Merge pull request #8130 from dyn888/master

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

diff --git a/youtube_dl/extractor/cbsnews.py b/youtube_dl/extractor/cbsnews.py
index 7e47960ab08f3ffba7a1e596b34c2fbfc7fd6c59..480435e26114cbdf4be2bf177a2dbc2c8381420e 100644 (file)
--- a/youtube_dl/extractor/cbsnews.py
+++ b/youtube_dl/extractor/cbsnews.py
@@ -5,6 +5,7 @@ import re
 import json
 
 from .common import InfoExtractor
+from ..utils import remove_start
 
 
 class CBSNewsIE(InfoExtractor):
@@ -32,8 +33,13 @@ class CBSNewsIE(InfoExtractor):
                 'id': 'fort-hood-shooting-army-downplays-mental-illness-as-cause-of-attack',
                 'ext': 'flv',
                 'title': 'Fort Hood shooting: Army downplays mental illness as cause of attack',
-                'thumbnail': 'http://cbsnews2.cbsistatic.com/hub/i/r/2014/04/04/0c9fbc66-576b-41ca-8069-02d122060dd2/thumbnail/140x90/6dad7a502f88875ceac38202984b6d58/en-0404-werner-replace-640x360.jpg',
+                'thumbnail': 're:^https?://.*\.jpg$',
                 'duration': 205,
+                'subtitles': {
+                    'en': [{
+                        'ext': 'ttml',
+                    }],
+                },
             },
             'params': {
                 # rtmp download
@@ -62,14 +68,20 @@ class CBSNewsIE(InfoExtractor):
             uri = item.get('media' + format_id + 'URI')
             if not uri:
                 continue
+            uri = remove_start(uri, '{manifest:none}')
             fmt = {
                 'url': uri,
                 'format_id': format_id,
             }
             if uri.startswith('rtmp'):
+                play_path = re.sub(
+                    r'{slistFilePath}', '',
+                    uri.split('<break>')[-1].split('{break}')[-1])
+                play_path = re.sub(
+                    r'{manifest:.+}.*$', '', play_path)
                 fmt.update({
                     'app': 'ondemand?auth=cbs',
-                    'play_path': 'mp4:' + uri.split('<break>')[-1],
+                    'play_path': 'mp4:' + play_path,
                     'player_url': 'http://www.cbsnews.com/[[IMPORT]]/vidtech.cbsinteractive.com/player/3_3_0/CBSI_PLAYER_HD.swf',
                     'page_url': 'http://www.cbsnews.com',
                     'ext': 'flv',
@@ -78,10 +90,18 @@ class CBSNewsIE(InfoExtractor):
                 fmt['ext'] = 'mp4'
             formats.append(fmt)
 
+        subtitles = {}
+        if 'mpxRefId' in video_info:
+            subtitles['en'] = [{
+                'ext': 'ttml',
+                'url': 'http://www.cbsnews.com/videos/captions/%s.adb_xml' % video_info['mpxRefId'],
+            }]
+
         return {
             'id': video_id,
             'title': title,
             'thumbnail': thumbnail,
             'duration': duration,
             'formats': formats,
+            'subtitles': subtitles,
         }