如何使用mocking/patching更改类方法中的本地目录变量，以避免将测试输出文件写入其中？

Question

我有一个对象，下面是一个to_json方法，我想为它写一个测试。

import networkx as nx

class Example:
    def __init__(self, graph):
        self.graph = graph

    def to_json(self):
        data = nx.readwrite.node_link_data(self.graph)
        filepath = os.path.join('~/.cache/product', 'some_file_name.txt')
        with open(filepath, "w") as outfile:
            json.dump(data, outfile)
        return filepath

在运行to_json的单元测试时，我不希望将输出文件写入~/.cache/product目录，而希望将其写入/tmp/product目录。我如何使用mocking/patching来实现这一点？

我当前的测试设置类似于：

def test_example_to_json():
    ex = Example(graph={}) # Some graph
    filepath = ex.to_json() # This is the problematic step since the file is being written to ~/.cache/product
    with open(filepath, "r") as infile:
        assert json.load(infile) == {}

这里的任何建议都将不胜感激。

Answer 1

不可能更改filepath变量，然后让函数通过模拟/修补继续其余的计算。

这里一个对我有效的解决方案如下:我定义了一个名为CACHE_DIR的环境变量，并使用它来获取缓存目录。Monkeypatch确实允许您使用monkeypatch.setenv("CACHE_DIR", "/tmp/product")。

import networkx as nx


class Example:
    def __init__(self, graph):
        self.graph = graph

    def to_json(self):
        data = nx.readwrite.node_link_data(self.graph)
        filepath = os.path.join(os.getenv('CACHE_DIR'), 'some_file_name.txt')
        with open(filepath, "w") as outfile:
            json.dump(data, outfile)
        return filepath

使用以下单元测试：

import pytest


def test_example_to_json(monkeypatch):
    ex = Example(graph={})

    monkeypatch.setenv("CACHE_DIR", "/tmp/product")
    filepath = ex.to_json() 
    with open(filepath, "r") as infile:
        assert json.load(infile) == {}