]> jfr.im git - yt-dlp.git/blobdiff - yt_dlp/extractor/europa.py
[ie/youtube] Suppress "Unavailable videos are hidden" warning (#10159)
[yt-dlp.git] / yt_dlp / extractor / europa.py
index 29daabe4a385bb1d813625318feeffd7a38beca1..aa8baf2f78f1f97b773c793f75a6f50bb87da2cd 100644 (file)
@@ -6,12 +6,14 @@
     parse_iso8601,
     parse_qs,
     qualities,
+    traverse_obj,
     unified_strdate,
-    xpath_text
+    xpath_text,
 )
 
 
 class EuropaIE(InfoExtractor):
+    _WORKING = False
     _VALID_URL = r'https?://ec\.europa\.eu/avservices/(?:video/player|audio/audioDetails)\.cfm\?.*?\bref=(?P<id>[A-Za-z0-9-]+)'
     _TESTS = [{
         'url': 'http://ec.europa.eu/avservices/video/player.cfm?ref=I107758',
@@ -26,7 +28,7 @@ class EuropaIE(InfoExtractor):
             'duration': 34,
             'view_count': int,
             'formats': 'mincount:3',
-        }
+        },
     }, {
         'url': 'http://ec.europa.eu/avservices/video/player.cfm?sitelang=en&ref=I107786',
         'only_matching': True,
@@ -39,11 +41,11 @@ def _real_extract(self, url):
         video_id = self._match_id(url)
 
         playlist = self._download_xml(
-            'http://ec.europa.eu/avservices/video/player/playlist.cfm?ID=%s' % video_id, video_id)
+            f'http://ec.europa.eu/avservices/video/player/playlist.cfm?ID={video_id}', video_id)
 
         def get_item(type_, preference):
             items = {}
-            for item in playlist.findall('./info/%s/item' % type_):
+            for item in playlist.findall(f'./info/{type_}/item'):
                 lang, label = xpath_text(item, 'lg', default=None), xpath_text(item, 'label', default=None)
                 if lang and label:
                     items[lang] = label.strip()
@@ -75,7 +77,7 @@ def get_item(type_, preference):
                 'url': video_url,
                 'format_id': lang,
                 'format_note': xpath_text(file_, './lglabel'),
-                'language_preference': language_preference(lang)
+                'language_preference': language_preference(lang),
             })
 
         return {
@@ -86,88 +88,103 @@ def get_item(type_, preference):
             'upload_date': upload_date,
             'duration': duration,
             'view_count': view_count,
-            'formats': formats
+            'formats': formats,
         }
 
 
 class EuroParlWebstreamIE(InfoExtractor):
     _VALID_URL = r'''(?x)
-        https?://(?:multimedia|webstreaming)\.europarl\.europa\.eu/[^/#?]+/
-        (?:embed/embed\.html\?event=|(?!video)[^/#?]+/[\w-]+_)(?P<id>[\w-]+)
+        https?://multimedia\.europarl\.europa\.eu/
+        (?:\w+/)?webstreaming/(?:[\w-]+_)?(?P<id>[\w-]+)
     '''
     _TESTS = [{
         'url': 'https://multimedia.europarl.europa.eu/pl/webstreaming/plenary-session_20220914-0900-PLENARY',
         'info_dict': {
-            'id': 'bcaa1db4-76ef-7e06-8da7-839bd0ad1dbe',
+            'id': '62388b15-d85b-4add-99aa-ba12ccf64f0d',
+            'display_id': '20220914-0900-PLENARY',
             'ext': 'mp4',
-            'release_timestamp': 1663137900,
             'title': 'Plenary session',
+            'release_timestamp': 1663139069,
             'release_date': '20220914',
         },
         'params': {
             'skip_download': True,
-        }
+        },
     }, {
-        'url': 'https://multimedia.europarl.europa.eu/pl/webstreaming/eu-cop27-un-climate-change-conference-in-sharm-el-sheikh-egypt-ep-delegation-meets-with-ngo-represen_20221114-1600-SPECIAL-OTHER',
+        # live webstream
+        'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/euroscola_20221115-1000-SPECIAL-EUROSCOLA',
         'info_dict': {
-            'id': 'a8428de8-b9cd-6a2e-11e4-3805d9c9ff5c',
             'ext': 'mp4',
-            'release_timestamp': 1668434400,
-            'release_date': '20221114',
-            'title': 'md5:d3550280c33cc70e0678652e3d52c028',
+            'id': '510eda7f-ba72-161b-7ee7-0e836cd2e715',
+            'release_timestamp': 1668502800,
+            'title': 'Euroscola 2022-11-15 19:21',
+            'release_date': '20221115',
+            'live_status': 'is_live',
         },
-        'params': {
-            'skip_download': True,
-        }
+        'skip': 'not live anymore',
     }, {
-        # embed webpage
-        'url': 'https://webstreaming.europarl.europa.eu/ep/embed/embed.html?event=20220914-0900-PLENARY&language=en&autoplay=true&logo=true',
+        'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/committee-on-culture-and-education_20230301-1130-COMMITTEE-CULT',
         'info_dict': {
-            'id': 'bcaa1db4-76ef-7e06-8da7-839bd0ad1dbe',
+            'id': '7355662c-8eac-445e-4bb9-08db14b0ddd7',
+            'display_id': '20230301-1130-COMMITTEE-CULT',
             'ext': 'mp4',
-            'title': 'Plenary session',
-            'release_date': '20220914',
-            'release_timestamp': 1663137900,
+            'release_date': '20230301',
+            'title': 'Committee on Culture and Education',
+            'release_timestamp': 1677666641,
         },
-        'params': {
-            'skip_download': True,
-        }
     }, {
-        # live webstream
-        'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/euroscola_20221115-1000-SPECIAL-EUROSCOLA',
+        # live stream
+        'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/committee-on-environment-public-health-and-food-safety_20230524-0900-COMMITTEE-ENVI',
         'info_dict': {
+            'id': 'e4255f56-10aa-4b3c-6530-08db56d5b0d9',
             'ext': 'mp4',
-            'id': '510eda7f-ba72-161b-7ee7-0e836cd2e715',
-            'release_timestamp': 1668502800,
-            'title': 'Euroscola 2022-11-15 19:21',
-            'release_date': '20221115',
+            'release_date': '20230524',
+            'title': r're:Committee on Environment, Public Health and Food Safety \d{4}-\d{2}-\d{2}\s\d{2}:\d{2}',
+            'release_timestamp': 1684911541,
             'live_status': 'is_live',
         },
-        'skip': 'not live anymore'
+        'skip': 'Not live anymore',
+    }, {
+        'url': 'https://multimedia.europarl.europa.eu/en/webstreaming/20240320-1345-SPECIAL-PRESSER',
+        'info_dict': {
+            'id': 'c1f11567-5b52-470a-f3e1-08dc3c216ace',
+            'display_id': '20240320-1345-SPECIAL-PRESSER',
+            'ext': 'mp4',
+            'release_date': '20240320',
+            'title': 'md5:7c6c814cac55dea5e2d87bf8d3db2234',
+            'release_timestamp': 1710939767,
+        },
+    }, {
+        'url': 'https://multimedia.europarl.europa.eu/webstreaming/briefing-for-media-on-2024-european-elections_20240429-1000-SPECIAL-OTHER',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
         display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+
+        webpage_nextjs = self._search_nextjs_data(webpage, display_id)['props']['pageProps']
 
         json_info = self._download_json(
-            'https://vis-api.vuplay.co.uk/event/external', display_id,
+            'https://acs-api.europarl.connectedviews.eu/api/FullMeeting', display_id,
             query={
-                'player_key': 'europarl|718f822c-a48c-4841-9947-c9cb9bb1743c',
-                'external_id': display_id,
+                'api-version': 1.0,
+                'tenantId': 'bae646ca-1fc8-4363-80ba-2c04f06b4968',
+                'externalReference': display_id,
             })
 
-        formats, subtitles = self._extract_mpd_formats_and_subtitles(json_info['streaming_url'], display_id)
-        fmts, subs = self._extract_m3u8_formats_and_subtitles(
-            json_info['streaming_url'].replace('.mpd', '.m3u8'), display_id)
-
-        formats.extend(fmts)
-        self._merge_subtitles(subs, target=subtitles)
+        formats, subtitles = [], {}
+        for hls_url in traverse_obj(json_info, ((('meetingVideo'), ('meetingVideos', ...)), 'hlsUrl')):
+            fmt, subs = self._extract_m3u8_formats_and_subtitles(hls_url, display_id)
+            formats.extend(fmt)
+            self._merge_subtitles(subs, target=subtitles)
 
         return {
             'id': json_info['id'],
-            'title': json_info.get('title'),
+            'display_id': display_id,
+            'title': traverse_obj(webpage_nextjs, (('mediaItem', 'title'), ('title', )), get_all=False),
             'formats': formats,
             'subtitles': subtitles,
-            'release_timestamp': parse_iso8601(json_info.get('published_start')),
-            'is_live': 'LIVE' in json_info.get('state', '')
+            'release_timestamp': parse_iso8601(json_info.get('startDateTime')),
+            'is_live': traverse_obj(webpage_nextjs, ('mediaItem', 'mediaSubType')) == 'Live',
         }