Merge pull request #485 from opennaslab/cherry-pick-480-release-0.6

Cherry pick PR(480)/fix: yutto downloader not recognize url from app

kubespider/download_provider/yutto_download_provider/provider.py

@@ -39,7 +39,7 @@ def send_general_task(self, task: Task) -> TypeError:
         data = {'dataSource': task.url, 'path': task.path}
         logging.info('Send general task:%s', json.dumps(data))
 
-        if not task.url.startswith('https://www.bilibili.com'):
+        if not (task.url.startswith('https://www.bilibili.com') or task.url.startswith('https://b23.tv')):
             return TypeError('yutto only support specific resource')
 
         # This downloading tasks is special, other download software could not handle

kubespider/source_provider/bilibili_source_provider/provider.py

@@ -2,6 +2,7 @@
 # Function: download single video link
 # encoding:utf-8
 import logging
+import re
 from urllib.parse import urlparse
 
 from source_provider import provider
@@ -52,7 +53,7 @@ def is_webhook_enable(self) -> bool:
         return self.webhook_enable
 
     def should_handle(self, event: Event) -> bool:
-        parse_url = urlparse(event.source)
+        parse_url = urlparse(re.findall(r"https?://\S+", event.source)[0])
         if parse_url.hostname in ('www.bilibili.com', 'b23.tv'):
             logging.info('%s belongs to BilibiliSourceProvider', event.source)
             return True
@@ -63,7 +64,7 @@ def get_links(self, event: Event) -> list[Resource]:
         event.put_extra_params(self.get_download_param())
 
         return [Resource(
-            url=event.source,
+            url=re.findall(r"https?://\S+", event.source)[0],
             path='',
             link_type=self.get_link_type(),
             file_type=types.FILE_TYPE_VIDEO_MIXED,