From 43bd80d3b6f3d74cca123dd6837dc785dc796f8d Mon Sep 17 00:00:00 2001
From: youpaishidifu <youpaishidifu@git.ypshidifu.cn>
Date: Thu, 30 Jul 2026 12:22:04 +0800
Subject: [PATCH] =?UTF-8?q?=E4=BF=AE=E5=A4=8D2=E4=B8=AAbug:=20stop()?=
 =?UTF-8?q?=E6=9C=AA=E6=9D=80EasyOCR=E5=AD=90=E8=BF=9B=E7=A8=8B=E6=AE=8B?=
 =?UTF-8?q?=E7=95=99=20+=20=5F/sdfl=E5=8E=86=E5=8F=B2=E6=B6=88=E6=81=AF?=
 =?UTF-8?q?=E8=AF=AF=E5=88=A4=E4=B8=BA=E6=96=B0=E6=B6=88=E6=81=AF?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Bug1 - 消息监控停止后EasyOCR子进程残留吃GPU导致冻电脑：
  stop()只清线程不杀子进程→GPU推理持续运行→系统冻结。
  修复：stop()末尾增加杀死_easyocr_proc + wait(timeout=3)

Bug2 - 用户问寻宝插件奖励配置，bot回复sdf1插件导入错误：
  微信OCR截到'_ /sdfl undo撤销导入昨天'(历史消息残影)，
  sdfl过滤器要求^[A-Za-z]字母开头，但该文本以下划线_开头→漏网→
  被判为新消息优先处理，真正的用户消息'寻宝玩法中支持哪些奖励配置？'
  被同poll第条跳过。
  修复：
  - sdfl过滤器去掉^锚点限制，只要含sdfl/sdf1即拦截
  - 新增_/下划线斜线残影过滤器^_[ /]拦截OCR气泡残影
---
 monitor.py | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/monitor.py b/monitor.py
index 707f825..902c9dc 100644
--- a/monitor.py
+++ b/monitor.py
@@ -269,6 +269,16 @@ class MessageMonitor:
                 pass
             self._emergency_thread.join(timeout=2)
             self._emergency_thread = None
+        # 杀死 EasyOCR 子进程（防止 GPU 推理残留吃资源/冻电脑）
+        try:
+            if self._easyocr_proc and self._easyocr_proc.poll() is None:
+                self._easyocr_proc.kill()
+                self._easyocr_proc.wait(timeout=3)
+                logger.info("[stop] EasyOCR 子进程已杀死")
+        except Exception as e:
+            logger.debug(f"[stop] 杀 EasyOCR 异常: {e}")
+        self._easyocr_proc = None
+
         # 释放钩子回调引用（防止 GC 延迟导致卸载后崩溃）
         if hasattr(self, '_esc_hook_proc'):
             del self._esc_hook_proc
@@ -3519,8 +3529,13 @@ class MessageMonitor:
             if re.match(r'^/[A-Za-z]', ts):
                 logger.debug(f"[{platform}] Poll过滤: 斜线英文命令 ts={ts!r}")
                 continue
-            # === OCR 误读斜线为 I/J："/sdfl"→"Isdfl", "/import"→"Jimport" ===
-            if re.match(r'^[A-Za-z]', ts) and ('sdfl' in ts.lower() or 'sdf1' in ts.lower()):
+            # === 下划线+斜线：OCR 左对齐气泡残影 "_ /sdfl undo" ===
+            if re.match(r'^_[ /]', ts):
+                logger.debug(f"[{platform}] Poll过滤: 下划线斜线残影 ts={ts!r}")
+                continue
+            # === OCR 误读斜线为 I/J："/sdfl"→"Isdfl", "_ /sdfl"→下划线前缀 ===
+            # 常见变体：/sdfl, _ /sdfl, Isdfl, sdfl undo...
+            if 'sdfl' in ts.lower() or 'sdf1' in ts.lower():
                 logger.debug(f"[{platform}] Poll过滤: 变形命令(sdfl) ts={ts!r}")
                 continue
             if re.match(r'^[A-Z][a-z]{4,}', ts) and re.search(r'[\u4e00-\u9fff]', ts) and len(ts) > 10: