Add `--force-generic-extractor`

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

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 8f2e530632faa10adf2aa49decc8ccf0e3c6c3be..3d672197c2c046a6291eeecbfa3bf27e3fc02c77 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -34,6 +34,7 @@ from .brightcove import BrightcoveIE
 from .nbc import NBCSportsVPlayerIE
 from .ooyala import OoyalaIE
 from .rutv import RUTVIE
+from .tvc import TVCIE
 from .sportbox import SportBoxEmbedIE
 from .smotri import SmotriIE
 from .condenast import CondeNastIE
@@ -59,7 +60,7 @@ class GenericIE(InfoExtractor):
                 'upload_date': '20100513',
             }
         },
-        # Direct link to a media delivered compressed (requires Accept-Encoding == *)
+        # Direct link to media delivered compressed (until Accept-Encoding is *)
         {
             'url': 'http://calimero.tk/muzik/FictionJunction-Parallel_Hearts.flac',
             'md5': '128c42e68b13950268b648275386fc74',
@@ -291,6 +292,15 @@ class GenericIE(InfoExtractor):
                 'skip_download': True,
             },
         },
+        # TVC embed
+        {
+            'url': 'http://sch1298sz.mskobr.ru/dou_edu/karamel_ki/filial_galleries/video/iframe_src_http_tvc_ru_video_iframe_id_55304_isplay_false_acc_video_id_channel_brand_id_11_show_episodes_episode_id_32307_frameb/',
+            'info_dict': {
+                'id': '55304',
+                'ext': 'mp4',
+                'title': 'Дошкольное воспитание',
+            },
+        },
         # SportBox embed
         {
             'url': 'http://www.vestifinance.ru/articles/25753',
@@ -789,6 +799,18 @@ class GenericIE(InfoExtractor):
                 # rtmpe downloads
                 'skip_download': True,
             }
+        },
+        # Brightcove URL in single quotes
+        {
+            'url': 'http://www.sportsnet.ca/baseball/mlb/sn-presents-russell-martin-world-citizen/',
+            'md5': '4ae374f1f8b91c889c4b9203c8c752af',
+            'info_dict': {
+                'id': '4255764656001',
+                'ext': 'mp4',
+                'title': 'SN Presents: Russell Martin, World Citizen',
+                'description': 'To understand why he was the Toronto Blue Jays’ top off-season priority is to appreciate his background and upbringing in Montreal, where he first developed his baseball skills. Written and narrated by Stephen Brunt.',
+                'uploader': 'Rogers Sportsnet',
+            },
         }
     ]
 
@@ -955,7 +977,9 @@ class GenericIE(InfoExtractor):
                 'upload_date': upload_date,
             }
 
-        if not self._downloader.params.get('test', False) and not is_intentional:
+        if (not self._downloader.params.get('test', False) and
+                not is_intentional and
+                not self._downloader.params.get('force_generic_extractor', False)):
             self._downloader.report_warning('Falling back on generic information extractor.')
 
         if not full_response:
@@ -1061,7 +1085,7 @@ class GenericIE(InfoExtractor):
 
         # Look for embedded rtl.nl player
         matches = re.findall(
-            r'<iframe\s+(?:[a-zA-Z-]+="[^"]+"\s+)*?src="((?:https?:)?//(?:www\.)?rtl\.nl/system/videoplayer/[^"]+video_embed[^"]+)"',
+            r'<iframe[^>]+?src="((?:https?:)?//(?:www\.)?rtl\.nl/system/videoplayer/[^"]+(?:video_)?embed[^"]+)"',
             webpage)
         if matches:
             return _playlist_from_matches(matches, ie='RtlNl')
@@ -1289,11 +1313,22 @@ class GenericIE(InfoExtractor):
         if rutv_url:
             return self.url_result(rutv_url, 'RUTV')
 
+        # Look for embedded TVC player
+        tvc_url = TVCIE._extract_url(webpage)
+        if tvc_url:
+            return self.url_result(tvc_url, 'TVC')
+
         # Look for embedded SportBox player
         sportbox_urls = SportBoxEmbedIE._extract_urls(webpage)
         if sportbox_urls:
             return _playlist_from_matches(sportbox_urls, ie='SportBoxEmbed')
 
+        # Look for embedded Tvigle player
+        mobj = re.search(
+            r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//cloud\.tvigle\.ru/video/.+?)\1', webpage)
+        if mobj is not None:
+            return self.url_result(mobj.group('url'), 'Tvigle')
+
         # Look for embedded TED player
         mobj = re.search(
             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed(?:-ssl)?\.ted\.com/.+?)\1', webpage)