondary_assets
, list):
208 for secondary_asset
in secondary_assets
:
209 if isinstance(secondary_asset
, dict):
210 assets
.append(secondary_asset
)
212 def hex_to_bytes(hex):
213 return binascii
.a2b_hex(hex.encode('ascii'))
218 data
= hex_to_bytes(e
[10:10 + n
])
219 key
= hashlib
.sha256(('%s:sRBzYNXBzkKgnjj8pGtkACch' % a
).encode('utf-8')).digest()
221 decrypted
= unpad_pkcs7(aes_cbc_decrypt_bytes(data
, key
, iv
))
222 return decrypted
.decode('utf-8').split('?')[0]
225 kind
= asset
.get('Kind')
227 thumbnail
= url_or_none(asset
.get('Uri'))
228 elif kind
in ('VideoResource', 'AudioResource'):
229 duration
= float_or_none(asset
.get('DurationInMilliseconds'), 1000)
230 restricted_to_denmark
= asset
.get('RestrictedToDenmark')
231 asset_target
= asset
.get('Target')
232 for link
in asset
.get('Links', []):
233 uri
= link
.get('Uri')
235 encrypted_uri
= link
.get('EncryptedUri')
236 if not encrypted_uri
:
239 uri
= decrypt_uri(encrypted_uri
)
242 'Unable to decrypt EncryptedUri', video_id
)
244 uri
= url_or_none(uri
)
247 target
= link
.get('Target')
248 format_id
= target
or ''
249 if asset_target
in ('SpokenSubtitles', 'SignLanguage', 'VisuallyInterpreted'):
251 format_id
+= '-%s' % asset_target
252 elif asset_target
== 'Default':
257 f4m_formats
= self
._extract
_f
4m
_formats
(
258 uri
+ '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43',
259 video_id
, preference
, f4m_id
=format_id
, fatal
=False)
260 if kind
== 'AudioResource':
261 for f
in f4m_formats
:
263 formats
.extend(f4m_formats
)
264 elif target
== 'HLS':
265 formats
.extend(self
._extract
_m
3u8_formats
(
266 uri
, video_id
, 'mp4', entry_protocol
='m3u8_native',
267 quality
=preference
, m3u8_id
=format_id
,
270 bitrate
= link
.get('Bitrate')
272 format_id
+= '-%s' % bitrate
275 'format_id': format_id
,
276 'tbr': int_or_none(bitrate
),
277 'ext': link
.get('FileFormat'),
278 'vcodec': 'none' if kind
== 'AudioResource' else None,
279 'quality': preference
,
281 subtitles_list
= asset
.get('SubtitlesList') or asset
.get('Subtitleslist')
282 if isinstance(subtitles_list
, list):
286 for subs
in subtitles_list
:
287 if not isinstance(subs
, dict):
289 sub_uri
= url_or_none(subs
.get('Uri'))
292 lang
= subs
.get('Language') or 'da'
293 subtitles
.setdefault(LANGS
.get(lang
, lang
), []).append({
295 'ext': mimetype2ext(subs
.get('MimeType')) or 'vtt'
298 if not formats
and restricted_to_denmark
:
299 self
.raise_geo_restricted(
300 'Unfortunately, DR is not allowed to show this program outside Denmark.',
301 countries
=self
._GEO
_COUNTRIES
)
303 self
._sort
_formats
(formats
)
308 'description': description
,
309 'thumbnail': thumbnail
,
310 'timestamp': timestamp
,
311 'duration': duration
,
313 'subtitles': subtitles
,
314 'series': str_or_none(data
.get('SeriesTitle')),
315 'season': str_or_none(data
.get('SeasonTitle')),
316 'season_number': int_or_none(data
.get('SeasonNumber')),
317 'season_id': str_or_none(data
.get('SeasonUrn')),
318 'episode': str_or_none(data
.get('EpisodeTitle')),
319 'episode_number': int_or_none(data
.get('EpisodeNumber')),
320 'release_year': int_or_none(data
.get('ProductionYear')),
324 class DRTVLiveIE(InfoExtractor
):
325 IE_NAME
= 'drtv:live'
326 _VALID_URL
= r
'https?://(?:www\.)?dr\.dk/(?:tv|TV)/live/(?P<id>[\da-z-]+)'
327 _GEO_COUNTRIES
= ['DK']
329 'url': 'https://www.dr.dk/tv/live/dr1',
333 'title': 're:^DR1 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
337 'skip_download': True,
341 def _real_extract(self
, url
):
342 channel_id
= self
._match
_id
(url
)
343 channel_data
= self
._download
_json
(
344 'https://www.dr.dk/mu-online/api/1.0/channel/' + channel_id
,
346 title
= channel_data
['Title']
349 for streaming_server
in channel_data
.get('StreamingServers', []):
350 server
= streaming_server
.get('Server')
353 link_type
= streaming_server
.get('LinkType')
354 for quality
in streaming_server
.get('Qualities', []):
355 for stream
in quality
.get('Streams', []):
356 stream_path
= stream
.get('Stream')
359 stream_url
= update_url_query(
360 '%s/%s' % (server
, stream_path
), {'b': ''}
)
361 if link_type
== 'HLS':
362 formats
.extend(self
._extract
_m
3u8_formats
(
363 stream_url
, channel_id
, 'mp4',
364 m3u8_id
=link_type
, fatal
=False, live
=True))
365 elif link_type
== 'HDS':
366 formats
.extend(self
._extract
_f
4m
_formats
(update_url_query(
367 '%s/%s' % (server
, stream_path
), {'hdcore': '3.7.0'}
),
368 channel_id
, f4m_id
=link_type
, fatal
=False))
369 self
._sort
_formats
(formats
)
374 'thumbnail': channel_data
.get('PrimaryImageUri'),