yt_dlp/extractor/tudou.py

@@ -4,10 +4,11 @@ import re
 
 
 class TudouIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:play\.)?tudou\.com/v_show/(?P<id>id_[\w=]+).html[\w\W]+'
+    _VALID_URL = r'https?://(?:play\.)?tudou\.com/v_show/(?P<id>id_[\w=.]+)'
     _TESTS = [{
-        'url': 'https://play.tudou.com/v_show/id_XNjAxNjI2OTU3Ng.html?',
+        'url': 'https://play.tudou.com/v_show/id_XNjAxNjI2OTU3Ng',
         'md5': 'to be updated',
+
         'info_dict': {
             'id': 'XNjAxNjI2OTU3Ng==',
             'ext': 'mp4',
@@ -34,19 +35,23 @@ class TudouIE(InfoExtractor):
         cna = re.findall(re_cna, urlh.headers['set-cookie'])[0].replace("; expires", "")
         ts1 = time.time() * 1000
 
+        # About params1{} (sometimes they call it querystring?)
+        # IS there a way to know which value is essential to make the request work?
         params1 = {
-            'vid': video_id,
+            "vid": video_id,
-            'ccode': '050F',
+            "ccode": "050F",
-            'client_ip': '192.168.1.1',
+            "client_ip": "192.168.1.1",
-            'utid': cna,
+            "utid": cna,
-            'client_ts': ts1,
+            "client_ts": ts1,
-            'ckey': '140#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',
+            "ckey": "140#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",
-            'vip': '0',
+            "vip": "0",
+            # "callback": "",  # youkuPlayer_call_1695807088862
+            # "_t": "08847696931775378"
         }
 
         headers1 = {
-            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/117.0',
+            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/117.0",
-            'Referer': 'https://play.tudou.com/'
+            "Referer": "https://play.tudou.com/"
         }
 
         # About video_url
@@ -54,27 +59,28 @@ class TudouIE(InfoExtractor):
         # This get.json file contains 4 cdn_url and 4 m3u8_url
         # I'm not sure what cdn_url stands for, but it also delivers the full video
         # I'll put the cdn_url here for now, if needed I can replace it with m3u8_url
+        # I probably need to do a bit more on stuff like cookies, to pass the authentication, I wonder how those values are calculated
         data_m3u8 = self._download_json(
             'https://ups.youku.com/ups/get.json', video_id,
             'Downloading JSON metadata',
             query=params1, headers=headers1)
 
-        list_of_video_data = []
+        list_of_url = []
         for item in data_m3u8['data']['stream']:
-            item_of_video_data = {
+            item_of_url = {
-                'url': item['segs'][0]['cdn_url'].replace('http://', 'https://'),
-                'ext': 'mp4',
-                'filesize': item['size'],
                 'width': item['width'],
                 'height': item['height'],
+                'size': item['size'],
+                'cdn_url': item['segs'][0]['cdn_url'].replace('http://', 'https://'),
             }
-            list_of_video_data.append(item_of_video_data)
+            list_of_url.append(item_of_url)
-        sorted_list_of_video_data = sorted(list_of_video_data, key=lambda d: d['filesize'], reverse=False)
+        # Sort urls in descending order, according to size, so the first one in the list will always be the best quality one
+        sorted_list_of_url = sorted(list_of_url, key=lambda d: d['size'], reverse=True)
 
         return {
             'id': video_id,
             'title': title,
-            'show_name': show_name,
             'ext': 'mp4',
-            'formats': sorted_list_of_video_data,
+            'url': sorted_list_of_url[0]['cdn_url'],
+            'show_name': show_name,
         }