[openload] fix extraction(closes #10408)

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

diff --git a/youtube_dl/extractor/rai.py b/youtube_dl/extractor/rai.py
index eb32bbdb60914271a3e8dbdbfed3d7538c2f8fd8..41afbd9afa5472fdbd782db06f392abd587cf570 100644 (file)
--- a/youtube_dl/extractor/rai.py
+++ b/youtube_dl/extractor/rai.py
@@ -20,17 +20,12 @@ class RaiBaseIE(InfoExtractor):
         formats = []
 
         for platform in ('mon', 'flash', 'native'):
-            headers = {}
-            # TODO: rename --cn-verification-proxy
-            cn_verification_proxy = self._downloader.params.get('cn_verification_proxy')
-            if cn_verification_proxy:
-                headers['Ytdl-request-proxy'] = cn_verification_proxy
-
             relinker = self._download_xml(
                 relinker_url, video_id,
                 note='Downloading XML metadata for platform %s' % platform,
                 transform_source=fix_xml_ampersands,
-                query={'output': 45, 'pl': platform}, headers=headers)
+                query={'output': 45, 'pl': platform},
+                headers=self.geo_verification_headers())
 
             media_url = find_xpath_attr(relinker, './url', 'type', 'content').text
             if media_url == 'http://download.rai.it/video_no_available.mp4':
@@ -125,7 +120,7 @@ class RaiTVIE(RaiBaseIE):
                 'description': 'md5:f27c544694cacb46a078db84ec35d2d9',
                 'upload_date': '20140407',
                 'duration': 6160,
-                'thumbnail': 're:^https?://.*\.jpg$',
+                'thumbnail': r're:^https?://.*\.jpg$',
             }
         },
         {
@@ -138,7 +133,7 @@ class RaiTVIE(RaiBaseIE):
                 'title': 'TG PRIMO TEMPO',
                 'upload_date': '20140612',
                 'duration': 1758,
-                'thumbnail': 're:^https?://.*\.jpg$',
+                'thumbnail': r're:^https?://.*\.jpg$',
             },
             'skip': 'Geo-restricted to Italy',
         },
@@ -174,7 +169,7 @@ class RaiTVIE(RaiBaseIE):
                 'description': 'md5:364b604f7db50594678f483353164fb8',
                 'upload_date': '20140923',
                 'duration': 386,
-                'thumbnail': 're:^https?://.*\.jpg$',
+                'thumbnail': r're:^https?://.*\.jpg$',
             }
         },
     ]
@@ -221,6 +216,15 @@ class RaiIE(RaiBaseIE):
                 'upload_date': '20160702',
             },
         },
+        {
+            'url': 'http://www.rainews.it/dl/rainews/live/ContentItem-3156f2f2-dc70-4953-8e2f-70d7489d4ce9.html',
+            # HDS live stream, MD5 is unstable
+            'info_dict': {
+                'id': '3156f2f2-dc70-4953-8e2f-70d7489d4ce9',
+                'ext': 'flv',
+                'title': 'La diretta di Rainews24',
+            },
+        },
     ]
 
     @classmethod
@@ -247,7 +251,7 @@ class RaiIE(RaiBaseIE):
             return self._extract_from_content_id(content_item_id, url)
 
         relinker_url = compat_urlparse.urljoin(url, self._search_regex(
-            r'var\s+videoURL\s*=\s*(?P<q1>[\'"])(?P<url>(https?:)?//mediapolis\.rai\.it/relinker/relinkerServlet\.htm\?cont=\d+)(?P=q1)',
+            r'(?:var\s+videoURL|mediaInfo\.mediaUri)\s*=\s*(?P<q1>[\'"])(?P<url>(https?:)?//mediapolis\.rai\.it/relinker/relinkerServlet\.htm\?cont=\d+)(?P=q1)',
             webpage, 'relinker URL', group='url'))
         formats = self._extract_relinker_formats(relinker_url, video_id)
         self._sort_formats(formats)