[extractor/generic] Add support for async wistia embeds (Closes #9549)

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

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 0f1eb7fa64dbf64b32000fd05ea6b65bb8ea502a..632d7b5f01a612d1fb5e33ddbfc845d09ca6e034 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -61,6 +61,7 @@ from .jwplatform import JWPlatformIE
 from .digiteka import DigitekaIE
 from .instagram import InstagramIE
 from .liveleak import LiveLeakIE
+from .threeqsdn import ThreeQSDNIE
 
 
 class GenericIE(InfoExtractor):
@@ -1427,7 +1428,8 @@ class GenericIE(InfoExtractor):
         #   Site Name | Video Title
         #   Video Title - Tagline | Site Name
         # and so on and so forth; it's just not practical
-        video_title = self._html_search_regex(
+        video_title = self._og_search_title(
+            webpage, default=None) or self._html_search_regex(
             r'(?s)<title>(.*?)</title>', webpage, 'video title',
             default='video')
 
@@ -1445,6 +1447,9 @@ class GenericIE(InfoExtractor):
         video_uploader = self._search_regex(
             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 
+        video_description = self._og_search_description(webpage, default=None)
+        video_thumbnail = self._og_search_thumbnail(webpage, default=None)
+
         # Helper method
         def _playlist_from_matches(matches, getter=None, ie=None):
             urlrs = orderedSet(
@@ -1558,6 +1563,15 @@ class GenericIE(InfoExtractor):
                 'id': match.group('id')
             }
 
+        match = re.search(
+            r'''(?sx)
+                <script[^>]+src=(["'])(?:https?:)?//fast\.wistia\.com/assets/external/E-v1\.js\1[^>]*>.*?
+                <div[^>]+class=(["']).*?\bwistia_async_(?P<id>[a-z0-9]+)\b.*?\2
+            ''', webpage)
+        if match:
+            return self.url_result(self._proto_relative_url(
+                'wistia:%s' % match.group('id')), 'Wistia')
+
         # Look for SVT player
         svt_url = SVTIE._extract_url(webpage)
         if svt_url:
@@ -1983,6 +1997,19 @@ class GenericIE(InfoExtractor):
         if liveleak_url:
             return self.url_result(liveleak_url, 'LiveLeak')
 
+        # Look for 3Q SDN embeds
+        threeqsdn_url = ThreeQSDNIE._extract_url(webpage)
+        if threeqsdn_url:
+            return {
+                '_type': 'url_transparent',
+                'ie_key': ThreeQSDNIE.ie_key(),
+                'url': self._proto_relative_url(threeqsdn_url),
+                'title': video_title,
+                'description': video_description,
+                'thumbnail': video_thumbnail,
+                'uploader': video_uploader,
+            }
+
         def check_video(vurl):
             if YoutubeIE.suitable(vurl):
                 return True