[mangomolo] add support for Mangomolo embeds

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

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 197ab95319419d29971979b94464aa191ef04a04..e0130594240f2dd6a73181f7066ec79f752af1fa 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -73,6 +73,7 @@ from .eagleplatform import EaglePlatformIE
 from .facebook import FacebookIE
 from .soundcloud import SoundcloudIE
 from .vbox7 import Vbox7IE
+from .dbtv import DBTVIE
 
 
 class GenericIE(InfoExtractor):
@@ -103,7 +104,8 @@ class GenericIE(InfoExtractor):
             },
             'expected_warnings': [
                 'URL could be a direct video link, returning it as such.'
-            ]
+            ],
+            'skip': 'URL invalid',
         },
         # Direct download with broken HEAD
         {
@@ -267,7 +269,8 @@ class GenericIE(InfoExtractor):
             'params': {
                 # m3u8 downloads
                 'skip_download': True,
-            }
+            },
+            'skip': 'video gone',
         },
         # m3u8 served with Content-Type: text/plain
         {
@@ -282,7 +285,8 @@ class GenericIE(InfoExtractor):
             'params': {
                 # m3u8 downloads
                 'skip_download': True,
-            }
+            },
+            'skip': 'video gone',
         },
         # google redirect
         {
@@ -367,6 +371,7 @@ class GenericIE(InfoExtractor):
                 'description': 'Mississauga resident David Farmer is still out of power as a result of the ice storm a month ago. To keep the house warm, Farmer cuts wood from his property for a wood burning stove downstairs.',
             },
             'add_ie': ['BrightcoveLegacy'],
+            'skip': 'video gone',
         },
         {
             'url': 'http://www.championat.com/video/football/v/87/87499.html',
@@ -420,6 +425,7 @@ class GenericIE(InfoExtractor):
             'params': {
                 'skip_download': True,
             },
+            'skip': 'movie expired',
         },
         # embed.ly video
         {
@@ -447,6 +453,8 @@ class GenericIE(InfoExtractor):
                 'title': 'Between Two Ferns with Zach Galifianakis: President Barack Obama',
                 'description': 'Episode 18: President Barack Obama sits down with Zach Galifianakis for his most memorable interview yet.',
             },
+            # HEAD requests lead to endless 301, while GET is OK
+            'expected_warnings': ['301'],
         },
         # RUTV embed
         {
@@ -521,6 +529,9 @@ class GenericIE(InfoExtractor):
                 'title': '[NSFL] [FM15] which pumiscer was this ( vid ) ( alfa as fuck srx )',
             },
             'playlist_mincount': 7,
+            # This forum does not allow <iframe> syntaxes anymore
+            # Now HTML tags are displayed as-is
+            'skip': 'No videos on this page',
         },
         # Embedded TED video
         {
@@ -569,7 +580,8 @@ class GenericIE(InfoExtractor):
             },
             'params': {
                 'skip_download': 'Requires rtmpdump'
-            }
+            },
+            'skip': 'video gone',
         },
         # francetv embed
         {
@@ -1357,6 +1369,11 @@ class GenericIE(InfoExtractor):
             },
             'add_ie': ['Vimeo'],
         },
+        {
+            # generic vimeo embed that requires original URL passed as Referer
+            'url': 'http://racing4everyone.eu/2016/07/30/formula-1-2016-round12-germany/',
+            'only_matching': True,
+        },
         {
             'url': 'https://support.arkena.com/display/PLAY/Ways+to+embed+your+video',
             'md5': 'b96f2f71b359a8ecd05ce4e1daa72365',
@@ -1386,6 +1403,15 @@ class GenericIE(InfoExtractor):
             },
             'add_ie': [Vbox7IE.ie_key()],
         },
+        {
+            # DBTV embeds
+            'url': 'http://www.dagbladet.no/2016/02/23/nyheter/nordlys/ski/troms/ver/43254897/',
+            'info_dict': {
+                'id': '43254897',
+                'title': 'Etter ett års planlegging, klaffet endelig alt: - Jeg måtte ta en liten dans',
+            },
+            'playlist_mincount': 3,
+        },
         # {
         #     # TODO: find another test
         #     # http://schema.org/VideoObject
@@ -2222,11 +2248,40 @@ class GenericIE(InfoExtractor):
 
         # Look for VODPlatform embeds
         mobj = re.search(
-            r'<iframe[^>]+src=[\'"]((?:https?:)?//(?:www\.)?vod-platform\.net/embed/[^/?#]+)',
+            r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?vod-platform\.net/[eE]mbed/.+?)\1',
             webpage)
         if mobj is not None:
             return self.url_result(
-                self._proto_relative_url(unescapeHTML(mobj.group(1))), 'VODPlatform')
+                self._proto_relative_url(unescapeHTML(mobj.group('url'))), 'VODPlatform')
+
+        # Look for Mangomolo embeds
+        mobj = re.search(
+            r'''(?x)<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?admin\.mangomolo.com/analytics/index\.php/customers/embed/
+                (?:
+                    video\?.*?\bid=(?P<video_id>\d+)|
+                    index\?.*?\bchannelid=(?P<channel_id>(?:[A-Za-z0-9+/=]|%2B|%2F|%3D)+)
+                ).+?)\1''', webpage)
+        if mobj is not None:
+            info = {
+                '_type': 'url_transparent',
+                'url': self._proto_relative_url(unescapeHTML(mobj.group('url'))),
+                'title': video_title,
+                'description': video_description,
+                'thumbnail': video_thumbnail,
+                'uploader': video_uploader,
+            }
+            video_id = mobj.group('video_id')
+            if video_id:
+                info.update({
+                    'ie_key': 'MangomoloVideo',
+                    'id': video_id,
+                })
+            else:
+                info.update({
+                    'ie_key': 'MangomoloLive',
+                    'id': mobj.group('channel_id'),
+                })
+            return info
 
         # Look for Instagram embeds
         instagram_embed_url = InstagramIE._extract_embed_url(webpage)
@@ -2257,6 +2312,11 @@ class GenericIE(InfoExtractor):
         if vbox7_url:
             return self.url_result(vbox7_url, Vbox7IE.ie_key())
 
+        # Look for DBTV embeds
+        dbtv_urls = DBTVIE._extract_urls(webpage)
+        if dbtv_urls:
+            return _playlist_from_matches(dbtv_urls, ie=DBTVIE.ie_key())
+
         # Looking for http://schema.org/VideoObject
         json_ld = self._search_json_ld(
             webpage, video_id, default={}, expected_type='VideoObject')