[mgtv] extract http formats

youtube_dl/extractor/mgtv.py

@@ -11,7 +11,7 @@ class MGTVIE(InfoExtractor):
 
     _TEST = {
         'url': 'http://www.mgtv.com/v/1/290525/f/3116640.html',
-        'md5': '',
+        'md5': '1bdadcf760a0b90946ca68ee9a2db41a',
         'info_dict': {
             'id': '3116640',
             'ext': 'mp4',
@@ -20,15 +20,6 @@ class MGTVIE(InfoExtractor):
             'duration': 7461,
             'thumbnail': 're:^https?://.*\.jpg$',
         },
-        'params': {
-            'skip_download': True,  # m3u8 download
-        },
-    }
-
-    _FORMAT_MAP = {
-        '标清': ('Standard', 0),
-        '高清': ('High', 1),
-        '超清': ('SuperHigh', 2),
     }
 
     def _real_extract(self, url):
@@ -40,17 +31,27 @@ def _real_extract(self, url):
 
         formats = []
         for idx, stream in enumerate(api_data['stream']):
-            format_name = stream.get('name')
+            stream_url = stream.get('url')
-            format_id, preference = self._FORMAT_MAP.get(format_name, (None, None))
+            if not stream_url:
+                continue
+            tbr = int_or_none(self._search_regex(
+                r'(\d+)\.mp4', stream_url, 'tbr', default=None))
+
+            def extract_format(stream_url, format_id, idx, query={}):
                 format_info = self._download_json(
-                stream['url'], video_id,
+                    stream_url, video_id,
-                note='Download video info for format %s' % format_id or '#%d' % idx)
+                    note='Download video info for format %s' % format_id or '#%d' % idx, query=query)
-            formats.append({
+                return {
                     'format_id': format_id,
                     'url': format_info['info'],
-                'ext': 'mp4',  # These are m3u8 playlists
+                    'ext': 'mp4',
-                'preference': preference,
+                    'tbr': tbr,
-            })
+                }
+
+            formats.append(extract_format(
+                stream_url, 'hls-%d' % tbr if tbr else None, idx * 2))
+            formats.append(extract_format(stream_url.replace(
+                '/playlist.m3u8', ''), 'http-%d' % tbr if tbr else None, idx * 2 + 1, {'pno': 1031}))
         self._sort_formats(formats)
 
         return {