app.py

app.py

@@ -293,14 +293,14 @@ def clean_dataset(file_path, question_column, model_choice, temperature, max_sam
         try:
             check_api_key()
         except ValueError as e:
-            return str(e), None, None, ""
+            return str(e), None, ""
         
         progress(0.05, desc="📁 读取数据文件...")
         df = pd.read_parquet(file_path)
         
         if question_column not in df.columns:
             available_columns = ", ".join(df.columns.tolist())
-            return f"❌ 列名 '{question_column}' 不存在！\n可用列名: {available_columns}", None, None, ""
+            return f"❌ 列名 '{question_column}' 不存在！\n可用列名: {available_columns}", None, ""
         
         data_ori = df[question_column].tolist()[:int(max_samples)]
         total = len(data_ori)
@@ -414,19 +414,14 @@ def clean_dataset(file_path, question_column, model_choice, temperature, max_sam
         # 生成带颜色的对比HTML
         preview_html = create_comparison_html(data_ori[:5], lst_final[:5])
         
-        preview_df = pd.DataFrame({
-            '原始问题': [str(x)[:100] for x in data_ori[:5]],
-            '清洗后问题': [str(x)[:100] for x in lst_final[:5]]
-        })
-        
         progress(1.0, desc="✅ 完成！")
         
-        return log_text, output_path, preview_df, preview_html
+        return log_text, output_path, preview_html
         
     except Exception as e:
         import traceback
         error_detail = traceback.format_exc()
-        return f"❌ 处理出错: {str(e)}\n\n详细错误:\n{error_detail}", None, None, ""
+        return f"❌ 处理出错: {str(e)}\n\n详细错误:\n{error_detail}", None, ""
 
 # ======================== 文本内容 ========================
 ABOUT_TEXT = """
@@ -610,7 +605,7 @@ with demo:
                     
                     download_file = gr.File(label="📥 下载去噪后的数据集")
             
-            # 新增：颜色对比预览区域
+            # 颜色对比预览区域
             gr.Markdown("### 🎨 清洗效果对比预览")
             gr.Markdown("""
             **颜色说明**: 
@@ -619,17 +614,12 @@ with demo:
             - ⚫ 黑色 = 未修改的正确部分
             """)
             
-            colored_preview = gr.HTML(label="带颜色标注的对比")
-            
-            preview_df = gr.Dataframe(
-                label="🔍 原始对比表格",
-                wrap=True
-            )
+            colored_preview = gr.HTML(label="")
             
             clean_btn.click(
                 fn=clean_dataset,
                 inputs=[file_input, question_column, model_choice, temperature, max_samples],
-                outputs=[output_text, download_file, preview_df, colored_preview]
+                outputs=[output_text, download_file, colored_preview]
             )
 
 if __name__ == "__main__":