[imdb] Fix extraction (closes #23443)

youtube_dl/extractor/imdb.py

@@ -1,5 +1,7 @@
 from __future__ import unicode_literals
 
+import base64
+import json
 import re
 
 from .common import InfoExtractor
@@ -8,6 +10,7 @@
     mimetype2ext,
     parse_duration,
     qualities,
+    try_get,
     url_or_none,
 )
 
@@ -15,15 +18,16 @@
 class ImdbIE(InfoExtractor):
     IE_NAME = 'imdb'
     IE_DESC = 'Internet Movie Database trailers'
-    _VALID_URL = r'https?://(?:www|m)\.imdb\.com/(?:video|title|list).+?[/-]vi(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:www|m)\.imdb\.com/(?:video|title|list).*?[/-]vi(?P<id>\d+)'
 
     _TESTS = [{
         'url': 'http://www.imdb.com/video/imdb/vi2524815897',
         'info_dict': {
             'id': '2524815897',
             'ext': 'mp4',
-            'title': 'No. 2 from Ice Age: Continental Drift (2012)',
+            'title': 'No. 2',
             'description': 'md5:87bd0bdc61e351f21f20d2d7441cb4e7',
+            'duration': 152,
         }
     }, {
         'url': 'http://www.imdb.com/video/_/vi2524815897',
@@ -47,21 +51,23 @@ class ImdbIE(InfoExtractor):
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
-        webpage = self._download_webpage(
+
-            'https://www.imdb.com/videoplayer/vi' + video_id, video_id)
+        data = self._download_json(
-        video_metadata = self._parse_json(self._search_regex(
+            'https://www.imdb.com/ve/data/VIDEO_PLAYBACK_DATA', video_id,
-            r'window\.IMDbReactInitialState\.push\(({.+?})\);', webpage,
+            query={
-            'video metadata'), video_id)['videos']['videoMetadata']['vi' + video_id]
+                'key': base64.b64encode(json.dumps({
-        title = self._html_search_meta(
+                    'type': 'VIDEO_PLAYER',
-            ['og:title', 'twitter:title'], webpage) or self._html_search_regex(
+                    'subType': 'FORCE_LEGACY',
-            r'<title>(.+?)</title>', webpage, 'title', fatal=False) or video_metadata['title']
+                    'id': 'vi%s' % video_id,
+                }).encode()).decode(),
+            })[0]
 
         quality = qualities(('SD', '480p', '720p', '1080p'))
         formats = []
-        for encoding in video_metadata.get('encodings', []):
+        for encoding in data['videoLegacyEncodings']:
             if not encoding or not isinstance(encoding, dict):
                 continue
-            video_url = url_or_none(encoding.get('videoUrl'))
+            video_url = url_or_none(encoding.get('url'))
             if not video_url:
                 continue
             ext = mimetype2ext(encoding.get(
@@ -69,7 +75,7 @@ def _real_extract(self, url):
             if ext == 'm3u8':
                 formats.extend(self._extract_m3u8_formats(
                     video_url, video_id, 'mp4', entry_protocol='m3u8_native',
-                    m3u8_id='hls', fatal=False))
+                    preference=1, m3u8_id='hls', fatal=False))
                 continue
             format_id = encoding.get('definition')
             formats.append({
@@ -80,13 +86,33 @@ def _real_extract(self, url):
             })
         self._sort_formats(formats)
 
+        webpage = self._download_webpage(
+            'https://www.imdb.com/video/vi' + video_id, video_id)
+        video_metadata = self._parse_json(self._search_regex(
+            r'args\.push\(\s*({.+?})\s*\)\s*;', webpage,
+            'video metadata'), video_id)
+
+        video_info = video_metadata.get('VIDEO_INFO')
+        if video_info and isinstance(video_info, dict):
+            info = try_get(
+                video_info, lambda x: x[list(video_info.keys())[0]][0], dict)
+        else:
+            info = {}
+
+        title = self._html_search_meta(
+            ['og:title', 'twitter:title'], webpage) or self._html_search_regex(
+            r'<title>(.+?)</title>', webpage, 'title',
+            default=None) or info['videoTitle']
+
         return {
             'id': video_id,
             'title': title,
+            'alt_title': info.get('videoSubTitle'),
             'formats': formats,
-            'description': video_metadata.get('description'),
+            'description': info.get('videoDescription'),
-            'thumbnail': video_metadata.get('slate', {}).get('url'),
+            'thumbnail': url_or_none(try_get(
-            'duration': parse_duration(video_metadata.get('duration')),
+                video_metadata, lambda x: x['videoSlate']['source'])),
+            'duration': parse_duration(info.get('videoRuntime')),
         }