新增B站创作者（UP主）信息爬取

base/base_crawler.py

@@ -48,7 +48,7 @@ class AbstractStore(ABC):
 
     # TODO support all platform
     # only xhs is supported, so @abstractmethod is commented
-    # @abstractmethod
+    @abstractmethod
     async def store_creator(self, creator: Dict):
         pass
 

media_platform/bilibili/core.py

@@ -127,6 +127,7 @@ class BilibiliCrawler(AbstractCrawler):
                     if video_item:
                         video_id_list.append(video_item.get("View").get("aid"))
                         await bilibili_store.update_bilibili_video(video_item)
+                        await bilibili_store.update_up_info(video_item)
                         await self.get_bilibili_video(video_item, semaphore)
                 page += 1
                 await self.batch_get_video_comments(video_id_list)

schema/tables.sql

@@ -46,6 +46,25 @@ CREATE TABLE `bilibili_video_comment` (
   KEY `idx_bilibili_vi_video_i_f22873` (`video_id`)
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci COMMENT='B 站视频评论';
 
+-- ----------------------------
+-- Table structure for bilibili_up_info
+-- ----------------------------
+DROP TABLE IF EXISTS `bilibili_up_info`;
+CREATE TABLE `bilibili_up_info` (
+  `id` int NOT NULL AUTO_INCREMENT COMMENT '自增ID',
+  `user_id` varchar(64) DEFAULT NULL COMMENT '用户ID',
+  `nickname` varchar(64) DEFAULT NULL COMMENT '用户昵称',
+  `avatar` varchar(255) DEFAULT NULL COMMENT '用户头像地址',
+  `add_ts` bigint NOT NULL COMMENT '记录添加时间戳',
+  `last_modify_ts` bigint NOT NULL COMMENT '记录最后修改时间戳',
+  `total_fans` bigint DEFAULT NULL COMMENT '粉丝数',
+  `total_liked` bigint DEFAULT NULL COMMENT '总获赞数',
+  `user_rank` int DEFAULT NULL COMMENT '用户等级',
+  `is_official` int DEFAULT NULL COMMENT '是否官号',
+  PRIMARY KEY (`id`),
+  KEY `idx_bilibili_vi_user_123456` (`user_id`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci COMMENT='B 站UP主信息';
+
 -- ----------------------------
 -- Table structure for douyin_aweme
 -- ----------------------------

store/bilibili/__init__.py

@@ -53,6 +53,24 @@ async def update_bilibili_video(video_item: Dict):
     await BiliStoreFactory.create_store().store_content(content_item=save_content_item)
 
 
+async def update_up_info(video_item: Dict):  
+    video_item_card_list: Dict = video_item.get("Card")
+    video_item_card: Dict = video_item_card_list.get("card") 
+    saver_up_info = {
+        "user_id": str(video_item_card.get("mid")), 
+        "nickname": video_item_card.get("name"),  
+        "avatar": video_item_card.get("face"), 
+        "last_modify_ts": utils.get_current_timestamp(),  
+        "total_fans": video_item_card.get("fans"), 
+        "total_liked": video_item_card_list.get("like_num"), 
+        "user_rank": video_item_card.get("level_info").get("current_level"),  
+        "is_official": video_item_card.get("official_verify").get("type"), 
+    }
+    utils.logger.info(
+        f"[store.bilibili.update_up_info] bilibili user_id:{video_item_card.get('mid')}")
+    await BiliStoreFactory.create_store().store_creator(creator=saver_up_info)
+    
+
 async def batch_update_bilibili_video_comments(video_id: str, comments: List[Dict]):
     if not comments:
         return

store/bilibili/bilibili_store_impl.py

@@ -85,6 +85,17 @@ class BiliCsvStoreImplement(AbstractStore):
         """
         await self.save_data_to_csv(save_item=comment_item, store_type="comments")
 
+    async def store_creator(self, creator: Dict):
+        """
+        Bilibili creator CSV storage implementation
+        Args:
+            creator: creator item dict
+
+        Returns:
+
+        """
+        await self.save_data_to_csv(save_item=creator, store_type="creators")
+
 
 class BiliDbStoreImplement(AbstractStore):
     async def store_content(self, content_item: Dict):
@@ -129,6 +140,27 @@ class BiliDbStoreImplement(AbstractStore):
         else:
             await update_comment_by_comment_id(comment_id, comment_item=comment_item)
 
+    async def store_creator(self, creator: Dict):
+        """
+        Bilibili creator DB storage implementation
+        Args:
+            creator: creator item dict
+
+        Returns:
+
+        """
+
+        from .bilibili_store_sql import (add_new_creator,
+                                         query_creator_by_creator_id,
+                                         update_creator_by_creator_id)
+        creator_id = creator.get("user_id")
+        creator_detail: Dict = await query_creator_by_creator_id(creator_id=creator_id)
+        if not creator_detail:
+            creator["add_ts"] = utils.get_current_timestamp()
+            await add_new_creator(creator)
+        else:
+            await update_creator_by_creator_id(creator_id,creator_item=creator)
+
 
 class BiliJsonStoreImplement(AbstractStore):
     json_store_path: str = "data/bilibili/json"
@@ -204,3 +236,14 @@ class BiliJsonStoreImplement(AbstractStore):
 
         """
         await self.save_data_to_json(comment_item, "comments")
+
+    async def store_creator(self, creator: Dict):
+        """
+        creator JSON storage implementatio
+        Args:
+            creator:
+
+        Returns:
+
+        """
+        await self.save_data_to_json(creator, "creators")

store/bilibili/bilibili_store_sql.py

@@ -100,3 +100,50 @@ async def update_comment_by_comment_id(comment_id: str, comment_item: Dict) -> i
     async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
     effect_row: int = await async_db_conn.update_table("bilibili_video_comment", comment_item, "comment_id", comment_id)
     return effect_row
+
+
+async def query_creator_by_creator_id(creator_id: str) -> Dict:
+    """
+    查询up主信息
+    Args:
+        creator_id:
+
+    Returns:
+
+    """
+    async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
+    sql: str = f"select * from bilibili_up_info where user_id = '{creator_id}'"
+    rows: List[Dict] = await async_db_conn.query(sql)
+    if len(rows) > 0:
+        return rows[0]
+    return dict()
+
+
+async def add_new_creator(creator_item: Dict) -> int:
+    """
+    新增up主信息
+    Args:
+        creator_item:
+
+    Returns:
+
+    """
+    async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
+    last_row_id: int = await async_db_conn.item_to_table("bilibili_up_info", creator_item)
+    return last_row_id
+
+
+async def update_creator_by_creator_id(creator_id: str, creator_item: Dict) -> int:
+    """
+    更新up主信息
+    Args:
+        creator_id:
+        creator_item:
+
+    Returns:
+
+    """
+    async_db_conn: AsyncMysqlDB = media_crawler_db_var.get()
+    effect_row: int = await async_db_conn.update_table("bilibili_up_info", creator_item, "user_id", creator_id)
+    return effect_row
+