[generic] add generic support for twitter:player embeds

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

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 1592a8a3a0f8a6a7478b4de880ddb3cfd7b2a1ee..c2a7f92025c88e69f405817ebb24836832767930 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -64,6 +64,8 @@ from .liveleak import LiveLeakIE
 from .threeqsdn import ThreeQSDNIE
 from .theplatform import ThePlatformIE
 from .vessel import VesselIE
+from .kaltura import KalturaIE
+from .eagleplatform import EaglePlatformIE
 
 
 class GenericIE(InfoExtractor):
@@ -920,6 +922,24 @@ class GenericIE(InfoExtractor):
             },
             'add_ie': ['Kaltura'],
         },
+        {
+            # Kaltura embedded via quoted entry_id
+            'url': 'https://www.oreilly.com/ideas/my-cloud-makes-pretty-pictures',
+            'info_dict': {
+                'id': '0_utuok90b',
+                'ext': 'mp4',
+                'title': '06_matthew_brender_raj_dutt',
+                'timestamp': 1466638791,
+                'upload_date': '20160622',
+            },
+            'add_ie': ['Kaltura'],
+            'expected_warnings': [
+                'Could not send HEAD request'
+            ],
+            'params': {
+                'skip_download': True,
+            }
+        },
         # Eagle.Platform embed (generic URL)
         {
             'url': 'http://lenta.ru/news/2015/03/06/navalny/',
@@ -1225,6 +1245,22 @@ class GenericIE(InfoExtractor):
                 'uploader': 'www.hudl.com',
             },
         },
+        # twitter:player embed
+        {
+            'url': 'http://www.theatlantic.com/video/index/484130/what-do-black-holes-sound-like/',
+            'md5': 'a3e0df96369831de324f0778e126653c',
+            'info_dict': {
+                'id': '4909620399001',
+                'ext': 'mp4',
+                'title': 'What Do Black Holes Sound Like?',
+                'description': 'what do black holes sound like',
+                'upload_date': '20160524',
+                'uploader_id': '29913724001',
+                'timestamp': 1464107587,
+                'uploader': 'TheAtlantic',
+            },
+            'add_ie': ['BrightcoveLegacy'],
+        }
     ]
 
     def report_following_redirect(self, new_url):
@@ -1908,18 +1944,14 @@ class GenericIE(InfoExtractor):
             return self.url_result(mobj.group('url'), 'Zapiks')
 
         # Look for Kaltura embeds
-        mobj = (re.search(r"(?s)kWidget\.(?:thumb)?[Ee]mbed\(\{.*?(?P<q1>['\"])wid(?P=q1)\s*:\s*(?P<q2>['\"])_?(?P<partner_id>[^'\"]+)(?P=q2),.*?(?P<q3>['\"])entry_?[Ii]d(?P=q3)\s*:\s*(?P<q4>['\"])(?P<id>[^'\"]+)(?P=q4),", webpage) or
-                re.search(r'(?s)(?P<q1>["\'])(?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/.*?(?:p|partner_id)/(?P<partner_id>\d+).*?(?P=q1).*?entry_?[Ii]d\s*:\s*(?P<q2>["\'])(?P<id>.+?)(?P=q2)', webpage))
-        if mobj is not None:
-            return self.url_result(smuggle_url(
-                'kaltura:%(partner_id)s:%(id)s' % mobj.groupdict(),
-                {'source_url': url}), 'Kaltura')
+        kaltura_url = KalturaIE._extract_url(webpage)
+        if kaltura_url:
+            return self.url_result(smuggle_url(kaltura_url, {'source_url': url}), KalturaIE.ie_key())
 
         # Look for Eagle.Platform embeds
-        mobj = re.search(
-            r'<iframe[^>]+src="(?P<url>https?://.+?\.media\.eagleplatform\.com/index/player\?.+?)"', webpage)
-        if mobj is not None:
-            return self.url_result(mobj.group('url'), 'EaglePlatform')
+        eagleplatform_url = EaglePlatformIE._extract_url(webpage)
+        if eagleplatform_url:
+            return self.url_result(eagleplatform_url, EaglePlatformIE.ie_key())
 
         # Look for ClipYou (uses Eagle.Platform) embeds
         mobj = re.search(
@@ -2065,6 +2097,11 @@ class GenericIE(InfoExtractor):
                 'uploader': video_uploader,
             }
 
+        # https://dev.twitter.com/cards/types/player#On_twitter.com_via_desktop_browser
+        embed_url = self._twitter_search_player(webpage)
+        if embed_url:
+            return self.url_result(embed_url)
+
         def check_video(vurl):
             if YoutubeIE.suitable(vurl):
                 return True