From 25abc25c9dcf2c2a8b4e0ccd4b6e38ba9c173280 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=A9=AC=E4=B8=80=E4=B8=81?= <1769123563@qq.com>
Date: Thu, 4 Dec 2025 15:29:00 +0800
Subject: [PATCH] Update Report Engine Log Filtering Strategy

---
 ReportEngine/agent.py           | 28 +++++++++++++++++++++++++++-
 ReportEngine/flask_interface.py | 29 +++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 1 deletion(-)

diff --git a/ReportEngine/agent.py b/ReportEngine/agent.py
index 5246787..a029622 100644
--- a/ReportEngine/agent.py
+++ b/ReportEngine/agent.py
@@ -245,6 +245,31 @@ class ReportAgent:
         log_dir = os.path.dirname(self.config.LOG_FILE)
         os.makedirs(log_dir, exist_ok=True)
 
+        def _exclude_other_engines(record):
+            """
+            过滤掉其他引擎(Insight/Media/Query/Forum)产生的日志，其余日志全部保留。
+
+            使用路径匹配为主，无法获取路径时退化到模块名。
+            """
+            excluded_keywords = ("InsightEngine", "MediaEngine", "QueryEngine", "ForumEngine")
+            try:
+                file_path = record["file"].path
+                if any(keyword in file_path for keyword in excluded_keywords):
+                    return False
+            except Exception:
+                pass
+
+            try:
+                module_name = record.get("module", "")
+                if isinstance(module_name, str):
+                    lowered = module_name.lower()
+                    if any(keyword.lower() in lowered for keyword in excluded_keywords):
+                        return False
+            except Exception:
+                pass
+
+            return True
+
         # 【修复】检查是否已经添加过这个文件的handler，避免重复
         # loguru会自动去重，但显式检查更安全
         log_file_path = str(Path(self.config.LOG_FILE).resolve())
@@ -274,7 +299,8 @@ class ReportAgent:
                 buffering=1,        # 行缓冲，每行立即写入
                 serialize=False,    # 普通文本格式，不序列化为JSON
                 encoding="utf-8",   # 明确UTF-8编码
-                mode="a"            # 追加模式
+                mode="a",           # 追加模式
+                filter=_exclude_other_engines # 过滤掉四个 Engine 的日志，保留其余信息
             )
             logger.debug(f"已添加日志handler (ID: {handler_id}): {self.config.LOG_FILE}")
 
diff --git a/ReportEngine/flask_interface.py b/ReportEngine/flask_interface.py
index fdadca1..331dc51 100644
--- a/ReportEngine/flask_interface.py
+++ b/ReportEngine/flask_interface.py
@@ -42,6 +42,33 @@ tasks_registry: Dict[str, 'ReportTask'] = {}
 LOG_STREAM_LEVELS = {"DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"}
 log_stream_handler_id: Optional[int] = None
 
+EXCLUDED_ENGINE_PATH_KEYWORDS = ("ForumEngine", "InsightEngine", "MediaEngine", "QueryEngine")
+
+def _is_excluded_engine_log(record: Dict[str, Any]) -> bool:
+    """
+    判断日志是否来自其他引擎（Insight/Media/Query/Forum），用于过滤混入的日志。
+
+    返回:
+        bool: True 表示应当过滤（即不写入/不转发）。
+    """
+    try:
+        file_path = record["file"].path
+        if any(keyword in file_path for keyword in EXCLUDED_ENGINE_PATH_KEYWORDS):
+            return True
+    except Exception:
+        pass
+
+    # 兜底：尝试按模块名过滤，防止file信息缺失时误混入
+    try:
+        module_name = record.get("module", "")
+        if isinstance(module_name, str):
+            lowered = module_name.lower()
+            return any(keyword.lower() in lowered for keyword in EXCLUDED_ENGINE_PATH_KEYWORDS)
+    except Exception:
+        pass
+
+    return False
+
 
 def _stream_log_to_task(message):
     """
@@ -54,6 +81,8 @@ def _stream_log_to_task(message):
         level_name = record["level"].name
         if level_name not in LOG_STREAM_LEVELS:
             return
+        if _is_excluded_engine_log(record):
+            return
 
         with task_lock:
             task = current_task