[cnbc] Add new extractor(closes #8012)

diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
index 1e4b078a4aa45e6b5ead9e5ad0956959f04d1a6c..bd1f7d293fcfa5cbe34162e4f23e06d81cec6c6a 100644 (file)
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@ -127,6 +127,7 @@ from .cloudy import CloudyIE
 from .clubic import ClubicIE
 from .clyp import ClypIE
 from .cmt import CMTIE
+from .cnbc import CNBCIE
 from .cnet import CNETIE
 from .cnn import (
     CNNIE,

diff --git a/youtube_dl/extractor/cnbc.py b/youtube_dl/extractor/cnbc.py
new file mode 100644 (file)
index 0000000..593e459
--- /dev/null
+++ b/youtube_dl/extractor/cnbc.py
@@ -0,0 +1,29 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+from ..utils import smuggle_url
+
+
+class CNBCIE(InfoExtractor):
+    _VALID_URL = r'https?://video\.cnbc\.com/gallery/\?video=(?P<id>[0-9]+)'
+    _TEST = {
+        'url': 'http://video.cnbc.com/gallery/?video=3000503714',
+        'md5': '',
+        'info_dict': {
+            'id': '3000503714',
+            'ext': 'mp4',
+            'title': 'Video title goes here',
+        }
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        return {
+            '_type': 'url_transparent',
+            'ie_key': 'ThePlatform',
+            'url': smuggle_url(
+                'http://link.theplatform.com/s/gZWlPC/media/guid/2408950221/%s?mbr=true&manifest=m3u' % video_id,
+                {'force_smil_url': True}),
+            'id': video_id,
+        }

diff --git a/youtube_dl/extractor/theplatform.py b/youtube_dl/extractor/theplatform.py
index 863914299234fb0d7dd279da51c0f0d8c18d6534..236c999724ee90f86ecad4bc2b771aefca0bf6f5 100644 (file)
--- a/youtube_dl/extractor/theplatform.py
+++ b/youtube_dl/extractor/theplatform.py
@@ -82,7 +82,7 @@ class ThePlatformBaseIE(OnceIE):
 class ThePlatformIE(ThePlatformBaseIE):
     _VALID_URL = r'''(?x)
         (?:https?://(?:link|player)\.theplatform\.com/[sp]/(?P<provider_id>[^/]+)/
-           (?:(?P<media>(?:(?:[^/]+/)+select/)?media/)|(?P<config>(?:[^/\?]+/(?:swf|config)|onsite)/select/))?
+           (?:(?:(?:[^/]+/)+select/)?(?P<media>media/(?:guid/\d+/)?)|(?P<config>(?:[^/\?]+/(?:swf|config)|onsite)/select/))?
          |theplatform:)(?P<id>[^/\?&]+)'''
 
     _TESTS = [{
@@ -170,10 +170,10 @@ class ThePlatformIE(ThePlatformBaseIE):
         if not provider_id:
             provider_id = 'dJ5BDC'
 
-        path = provider_id
+        path = provider_id + '/'
         if mobj.group('media'):
-            path += '/media'
-        path += '/' + video_id
+            path += mobj.group('media')
+        path += video_id
 
         qs_dict = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
         if 'guid' in qs_dict: