Procházet zdrojové kódy

feat: Add functionality to extract specific column data from CSV starting from a defined row

mrh (aider) před 1 rokem
rodič
revize
f1333a5ea3
1 změnil soubory, kde provedl 22 přidání a 6 odebrání
  1. 22 6
      mylib/translate_utils.py

+ 22 - 6
mylib/translate_utils.py

@@ -102,11 +102,27 @@ def extract_column_data(df: pd.DataFrame, column_identifier: Union[str, int], st
         logger.error(f"提取列数据时出错: {e}")
         raise
 
-def test_column_extraction():
-    output_dir = Path('temp')
-    input_file = output_dir/"测试.csv"
-    output_file = output_dir/"processed_测试.csv"
-
+def test_column_extraction(input_file: str):
+    """测试列提取功能
+    
+    Args:
+        input_file: 输入CSV文件路径
+    """
+    try:
+        # 读取CSV文件
+        df = read_csv_with_header(input_file, header_row=1)
+        
+        # 提取第二列的数据,从第三行开始
+        column_data = extract_column_data(df, column_identifier=1, start_row=2, header_row=1)
+        
+        # 打印提取的数据
+        print("提取的列数据:")
+        print(column_data)
+        
+    except Exception as e:
+        logger.error(f"测试列提取时出错: {e}")
 
 if __name__ == '__main__':
-    test_column_extraction()
+    output_dir = Path('temp')
+    input_file = output_dir / "测试.csv"
+    test_column_extraction(str(input_file))