From dcd3256cc71c50bb767b3416b9f91eae902152eb Mon Sep 17 00:00:00 2001
From: zhangwei13 <zhangwei13@xiaohongshu.com>
Date: Thu, 7 Aug 2025 16:22:23 +0800
Subject: [PATCH] fix clean text bug

---
 dots_ocr/utils/format_transformer.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/dots_ocr/utils/format_transformer.py b/dots_ocr/utils/format_transformer.py
index b2a2123..de4b3e0 100755
--- a/dots_ocr/utils/format_transformer.py
+++ b/dots_ocr/utils/format_transformer.py
@@ -136,7 +136,8 @@ def clean_text(text: str) -> str:
     text = text.strip()
     
     # Replace multiple consecutive whitespace characters with a single space
-    text = re.sub(r'\s+', ' ', text)
+    if text[:2] == '`$' and text[-2:] == '$`':
+        text = text[1:-1]
     
     return text