[ie/medici] Fix extractor (#9518)

Closes #8813
Authored by: Offert4324

yt_dlp/extractor/generic.py

@@ -2104,22 +2104,6 @@ class GenericIE(InfoExtractor):
                 'age_limit': 0,
             },
         },
-        {
-            'note': 'JW Player embed with unicode-escape sequences in URL',
-            'url': 'https://www.medici.tv/en/concerts/lahav-shani-mozart-mahler-israel-philharmonic-abu-dhabi-classics',
-            'info_dict': {
-                'id': 'm',
-                'ext': 'mp4',
-                'title': 'Lahav Shani conducts the Israel Philharmonic\'s first-ever concert in Abu Dhabi',
-                'description': 'Mahler\'s ',
-                'uploader': 'www.medici.tv',
-                'age_limit': 0,
-                'thumbnail': r're:^https?://.+\.jpg',
-            },
-            'params': {
-                'skip_download': True,
-            },
-        },
         {
             'url': 'https://shooshtime.com/videos/284002/just-out-of-the-shower-joi/',
             'md5': 'e2f0a4c329f7986280b7328e24036d60',

yt_dlp/extractor/medici.py

@@ -1,67 +1,153 @@
+import urllib.parse
+
 from .common import InfoExtractor
 from ..utils import (
-    unified_strdate,
+    filter_dict,
-    update_url_query,
+    parse_iso8601,
-    urlencode_postdata,
+    traverse_obj,
+    try_call,
+    url_or_none,
 )
 
 
 class MediciIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?medici\.tv/#!/(?P<id>[^?#&]+)'
+    _VALID_URL = r'https?://(?:(?P<sub>www|edu)\.)?medici\.tv/[a-z]{2}/[\w.-]+/(?P<id>[^/?#&]+)'
-    _TEST = {
+    _TESTS = [{
-        'url': 'http://www.medici.tv/#!/daniel-harding-frans-helmerson-verbier-festival-music-camp',
+        'url': 'https://www.medici.tv/en/operas/thomas-ades-the-exterminating-angel-calixto-bieito-opera-bastille-paris',
-        'md5': '004c21bb0a57248085b6ff3fec72719d',
+        'md5': 'd483f74e7a7a9eac0dbe152ab189050d',
         'info_dict': {
-            'id': '3059',
+            'id': '8032',
-            'ext': 'flv',
+            'ext': 'mp4',
-            'title': 'Daniel Harding conducts the Verbier Festival Music Camp \u2013 With Frans Helmerson',
+            'title': 'Thomas Adès\'s The Exterminating Angel',
-            'description': 'md5:322a1e952bafb725174fd8c1a8212f58',
+            'description': 'md5:708ae6350dadc604225b4a6e32482bab',
-            'thumbnail': r're:^https?://.*\.jpg$',
+            'thumbnail': r're:https://.+/.+\.jpg',
-            'upload_date': '20170408',
+            'upload_date': '20240304',
+            'timestamp': 1709561766,
+            'display_id': 'thomas-ades-the-exterminating-angel-calixto-bieito-opera-bastille-paris',
         },
-    }
+        'expected_warnings': [r'preview'],
+    }, {
+        'url': 'https://edu.medici.tv/en/operas/wagner-lohengrin-paris-opera-kirill-serebrennikov-piotr-beczala-kwangchul-youn-johanni-van-oostrum',
+        'md5': '4ef3f4079a6e1c617584463a9eb84f99',
+        'info_dict': {
+            'id': '7900',
+            'ext': 'mp4',
+            'title': 'Wagner\'s Lohengrin',
+            'description': 'md5:a384a62937866101f86902f21752cd89',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'upload_date': '20231017',
+            'timestamp': 1697554771,
+            'display_id': 'wagner-lohengrin-paris-opera-kirill-serebrennikov-piotr-beczala-kwangchul-youn-johanni-van-oostrum',
+        },
+        'expected_warnings': [r'preview'],
+    }, {
+        'url': 'https://www.medici.tv/en/concerts/sergey-smbatyan-conducts-mansurian-chouchane-siranossian-mario-brunello',
+        'md5': '9dd757e53b22b2511e85ea9ea60e4815',
+        'info_dict': {
+            'id': '5712',
+            'ext': 'mp4',
+            'title': 'Sergey Smbatyan conducts Tigran Mansurian — With Chouchane Siranossian and Mario Brunello',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'description': 'md5:9411fe44c874bb10e9af288c65816e41',
+            'upload_date': '20200323',
+            'timestamp': 1584975600,
+            'display_id': 'sergey-smbatyan-conducts-mansurian-chouchane-siranossian-mario-brunello',
+        },
+        'expected_warnings': [r'preview'],
+    }, {
+        'url': 'https://www.medici.tv/en/ballets/carmen-ballet-choregraphie-de-jiri-bubenicek-teatro-dellopera-di-roma',
+        'md5': '40f5e76cb701a97a6d7ba23b62c49990',
+        'info_dict': {
+            'id': '7857',
+            'ext': 'mp4',
+            'title': 'Carmen by Jiří Bubeníček after Roland Petit, music by Bizet, de Falla, Castelnuovo-Tedesco, and Bonolis',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'description': 'md5:0f15a15611ed748020c769873e10a8bb',
+            'upload_date': '20240223',
+            'timestamp': 1708707600,
+            'display_id': 'carmen-ballet-choregraphie-de-jiri-bubenicek-teatro-dellopera-di-roma',
+        },
+        'expected_warnings': [r'preview'],
+    }, {
+        'url': 'https://www.medici.tv/en/documentaries/la-sonnambula-liege-2023-documentaire',
+        'md5': '87ff198018ce79a34757ab0dd6f21080',
+        'info_dict': {
+            'id': '7513',
+            'ext': 'mp4',
+            'title': 'La Sonnambula',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'description': 'md5:0caf9109a860fd50cd018df062a67f34',
+            'upload_date': '20231103',
+            'timestamp': 1699010830,
+            'display_id': 'la-sonnambula-liege-2023-documentaire',
+        },
+        'expected_warnings': [r'preview'],
+    }, {
+        'url': 'https://edu.medici.tv/en/masterclasses/yvonne-loriod-olivier-messiaen',
+        'md5': 'fb5dcec46d76ad20fbdbaabb01da191d',
+        'info_dict': {
+            'id': '3024',
+            'ext': 'mp4',
+            'title': 'Olivier Messiaen and Yvonne Loriod, pianists and teachers',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'description': 'md5:aab948e2f7690214b5c28896c83f1fc1',
+            'upload_date': '20150223',
+            'timestamp': 1424706608,
+            'display_id': 'yvonne-loriod-olivier-messiaen',
+        },
+        'skip': 'Requires authentication; preview starts in the middle',
+    }, {
+        'url': 'https://www.medici.tv/en/jazz/makaya-mccraven-la-rochelle',
+        'md5': '4cc279a8b06609782747c8f50beea2b3',
+        'info_dict': {
+            'id': '7922',
+            'ext': 'mp4',
+            'title': 'NEW: Makaya McCraven in La Rochelle',
+            'thumbnail': r're:https://.+/.+\.jpg',
+            'description': 'md5:b5a8aaeb6993d8ccb18bde8abb8aa8d2',
+            'upload_date': '20231228',
+            'timestamp': 1703754863,
+            'display_id': 'makaya-mccraven-la-rochelle',
+        },
+        'expected_warnings': [r'preview'],
+    }]
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        display_id, subdomain = self._match_valid_url(url).group('id', 'sub')
-
+        self._request_webpage(url, display_id, 'Requesting CSRF token cookie')
-        # Sets csrftoken cookie
-        self._download_webpage(url, video_id)
 
-        MEDICI_URL = 'http://www.medici.tv/'
+        subdomain = 'edu-' if subdomain == 'edu' else ''
+        origin = f'https://{urllib.parse.urlparse(url).hostname}'
 
         data = self._download_json(
-            MEDICI_URL, video_id,
+            f'https://api.medici.tv/{subdomain}satie/edito/movie-file/{display_id}/', display_id,
-            data=urlencode_postdata({
+            headers=filter_dict({
-                'json': 'true',
+                'Authorization': try_call(
-                'page': '/%s' % video_id,
+                    lambda: urllib.parse.unquote(self._get_cookies(url)['auth._token.mAuth'].value)),
-                'timezone_offset': -420,
+                'Device-Type': 'web',
-            }), headers={
+                'Origin': origin,
-                'X-CSRFToken': self._get_cookies(url)['csrftoken'].value,
+                'Referer': f'{origin}/',
-                'X-Requested-With': 'XMLHttpRequest',
+                'Accept': 'application/json, text/plain, */*',
-                'Referer': MEDICI_URL,
+            }))
-                'Content-Type': 'application/x-www-form-urlencoded',
-            })
-
-        video = data['video']['videos']['video1']
-
-        title = video.get('nom') or data['title']
 
-        video_id = video.get('id') or video_id
+        if not traverse_obj(data, ('video', 'is_full_video')) and traverse_obj(
-        formats = self._extract_f4m_formats(
+                data, ('video', 'is_limited_by_user_access')):
-            update_url_query(video['url_akamai'], {
+            self.report_warning(
-                'hdcore': '3.1.0',
+                'The full video is for subscribers only. Only previews will be downloaded. If you '
-                'plugin=aasp': '3.1.0.43.124',
+                'have used the --cookies-from-browser option, try using the --cookies option instead')
-            }), video_id, f4m_id='hds')
 
-        description = data.get('meta_description')
+        formats, subtitles = self._extract_m3u8_formats_and_subtitles(
-        thumbnail = video.get('url_thumbnail') or data.get('main_image')
+            data['video']['video_url'], display_id, 'mp4')
-        upload_date = unified_strdate(data['video'].get('date'))
 
         return {
-            'id': video_id,
+            'id': str(data['id']),
-            'title': title,
+            'display_id': display_id,
-            'description': description,
-            'thumbnail': thumbnail,
-            'upload_date': upload_date,
             'formats': formats,
+            'subtitles': subtitles,
+            **traverse_obj(data, {
+                'title': ('title', {str}),
+                'description': ('subtitle', {str}),
+                'thumbnail': ('picture', {url_or_none}),
+                'timestamp': ('date_publish', {parse_iso8601}),
+            }),
         }