kumi/yt-dlp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[ie/ARDBetaMediathek] Fix series extraction (#8687)

Browse source 
Closes #7666
Authored by: lstrojny
This commit is contained in:
Lars Strojny 2023-12-24 23:38:21 +01:00 committed by GitHub
parent 00cdda4f6f
commit 1f8bd8eba8
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 27 additions and 23 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

50
yt_dlp/extractor/ard.py
View file

@ -292,7 +292,7 @@ class ARDIE(InfoExtractor):
_TESTS = [{ _TESTS = [{
# available till 7.12.2023 # available till 7.12.2023
'url': 'https://www.daserste.de/information/talk/maischberger/videos/maischberger-video-424.html', 'url': 'https://www.daserste.de/information/talk/maischberger/videos/maischberger-video-424.html',
'md5': 'a438f671e87a7eba04000336a119ccc4', 'md5': '94812e6438488fb923c361a44469614b',
'info_dict': { 'info_dict': {
'id': 'maischberger-video-424', 'id': 'maischberger-video-424',
'display_id': 'maischberger-video-424', 'display_id': 'maischberger-video-424',
@ -403,26 +403,25 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
_VALID_URL = r'''(?x)https:// _VALID_URL = r'''(?x)https://
(?:(?:beta|www)\.)?ardmediathek\.de/ (?:(?:beta|www)\.)?ardmediathek\.de/
(?:(?P<client>[^/]+)/)? (?:(?P<client>[^/]+)/)?
(?:player|live|video|(?P<playlist>sendung|sammlung))/ (?:player|live|video|(?P<playlist>sendung|serie|sammlung))/
(?:(?P<display_id>(?(playlist)[^?#]+?|[^?#]+))/)? (?:(?P<display_id>(?(playlist)[^?#]+?|[^?#]+))/)?
(?P<id>(?(playlist)|Y3JpZDovL)[a-zA-Z0-9]+) (?P<id>(?(playlist)|Y3JpZDovL)[a-zA-Z0-9]+)
(?(playlist)/(?P<season>\d+)?/?(?:[?#]|$))''' (?(playlist)/(?P<season>\d+)?/?(?:[?#]|$))'''
_TESTS = [{ _TESTS = [{
'url': 'https://www.ardmediathek.de/video/filme-im-mdr/wolfsland-die-traurigen-schwestern/mdr-fernsehen/Y3JpZDovL21kci5kZS9iZWl0cmFnL2Ntcy8xZGY0ZGJmZS00ZWQwLTRmMGItYjhhYy0wOGQ4ZmYxNjVhZDI', 'url': 'https://www.ardmediathek.de/video/filme-im-mdr/liebe-auf-vier-pfoten/mdr-fernsehen/Y3JpZDovL21kci5kZS9zZW5kdW5nLzI4MjA0MC80MjIwOTEtNDAyNTM0',
'md5': '3fd5fead7a370a819341129c8d713136', 'md5': 'b6e8ab03f2bcc6e1f9e6cef25fcc03c4',
'info_dict': { 'info_dict': {
'display_id': 'filme-im-mdr/wolfsland-die-traurigen-schwestern/mdr-fernsehen', 'display_id': 'filme-im-mdr/liebe-auf-vier-pfoten/mdr-fernsehen',
'id': '12172961', 'id': '12939099',
'title': 'Wolfsland - Die traurigen Schwestern', 'title': 'Liebe auf vier Pfoten',
'description': r're:^Als der Polizeiobermeister Raaben', 'description': r're:^Claudia Schmitt, Anwältin in Salzburg',
'duration': 5241, 'duration': 5222,
'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:efa186f7b0054957', 'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:aee7cbf8f06de976?w=960&ch=ae4d0f2ee47d8b9b',
'timestamp': 1670710500, 'timestamp': 1701343800,
'upload_date': '20221210', 'upload_date': '20231130',
'ext': 'mp4', 'ext': 'mp4',
'age_limit': 12, 'episode': 'Liebe auf vier Pfoten',
'episode': 'Wolfsland - Die traurigen Schwestern',
'series': 'Filme im MDR' 'series': 'Filme im MDR'
}, },
}, { }, {
@ -454,7 +453,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
'duration': 915, 'duration': 915,
'episode': 'tagesschau, 20:00 Uhr', 'episode': 'tagesschau, 20:00 Uhr',
'series': 'tagesschau', 'series': 'tagesschau',
'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:fbb21142783b0a49', 'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:fbb21142783b0a49?w=960&ch=ee69108ae344f678',
}, },
}, { }, {
'url': 'https://beta.ardmediathek.de/ard/video/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE', 'url': 'https://beta.ardmediathek.de/ard/video/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE',
@ -475,6 +474,10 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
# playlist of type 'sendung' # playlist of type 'sendung'
'url': 'https://www.ardmediathek.de/ard/sendung/doctor-who/Y3JpZDovL3dkci5kZS9vbmUvZG9jdG9yIHdobw/', 'url': 'https://www.ardmediathek.de/ard/sendung/doctor-who/Y3JpZDovL3dkci5kZS9vbmUvZG9jdG9yIHdobw/',
'only_matching': True, 'only_matching': True,
}, {
# playlist of type 'serie'
'url': 'https://www.ardmediathek.de/serie/nachtstreife/staffel-1/Y3JpZDovL3N3ci5kZS9zZGIvc3RJZC8xMjQy/1',
'only_matching': True,
}, { }, {
# playlist of type 'sammlung' # playlist of type 'sammlung'
'url': 'https://www.ardmediathek.de/ard/sammlung/team-muenster/5JpTzLSbWUAK8184IOvEir/', 'url': 'https://www.ardmediathek.de/ard/sammlung/team-muenster/5JpTzLSbWUAK8184IOvEir/',
@ -487,10 +490,11 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
'only_matching': True, 'only_matching': True,
}] }]
def _ARD_load_playlist_snipped(self, playlist_id, display_id, client, mode, pageNumber): def _ARD_load_playlist_snippet(self, playlist_id, display_id, client, mode, page_number):
""" Query the ARD server for playlist information """ Query the ARD server for playlist information
and returns the data in "raw" format """ and returns the data in "raw" format """
if mode == 'sendung': assert mode in ('sendung', 'serie', 'sammlung')
if mode in ('sendung', 'serie'):
graphQL = json.dumps({ graphQL = json.dumps({
'query': '''{ 'query': '''{
showPage( showPage(
@ -507,7 +511,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
links { target { id href title } } links { target { id href title } }
type type
} }
}}''' % (client, playlist_id, pageNumber), }}''' % (client, playlist_id, page_number),
}).encode() }).encode()
else: # mode == 'sammlung' else: # mode == 'sammlung'
graphQL = json.dumps({ graphQL = json.dumps({
@ -528,7 +532,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
type type
} }
} }
}}''' % (client, playlist_id, pageNumber), }}''' % (client, playlist_id, page_number),
}).encode() }).encode()
# Ressources for ARD graphQL debugging: # Ressources for ARD graphQL debugging:
# https://api-test.ardmediathek.de/public-gateway # https://api-test.ardmediathek.de/public-gateway
@ -538,7 +542,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
data=graphQL, data=graphQL,
headers={'Content-Type': 'application/json'})['data'] headers={'Content-Type': 'application/json'})['data']
# align the structure of the returned data: # align the structure of the returned data:
if mode == 'sendung': if mode in ('sendung', 'serie'):
show_page = show_page['showPage'] show_page = show_page['showPage']
else: # mode == 'sammlung' else: # mode == 'sammlung'
show_page = show_page['morePage']['widget'] show_page = show_page['morePage']['widget']
@ -546,12 +550,12 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
def _ARD_extract_playlist(self, url, playlist_id, display_id, client, mode): def _ARD_extract_playlist(self, url, playlist_id, display_id, client, mode):
""" Collects all playlist entries and returns them as info dict. """ Collects all playlist entries and returns them as info dict.
Supports playlists of mode 'sendung' and 'sammlung', and also nested Supports playlists of mode 'sendung', 'serie', and 'sammlung',
playlists. """ as well as nested playlists. """
entries = [] entries = []
pageNumber = 0 pageNumber = 0
while True: # iterate by pageNumber while True: # iterate by pageNumber
show_page = self._ARD_load_playlist_snipped( show_page = self._ARD_load_playlist_snippet(
playlist_id, display_id, client, mode, pageNumber) playlist_id, display_id, client, mode, pageNumber)
for teaser in show_page['teasers']: # process playlist items for teaser in show_page['teasers']: # process playlist items
if '/compilation/' in teaser['links']['target']['href']: if '/compilation/' in teaser['links']['target']['href']: