[ie/AmazonMiniTV] Fix extractor (#8103)

Closes #7817
Authored by: Aniruddh-J
This commit is contained in:
Aniruddh Joshi 2023-09-17 02:28:21 +05:30 committed by GitHub
parent 5336bf57a7
commit 9d376c4dae
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -37,7 +37,7 @@ class AmazonMiniTVBaseIE(InfoExtractor):
return resp['data'][data['operationName']] return resp['data'][data['operationName']]
class AmazonMiniTVIE(AmazonMiniTVBaseIE): class AmazonMiniTVIE(InfoExtractor):
_VALID_URL = r'(?:https?://(?:www\.)?amazon\.in/minitv/tp/|amazonminitv:(?:amzn1\.dv\.gti\.)?)(?P<id>[a-f0-9-]+)' _VALID_URL = r'(?:https?://(?:www\.)?amazon\.in/minitv/tp/|amazonminitv:(?:amzn1\.dv\.gti\.)?)(?P<id>[a-f0-9-]+)'
_TESTS = [{ _TESTS = [{
'url': 'https://www.amazon.in/minitv/tp/75fe3a75-b8fe-4499-8100-5c9424344840?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv', 'url': 'https://www.amazon.in/minitv/tp/75fe3a75-b8fe-4499-8100-5c9424344840?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
@ -86,56 +86,14 @@ class AmazonMiniTVIE(AmazonMiniTVBaseIE):
'only_matching': True, 'only_matching': True,
}] }]
_GRAPHQL_QUERY_CONTENT = '''
query content($sessionIdToken: String!, $deviceLocale: String, $contentId: ID!, $contentType: ContentType!, $clientId: String) {
content(
applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
contentId: $contentId
contentType: $contentType
) {
contentId
name
... on Episode {
contentId
vodType
name
images
description {
synopsis
contentLengthInSeconds
}
publicReleaseDateUTC
audioTracks
seasonId
seriesId
seriesName
seasonNumber
episodeNumber
timecode {
endCreditsTime
}
}
... on MovieContent {
contentId
vodType
name
description {
synopsis
contentLengthInSeconds
}
images
publicReleaseDateUTC
audioTracks
}
}
}'''
def _real_extract(self, url): def _real_extract(self, url):
asin = f'amzn1.dv.gti.{self._match_id(url)}' video_uuid = self._match_id(url)
prs = self._call_api(asin, note='Downloading playback info') asin = f'amzn1.dv.gti.{video_uuid}'
webpage = self._download_webpage(f'https://www.amazon.in/minitv/tp/{video_uuid}', asin)
data = self._search_nextjs_data(webpage, asin)['props']['pageProps']['ssrProps']
formats, subtitles = [], {} formats, subtitles = [], {}
for type_, asset in prs['playbackAssets'].items(): for type_, asset in traverse_obj(data, ('playbackData', 'playbackAssets', {dict.items}, ...)):
if not traverse_obj(asset, 'manifestUrl'): if not traverse_obj(asset, 'manifestUrl'):
continue continue
if type_ == 'hls': if type_ == 'hls':
@ -152,12 +110,7 @@ query content($sessionIdToken: String!, $deviceLocale: String, $contentId: ID!,
else: else:
self.report_warning(f'Unknown asset type: {type_}') self.report_warning(f'Unknown asset type: {type_}')
title_info = self._call_api( title_info = traverse_obj(data, ('contentData', {dict})) or {}
asin, note='Downloading title info', data={
'operationName': 'content',
'variables': {'contentId': asin},
'query': self._GRAPHQL_QUERY_CONTENT,
})
credits_time = try_get(title_info, lambda x: x['timecode']['endCreditsTime'] / 1000) credits_time = try_get(title_info, lambda x: x['timecode']['endCreditsTime'] / 1000)
is_episode = title_info.get('vodType') == 'EPISODE' is_episode = title_info.get('vodType') == 'EPISODE'
@ -192,6 +145,7 @@ class AmazonMiniTVSeasonIE(AmazonMiniTVBaseIE):
IE_NAME = 'amazonminitv:season' IE_NAME = 'amazonminitv:season'
_VALID_URL = r'amazonminitv:season:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)' _VALID_URL = r'amazonminitv:season:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
IE_DESC = 'Amazon MiniTV Season, "minitv:season:" prefix' IE_DESC = 'Amazon MiniTV Season, "minitv:season:" prefix'
_WORKING = False
_TESTS = [{ _TESTS = [{
'url': 'amazonminitv:season:amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0', 'url': 'amazonminitv:season:amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
'playlist_mincount': 6, 'playlist_mincount': 6,
@ -251,6 +205,7 @@ class AmazonMiniTVSeriesIE(AmazonMiniTVBaseIE):
IE_NAME = 'amazonminitv:series' IE_NAME = 'amazonminitv:series'
_VALID_URL = r'amazonminitv:series:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)' _VALID_URL = r'amazonminitv:series:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
IE_DESC = 'Amazon MiniTV Series, "minitv:series:" prefix' IE_DESC = 'Amazon MiniTV Series, "minitv:series:" prefix'
_WORKING = False
_TESTS = [{ _TESTS = [{
'url': 'amazonminitv:series:amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0', 'url': 'amazonminitv:series:amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
'playlist_mincount': 3, 'playlist_mincount': 3,