[brightcove:legacy] Improve embeds detection (closes #11523)

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

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index f9707c155c60841d8acd819ddbab2151c0039807..79d10a1d1e3ba9b6665ad83cc63ee32419608cf2 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -56,10 +56,10 @@ from .dailymotion import (
 )
 from .onionstudios import OnionStudiosIE
 from .viewlift import ViewLiftEmbedIE
-from .screenwavemedia import ScreenwaveMediaIE
 from .mtv import MTVServicesEmbeddedIE
 from .pladform import PladformIE
 from .videomore import VideomoreIE
+from .webcaster import WebcasterFeedIE
 from .googledrive import GoogleDriveIE
 from .jwplatform import JWPlatformIE
 from .digiteka import DigitekaIE
@@ -75,6 +75,7 @@ from .facebook import FacebookIE
 from .soundcloud import SoundcloudIE
 from .vbox7 import Vbox7IE
 from .dbtv import DBTVIE
+from .piksel import PikselIE
 
 
 class GenericIE(InfoExtractor):
@@ -343,10 +344,10 @@ class GenericIE(InfoExtractor):
             },
             'skip': 'There is a limit of 200 free downloads / month for the test song',
         },
-        # embedded brightcove video
-        # it also tests brightcove videos that need to set the 'Referer' in the
-        # http requests
         {
+            # embedded brightcove video
+            # it also tests brightcove videos that need to set the 'Referer'
+            # in the http requests
             'add_ie': ['BrightcoveLegacy'],
             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
             'info_dict': {
@@ -360,6 +361,24 @@ class GenericIE(InfoExtractor):
                 'skip_download': True,
             },
         },
+        {
+            # embedded with itemprop embedURL and video id spelled as `idVideo`
+            'add_id': ['BrightcoveLegacy'],
+            'url': 'http://bfmbusiness.bfmtv.com/mediaplayer/chroniques/olivier-delamarche/',
+            'info_dict': {
+                'id': '5255628253001',
+                'ext': 'mp4',
+                'title': 'md5:37c519b1128915607601e75a87995fc0',
+                'description': 'md5:37f7f888b434bb8f8cc8dbd4f7a4cf26',
+                'uploader': 'BFM BUSINESS',
+                'uploader_id': '876450612001',
+                'timestamp': 1482255315,
+                'upload_date': '20161220',
+            },
+            'params': {
+                'skip_download': True,
+            },
+        },
         {
             # https://github.com/rg3/youtube-dl/issues/2253
             'url': 'http://bcove.me/i6nfkrc3',
@@ -972,6 +991,20 @@ class GenericIE(InfoExtractor):
                 'skip_download': True,
             }
         },
+        {
+            # Kaltura embedded, some fileExt broken (#11480)
+            'url': 'http://www.cornell.edu/video/nima-arkani-hamed-standard-models-of-particle-physics',
+            'info_dict': {
+                'id': '1_sgtvehim',
+                'ext': 'mp4',
+                'title': 'Our "Standard Models" of particle physics and cosmology',
+                'description': 'md5:67ea74807b8c4fea92a6f38d6d323861',
+                'timestamp': 1321158993,
+                'upload_date': '20111113',
+                'uploader_id': 'kps1',
+            },
+            'add_ie': ['Kaltura'],
+        },
         # Eagle.Platform embed (generic URL)
         {
             'url': 'http://lenta.ru/news/2015/03/06/navalny/',
@@ -1189,16 +1222,6 @@ class GenericIE(InfoExtractor):
                 'duration': 248.667,
             },
         },
-        # ScreenwaveMedia embed
-        {
-            'url': 'http://www.thecinemasnob.com/the-cinema-snob/a-nightmare-on-elm-street-2-freddys-revenge1',
-            'md5': '24ace5baba0d35d55c6810b51f34e9e0',
-            'info_dict': {
-                'id': 'cinemasnob-55d26273809dd',
-                'ext': 'mp4',
-                'title': 'cinemasnob',
-            },
-        },
         # BrightcoveInPageEmbed embed
         {
             'url': 'http://www.geekandsundry.com/tabletop-bonus-wils-final-thoughts-on-dread/',
@@ -2140,6 +2163,11 @@ class GenericIE(InfoExtractor):
         if videomore_url:
             return self.url_result(videomore_url)
 
+        # Look for Webcaster embeds
+        webcaster_url = WebcasterFeedIE._extract_url(self, webpage)
+        if webcaster_url:
+            return self.url_result(webcaster_url, ie=WebcasterFeedIE.ie_key())
+
         # Look for Playwire embeds
         mobj = re.search(
             r'<script[^>]+data-config=(["\'])(?P<url>(?:https?:)?//config\.playwire\.com/.+?)\1', webpage)
@@ -2206,11 +2234,6 @@ class GenericIE(InfoExtractor):
         if jwplatform_url:
             return self.url_result(jwplatform_url, 'JWPlatform')
 
-        # Look for ScreenwaveMedia embeds
-        mobj = re.search(ScreenwaveMediaIE.EMBED_PATTERN, webpage)
-        if mobj is not None:
-            return self.url_result(unescapeHTML(mobj.group('url')), 'ScreenwaveMedia')
-
         # Look for Digiteka embeds
         digiteka_url = DigitekaIE._extract_url(webpage)
         if digiteka_url:
@@ -2221,6 +2244,11 @@ class GenericIE(InfoExtractor):
         if arkena_url:
             return self.url_result(arkena_url, ArkenaIE.ie_key())
 
+        # Look for Piksel embeds
+        piksel_url = PikselIE._extract_url(webpage)
+        if piksel_url:
+            return self.url_result(piksel_url, PikselIE.ie_key())
+
         # Look for Limelight embeds
         mobj = re.search(r'LimelightPlayer\.doLoad(Media|Channel|ChannelList)\(["\'](?P<id>[a-z0-9]{32})', webpage)
         if mobj: