debian/rules: Ignore tests temporarily.

[youtubedl] / youtube_dl / extractor / cbs.py

diff --git a/youtube_dl/extractor/cbs.py b/youtube_dl/extractor/cbs.py
index e43756ec69b1d7f1872e45cc6901b41752ec6ef6..a23173d6f1a9570225242692ee74d68fc061fb3d 100644 (file)
--- a/youtube_dl/extractor/cbs.py
+++ b/youtube_dl/extractor/cbs.py
@@ -1,48 +1,63 @@
 from __future__ import unicode_literals
 
 from __future__ import unicode_literals
 

-import re
+from .theplatform import ThePlatformFeedIE
+from ..utils import (
+    int_or_none,
+    find_xpath_attr,
+)

 
 

-from .common import InfoExtractor

 
 

+class CBSBaseIE(ThePlatformFeedIE):
+    def _parse_smil_subtitles(self, smil, namespace=None, subtitles_lang='en'):
+        closed_caption_e = find_xpath_attr(smil, self._xpath_ns('.//param', namespace), 'name', 'ClosedCaptionURL')
+        return {
+            'en': [{
+                'ext': 'ttml',
+                'url': closed_caption_e.attrib['value'],
+            }]
+        } if closed_caption_e is not None and closed_caption_e.attrib.get('value') else []

 
 

-class CBSIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?cbs\.com/shows/[^/]+/(?:video|artist)/(?P<id>[^/]+)/.*'
+    def _extract_video_info(self, filter_query, video_id):
+        return self._extract_feed_info(
+            'dJ5BDC', 'VxxJg8Ymh8sE', filter_query, video_id, lambda entry: {
+                'series': entry.get('cbs$SeriesTitle'),
+                'season_number': int_or_none(entry.get('cbs$SeasonNumber')),
+                'episode': entry.get('cbs$EpisodeTitle'),
+                'episode_number': int_or_none(entry.get('cbs$EpisodeNumber')),
+            }, {
+                'StreamPack': {
+                    'manifest': 'm3u',
+                }
+            })
+
+
+class CBSIE(CBSBaseIE):
+    _VALID_URL = r'(?:cbs:|https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/video|colbertlateshow\.com/(?:video|podcasts))/)(?P<id>[\w-]+)'

 
     _TESTS = [{
         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
         'info_dict': {
 
     _TESTS = [{
         'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/',
         'info_dict': {

-            'id': '4JUVEwq3wUT7',
-            'ext': 'flv',
+            'id': '_u7W953k6la293J7EPTd9oHkSPs6Xn6_',
+            'display_id': 'connect-chat-feat-garth-brooks',
+            'ext': 'mp4',

             'title': 'Connect Chat feat. Garth Brooks',
             'description': 'Connect with country music singer Garth Brooks, as he chats with fans on Wednesday November 27, 2013. Be sure to tune in to Garth Brooks: Live from Las Vegas, Friday November 29, at 9/8c on CBS!',
             'duration': 1495,
             'title': 'Connect Chat feat. Garth Brooks',
             'description': 'Connect with country music singer Garth Brooks, as he chats with fans on Wednesday November 27, 2013. Be sure to tune in to Garth Brooks: Live from Las Vegas, Friday November 29, at 9/8c on CBS!',
             'duration': 1495,

+            'timestamp': 1385585425,
+            'upload_date': '20131127',
+            'uploader': 'CBSI-NEW',

         },
         },

-        'params': {
-            # rtmp download
-            'skip_download': True,
-        },
+        'expected_warnings': ['Failed to download m3u8 information'],

         '_skip': 'Blocked outside the US',
     }, {
         '_skip': 'Blocked outside the US',
     }, {

-        'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/',
-        'info_dict': {
-            'id': 'WWF_5KqY3PK1',
-            'ext': 'flv',
-            'title': 'Live on Letterman - St. Vincent',
-            'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.',
-            'duration': 3221,
-        },
-        'params': {
-            # rtmp download
-            'skip_download': True,
-        },
-        '_skip': 'Blocked outside the US',
+        'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/',
+        'only_matching': True,

     }]
     }]

+    TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/dJ5BDC/%s?mbr=true'

 
     def _real_extract(self, url):
 
     def _real_extract(self, url):

-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-        webpage = self._download_webpage(url, video_id)
-        real_id = self._search_regex(
-            r"video\.settings\.pid\s*=\s*'([^']+)';",
-            webpage, 'real video ID')
-        return self.url_result('theplatform:%s' % real_id)
+        content_id = self._match_id(url)
+        return self._extract_video_info('byGuid=%s' % content_id, content_id)