retriever/forms.py

@@ -12,4 +12,4 @@ class SpaceSeparatedField(forms.CharField):
 
 class UploadForm(forms.Form):
     attachments = MultiFileField(min_num=1, max_num=10, max_file_size=1024 * 1024 * 64,
-                                 attrs={'class': 'file-input is-primary', 'accept': '.docx, .doc, .dot, .pptx, .ppt, .pdf, .xls'})
+                                 attrs={'class': 'file-input is-primary'})

retriever/tools/keyword_find.py

@@ -29,7 +29,6 @@ def docx_find(file_path: str, keyword_list: list) -> dict:
     paragraph_keyword = []
 
     for para in doc_text.split('\n'):
-        para = re.sub(r'\s+', '', para)  # 删除段落中空格
         this_para_keyword = [keyword for keyword in keyword_list if keyword in para]  # 查找该段落中的敏感词
 
         if this_para_keyword: