Python 如何从pandas的rame类继承并使用filepath实例化

发布于04月08日

我想创建一个继承自pandas的类.我想用CSV文件路径实例化对象，但不断出错.我找到了这篇文章:Use read_csv from pandas to create an instance of a child class of pandas.DataFrame in Python，这是非常相似的，但需要先读取CSV，然后将对象包装在它周围，然而，我想基本上在一行中完成，更像下面这样:

class my_class(pd.DataFrame):
    def __new__(cls, *args, **kwargs):
        return super().__new__(cls)
        
    def __init__(self, save_path):
        self.save_path = save_path
        self.__new__(pd.read_csv(save_path))

    def my_func(self):
        print('New function!')

save_path = 'test_my_class.csv'

pd.DataFrame({'test_col': [1,2,3]}).to_csv(save_path)

mc = my_class(save_path)

我已经try 过这个方法以及其他一些变体，但是不断得到最大的递归错误，我相信来自循环调用new和init.

谢谢你的帮助！

推荐答案

这将修复您的代码:

import pandas as pd

save_path = 'test.csv'
pd.DataFrame({'test_col': [1,2,3]}).to_csv(save_path)

class my_class(pd.DataFrame):
    def __new__(cls, *args, **kwargs):
        return super().__new__(cls)
        
    def __init__(self, save_path):
        super().__init__(pd.read_csv(save_path))
        self.save_path = save_path
        
    def my_func(self):
        print('New function!')

mc = my_class(save_path)

mc.my_func()
mc.head()

You can actually drop the __ new__ implementation, it is not required. When I run your code I also get the max recursion error at the end, however you have scroll to the middle of the traceback stack to get what is wrong: