[ard] Change subtitles extension to ttml (Closes #9169)

[youtube-dl] / youtube_dl / extractor / tmz.py

diff --git a/youtube_dl/extractor/tmz.py b/youtube_dl/extractor/tmz.py
index b82dc8233f6201a0a94f550b9b4c91debb247701..7dbe68b5c228ea6d3fa20d835a60ecf38b820816 100644 (file)
--- a/youtube_dl/extractor/tmz.py
+++ b/youtube_dl/extractor/tmz.py
@@ -15,25 +15,46 @@ class TMZIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Kim Kardashian\'s Boobs Unlock a Mystery!',
             'description': 'Did Kim Kardasain try to one-up Khloe by one-upping Kylie???  Or is she just showing off her amazing boobs?',
-            'thumbnail': 'http://cdnbakmi.kaltura.com/p/591531/sp/59153100/thumbnail/entry_id/0_okj015ty/version/100002/acv/182/width/640',
+            'thumbnail': r're:http://cdnbakmi\.kaltura\.com/.*thumbnail.*',
         }
     }
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
-
         webpage = self._download_webpage(url, video_id)
 
-        url = self._html_search_regex(r'<meta name="VideoURL" content="(.+)"', webpage, 'url')
-        title = self._html_search_regex(r'<meta property="og:title" content="(.+)"', webpage, 'title')
-        description = self._html_search_regex(r'<meta property="og:description" content="(.+)"', webpage, 'description')
-        thumbnail_url = self._html_search_regex(r'<meta name="ThumbURL" content="(.+)"', webpage, 'thumbnail url')
-
         return {
             'id': video_id,
-            'url': url,
+            'url': self._html_search_meta('VideoURL', webpage, fatal=True),
+            'title': self._og_search_title(webpage),
+            'description': self._og_search_description(webpage),
+            'thumbnail': self._html_search_meta('ThumbURL', webpage),
+        }
+
+
+class TMZArticleIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?tmz\.com/\d{4}/\d{2}/\d{2}/(?P<id>[^/]+)/?'
+    _TEST = {
+        'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
+        'md5': 'e482a414a38db73087450e3a6ce69d00',
+        'info_dict': {
+            'id': '0_6snoelag',
             'ext': 'mp4',
-            'title': title,
-            'description': description,
-            'thumbnail': thumbnail_url,
+            'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake',
+            'description': 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake.  She\'s watching me."',
         }
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, video_id)
+        embedded_video_info_str = self._html_search_regex(
+            r'tmzVideoEmbedV2\("([^)]+)"\);', webpage, 'embedded video info')
+
+        embedded_video_info = self._parse_json(
+            embedded_video_info_str, video_id,
+            transform_source=lambda s: s.replace('\\', ''))
+
+        return self.url_result(
+            'http://www.tmz.com/videos/%s/' % embedded_video_info['id'])