fix: 改用 start_time 增量轮询+分页，彻底修复新消息漏检

image_daemon.py

@@ -58,29 +58,43 @@ def get_bot_chats(token):
         log(f"❌ get_bot_chats 异常: {type(e).__name__}: {e}")
         return []
 
-def get_recent_messages(token, chat_id, limit=50):
-    """获取聊天中最近的消息"""
+def get_recent_messages(token, chat_id, start_time=None, limit=50):
+    """获取聊天中的消息，支持 start_time 增量拉取和分页"""
+    all_items = []
+    page_token = None
     try:
         headers = {"Authorization": f"Bearer {token}"}
-        resp = requests.get(f"{FEISHU_BASE}/im/v1/messages",
-            headers=headers,
-            params={
+        while True:
+            params = {
                 "container_id_type": "chat",
                 "container_id": chat_id,
-                "sort_type": "ByCreateTimeDesc",
+                "sort_type": "ByCreateTimeAsc",
                 "page_size": limit
-            }, timeout=10)
-        data = resp.json()
-        code = data.get("code", -1)
-        if code != 0:
-            msg = data.get("msg", "unknown")
-            log(f"❌ 获取消息失败 chat={chat_id} (code={code}): {msg}")
-            log(f"   需要权限: im:message / im:message:readonly")
-            return []
-        return data.get("data", {}).get("items", [])
+            }
+            if start_time:
+                params["start_time"] = start_time
+            if page_token:
+                params["page_token"] = page_token
+            resp = requests.get(f"{FEISHU_BASE}/im/v1/messages",
+                headers=headers, params=params, timeout=10)
+            data = resp.json()
+            code = data.get("code", -1)
+            if code != 0:
+                msg = data.get("msg", "unknown")
+                log(f"❌ 获取消息失败 chat={chat_id} (code={code}): {msg}")
+                log(f"   需要权限: im:message / im:message:readonly")
+                return all_items
+            items = data.get("data", {}).get("items", [])
+            all_items.extend(items)
+            # 检查是否有下一页
+            has_more = data.get("data", {}).get("has_more", False)
+            page_token = data.get("data", {}).get("page_token")
+            if not has_more or not page_token:
+                break
+        return all_items
     except Exception as e:
         log(f"❌ get_recent_messages 异常: {type(e).__name__}: {e}")
-        return []
+        return all_items
 
 def extract_image_keys(msg):
     """从消息中提取所有 image_key"""
@@ -256,6 +270,8 @@ def main():
     token = None
     token_time = 0
     cycle = 0
+    # 每个 chat 的最后已知消息时间戳（毫秒字符串）
+    chat_last_time = {}
 
     while True:
         try:
@@ -287,7 +303,9 @@ def main():
 
             new_imgs = 0
             for chat_id in chats:
-                messages = get_recent_messages(token, chat_id)
+                # 增量拉取：首轮不带 start_time（拉最近 50 条），后续用 start_time
+                start_time = chat_last_time.get(chat_id)
+                messages = get_recent_messages(token, chat_id, start_time=start_time)
 
                 # 首轮打印所有消息类型统计
                 if cycle == 1:
@@ -301,6 +319,13 @@ def main():
                 new_in_chat = 0
                 for msg in messages:
                     msg_id = msg.get("message_id", "")
+                    # 更新该 chat 的最新消息时间
+                    create_time = msg.get("create_time", "")
+                    if create_time:
+                        old = chat_last_time.get(chat_id, "0")
+                        if create_time > old:
+                            chat_last_time[chat_id] = create_time
+
                     if msg_id in processed:
                         continue
                     processed.add(msg_id)
@@ -319,7 +344,7 @@ def main():
                         new_imgs += 1
 
                 if new_in_chat > 0 or cycle % 20 == 0:
-                    log(f"📊 chat={chat_id[:12]}... 拉取 {len(messages)} 条, 新 {new_in_chat} 条, 已知 {len(processed)} 条")
+                    log(f"📊 chat={chat_id[:12]}... 拉取 {len(messages)} 条, 新 {new_in_chat} 条, 已知 {len(processed)} 条, start_time={start_time}")
 
 
             if new_imgs > 0: