Browse Source

fix clean text bug

zhangwei13 3 months ago
parent
commit
dcd3256cc7
1 changed files with 2 additions and 1 deletions
  1. 2 1
      dots_ocr/utils/format_transformer.py

+ 2 - 1
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