refactor: 数据存储重构，分离不同类型的存储实现

2025-11-25 03:15:17 +08:00 · 2024-01-14 22:06:31 +08:00
parent e31aebbdfb
commit 894dabcf63
37 changed files with 1427 additions and 864 deletions
--- a/store/xhs/init.py
+++ b/store/xhs/init.py
@@ -0,0 +1,74 @@
+# -*- coding: utf-8 -*-
+# @Author  : relakkes@gmail.com
+# @Time    : 2024/1/14 17:34
+# @Desc    :
+from typing import List
+
+import config
+
+from . import xhs_store_impl
+from .xhs_store_db_types import *
+from .xhs_store_impl import *
+
+
+class XhsStoreFactory:
+    STORES = {
+        "csv": XhsCsvStoreImplement,
+        "db": XhsDbStoreImplement,
+        "json": XhsJsonStoreImplement
+    }
+
+    @staticmethod
+    def create_store() -> AbstractStore:
+        store_class = XhsStoreFactory.STORES.get(config.SAVE_DATA_OPTION)
+        if not store_class:
+            raise ValueError("[XhsStoreFactory.create_store] Invalid save option only supported csv or db or json ...")
+        return store_class()
+
+
+async def update_xhs_note(note_item: Dict):
+    note_id = note_item.get("note_id")
+    user_info = note_item.get("user", {})
+    interact_info = note_item.get("interact_info", {})
+    image_list: List[Dict] = note_item.get("image_list", [])
+
+    local_db_item = {
+        "note_id": note_item.get("note_id"),
+        "type": note_item.get("type"),
+        "title": note_item.get("title") or note_item.get("desc", "")[:255],
+        "desc": note_item.get("desc", ""),
+        "time": note_item.get("time"),
+        "last_update_time": note_item.get("last_update_time", 0),
+        "user_id": user_info.get("user_id"),
+        "nickname": user_info.get("nickname"),
+        "avatar": user_info.get("avatar"),
+        "liked_count": interact_info.get("liked_count"),
+        "collected_count": interact_info.get("collected_count"),
+        "comment_count": interact_info.get("comment_count"),
+        "share_count": interact_info.get("share_count"),
+        "ip_location": note_item.get("ip_location", ""),
+        "image_list": ','.join([img.get('url', '') for img in image_list]),
+        "last_modify_ts": utils.get_current_timestamp(),
+        "note_url": f"https://www.xiaohongshu.com/explore/{note_id}"
+    }
+    utils.logger.info(f"[store.xhs.update_xhs_note] xhs note: {local_db_item}")
+    await XhsStoreFactory.create_store().store_content(local_db_item)
+
+
+async def update_xhs_note_comment(note_id: str, comment_item: Dict):
+    user_info = comment_item.get("user_info", {})
+    comment_id = comment_item.get("id")
+    local_db_item = {
+        "comment_id": comment_id,
+        "create_time": comment_item.get("create_time"),
+        "ip_location": comment_item.get("ip_location"),
+        "note_id": note_id,
+        "content": comment_item.get("content"),
+        "user_id": user_info.get("user_id"),
+        "nickname": user_info.get("nickname"),
+        "avatar": user_info.get("image"),
+        "sub_comment_count": comment_item.get("sub_comment_count"),
+        "last_modify_ts": utils.get_current_timestamp(),
+    }
+    utils.logger.info(f"[store.xhs.update_xhs_note_comment] xhs note comment:{local_db_item}")
+    await XhsStoreFactory.create_store().store_comment(local_db_item)