Merge 78ea2ecb7d into 64766459e3

yt_dlp/extractor/epoch.py

@@ -1,5 +1,14 @@
+import re
+
 from .common import InfoExtractor
-from ..utils import extract_attributes, get_element_html_by_id
+from ..utils import (
+    ExtractorError,
+    clean_html,
+    extract_attributes,
+    get_element_by_class,
+    get_element_html_by_class,
+    try_get,
+)
 
 
 class EpochIE(InfoExtractor):
@@ -11,6 +20,8 @@ class EpochIE(InfoExtractor):
                 'id': 'a3dd732c-4750-4bc8-8156-69180668bda1',
                 'ext': 'mp4',
                 'title': '‘They Can Do Audio, Video, Physical Surveillance on You 24H/365D a Year’: Rex Lee on Intrusive Apps',
+                'description': 'md5:00f32d1e821481a88698e6f450a04e1b',
+                'thumbnail': r're:https://.*\.jpg',
             }
         },
         {
@@ -19,6 +30,8 @@ class EpochIE(InfoExtractor):
                 'id': '276c7f46-3bbf-475d-9934-b9bbe827cf0a',
                 'ext': 'mp4',
                 'title': 'The Communist Party’s Cyberattacks on America Explained; Rex Lee Talks Tech Hybrid Warfare',
+                'description': 'md5:c993752ccec901a6b84e12117c5cc8f8',
+                'thumbnail': r're:https://.*\.jpg',
             }
         },
         {
@@ -27,6 +40,8 @@ class EpochIE(InfoExtractor):
                 'id': 'aa9ceecd-a127-453d-a2de-7153d6fd69b6',
                 'ext': 'mp4',
                 'title': 'Kash Patel: A ‘6-Year-Saga’ of Government Corruption, From Russiagate to Mar-a-Lago',
+                'description': 'md5:1aa6aa4b99934f985785837a34707753',
+                'thumbnail': r're:https://.*\.jpg',
             }
         },
         {
@@ -35,21 +50,57 @@ class EpochIE(InfoExtractor):
                 'id': '9489f994-2a20-4812-b233-ac0e5c345632',
                 'ext': 'mp4',
                 'title': 'Dick Morris Discusses His Book ‘The Return: Trump’s Big 2024 Comeback’',
+                'description': 'md5:ef6f45dd925b5d23e2a862487b42030e',
+                'thumbnail': r're:https://.*\.jpg',
             }
         },
+        {
+            'note': 'Georestricted to US',
+            'url': 'https://www.theepochtimes.com/silence-patton-documentary_4849547.html',
+            'info_dict': {
+                'id': '4849547',
+                'ext': 'mp4',
+                'title': 'Silence Patton | Documentary',
+                'description': 'Why was General Patton silenced during his service in World War II?',
+                'thumbnail': r're:https://.*\.jpg',
+            },
+            'params': {'skip_download': True},
+            'expected_warnings': [
+                'This film is only available in the United States because of territorial licensing.'
+            ],
+        },
     ]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
+        raw_description = clean_html(get_element_by_class('article_content', webpage))
+        description_lines_iter = iter(raw_description.splitlines())
+        description = next(description_lines_iter, None)
+        if re.match(r'This film is [a-z ]*available ', description or ''):
+            self.report_warning(description, video_id)
+            description = next(description_lines_iter, None)
+
+        player = extract_attributes(get_element_html_by_class('player-container', webpage) or '<>')
+        youmaker_video_id = try_get(
+            player, lambda x: re.fullmatch(r'player-([0-9a-f-]+)', x['data-id']).group(1))
+
+        if youmaker_video_id:
+            video_id = youmaker_video_id
+            manifest_url = f'http://vs1.youmaker.com/assets/{youmaker_video_id}/playlist.m3u8'
+        elif player.get('data-source', '').endswith('.m3u8'):
+            manifest_url = player['data-source']
+        else:
+            raise ExtractorError('No video found.')
 
-        youmaker_video_id = extract_attributes(get_element_html_by_id('videobox', webpage))['data-id']
         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
-            f'http://vs1.youmaker.com/assets/{youmaker_video_id}/playlist.m3u8', video_id, 'mp4', m3u8_id='hls')
+            manifest_url, video_id, 'mp4', m3u8_id='hls')
 
         return {
-            'id': youmaker_video_id,
+            'id': video_id,
             'formats': formats,
             'subtitles': subtitles,
-            'title': self._html_extract_title(webpage)
+            'title': self._html_extract_title(webpage),
+            'description': description,
+            'thumbnail': self._og_search_thumbnail(webpage),
         }