]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/mgtv.py
debian/patches: Drop patch included in version 2017.02.07.
[youtubedl] / youtube_dl / extractor / mgtv.py
index 9fbc74f5d52e819ca3213623a30b6d9cd2e6ec08..659ede8c2254d6ce524953c298c8a69b0b13d745 100644 (file)
@@ -6,10 +6,10 @@ from ..utils import int_or_none
 
 
 class MGTVIE(InfoExtractor):
-    _VALID_URL = r'https?://www\.mgtv\.com/v/(?:[^/]+/)*(?P<id>\d+)\.html'
+    _VALID_URL = r'https?://(?:www\.)?mgtv\.com/v/(?:[^/]+/)*(?P<id>\d+)\.html'
     IE_DESC = '芒果TV'
 
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.mgtv.com/v/1/290525/f/3116640.html',
         'md5': '1bdadcf760a0b90946ca68ee9a2db41a',
         'info_dict': {
@@ -18,15 +18,20 @@ class MGTVIE(InfoExtractor):
             'title': '我是歌手第四季双年巅峰会:韩红李玟“双王”领军对抗',
             'description': '我是歌手第四季双年巅峰会',
             'duration': 7461,
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'thumbnail': r're:^https?://.*\.jpg$',
         },
-    }
+    }, {
+        # no tbr extracted from stream_url
+        'url': 'http://www.mgtv.com/v/1/1/f/3324755.html',
+        'only_matching': True,
+    }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
         api_data = self._download_json(
             'http://v.api.mgtv.com/player/video', video_id,
-            query={'video_id': video_id})['data']
+            query={'video_id': video_id},
+            headers=self.geo_verification_headers())['data']
         info = api_data['info']
 
         formats = []
@@ -40,7 +45,8 @@ class MGTVIE(InfoExtractor):
             def extract_format(stream_url, format_id, idx, query={}):
                 format_info = self._download_json(
                     stream_url, video_id,
-                    note='Download video info for format %s' % format_id or '#%d' % idx, query=query)
+                    note='Download video info for format %s' % (format_id or '#%d' % idx),
+                    query=query)
                 return {
                     'format_id': format_id,
                     'url': format_info['info'],