[extractor/txxx] Add extractors (#5240)

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

diff --git a/yt_dlp/extractor/generic.py b/yt_dlp/extractor/generic.py
index 14d492f075580ba6f93710029144cd7fac24cdbb..55e55d5248bd52dfa8cdf010de7a2ef0940350a7 100644 (file)
--- a/yt_dlp/extractor/generic.py
+++ b/yt_dlp/extractor/generic.py
@@ -864,20 +864,6 @@ class GenericIE(InfoExtractor):
                 'thumbnail': r're:^https?://.*\.jpg$',
             },
         },
-        {
-            # JWPlayer config passed as variable
-            'url': 'http://www.txxx.com/videos/3326530/ariele/',
-            'info_dict': {
-                'id': '3326530_hq',
-                'ext': 'mp4',
-                'title': 'ARIELE | Tube Cup',
-                'uploader': 'www.txxx.com',
-                'age_limit': 18,
-            },
-            'params': {
-                'skip_download': True,
-            }
-        },
         {
             # Video.js embed, multiple formats
             'url': 'http://ortcam.com/solidworks-урок-6-настройка-чертежа_33f9b7351.html',
@@ -2135,7 +2121,8 @@ class GenericIE(InfoExtractor):
                 'age_limit': 0,
                 'direct': True,
             }
-        }, {
+        },
+        {
             'note': 'server returns data in brotli compression by default if `accept-encoding: *` is specified.',
             'url': 'https://www.extra.cz/cauky-lidi-70-dil-babis-predstavil-pohadky-prymulanek-nebo-andrejovy-nove-saty-ac867',
             'info_dict': {
@@ -2149,7 +2136,23 @@ class GenericIE(InfoExtractor):
                 'duration': 318.0,
                 'direct': True,
                 '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/',
@@ -2620,11 +2623,11 @@ def _extract_embeds(self, url, webpage, *, urlh=None, info_dict={}):
 
         # Look for generic KVS player (before json-ld bc of some urls that break otherwise)
         found = self._search_regex((
-            r'<script\b[^>]+?\bsrc\s*=\s*(["\'])https?://(?:\S+?/)+kt_player\.js\?v=(?P<ver>\d+(?:\.\d+)+)\1[^>]*>',
-            r'kt_player\s*\(\s*(["\'])(?:(?!\1)[\w\W])+\1\s*,\s*(["\'])https?://(?:\S+?/)+kt_player\.swf\?v=(?P<ver>\d+(?:\.\d+)+)\2\s*,',
+            r'<script\b[^>]+?\bsrc\s*=\s*(["\'])https?://(?:(?!\1)[^?#])+/kt_player\.js\?v=(?P<ver>\d+(?:\.\d+)+)\1[^>]*>',
+            r'kt_player\s*\(\s*(["\'])(?:(?!\1)[\w\W])+\1\s*,\s*(["\'])https?://(?:(?!\2)[^?#])+/kt_player\.swf\?v=(?P<ver>\d+(?:\.\d+)+)\2\s*,',
         ), webpage, 'KVS player', group='ver', default=False)
         if found:
-            self.report_detected('KWS Player')
+            self.report_detected('KVS Player')
             if found.split('.')[0] not in ('4', '5', '6'):
                 self.report_warning(f'Untested major version ({found}) in player engine - download may fail.')
             return [self._extract_kvs(url, webpage, video_id)]
@@ -2751,6 +2754,7 @@ def filter_video(urls):
 
         entries = []
         for video_url in orderedSet(found):
+            video_url = video_url.encode().decode('unicode-escape')
             video_url = unescapeHTML(video_url)
             video_url = video_url.replace('\\/', '/')
             video_url = urllib.parse.urljoin(url, video_url)