在属性的属性更改时触发操作

Question

我有一个名为 Data 的对象，它有一个 pandas DataFrame (df) 作为属性。我想要做的是在 df 的更改上触发操作。具体来说，添加一个新列。

现在我通过重新设置属性来使其工作，但它没有捕获其属性的任何更改。这是一个示例：

class Data:
    def __init__(self, df: Optional[DataFrame] = None) -> None:
        self.sets = 0
        self.updates = 0
        self._df = df if df is not None else pd.DataFrame

    @property
    def df(self) -> DataFrame:
        return self._df

    @df.setter
    def df(self, new_df: DataFrame) -> None:
        if self._df.empty:
            self.sets += 1
            # do something...
        else:
            self.updates += 1
            # do somthing else...
        self._df = new_df.copy(deep=True)

---

行为

初始化

d = Data()
# d.sets == 0
# d.updates == 0

设置df

d.df = pd.DataFrame({
    'a': [1, 1, 1],
    'b': [2, 2, 2],
})
# d.sets == 1
# d.updates == 0

更新df

d.df = pd.DataFrame({
    'a': [1, 1, 1],
    'b': [2, 2, 2],
    'c': [3, 3, 3],
})
# d.sets == 1
# d.updates == 1

添加新列（不工作）

d.df['d'] = [4, 4, 4]
# d.sets == 1
# d.updates == 1 (EXPECTED 2)

Answer 1

一种方法是定义一个辅助方法来获取集合和更新计数，而不是直接访问属性，如下所示

from typing import Optional

import pandas as pd
from pandas.testing import assert_frame_equal


class Data:
    def __init__(self, df: Optional[pd.DataFrame] = None) -> None:
        self.sets = 0
        self.updates = 0
        self._df = df or pd.DataFrame  # you don't need an if/else expression
        self.previous = self._df  # new attribute to track last known _df

    @property
    def df(self) -> pd.DataFrame:
        return self._df

    @df.setter
    def df(self, new_df: pd.DataFrame) -> None:
        if self._df.empty:
            self.sets += 1
            # do something...
        else:
            self.updates += 1
            # do somthing else...
        self._df = new_df.copy(deep=True)
        self.previous = new_df.copy(deep=True)

    @property
    def count_changes(self) -> dict[str, int]:
        # check if _df has been modified
        try:
            assert_frame_equal(self._df, self.previous)
        except AssertionError:
            self.updates += 1
        return {"sets": self.sets, "updates": self.updates}

：

d = Data()
d.df = pd.DataFrame(
    {
        "a": [1, 1, 1],
        "b": [2, 2, 2],
    }
)
d.df = pd.DataFrame(
    {
        "a": [1, 1, 1],
        "b": [2, 2, 2],
        "c": [3, 3, 3],
    }
)
d.df["d"] = [4, 4, 4]

print(d.count_changes)
# Output
{'sets': 1, 'updates': 2}  # as expected