[yt-dlp.git] / yt_dlp / extractor / amazonminitv.py

import json

from .common import InfoExtractor
from ..utils import ExtractorError, int_or_none, traverse_obj, try_get


class AmazonMiniTVBaseIE(InfoExtractor):
    def _real_initialize(self):
        self._download_webpage(
            'https://www.amazon.in/minitv', None,
            note='Fetching guest session cookies')
        AmazonMiniTVBaseIE.session_id = self._get_cookies('https://www.amazon.in')['session-id'].value

    def _call_api(self, asin, data=None, note=None):
        device = {'clientId': 'ATVIN', 'deviceLocale': 'en_GB'}
        if data:
            data['variables'].update({
                'contentType': 'VOD',
                'sessionIdToken': self.session_id,
                **device,
            })

        resp = self._download_json(
            f'https://www.amazon.in/minitv/api/web/{"graphql" if data else "prs"}',
            asin, note=note, headers={'Content-Type': 'application/json'},
            data=json.dumps(data).encode() if data else None,
            query=None if data else {
                'deviceType': 'A1WMMUXPCUJL4N',
                'contentId': asin,
                **device,
            })

        if resp.get('errors'):
            raise ExtractorError(f'MiniTV said: {resp["errors"][0]["message"]}')
        elif not data:
            return resp
        return resp['data'][data['operationName']]


class AmazonMiniTVIE(AmazonMiniTVBaseIE):
    _VALID_URL = r'(?:https?://(?:www\.)?amazon\.in/minitv/tp/|amazonminitv:(?:amzn1\.dv\.gti\.)?)(?P<id>[a-f0-9-]+)'
    _TESTS = [{
        'url': 'https://www.amazon.in/minitv/tp/75fe3a75-b8fe-4499-8100-5c9424344840?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
        'info_dict': {
            'id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
            'ext': 'mp4',
            'title': 'May I Kiss You?',
            'language': 'Hindi',
            'thumbnail': r're:^https?://.*\.jpg$',
            'description': 'md5:a549bfc747973e04feb707833474e59d',
            'release_timestamp': 1644710400,
            'release_date': '20220213',
            'duration': 846,
            'chapters': 'count:2',
            'series': 'Couple Goals',
            'series_id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
            'season': 'Season 3',
            'season_number': 3,
            'season_id': 'amzn1.dv.gti.20331016-d9b9-4968-b991-c89fa4927a36',
            'episode': 'May I Kiss You?',
            'episode_number': 2,
            'episode_id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
        },
    }, {
        'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
        'info_dict': {
            'id': 'amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
            'ext': 'mp4',
            'title': 'Jahaan',
            'language': 'Hindi',
            'thumbnail': r're:^https?://.*\.jpg',
            'description': 'md5:05eb765a77bf703f322f120ec6867339',
            'release_timestamp': 1647475200,
            'release_date': '20220317',
            'duration': 783,
            'chapters': [],
        },
    }, {
        'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab',
        'only_matching': True,
    }, {
        'url': 'amazonminitv:amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
        'only_matching': True,
    }, {
        'url': 'amazonminitv:280d2564-584f-452f-9c98-7baf906e01ab',
        '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):
        asin = f'amzn1.dv.gti.{self._match_id(url)}'
        prs = self._call_api(asin, note='Downloading playback info')

        formats, subtitles = [], {}
        for type_, asset in prs['playbackAssets'].items():
            if not traverse_obj(asset, 'manifestUrl'):
                continue
            if type_ == 'hls':
                m3u8_fmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
                    asset['manifestUrl'], asin, ext='mp4', entry_protocol='m3u8_native',
                    m3u8_id=type_, fatal=False)
                formats.extend(m3u8_fmts)
                subtitles = self._merge_subtitles(subtitles, m3u8_subs)
            elif type_ == 'dash':
                mpd_fmts, mpd_subs = self._extract_mpd_formats_and_subtitles(
                    asset['manifestUrl'], asin, mpd_id=type_, fatal=False)
                formats.extend(mpd_fmts)
                subtitles = self._merge_subtitles(subtitles, mpd_subs)
            else:
                self.report_warning(f'Unknown asset type: {type_}')

        title_info = self._call_api(
            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)
        is_episode = title_info.get('vodType') == 'EPISODE'

        return {
            'id': asin,
            'title': title_info.get('name'),
            'formats': formats,
            'subtitles': subtitles,
            'language': traverse_obj(title_info, ('audioTracks', 0)),
            'thumbnails': [{
                'id': type_,
                'url': url,
            } for type_, url in (title_info.get('images') or {}).items()],
            'description': traverse_obj(title_info, ('description', 'synopsis')),
            'release_timestamp': int_or_none(try_get(title_info, lambda x: x['publicReleaseDateUTC'] / 1000)),
            'duration': traverse_obj(title_info, ('description', 'contentLengthInSeconds')),
            'chapters': [{
                'start_time': credits_time,
                'title': 'End Credits',
            }] if credits_time else [],
            'series': title_info.get('seriesName'),
            'series_id': title_info.get('seriesId'),
            'season_number': title_info.get('seasonNumber'),
            'season_id': title_info.get('seasonId'),
            'episode': title_info.get('name') if is_episode else None,
            'episode_number': title_info.get('episodeNumber'),
            'episode_id': asin if is_episode else None,
        }


class AmazonMiniTVSeasonIE(AmazonMiniTVBaseIE):
    IE_NAME = 'amazonminitv:season'
    _VALID_URL = r'amazonminitv:season:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
    IE_DESC = 'Amazon MiniTV Season, "minitv:season:" prefix'
    _TESTS = [{
        'url': 'amazonminitv:season:amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
        'playlist_mincount': 6,
        'info_dict': {
            'id': 'amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
        },
    }, {
        'url': 'amazonminitv:season:0aa996eb-6a1b-4886-a342-387fbd2f1db0',
        'only_matching': True,
    }]

    _GRAPHQL_QUERY = '''
query getEpisodes($sessionIdToken: String!, $clientId: String, $episodeOrSeasonId: ID!, $deviceLocale: String) {
  getEpisodes(
    applicationContextInput: {sessionIdToken: $sessionIdToken, deviceLocale: $deviceLocale, clientId: $clientId}
    episodeOrSeasonId: $episodeOrSeasonId
  ) {
    episodes {
      ... on Episode {
        contentId
        name
        images
        seriesName
        seasonId
        seriesId
        seasonNumber
        episodeNumber
        description {
          synopsis
          contentLengthInSeconds
        }
        publicReleaseDateUTC
      }
    }
  }
}
'''

    def _entries(self, asin):
        season_info = self._call_api(
            asin, note='Downloading season info', data={
                'operationName': 'getEpisodes',
                'variables': {'episodeOrSeasonId': asin},
                'query': self._GRAPHQL_QUERY,
            })

        for episode in season_info['episodes']:
            yield self.url_result(
                f'amazonminitv:{episode["contentId"]}', AmazonMiniTVIE, episode['contentId'])

    def _real_extract(self, url):
        asin = f'amzn1.dv.gti.{self._match_id(url)}'
        return self.playlist_result(self._entries(asin), asin)


class AmazonMiniTVSeriesIE(AmazonMiniTVBaseIE):
    IE_NAME = 'amazonminitv:series'
    _VALID_URL = r'amazonminitv:series:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
    IE_DESC = 'Amazon MiniTV Series, "minitv:series:" prefix'
    _TESTS = [{
        'url': 'amazonminitv:series:amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
        'playlist_mincount': 3,
        'info_dict': {
            'id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
        },
    }, {
        'url': 'amazonminitv:series:56521d46-b040-4fd5-872e-3e70476a04b0',
        'only_matching': True,
    }]

    _GRAPHQL_QUERY = '''
query getSeasons($sessionIdToken: String!, $deviceLocale: String, $episodeOrSeasonOrSeriesId: ID!, $clientId: String) {
  getSeasons(
    applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
    episodeOrSeasonOrSeriesId: $episodeOrSeasonOrSeriesId
  ) {
    seasons {
      seasonId
    }
  }
}
'''

    def _entries(self, asin):
        season_info = self._call_api(
            asin, note='Downloading series info', data={
                'operationName': 'getSeasons',
                'variables': {'episodeOrSeasonOrSeriesId': asin},
                'query': self._GRAPHQL_QUERY,
            })

        for season in season_info['seasons']:
            yield self.url_result(f'amazonminitv:season:{season["seasonId"]}', AmazonMiniTVSeasonIE, season['seasonId'])

    def _real_extract(self, url):
        asin = f'amzn1.dv.gti.{self._match_id(url)}'
        return self.playlist_result(self._entries(asin), asin)
Commit	Line	Data
48652590	1	import json
	2
	3	from .common import InfoExtractor
	4	from ..utils import ExtractorError, int_or_none, traverse_obj, try_get
	5
	6
a9d069f5	7	class AmazonMiniTVBaseIE(InfoExtractor):
	8	def _real_initialize(self):
	9	self._download_webpage(
	10	'https://www.amazon.in/minitv', None,
	11	note='Fetching guest session cookies')
	12	AmazonMiniTVBaseIE.session_id = self._get_cookies('https://www.amazon.in')['session-id'].value
	13
	14	def _call_api(self, asin, data=None, note=None):
	15	device = {'clientId': 'ATVIN', 'deviceLocale': 'en_GB'}
	16	if data:
	17	data['variables'].update({
	18	'contentType': 'VOD',
	19	'sessionIdToken': self.session_id,
	20	**device,
	21	})
	22
	23	resp = self._download_json(
	24	f'https://www.amazon.in/minitv/api/web/{"graphql" if data else "prs"}',
	25	asin, note=note, headers={'Content-Type': 'application/json'},
	26	data=json.dumps(data).encode() if data else None,
	27	query=None if data else {
	28	'deviceType': 'A1WMMUXPCUJL4N',
	29	'contentId': asin,
	30	**device,
	31	})
	32
	33	if resp.get('errors'):
	34	raise ExtractorError(f'MiniTV said: {resp["errors"][0]["message"]}')
	35	elif not data:
	36	return resp
	37	return resp['data'][data['operationName']]
	38
	39
	40	class AmazonMiniTVIE(AmazonMiniTVBaseIE):
48652590	41	_VALID_URL = r'(?:https?://(?:www\.)?amazon\.in/minitv/tp/\|amazonminitv:(?:amzn1\.dv\.gti\.)?)(?P<id>[a-f0-9-]+)'
48652590	42	_TESTS = [{
48652590	43	'url': 'https://www.amazon.in/minitv/tp/75fe3a75-b8fe-4499-8100-5c9424344840?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
48652590	44	'info_dict': {
	45	'id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
	46	'ext': 'mp4',
	47	'title': 'May I Kiss You?',
	48	'language': 'Hindi',
	49	'thumbnail': r're:^https?://.*\.jpg$',
	50	'description': 'md5:a549bfc747973e04feb707833474e59d',
	51	'release_timestamp': 1644710400,
	52	'release_date': '20220213',
	53	'duration': 846,
a9d069f5	54	'chapters': 'count:2',
48652590	55	'series': 'Couple Goals',
	56	'series_id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	57	'season': 'Season 3',
	58	'season_number': 3,
	59	'season_id': 'amzn1.dv.gti.20331016-d9b9-4968-b991-c89fa4927a36',
	60	'episode': 'May I Kiss You?',
	61	'episode_number': 2,
	62	'episode_id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
	63	},
	64	}, {
	65	'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
48652590	66	'info_dict': {
	67	'id': 'amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
	68	'ext': 'mp4',
	69	'title': 'Jahaan',
	70	'language': 'Hindi',
	71	'thumbnail': r're:^https?://.*\.jpg',
	72	'description': 'md5:05eb765a77bf703f322f120ec6867339',
	73	'release_timestamp': 1647475200,
	74	'release_date': '20220317',
	75	'duration': 783,
	76	'chapters': [],
	77	},
	78	}, {
	79	'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab',
	80	'only_matching': True,
	81	}, {
	82	'url': 'amazonminitv:amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
	83	'only_matching': True,
	84	}, {
	85	'url': 'amazonminitv:280d2564-584f-452f-9c98-7baf906e01ab',
	86	'only_matching': True,
	87	}]
a9d069f5	88
48652590	89	_GRAPHQL_QUERY_CONTENT = '''
	90	query content($sessionIdToken: String!, $deviceLocale: String, $contentId: ID!, $contentType: ContentType!, $clientId: String) {
	91	content(
	92	applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
	93	contentId: $contentId
	94	contentType: $contentType
	95	) {
	96	contentId
	97	name
	98	... on Episode {
	99	contentId
	100	vodType
	101	name
	102	images
	103	description {
	104	synopsis
	105	contentLengthInSeconds
	106	}
	107	publicReleaseDateUTC
	108	audioTracks
	109	seasonId
	110	seriesId
	111	seriesName
	112	seasonNumber
	113	episodeNumber
	114	timecode {
	115	endCreditsTime
	116	}
	117	}
	118	... on MovieContent {
	119	contentId
	120	vodType
	121	name
	122	description {
	123	synopsis
	124	contentLengthInSeconds
	125	}
	126	images
	127	publicReleaseDateUTC
	128	audioTracks
	129	}
	130	}
	131	}'''
	132
48652590	133	def _real_extract(self, url):
48652590	134	asin = f'amzn1.dv.gti.{self._match_id(url)}'
48652590	135	prs = self._call_api(asin, note='Downloading playback info')
48652590	136
a9d069f5	137	formats, subtitles = [], {}
48652590	138	for type_, asset in prs['playbackAssets'].items():
a9d069f5	139	if not traverse_obj(asset, 'manifestUrl'):
48652590	140	continue
	141	if type_ == 'hls':
	142	m3u8_fmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
	143	asset['manifestUrl'], asin, ext='mp4', entry_protocol='m3u8_native',
	144	m3u8_id=type_, fatal=False)
	145	formats.extend(m3u8_fmts)
	146	subtitles = self._merge_subtitles(subtitles, m3u8_subs)
	147	elif type_ == 'dash':
	148	mpd_fmts, mpd_subs = self._extract_mpd_formats_and_subtitles(
	149	asset['manifestUrl'], asin, mpd_id=type_, fatal=False)
	150	formats.extend(mpd_fmts)
	151	subtitles = self._merge_subtitles(subtitles, mpd_subs)
a9d069f5	152	else:
a9d069f5	153	self.report_warning(f'Unknown asset type: {type_}')
48652590	154
a9d069f5	155	title_info = self._call_api(
	156	asin, note='Downloading title info', data={
	157	'operationName': 'content',
	158	'variables': {'contentId': asin},
	159	'query': self._GRAPHQL_QUERY_CONTENT,
	160	})
48652590	161	credits_time = try_get(title_info, lambda x: x['timecode']['endCreditsTime'] / 1000)
48652590	162	is_episode = title_info.get('vodType') == 'EPISODE'
	163
	164	return {
	165	'id': asin,
	166	'title': title_info.get('name'),
	167	'formats': formats,
	168	'subtitles': subtitles,
	169	'language': traverse_obj(title_info, ('audioTracks', 0)),
	170	'thumbnails': [{
	171	'id': type_,
	172	'url': url,
	173	} for type_, url in (title_info.get('images') or {}).items()],
	174	'description': traverse_obj(title_info, ('description', 'synopsis')),
	175	'release_timestamp': int_or_none(try_get(title_info, lambda x: x['publicReleaseDateUTC'] / 1000)),
a9d069f5	176	'duration': traverse_obj(title_info, ('description', 'contentLengthInSeconds')),
	177	'chapters': [{
	178	'start_time': credits_time,
	179	'title': 'End Credits',
	180	}] if credits_time else [],
48652590	181	'series': title_info.get('seriesName'),
	182	'series_id': title_info.get('seriesId'),
	183	'season_number': title_info.get('seasonNumber'),
	184	'season_id': title_info.get('seasonId'),
	185	'episode': title_info.get('name') if is_episode else None,
	186	'episode_number': title_info.get('episodeNumber'),
	187	'episode_id': asin if is_episode else None,
	188	}
	189
	190
a9d069f5	191	class AmazonMiniTVSeasonIE(AmazonMiniTVBaseIE):
48652590	192	IE_NAME = 'amazonminitv:season'
48652590	193	_VALID_URL = r'amazonminitv:season:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
a5387729	194	IE_DESC = 'Amazon MiniTV Season, "minitv:season:" prefix'
48652590	195	_TESTS = [{
	196	'url': 'amazonminitv:season:amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	197	'playlist_mincount': 6,
	198	'info_dict': {
	199	'id': 'amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	200	},
	201	}, {
	202	'url': 'amazonminitv:season:0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	203	'only_matching': True,
	204	}]
a9d069f5	205
48652590	206	_GRAPHQL_QUERY = '''
	207	query getEpisodes($sessionIdToken: String!, $clientId: String, $episodeOrSeasonId: ID!, $deviceLocale: String) {
	208	getEpisodes(
	209	applicationContextInput: {sessionIdToken: $sessionIdToken, deviceLocale: $deviceLocale, clientId: $clientId}
	210	episodeOrSeasonId: $episodeOrSeasonId
	211	) {
	212	episodes {
	213	... on Episode {
	214	contentId
	215	name
	216	images
	217	seriesName
	218	seasonId
	219	seriesId
	220	seasonNumber
	221	episodeNumber
	222	description {
	223	synopsis
	224	contentLengthInSeconds
	225	}
	226	publicReleaseDateUTC
	227	}
	228	}
	229	}
	230	}
	231	'''
	232
	233	def _entries(self, asin):
	234	season_info = self._call_api(
a9d069f5	235	asin, note='Downloading season info', data={
48652590	236	'operationName': 'getEpisodes',
a9d069f5	237	'variables': {'episodeOrSeasonId': asin},
48652590	238	'query': self._GRAPHQL_QUERY,
a9d069f5	239	})
48652590	240
48652590	241	for episode in season_info['episodes']:
a9d069f5	242	yield self.url_result(
a9d069f5	243	f'amazonminitv:{episode["contentId"]}', AmazonMiniTVIE, episode['contentId'])
48652590	244
	245	def _real_extract(self, url):
	246	asin = f'amzn1.dv.gti.{self._match_id(url)}'
a9d069f5	247	return self.playlist_result(self._entries(asin), asin)
48652590	248
48652590	249
a9d069f5	250	class AmazonMiniTVSeriesIE(AmazonMiniTVBaseIE):
48652590	251	IE_NAME = 'amazonminitv:series'
48652590	252	_VALID_URL = r'amazonminitv:series:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
a5387729	253	IE_DESC = 'Amazon MiniTV Series, "minitv:series:" prefix'
48652590	254	_TESTS = [{
	255	'url': 'amazonminitv:series:amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	256	'playlist_mincount': 3,
	257	'info_dict': {
	258	'id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	259	},
	260	}, {
	261	'url': 'amazonminitv:series:56521d46-b040-4fd5-872e-3e70476a04b0',
	262	'only_matching': True,
	263	}]
a9d069f5	264
48652590	265	_GRAPHQL_QUERY = '''
	266	query getSeasons($sessionIdToken: String!, $deviceLocale: String, $episodeOrSeasonOrSeriesId: ID!, $clientId: String) {
	267	getSeasons(
	268	applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
	269	episodeOrSeasonOrSeriesId: $episodeOrSeasonOrSeriesId
	270	) {
	271	seasons {
	272	seasonId
	273	}
	274	}
	275	}
	276	'''
	277
	278	def _entries(self, asin):
	279	season_info = self._call_api(
a9d069f5	280	asin, note='Downloading series info', data={
48652590	281	'operationName': 'getSeasons',
a9d069f5	282	'variables': {'episodeOrSeasonOrSeriesId': asin},
48652590	283	'query': self._GRAPHQL_QUERY,
a9d069f5	284	})
48652590	285
	286	for season in season_info['seasons']:
	287	yield self.url_result(f'amazonminitv:season:{season["seasonId"]}', AmazonMiniTVSeasonIE, season['seasonId'])
	288
	289	def _real_extract(self, url):
	290	asin = f'amzn1.dv.gti.{self._match_id(url)}'
a9d069f5	291	return self.playlist_result(self._entries(asin), asin)