[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',
                'currentpageurl': '/',
                'currentplatform': 'dWeb'
            }, 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|png)$',
            '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|png)',
            '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"}',
538d3767	25	asin, note=note, headers={
	26	'Content-Type': 'application/json',
	27	'currentpageurl': '/',
	28	'currentplatform': 'dWeb'
	29	}, data=json.dumps(data).encode() if data else None,
a9d069f5	30	query=None if data else {
	31	'deviceType': 'A1WMMUXPCUJL4N',
	32	'contentId': asin,
	33	**device,
	34	})
	35
	36	if resp.get('errors'):
	37	raise ExtractorError(f'MiniTV said: {resp["errors"][0]["message"]}')
	38	elif not data:
	39	return resp
	40	return resp['data'][data['operationName']]
	41
	42
2da7bcca	43	class AmazonMiniTVIE(AmazonMiniTVBaseIE):
48652590	44	_VALID_URL = r'(?:https?://(?:www\.)?amazon\.in/minitv/tp/\|amazonminitv:(?:amzn1\.dv\.gti\.)?)(?P<id>[a-f0-9-]+)'
48652590	45	_TESTS = [{
48652590	46	'url': 'https://www.amazon.in/minitv/tp/75fe3a75-b8fe-4499-8100-5c9424344840?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
48652590	47	'info_dict': {
	48	'id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
	49	'ext': 'mp4',
	50	'title': 'May I Kiss You?',
	51	'language': 'Hindi',
538d3767	52	'thumbnail': r're:^https?://.*\.(?:jpg\|png)$',
48652590	53	'description': 'md5:a549bfc747973e04feb707833474e59d',
	54	'release_timestamp': 1644710400,
	55	'release_date': '20220213',
	56	'duration': 846,
a9d069f5	57	'chapters': 'count:2',
48652590	58	'series': 'Couple Goals',
	59	'series_id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	60	'season': 'Season 3',
	61	'season_number': 3,
	62	'season_id': 'amzn1.dv.gti.20331016-d9b9-4968-b991-c89fa4927a36',
	63	'episode': 'May I Kiss You?',
	64	'episode_number': 2,
	65	'episode_id': 'amzn1.dv.gti.75fe3a75-b8fe-4499-8100-5c9424344840',
	66	},
	67	}, {
	68	'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab?referrer=https%3A%2F%2Fwww.amazon.in%2Fminitv',
48652590	69	'info_dict': {
	70	'id': 'amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
	71	'ext': 'mp4',
	72	'title': 'Jahaan',
	73	'language': 'Hindi',
538d3767	74	'thumbnail': r're:^https?://.*\.(?:jpg\|png)',
48652590	75	'description': 'md5:05eb765a77bf703f322f120ec6867339',
	76	'release_timestamp': 1647475200,
	77	'release_date': '20220317',
	78	'duration': 783,
	79	'chapters': [],
	80	},
	81	}, {
	82	'url': 'https://www.amazon.in/minitv/tp/280d2564-584f-452f-9c98-7baf906e01ab',
	83	'only_matching': True,
	84	}, {
	85	'url': 'amazonminitv:amzn1.dv.gti.280d2564-584f-452f-9c98-7baf906e01ab',
	86	'only_matching': True,
	87	}, {
	88	'url': 'amazonminitv:280d2564-584f-452f-9c98-7baf906e01ab',
	89	'only_matching': True,
	90	}]
a9d069f5	91
2da7bcca	92	_GRAPHQL_QUERY_CONTENT = '''
	93	query content($sessionIdToken: String!, $deviceLocale: String, $contentId: ID!, $contentType: ContentType!, $clientId: String) {
	94	content(
	95	applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
	96	contentId: $contentId
	97	contentType: $contentType
	98	) {
	99	contentId
	100	name
	101	... on Episode {
	102	contentId
	103	vodType
	104	name
	105	images
	106	description {
	107	synopsis
	108	contentLengthInSeconds
	109	}
	110	publicReleaseDateUTC
	111	audioTracks
	112	seasonId
	113	seriesId
	114	seriesName
	115	seasonNumber
	116	episodeNumber
	117	timecode {
	118	endCreditsTime
	119	}
	120	}
	121	... on MovieContent {
	122	contentId
	123	vodType
	124	name
	125	description {
	126	synopsis
	127	contentLengthInSeconds
	128	}
	129	images
	130	publicReleaseDateUTC
	131	audioTracks
	132	}
	133	}
	134	}'''
	135
48652590	136	def _real_extract(self, url):
2da7bcca	137	asin = f'amzn1.dv.gti.{self._match_id(url)}'
2da7bcca	138	prs = self._call_api(asin, note='Downloading playback info')
48652590	139
a9d069f5	140	formats, subtitles = [], {}
2da7bcca	141	for type_, asset in prs['playbackAssets'].items():
a9d069f5	142	if not traverse_obj(asset, 'manifestUrl'):
48652590	143	continue
	144	if type_ == 'hls':
	145	m3u8_fmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
	146	asset['manifestUrl'], asin, ext='mp4', entry_protocol='m3u8_native',
	147	m3u8_id=type_, fatal=False)
	148	formats.extend(m3u8_fmts)
	149	subtitles = self._merge_subtitles(subtitles, m3u8_subs)
	150	elif type_ == 'dash':
	151	mpd_fmts, mpd_subs = self._extract_mpd_formats_and_subtitles(
	152	asset['manifestUrl'], asin, mpd_id=type_, fatal=False)
	153	formats.extend(mpd_fmts)
	154	subtitles = self._merge_subtitles(subtitles, mpd_subs)
a9d069f5	155	else:
a9d069f5	156	self.report_warning(f'Unknown asset type: {type_}')
48652590	157
2da7bcca	158	title_info = self._call_api(
	159	asin, note='Downloading title info', data={
	160	'operationName': 'content',
	161	'variables': {'contentId': asin},
	162	'query': self._GRAPHQL_QUERY_CONTENT,
	163	})
48652590	164	credits_time = try_get(title_info, lambda x: x['timecode']['endCreditsTime'] / 1000)
48652590	165	is_episode = title_info.get('vodType') == 'EPISODE'
	166
	167	return {
	168	'id': asin,
	169	'title': title_info.get('name'),
	170	'formats': formats,
	171	'subtitles': subtitles,
	172	'language': traverse_obj(title_info, ('audioTracks', 0)),
	173	'thumbnails': [{
	174	'id': type_,
	175	'url': url,
	176	} for type_, url in (title_info.get('images') or {}).items()],
	177	'description': traverse_obj(title_info, ('description', 'synopsis')),
	178	'release_timestamp': int_or_none(try_get(title_info, lambda x: x['publicReleaseDateUTC'] / 1000)),
a9d069f5	179	'duration': traverse_obj(title_info, ('description', 'contentLengthInSeconds')),
	180	'chapters': [{
	181	'start_time': credits_time,
	182	'title': 'End Credits',
	183	}] if credits_time else [],
48652590	184	'series': title_info.get('seriesName'),
	185	'series_id': title_info.get('seriesId'),
	186	'season_number': title_info.get('seasonNumber'),
	187	'season_id': title_info.get('seasonId'),
	188	'episode': title_info.get('name') if is_episode else None,
	189	'episode_number': title_info.get('episodeNumber'),
	190	'episode_id': asin if is_episode else None,
	191	}
	192
	193
a9d069f5	194	class AmazonMiniTVSeasonIE(AmazonMiniTVBaseIE):
48652590	195	IE_NAME = 'amazonminitv:season'
48652590	196	_VALID_URL = r'amazonminitv:season:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
a5387729	197	IE_DESC = 'Amazon MiniTV Season, "minitv:season:" prefix'
48652590	198	_TESTS = [{
	199	'url': 'amazonminitv:season:amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	200	'playlist_mincount': 6,
	201	'info_dict': {
	202	'id': 'amzn1.dv.gti.0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	203	},
	204	}, {
	205	'url': 'amazonminitv:season:0aa996eb-6a1b-4886-a342-387fbd2f1db0',
	206	'only_matching': True,
	207	}]
a9d069f5	208
48652590	209	_GRAPHQL_QUERY = '''
	210	query getEpisodes($sessionIdToken: String!, $clientId: String, $episodeOrSeasonId: ID!, $deviceLocale: String) {
	211	getEpisodes(
	212	applicationContextInput: {sessionIdToken: $sessionIdToken, deviceLocale: $deviceLocale, clientId: $clientId}
	213	episodeOrSeasonId: $episodeOrSeasonId
	214	) {
	215	episodes {
	216	... on Episode {
	217	contentId
	218	name
	219	images
	220	seriesName
	221	seasonId
	222	seriesId
	223	seasonNumber
	224	episodeNumber
	225	description {
	226	synopsis
	227	contentLengthInSeconds
	228	}
	229	publicReleaseDateUTC
	230	}
	231	}
	232	}
	233	}
	234	'''
	235
	236	def _entries(self, asin):
	237	season_info = self._call_api(
a9d069f5	238	asin, note='Downloading season info', data={
48652590	239	'operationName': 'getEpisodes',
a9d069f5	240	'variables': {'episodeOrSeasonId': asin},
48652590	241	'query': self._GRAPHQL_QUERY,
a9d069f5	242	})
48652590	243
48652590	244	for episode in season_info['episodes']:
a9d069f5	245	yield self.url_result(
a9d069f5	246	f'amazonminitv:{episode["contentId"]}', AmazonMiniTVIE, episode['contentId'])
48652590	247
	248	def _real_extract(self, url):
	249	asin = f'amzn1.dv.gti.{self._match_id(url)}'
a9d069f5	250	return self.playlist_result(self._entries(asin), asin)
48652590	251
48652590	252
a9d069f5	253	class AmazonMiniTVSeriesIE(AmazonMiniTVBaseIE):
48652590	254	IE_NAME = 'amazonminitv:series'
48652590	255	_VALID_URL = r'amazonminitv:series:(?:amzn1\.dv\.gti\.)?(?P<id>[a-f0-9-]+)'
a5387729	256	IE_DESC = 'Amazon MiniTV Series, "minitv:series:" prefix'
48652590	257	_TESTS = [{
	258	'url': 'amazonminitv:series:amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	259	'playlist_mincount': 3,
	260	'info_dict': {
	261	'id': 'amzn1.dv.gti.56521d46-b040-4fd5-872e-3e70476a04b0',
	262	},
	263	}, {
	264	'url': 'amazonminitv:series:56521d46-b040-4fd5-872e-3e70476a04b0',
	265	'only_matching': True,
	266	}]
a9d069f5	267
48652590	268	_GRAPHQL_QUERY = '''
	269	query getSeasons($sessionIdToken: String!, $deviceLocale: String, $episodeOrSeasonOrSeriesId: ID!, $clientId: String) {
	270	getSeasons(
	271	applicationContextInput: {deviceLocale: $deviceLocale, sessionIdToken: $sessionIdToken, clientId: $clientId}
	272	episodeOrSeasonOrSeriesId: $episodeOrSeasonOrSeriesId
	273	) {
	274	seasons {
	275	seasonId
	276	}
	277	}
	278	}
	279	'''
	280
	281	def _entries(self, asin):
	282	season_info = self._call_api(
a9d069f5	283	asin, note='Downloading series info', data={
48652590	284	'operationName': 'getSeasons',
a9d069f5	285	'variables': {'episodeOrSeasonOrSeriesId': asin},
48652590	286	'query': self._GRAPHQL_QUERY,
a9d069f5	287	})
48652590	288
	289	for season in season_info['seasons']:
	290	yield self.url_result(f'amazonminitv:season:{season["seasonId"]}', AmazonMiniTVSeasonIE, season['seasonId'])
	291
	292	def _real_extract(self, url):
	293	asin = f'amzn1.dv.gti.{self._match_id(url)}'
a9d069f5	294	return self.playlist_result(self._entries(asin), asin)