3 from .common
import InfoExtractor
4 from ..compat
import compat_str
18 class TVNowBaseIE(InfoExtractor
):
20 'id', 'title', 'free', 'geoblocked', 'articleLong', 'articleShort',
21 'broadcastStartDate', 'isDrm', 'duration', 'season', 'episode',
22 'manifest.dashclear', 'manifest.hlsclear', 'manifest.smoothclear',
23 'format.title', 'format.defaultImage169Format', 'format.defaultImage169Logo')
25 def _call_api(self
, path
, video_id
, query
):
26 return self
._download
_json
(
27 'https://api.tvnow.de/v3/' + path
, video_id
, query
=query
)
29 def _extract_video(self
, info
, display_id
):
30 video_id
= compat_str(info
['id'])
34 for manifest_url
in (info
.get('manifest') or {}).values():
37 manifest_url
= update_url_query(manifest_url
, {'filter': ''}
)
38 path
= self
._search
_regex
(r
'https?://[^/]+/(.+?)\.ism/', manifest_url
, 'path')
43 def url_repl(proto
, suffix
):
45 r
'(?:hls|dash|hss)([.-])', proto
+ r
'\1', re
.sub(
46 r
'\.ism/(?:[^.]*\.(?:m3u8|mpd)|[Mm]anifest)',
47 '.ism/' + suffix
, manifest_url
))
49 def make_urls(proto
, suffix
):
50 urls
= [url_repl(proto
, suffix
)]
51 hd_url
= urls
[0].replace('/manifest/', '/ngvod/')
56 for man_url
in make_urls('dash', '.mpd'):
57 formats
= self
._extract
_mpd
_formats
(
58 man_url
, video_id
, mpd_id
='dash', fatal
=False)
59 for man_url
in make_urls('hss', 'Manifest'):
60 formats
.extend(self
._extract
_ism
_formats
(
61 man_url
, video_id
, ism_id
='mss', fatal
=False))
62 for man_url
in make_urls('hls', '.m3u8'):
63 formats
.extend(self
._extract
_m
3u8_formats
(
64 man_url
, video_id
, 'mp4', 'm3u8_native', m3u8_id
='hls',
69 if not self
.get_param('allow_unplayable_formats') and info
.get('isDrm'):
71 'Video %s is DRM protected' % video_id
, expected
=True)
72 if info
.get('geoblocked'):
73 raise self
.raise_geo_restricted()
74 if not info
.get('free', True):
76 'Video %s is not available for free' % video_id
, expected
=True)
78 description
= info
.get('articleLong') or info
.get('articleShort')
79 timestamp
= parse_iso8601(info
.get('broadcastStartDate'), ' ')
80 duration
= parse_duration(info
.get('duration'))
82 f
= info
.get('format', {})
85 'url': 'https://aistvnow-a.akamaihd.net/tvnow/movie/%s' % video_id
,
87 thumbnail
= f
.get('defaultImage169Format') or f
.get('defaultImage169Logo')
95 'display_id': display_id
,
97 'description': description
,
98 'thumbnails': thumbnails
,
99 'timestamp': timestamp
,
100 'duration': duration
,
101 'series': f
.get('title'),
102 'season_number': int_or_none(info
.get('season')),
103 'episode_number': int_or_none(info
.get('episode')),
109 class TVNowIE(TVNowBaseIE
):
110 _VALID_URL
= r
'''(?x)
112 (?:www\.)?tvnow\.(?:de|at|ch)/(?P<station>[^/]+)/
114 (?!(?:list|jahr)(?:/|$))(?P<id>[^/?\#&]+)
118 def suitable(cls
, url
):
119 return (False if TVNowNewIE
.suitable(url
) or TVNowSeasonIE
.suitable(url
) or TVNowAnnualIE
.suitable(url
) or TVNowShowIE
.suitable(url
)
120 else super(TVNowIE
, cls
).suitable(url
))
123 'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/der-neue-porsche-911-gt-3/player',
126 'display_id': 'grip-das-motormagazin/der-neue-porsche-911-gt-3',
128 'title': 'Der neue Porsche 911 GT 3',
129 'description': 'md5:6143220c661f9b0aae73b245e5d898bb',
130 'timestamp': 1495994400,
131 'upload_date': '20170528',
133 'series': 'GRIP - Das Motormagazin',
135 'episode_number': 405,
136 'episode': 'Der neue Porsche 911 GT 3',
140 'url': 'https://www.tvnow.de/rtl2/armes-deutschland/episode-0008/player',
141 'only_matching': True,
144 'url': 'https://www.tvnow.de/nitro/alarm-fuer-cobra-11-die-autobahnpolizei/auf-eigene-faust-pilot/player',
145 'only_matching': True,
148 'url': 'https://www.tvnow.de/superrtl/die-lustigsten-schlamassel-der-welt/u-a-ketchup-effekt/player',
149 'only_matching': True,
152 'url': 'https://www.tvnow.de/ntv/startup-news/goetter-in-weiss/player',
153 'only_matching': True,
156 'url': 'https://www.tvnow.de/vox/auto-mobil/neues-vom-automobilmarkt-2017-11-19-17-00-00/player',
157 'only_matching': True,
160 'url': 'https://www.tvnow.de/rtlplus/op-ruft-dr-bruckner/die-vernaehte-frau/player',
161 'only_matching': True,
163 'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/der-neue-porsche-911-gt-3',
164 'only_matching': True,
167 def _real_extract(self
, url
):
168 mobj
= self
._match
_valid
_url
(url
)
169 display_id
= '%s/%s' % mobj
.group(2, 3)
171 info
= self
._call
_api
(
172 'movies/' + display_id
, display_id
, query
={
173 'fields': ','.join(self
._VIDEO
_FIELDS
),
176 return self
._extract
_video
(info
, display_id
)
179 class TVNowNewIE(InfoExtractor
):
180 _VALID_URL
= r
'''(?x)
181 (?P<base_url>https?://
182 (?:www\.)?tvnow\.(?:de|at|ch)/
186 episode-\d+-(?P<episode>[^/?$&]+)-(?P<id>\d+)
190 'url': 'https://www.tvnow.de/shows/grip-das-motormagazin-1669/2017-05/episode-405-der-neue-porsche-911-gt-3-331082',
191 'only_matching': True,
194 def _real_extract(self
, url
):
195 mobj
= self
._match
_valid
_url
(url
)
196 base_url
= re
.sub(r
'(?:shows|serien)', '_', mobj
.group('base_url'))
197 show
, episode
= mobj
.group('show', 'episode')
198 return self
.url_result(
199 # Rewrite new URLs to the old format and use extraction via old API
200 # at api.tvnow.de as a loophole for bypassing premium content checks
201 '%s/%s/%s' % (base_url
, show
, episode
),
202 ie
=TVNowIE
.ie_key(), video_id
=mobj
.group('id'))
205 class TVNowFilmIE(TVNowBaseIE
):
206 _VALID_URL
= r
'''(?x)
207 (?P<base_url>https?://
208 (?:www\.)?tvnow\.(?:de|at|ch)/
210 (?P<title>[^/?$&]+)-(?P<id>\d+)
213 'url': 'https://www.tvnow.de/filme/lord-of-war-haendler-des-todes-7959',
216 'display_id': 'lord-of-war-haendler-des-todes',
218 'title': 'Lord of War',
219 'description': 'md5:5eda15c0d5b8cb70dac724c8a0ff89a9',
220 'timestamp': 1550010000,
221 'upload_date': '20190212',
225 'url': 'https://www.tvnow.de/filme/the-machinist-12157',
228 'display_id': 'the-machinist',
230 'title': 'The Machinist',
231 'description': 'md5:9a0e363fdd74b3a9e1cdd9e21d0ecc28',
232 'timestamp': 1496469720,
233 'upload_date': '20170603',
237 'url': 'https://www.tvnow.de/filme/horst-schlaemmer-isch-kandidiere-17777',
238 'only_matching': True, # DRM protected
241 def _real_extract(self
, url
):
242 mobj
= self
._match
_valid
_url
(url
)
243 display_id
= mobj
.group('title')
245 webpage
= self
._download
_webpage
(url
, display_id
, fatal
=False)
247 raise ExtractorError('Cannot download "%s"' % url
, expected
=True)
249 json_text
= get_element_by_id('now-web-state', webpage
)
251 raise ExtractorError('Cannot read video data', expected
=True)
253 json_data
= self
._parse
_json
(
256 transform_source
=lambda x
: x
.replace('&q;', '"'),
259 raise ExtractorError('Cannot read video data', expected
=True)
262 (key
for key
in json_data
.keys() if 'module/player' in key
),
265 (key
for key
in json_data
.keys() if 'page/filme' in key
),
270 lambda x
: x
[player_key
]['body']['id'],
271 lambda x
: x
[page_key
]['body']['modules'][0]['id'],
272 lambda x
: x
[page_key
]['body']['modules'][1]['id']],
275 raise ExtractorError('Cannot extract movie ID', expected
=True)
277 info
= self
._call
_api
(
278 'movies/%d' % movie_id
,
280 query
={'fields': ','.join(self._VIDEO_FIELDS)}
)
282 return self
._extract
_video
(info
, display_id
)
285 class TVNowNewBaseIE(InfoExtractor
):
286 def _call_api(self
, path
, video_id
, query
={}):
287 result
= self
._download
_json
(
288 'https://apigw.tvnow.de/module/' + path
, video_id
, query
=query
)
289 error
= result
.get('error')
291 raise ExtractorError(
292 '%s said: %s' % (self
.IE_NAME
, error
), expected
=True)
297 TODO: new apigw.tvnow.de based version of TVNowIE. Replace old TVNowIE with it
298 when api.tvnow.de is shut down. This version can't bypass premium checks though.
299 class TVNowIE(TVNowNewBaseIE):
300 _VALID_URL = r'''(?x)
302 (?:www\.)?tvnow\.(?:de|at|ch)/
303 (?:shows|serien)/[^/]+/
305 (?P<display_id>[^/?$&]+)-(?P<id>\d+)
309 # episode with annual navigation
310 'url': 'https://www.tvnow.de/shows/grip-das-motormagazin-1669/2017-05/episode-405-der-neue-porsche-911-gt-3-331082',
313 'display_id': 'grip-das-motormagazin/der-neue-porsche-911-gt-3',
315 'title': 'Der neue Porsche 911 GT 3',
316 'description': 'md5:6143220c661f9b0aae73b245e5d898bb',
317 'thumbnail': r're:^https?://.*\.jpg$',
318 'timestamp': 1495994400,
319 'upload_date': '20170528',
321 'series': 'GRIP - Das Motormagazin',
323 'episode_number': 405,
324 'episode': 'Der neue Porsche 911 GT 3',
327 # rtl2, episode with season navigation
328 'url': 'https://www.tvnow.de/shows/armes-deutschland-11471/staffel-3/episode-14-bernd-steht-seit-der-trennung-von-seiner-frau-allein-da-526124',
329 'only_matching': True,
332 'url': 'https://www.tvnow.de/serien/alarm-fuer-cobra-11-die-autobahnpolizei-1815/staffel-13/episode-5-auf-eigene-faust-pilot-366822',
333 'only_matching': True,
336 'url': 'https://www.tvnow.de/shows/die-lustigsten-schlamassel-der-welt-1221/staffel-2/episode-14-u-a-ketchup-effekt-364120',
337 'only_matching': True,
340 'url': 'https://www.tvnow.de/shows/startup-news-10674/staffel-2/episode-39-goetter-in-weiss-387630',
341 'only_matching': True,
344 'url': 'https://www.tvnow.de/shows/auto-mobil-174/2017-11/episode-46-neues-vom-automobilmarkt-2017-11-19-17-00-00-380072',
345 'only_matching': True,
347 'url': 'https://www.tvnow.de/shows/grip-das-motormagazin-1669/2017-05/episode-405-der-neue-porsche-911-gt-3-331082',
348 'only_matching': True,
351 def _extract_video(self, info, url, display_id):
352 config = info['config']
353 source = config['source']
355 video_id = compat_str(info.get('id') or source['videoId'])
356 title = source['title'].strip()
359 for manifest_url in (info.get('manifest') or {}).values():
362 manifest_url = update_url_query(manifest_url, {'filter': ''})
363 path = self._search_regex(r'https?://[^/]+/(.+?)\.ism/', manifest_url, 'path')
368 def url_repl(proto, suffix):
370 r'(?:hls|dash|hss)([.-])', proto + r'\1', re.sub(
371 r'\.ism/(?:[^.]*\.(?:m3u8|mpd)|[Mm]anifest)',
372 '.ism/' + suffix, manifest_url))
374 formats = self._extract_mpd_formats(
375 url_repl('dash', '.mpd'), video_id,
376 mpd_id='dash', fatal=False)
377 formats.extend(self._extract_ism_formats(
378 url_repl('hss', 'Manifest'),
379 video_id, ism_id='mss', fatal=False))
380 formats.extend(self._extract_m3u8_formats(
381 url_repl('hls', '.m3u8'), video_id, 'mp4',
382 'm3u8_native', m3u8_id='hls', fatal=False))
386 if try_get(info, lambda x: x['rights']['isDrm']):
387 raise ExtractorError(
388 'Video %s is DRM protected' % video_id, expected=True)
389 if try_get(config, lambda x: x['boards']['geoBlocking']['block']):
390 raise self.raise_geo_restricted()
391 if not info.get('free', True):
392 raise ExtractorError(
393 'Video %s is not available for free' % video_id, expected=True)
395 description = source.get('description')
396 thumbnail = url_or_none(source.get('poster'))
397 timestamp = unified_timestamp(source.get('previewStart'))
398 duration = parse_duration(source.get('length'))
400 series = source.get('format')
401 season_number = int_or_none(self._search_regex(
402 r'staffel-(\d+)', url, 'season number', default=None))
403 episode_number = int_or_none(self._search_regex(
404 r'episode-(\d+)', url, 'episode number', default=None))
408 'display_id': display_id,
410 'description': description,
411 'thumbnail': thumbnail,
412 'timestamp': timestamp,
413 'duration': duration,
415 'season_number': season_number,
416 'episode_number': episode_number,
421 def _real_extract(self, url):
422 display_id, video_id = self._match_valid_url(url).groups()
423 info = self._call_api('player/' + video_id, video_id)
424 return self._extract_video(info, video_id, display_id)
427 class TVNowFilmIE(TVNowIE): # XXX: Do not subclass from concrete IE
428 _VALID_URL = r'''(?x)
429 (?P<base_url>https?://
430 (?:www\.)?tvnow\.(?:de|at|ch)/
432 (?P<title>[^/?$&]+)-(?P<id>\d+)
435 'url': 'https://www.tvnow.de/filme/lord-of-war-haendler-des-todes-7959',
438 'display_id': 'lord-of-war-haendler-des-todes',
440 'title': 'Lord of War',
441 'description': 'md5:5eda15c0d5b8cb70dac724c8a0ff89a9',
442 'timestamp': 1550010000,
443 'upload_date': '20190212',
447 'url': 'https://www.tvnow.de/filme/the-machinist-12157',
450 'display_id': 'the-machinist',
452 'title': 'The Machinist',
453 'description': 'md5:9a0e363fdd74b3a9e1cdd9e21d0ecc28',
454 'timestamp': 1496469720,
455 'upload_date': '20170603',
459 'url': 'https://www.tvnow.de/filme/horst-schlaemmer-isch-kandidiere-17777',
460 'only_matching': True, # DRM protected
463 def _real_extract(self, url):
464 mobj = self._match_valid_url(url)
465 display_id = mobj.group('title')
467 webpage = self._download_webpage(url, display_id, fatal=False)
469 raise ExtractorError('Cannot download "%s"' % url, expected=True)
471 json_text = get_element_by_id('now-web-state', webpage)
473 raise ExtractorError('Cannot read video data', expected=True)
475 json_data = self._parse_json(
478 transform_source=lambda x: x.replace('&q;', '"'),
481 raise ExtractorError('Cannot read video data', expected=True)
484 (key for key in json_data.keys() if 'module/player' in key),
487 (key for key in json_data.keys() if 'page/filme' in key),
492 lambda x: x[player_key]['body']['id'],
493 lambda x: x[page_key]['body']['modules'][0]['id'],
494 lambda x: x[page_key]['body']['modules'][1]['id']],
497 raise ExtractorError('Cannot extract movie ID', expected=True)
499 info = self._call_api('player/%d' % movie_id, display_id)
500 return self._extract_video(info, url, display_id)
504 class TVNowListBaseIE(TVNowNewBaseIE
):
505 _SHOW_VALID_URL
= r
'''(?x)
508 (?:www\.)?tvnow\.(?:de|at|ch)/(?:shows|serien)/
509 [^/?#&]+-(?P<show_id>\d+)
514 def suitable(cls
, url
):
515 return (False if TVNowNewIE
.suitable(url
)
516 else super(TVNowListBaseIE
, cls
).suitable(url
))
518 def _extract_items(self
, url
, show_id
, list_id
, query
):
519 items
= self
._call
_api
(
520 'teaserrow/format/episode/' + show_id
, list_id
,
521 query
=query
)['items']
525 if not isinstance(item
, dict):
527 item_url
= urljoin(url
, item
.get('url'))
530 video_id
= str_or_none(item
.get('id') or item
.get('videoId'))
531 item_title
= item
.get('subheadline') or item
.get('text')
532 entries
.append(self
.url_result(
533 item_url
, ie
=TVNowNewIE
.ie_key(), video_id
=video_id
,
534 video_title
=item_title
))
536 return self
.playlist_result(entries
, '%s/%s' % (show_id
, list_id
))
539 class TVNowSeasonIE(TVNowListBaseIE
):
540 _VALID_URL
= r
'%s/staffel-(?P<id>\d+)' % TVNowListBaseIE
._SHOW
_VALID
_URL
542 'url': 'https://www.tvnow.de/serien/alarm-fuer-cobra-11-die-autobahnpolizei-1815/staffel-13',
546 'playlist_mincount': 22,
549 def _real_extract(self
, url
):
550 _
, show_id
, season_id
= self
._match
_valid
_url
(url
).groups()
551 return self
._extract
_items
(
552 url
, show_id
, season_id
, {'season': season_id}
)
555 class TVNowAnnualIE(TVNowListBaseIE
):
556 _VALID_URL
= r
'%s/(?P<year>\d{4})-(?P<month>\d{2})' % TVNowListBaseIE
._SHOW
_VALID
_URL
558 'url': 'https://www.tvnow.de/shows/grip-das-motormagazin-1669/2017-05',
560 'id': '1669/2017-05',
562 'playlist_mincount': 2,
565 def _real_extract(self
, url
):
566 _
, show_id
, year
, month
= self
._match
_valid
_url
(url
).groups()
567 return self
._extract
_items
(
568 url
, show_id
, '%s-%s' % (year
, month
), {
574 class TVNowShowIE(TVNowListBaseIE
):
575 _VALID_URL
= TVNowListBaseIE
._SHOW
_VALID
_URL
577 # annual navigationType
578 'url': 'https://www.tvnow.de/shows/grip-das-motormagazin-1669',
582 'playlist_mincount': 73,
584 # season navigationType
585 'url': 'https://www.tvnow.de/shows/armes-deutschland-11471',
589 'playlist_mincount': 3,
593 def suitable(cls
, url
):
594 return (False if TVNowNewIE
.suitable(url
) or TVNowSeasonIE
.suitable(url
) or TVNowAnnualIE
.suitable(url
)
595 else super(TVNowShowIE
, cls
).suitable(url
))
597 def _real_extract(self
, url
):
598 base_url
, show_id
= self
._match
_valid
_url
(url
).groups()
600 result
= self
._call
_api
(
601 'teaserrow/format/navigation/' + show_id
, show_id
)
603 items
= result
['items']
606 navigation
= result
.get('navigationType')
607 if navigation
== 'annual':
609 if not isinstance(item
, dict):
611 year
= int_or_none(item
.get('year'))
614 months
= item
.get('months')
615 if not isinstance(months
, list):
617 for month_dict
in months
:
618 if not isinstance(month_dict
, dict) or not month_dict
:
620 month_number
= int_or_none(list(month_dict
.keys())[0])
621 if month_number
is None:
623 entries
.append(self
.url_result(
624 '%s/%04d-%02d' % (base_url
, year
, month_number
),
625 ie
=TVNowAnnualIE
.ie_key()))
626 elif navigation
== 'season':
628 if not isinstance(item
, dict):
630 season_number
= int_or_none(item
.get('season'))
631 if season_number
is None:
633 entries
.append(self
.url_result(
634 '%s/staffel-%d' % (base_url
, season_number
),
635 ie
=TVNowSeasonIE
.ie_key()))
637 raise ExtractorError('Unknown navigationType')
639 return self
.playlist_result(entries
, show_id
)