[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / sportdeutschland.py
index 185353bef7eb363ad9aa74487d6c384e975560f9..378fc75686313f92a846aaa30579049e9a29eccc 100644 (file)
@@ -5,42 +5,26 @@ import re
 
 from .common import InfoExtractor
 from ..utils import (
-    compat_urllib_request,
     parse_iso8601,
+    sanitized_Request,
 )
 
 
 class SportDeutschlandIE(InfoExtractor):
     _VALID_URL = r'https?://sportdeutschland\.tv/(?P<sport>[^/?#]+)/(?P<id>[^?#/]+)(?:$|[?#])'
     _TESTS = [{
-        'url': 'http://sportdeutschland.tv/badminton/live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
+        'url': 'https://sportdeutschland.tv/badminton/re-live-deutsche-meisterschaften-2020-halbfinals?playlistId=0',
         'info_dict': {
-            'id': 'live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
+            'id': 're-live-deutsche-meisterschaften-2020-halbfinals',
             'ext': 'mp4',
-            'title': 'LIVE: Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen',
-            'categories': ['Badminton'],
+            'title': 're:Re-live: Deutsche Meisterschaften 2020.*Halbfinals',
+            'categories': ['Badminton-Deutschland'],
             'view_count': int,
-            'thumbnail': 're:^https?://.*\.jpg$',
-            'description': 're:^Die Badminton-WM 2014 aus Kopenhagen LIVE',
+            'thumbnail': r're:^https?://.*\.(?:jpg|png)$',
             'timestamp': int,
-            'upload_date': 're:^201408[23][0-9]$',
+            'upload_date': '20200201',
+            'description': 're:.*',  # meaningless description for THIS video
         },
-        'params': {
-            'skip_download': 'Live stream',
-        },
-    }, {
-        'url': 'http://sportdeutschland.tv/li-ning-badminton-wm-2014/lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
-        'info_dict': {
-            'id': 'lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
-            'ext': 'mp4',
-            'upload_date': '20140825',
-            'description': 'md5:60a20536b57cee7d9a4ec005e8687504',
-            'timestamp': 1408976060,
-            'title': 'Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen: Herren Einzel, Wei Lee vs. Keun Lee',
-            'thumbnail': 're:^https?://.*\.jpg$',
-            'view_count': int,
-            'categories': ['Li-Ning Badminton WM 2014'],
-        }
     }]
 
     def _real_extract(self, url):
@@ -48,16 +32,16 @@ class SportDeutschlandIE(InfoExtractor):
         video_id = mobj.group('id')
         sport_id = mobj.group('sport')
 
-        api_url = 'http://splink.tv/api/permalinks/%s/%s' % (
+        api_url = 'https://proxy.vidibusdynamic.net/ssl/backend.sportdeutschland.tv/api/permalinks/%s/%s?access_token=true' % (
             sport_id, video_id)
-        req = compat_urllib_request.Request(api_url, headers={
+        req = sanitized_Request(api_url, headers={
             'Accept': 'application/vnd.vidibus.v2.html+json',
             'Referer': url,
         })
         data = self._download_json(req, video_id)
 
-        categories = list(data.get('section', {}).get('tags', {}).values())
         asset = data['asset']
+        categories = [data['section']['title']]
 
         formats = []
         smil_url = asset['video']
@@ -68,10 +52,12 @@ class SportDeutschlandIE(InfoExtractor):
 
             smil_doc = self._download_xml(
                 smil_url, video_id, note='Downloading SMIL metadata')
-            base_url = smil_doc.find('./head/meta').attrib['base']
+            base_url_el = smil_doc.find('./head/meta')
+            if base_url_el:
+                base_url = base_url_el.attrib['base']
             formats.extend([{
                 'format_id': 'rmtp',
-                'url': base_url,
+                'url': base_url if base_url_el else n.attrib['src'],
                 'play_path': n.attrib['src'],
                 'ext': 'flv',
                 'preference': -100,
@@ -88,9 +74,9 @@ class SportDeutschlandIE(InfoExtractor):
             'title': asset['title'],
             'thumbnail': asset.get('image'),
             'description': asset.get('teaser'),
+            'duration': asset.get('duration'),
             'categories': categories,
             'view_count': asset.get('views'),
             'rtmp_live': asset.get('live'),
             'timestamp': parse_iso8601(asset.get('date')),
         }
-