yt_dlp/extractor/medici.py

   1 import urllib.parse
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     filter_dict,
   6     parse_iso8601,
   7     traverse_obj,
   8     try_call,
   9     url_or_none,
  10 )
  11
  12
  13 class MediciIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:(?P<sub>www|edu)\.)?medici\.tv/[a-z]{2}/[\w.-]+/(?P<id>[^/?#&]+)'
  15     _TESTS = [{
  16         'url': 'https://www.medici.tv/en/operas/thomas-ades-the-exterminating-angel-calixto-bieito-opera-bastille-paris',
  17         'md5': 'd483f74e7a7a9eac0dbe152ab189050d',
  18         'info_dict': {
  19             'id': '8032',
  20             'ext': 'mp4',
  21             'title': 'Thomas Adès\'s The Exterminating Angel',
  22             'description': 'md5:708ae6350dadc604225b4a6e32482bab',
  23             'thumbnail': r're:https://.+/.+\.jpg',
  24             'upload_date': '20240304',
  25             'timestamp': 1709561766,
  26             'display_id': 'thomas-ades-the-exterminating-angel-calixto-bieito-opera-bastille-paris',
  27         },
  28         'expected_warnings': [r'preview'],
  29     }, {
  30         'url': 'https://edu.medici.tv/en/operas/wagner-lohengrin-paris-opera-kirill-serebrennikov-piotr-beczala-kwangchul-youn-johanni-van-oostrum',
  31         'md5': '4ef3f4079a6e1c617584463a9eb84f99',
  32         'info_dict': {
  33             'id': '7900',
  34             'ext': 'mp4',
  35             'title': 'Wagner\'s Lohengrin',
  36             'description': 'md5:a384a62937866101f86902f21752cd89',
  37             'thumbnail': r're:https://.+/.+\.jpg',
  38             'upload_date': '20231017',
  39             'timestamp': 1697554771,
  40             'display_id': 'wagner-lohengrin-paris-opera-kirill-serebrennikov-piotr-beczala-kwangchul-youn-johanni-van-oostrum',
  41         },
  42         'expected_warnings': [r'preview'],
  43     }, {
  44         'url': 'https://www.medici.tv/en/concerts/sergey-smbatyan-conducts-mansurian-chouchane-siranossian-mario-brunello',
  45         'md5': '9dd757e53b22b2511e85ea9ea60e4815',
  46         'info_dict': {
  47             'id': '5712',
  48             'ext': 'mp4',
  49             'title': 'Sergey Smbatyan conducts Tigran Mansurian — With Chouchane Siranossian and Mario Brunello',
  50             'thumbnail': r're:https://.+/.+\.jpg',
  51             'description': 'md5:9411fe44c874bb10e9af288c65816e41',
  52             'upload_date': '20200323',
  53             'timestamp': 1584975600,
  54             'display_id': 'sergey-smbatyan-conducts-mansurian-chouchane-siranossian-mario-brunello',
  55         },
  56         'expected_warnings': [r'preview'],
  57     }, {
  58         'url': 'https://www.medici.tv/en/ballets/carmen-ballet-choregraphie-de-jiri-bubenicek-teatro-dellopera-di-roma',
  59         'md5': '40f5e76cb701a97a6d7ba23b62c49990',
  60         'info_dict': {
  61             'id': '7857',
  62             'ext': 'mp4',
  63             'title': 'Carmen by Jiří Bubeníček after Roland Petit, music by Bizet, de Falla, Castelnuovo-Tedesco, and Bonolis',
  64             'thumbnail': r're:https://.+/.+\.jpg',
  65             'description': 'md5:0f15a15611ed748020c769873e10a8bb',
  66             'upload_date': '20240223',
  67             'timestamp': 1708707600,
  68             'display_id': 'carmen-ballet-choregraphie-de-jiri-bubenicek-teatro-dellopera-di-roma',
  69         },
  70         'expected_warnings': [r'preview'],
  71     }, {
  72         'url': 'https://www.medici.tv/en/documentaries/la-sonnambula-liege-2023-documentaire',
  73         'md5': '87ff198018ce79a34757ab0dd6f21080',
  74         'info_dict': {
  75             'id': '7513',
  76             'ext': 'mp4',
  77             'title': 'La Sonnambula',
  78             'thumbnail': r're:https://.+/.+\.jpg',
  79             'description': 'md5:0caf9109a860fd50cd018df062a67f34',
  80             'upload_date': '20231103',
  81             'timestamp': 1699010830,
  82             'display_id': 'la-sonnambula-liege-2023-documentaire',
  83         },
  84         'expected_warnings': [r'preview'],
  85     }, {
  86         'url': 'https://edu.medici.tv/en/masterclasses/yvonne-loriod-olivier-messiaen',
  87         'md5': 'fb5dcec46d76ad20fbdbaabb01da191d',
  88         'info_dict': {
  89             'id': '3024',
  90             'ext': 'mp4',
  91             'title': 'Olivier Messiaen and Yvonne Loriod, pianists and teachers',
  92             'thumbnail': r're:https://.+/.+\.jpg',
  93             'description': 'md5:aab948e2f7690214b5c28896c83f1fc1',
  94             'upload_date': '20150223',
  95             'timestamp': 1424706608,
  96             'display_id': 'yvonne-loriod-olivier-messiaen',
  97         },
  98         'skip': 'Requires authentication; preview starts in the middle',
  99     }, {
 100         'url': 'https://www.medici.tv/en/jazz/makaya-mccraven-la-rochelle',
 101         'md5': '4cc279a8b06609782747c8f50beea2b3',
 102         'info_dict': {
 103             'id': '7922',
 104             'ext': 'mp4',
 105             'title': 'NEW: Makaya McCraven in La Rochelle',
 106             'thumbnail': r're:https://.+/.+\.jpg',
 107             'description': 'md5:b5a8aaeb6993d8ccb18bde8abb8aa8d2',
 108             'upload_date': '20231228',
 109             'timestamp': 1703754863,
 110             'display_id': 'makaya-mccraven-la-rochelle',
 111         },
 112         'expected_warnings': [r'preview'],
 113     }]
 114
 115     def _real_extract(self, url):
 116         display_id, subdomain = self._match_valid_url(url).group('id', 'sub')
 117         self._request_webpage(url, display_id, 'Requesting CSRF token cookie')
 118
 119         subdomain = 'edu-' if subdomain == 'edu' else ''
 120         origin = f'https://{urllib.parse.urlparse(url).hostname}'
 121
 122         data = self._download_json(
 123             f'https://api.medici.tv/{subdomain}satie/edito/movie-file/{display_id}/', display_id,
 124             headers=filter_dict({
 125                 'Authorization': try_call(
 126                     lambda: urllib.parse.unquote(self._get_cookies(url)['auth._token.mAuth'].value)),
 127                 'Device-Type': 'web',
 128                 'Origin': origin,
 129                 'Referer': f'{origin}/',
 130                 'Accept': 'application/json, text/plain, */*',
 131             }))
 132
 133         if not traverse_obj(data, ('video', 'is_full_video')) and traverse_obj(
 134                 data, ('video', 'is_limited_by_user_access')):
 135             self.report_warning(
 136                 'The full video is for subscribers only. Only previews will be downloaded. If you '
 137                 'have used the --cookies-from-browser option, try using the --cookies option instead')
 138
 139         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 140             data['video']['video_url'], display_id, 'mp4')
 141
 142         return {
 143             'id': str(data['id']),
 144             'display_id': display_id,
 145             'formats': formats,
 146             'subtitles': subtitles,
 147             **traverse_obj(data, {
 148                 'title': ('title', {str}),
 149                 'description': ('subtitle', {str}),
 150                 'thumbnail': ('picture', {url_or_none}),
 151                 'timestamp': ('date_publish', {parse_iso8601}),
 152             }),
 153         }