feat: ks comment api upgrade to v2

media_platform/kuaishou/client.py

@@ -54,6 +54,7 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
         self.timeout = timeout
         self.headers = headers
         self._host = "https://www.kuaishou.com/graphql"
+        self._rest_host = "https://www.kuaishou.com"
         self.playwright_page = playwright_page
         self.cookie_dict = cookie_dict
         self.graphql = KuaiShouGraphQL()
@@ -86,6 +87,29 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
             method="POST", url=f"{self._host}{uri}", data=json_str, headers=self.headers
         )
 
+    async def request_rest_v2(self, uri: str, data: dict) -> Dict:
+        """
+        Make REST API V2 request (for comment endpoints)
+        :param uri: API endpoint path
+        :param data: request body
+        :return: response data
+        """
+        await self._refresh_proxy_if_expired()
+
+        json_str = json.dumps(data, separators=(",", ":"), ensure_ascii=False)
+        async with httpx.AsyncClient(proxy=self.proxy) as client:
+            response = await client.request(
+                method="POST",
+                url=f"{self._rest_host}{uri}",
+                data=json_str,
+                timeout=self.timeout,
+                headers=self.headers,
+            )
+        result: Dict = response.json()
+        if result.get("result") != 1:
+            raise DataFetchError(f"REST API V2 error: {result}")
+        return result
+
     async def pong(self) -> bool:
         """get a note to check if login state is ok"""
         utils.logger.info("[KuaiShouClient.pong] Begin pong kuaishou...")
@@ -149,36 +173,32 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
         return await self.post("", post_data)
 
     async def get_video_comments(self, photo_id: str, pcursor: str = "") -> Dict:
-        """get video comments
+        """Get video first-level comments using REST API V2
-        :param photo_id: photo id you want to fetch
+        :param photo_id: video id you want to fetch
-        :param pcursor: last you get pcursor, defaults to ""
+        :param pcursor: pagination cursor, defaults to ""
-        :return:
+        :return: dict with rootCommentsV2, pcursorV2, commentCountV2
         """
         post_data = {
-            "operationName": "commentListQuery",
+            "photoId": photo_id,
-            "variables": {"photoId": photo_id, "pcursor": pcursor},
+            "pcursor": pcursor,
-            "query": self.graphql.get("comment_list"),
         }
-        return await self.post("", post_data)
+        return await self.request_rest_v2("/rest/v/photo/comment/list", post_data)
 
     async def get_video_sub_comments(
-        self, photo_id: str, rootCommentId: str, pcursor: str = ""
+        self, photo_id: str, root_comment_id: int, pcursor: str = ""
     ) -> Dict:
-        """get video sub comments
+        """Get video second-level comments using REST API V2
-        :param photo_id: photo id you want to fetch
+        :param photo_id: video id you want to fetch
-        :param pcursor: last you get pcursor, defaults to ""
+        :param root_comment_id: parent comment id (must be int type)
-        :return:
+        :param pcursor: pagination cursor, defaults to ""
+        :return: dict with subCommentsV2, pcursorV2
         """
         post_data = {
-            "operationName": "visionSubCommentList",
+            "photoId": photo_id,
-            "variables": {
+            "pcursor": pcursor,
-                "photoId": photo_id,
+            "rootCommentId": root_comment_id,  # Must be int type for V2 API
-                "pcursor": pcursor,
-                "rootCommentId": rootCommentId,
-            },
-            "query": self.graphql.get("vision_sub_comment_list"),
         }
-        return await self.post("", post_data)
+        return await self.request_rest_v2("/rest/v/photo/comment/sublist", post_data)
 
     async def get_creator_profile(self, userId: str) -> Dict:
         post_data = {
@@ -204,12 +224,12 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
         max_count: int = 10,
     ):
         """
-        get video all comments include sub comments
+        Get video all comments including sub comments (V2 REST API)
-        :param photo_id:
+        :param photo_id: video id
-        :param crawl_interval:
+        :param crawl_interval: delay between requests (seconds)
-        :param callback:
+        :param callback: callback function for processing comments
-        :param max_count:
+        :param max_count: max number of comments to fetch
-        :return:
+        :return: list of all comments
         """
 
         result = []
@@ -217,9 +237,9 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
 
         while pcursor != "no_more" and len(result) < max_count:
             comments_res = await self.get_video_comments(photo_id, pcursor)
-            vision_commen_list = comments_res.get("visionCommentList", {})
+            # V2 API returns data at top level, not nested in visionCommentList
-            pcursor = vision_commen_list.get("pcursor", "")
+            pcursor = comments_res.get("pcursorV2", "no_more")
-            comments = vision_commen_list.get("rootComments", [])
+            comments = comments_res.get("rootCommentsV2", [])
             if len(result) + len(comments) > max_count:
                 comments = comments[: max_count - len(result)]
             if callback:  # If there is a callback function, execute the callback function
@@ -240,14 +260,14 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
         callback: Optional[Callable] = None,
     ) -> List[Dict]:
         """
-        Get all second-level comments under specified first-level comments, this method will continue to find all second-level comment information under first-level comments
+        Get all second-level comments under specified first-level comments (V2 REST API)
         Args:
             comments: Comment list
             photo_id: Video ID
             crawl_interval: Delay unit for crawling comments once (seconds)
             callback: Callback after one comment crawl ends
         Returns:
-
+            List of sub comments
         """
         if not config.ENABLE_GET_SUB_COMMENTS:
             utils.logger.info(
@@ -257,29 +277,30 @@ class KuaiShouClient(AbstractApiClient, ProxyRefreshMixin):
 
         result = []
         for comment in comments:
-            sub_comments = comment.get("subComments")
+            # V2 API uses hasSubComments (boolean) instead of subCommentsPcursor (string)
-            if sub_comments and callback:
+            has_sub_comments = comment.get("hasSubComments", False)
-                await callback(photo_id, sub_comments)
+            if not has_sub_comments:
-
+                continue
-            sub_comment_pcursor = comment.get("subCommentsPcursor")
+
-            if sub_comment_pcursor == "no_more":
+            # V2 API uses comment_id (int) instead of commentId (string)
+            root_comment_id = comment.get("comment_id")
+            if not root_comment_id:
                 continue
 
-            root_comment_id = comment.get("commentId")
             sub_comment_pcursor = ""
 
             while sub_comment_pcursor != "no_more":
                 comments_res = await self.get_video_sub_comments(
                     photo_id, root_comment_id, sub_comment_pcursor
                 )
-                vision_sub_comment_list = comments_res.get("visionSubCommentList", {})
+                # V2 API returns data at top level
-                sub_comment_pcursor = vision_sub_comment_list.get("pcursor", "no_more")
+                sub_comment_pcursor = comments_res.get("pcursorV2", "no_more")
+                sub_comments = comments_res.get("subCommentsV2", [])
 
-                comments = vision_sub_comment_list.get("subComments", {})
+                if callback and sub_comments:
-                if callback:
+                    await callback(photo_id, sub_comments)
-                    await callback(photo_id, comments)
                 await asyncio.sleep(crawl_interval)
-                result.extend(comments)
+                result.extend(sub_comments)
         return result
 
     async def get_creator_info(self, user_id: str) -> Dict:

store/kuaishou/__init__.py

@@ -87,16 +87,22 @@ async def batch_update_ks_video_comments(video_id: str, comments: List[Dict]):
 
 
 async def update_ks_video_comment(video_id: str, comment_item: Dict):
-    comment_id = comment_item.get("commentId")
+    # V2 API uses snake_case field names and comment_id is int type
+    # Old GraphQL API used camelCase field names
+    # Support both formats for backward compatibility
+    comment_id = comment_item.get("comment_id") or comment_item.get("commentId")
     save_comment_item = {
-        "comment_id": comment_id,
+        "comment_id": str(comment_id) if comment_id else None,  # Convert to string for storage
         "create_time": comment_item.get("timestamp"),
         "video_id": video_id,
         "content": comment_item.get("content"),
-        "user_id": comment_item.get("authorId"),
+        # V2: author_id, Old: authorId
-        "nickname": comment_item.get("authorName"),
+        "user_id": comment_item.get("author_id") or comment_item.get("authorId"),
+        # V2: author_name, Old: authorName
+        "nickname": comment_item.get("author_name") or comment_item.get("authorName"),
         "avatar": comment_item.get("headurl"),
-        "sub_comment_count": str(comment_item.get("subCommentCount", 0)),
+        # V2: commentCount, Old: subCommentCount
+        "sub_comment_count": str(comment_item.get("commentCount") or comment_item.get("subCommentCount", 0)),
         "last_modify_ts": utils.get_current_timestamp(),
     }
     utils.logger.info(