fix: 忽略OCR末尾不完整括号噪声,避免误判为括号不匹配
This commit is contained in:
+4
-2
@@ -5380,9 +5380,11 @@ class MessageMonitor:
|
||||
logger.debug(f"[ocr] 过滤器2结果: 检测到短截断片段,返回True")
|
||||
return True
|
||||
# 括号不匹配(可能OCR截断)
|
||||
if text.count('[') != text.count(']') or text.count('(') != text.count(')') or text.count('(') != text.count(')'):
|
||||
# 清理末尾的不完整括号(OCR噪声),如"消息内容['"
|
||||
text_clean = re.sub(r'[\[((]+$', '', text)
|
||||
if text_clean.count('[') != text_clean.count(']') or text_clean.count('(') != text_clean.count(')') or text_clean.count('(') != text_clean.count(')'):
|
||||
# 仅当文本较短时才判定为病句(避免误杀正常聊天)
|
||||
if len(text) < 30:
|
||||
if len(text_clean) < 30:
|
||||
logger.debug(f"[ocr] 过滤器2结果: 检测到括号不匹配,返回True")
|
||||
return True
|
||||
# ⭐ 用户协议/服务条款文档内容检测(牛头不对马嘴修复)
|
||||
|
||||
Reference in New Issue
Block a user