我从Johannes Farhenkrug的答案开始，它对我很有效。我需要模拟请求库，因为我的目标是隔离我的应用程序，不测试任何第三方资源。

然后我读了更多关于python的Mock库，我意识到我可以用python Mock类替换MockResponse类，你可能称之为“Test Double”或“Fake”。

这样做的好处是可以访问assert_called_with, call_args等。不需要额外的库。额外的好处，如“可读性”或“更python化”是主观的，所以它们可能对你有影响，也可能没有。

这是我的版本，更新使用python的Mock而不是test double:

import json
import requests
from unittest import mock

# defube stubs
AUTH_TOKEN = '{"prop": "value"}'
LIST_OF_WIDGETS = '{"widgets": ["widget1", "widget2"]}'
PURCHASED_WIDGETS = '{"widgets": ["purchased_widget"]}'


# exception class when an unknown URL is mocked
class MockNotSupported(Exception):
  pass


# factory method that cranks out the Mocks
def mock_requests_factory(response_stub: str, status_code: int = 200):
    return mock.Mock(**{
        'json.return_value': json.loads(response_stub),
        'text.return_value': response_stub,
        'status_code': status_code,
        'ok': status_code == 200
    })


# side effect mock function
def mock_requests_post(*args, **kwargs):
    if args[0].endswith('/api/v1/get_auth_token'):
        return mock_requests_factory(AUTH_TOKEN)
    elif args[0].endswith('/api/v1/get_widgets'):
        return mock_requests_factory(LIST_OF_WIDGETS)
    elif args[0].endswith('/api/v1/purchased_widgets'):
        return mock_requests_factory(PURCHASED_WIDGETS)
    
    raise MockNotSupported


# patch requests.post and run tests
with mock.patch('requests.post') as requests_post_mock:
  requests_post_mock.side_effect = mock_requests_post
  response = requests.post('https://myserver/api/v1/get_widgets')
  assert response.ok is True
  assert response.status_code == 200
  assert 'widgets' in response.json()
  
  # now I can also do this
  requests_post_mock.assert_called_with('https://myserver/api/v1/get_widgets')

Repl。它的链接:

https://repl.it/@abkonsta/Using-unittestMock-for-requestspost#main.py

https://repl.it/@abkonsta/Using-test-double-for-requestspost#main.py

我从Johannes Farhenkrug的答案开始，它对我很有效。我需要模拟请求库，因为我的目标是隔离我的应用程序，不测试任何第三方资源。

然后我读了更多关于python的Mock库，我意识到我可以用python Mock类替换MockResponse类，你可能称之为“Test Double”或“Fake”。

这样做的好处是可以访问assert_called_with, call_args等。不需要额外的库。额外的好处，如“可读性”或“更python化”是主观的，所以它们可能对你有影响，也可能没有。

这是我的版本，更新使用python的Mock而不是test double:

import json
import requests
from unittest import mock

# defube stubs
AUTH_TOKEN = '{"prop": "value"}'
LIST_OF_WIDGETS = '{"widgets": ["widget1", "widget2"]}'
PURCHASED_WIDGETS = '{"widgets": ["purchased_widget"]}'


# exception class when an unknown URL is mocked
class MockNotSupported(Exception):
  pass


# factory method that cranks out the Mocks
def mock_requests_factory(response_stub: str, status_code: int = 200):
    return mock.Mock(**{
        'json.return_value': json.loads(response_stub),
        'text.return_value': response_stub,
        'status_code': status_code,
        'ok': status_code == 200
    })


# side effect mock function
def mock_requests_post(*args, **kwargs):
    if args[0].endswith('/api/v1/get_auth_token'):
        return mock_requests_factory(AUTH_TOKEN)
    elif args[0].endswith('/api/v1/get_widgets'):
        return mock_requests_factory(LIST_OF_WIDGETS)
    elif args[0].endswith('/api/v1/purchased_widgets'):
        return mock_requests_factory(PURCHASED_WIDGETS)
    
    raise MockNotSupported


# patch requests.post and run tests
with mock.patch('requests.post') as requests_post_mock:
  requests_post_mock.side_effect = mock_requests_post
  response = requests.post('https://myserver/api/v1/get_widgets')
  assert response.ok is True
  assert response.status_code == 200
  assert 'widgets' in response.json()
  
  # now I can also do this
  requests_post_mock.assert_called_with('https://myserver/api/v1/get_widgets')

Repl。它的链接:

https://repl.it/@abkonsta/Using-unittestMock-for-requestspost#main.py

https://repl.it/@abkonsta/Using-test-double-for-requestspost#main.py

这就是模拟请求的方法。Post，将其更改为HTTP方法

@patch.object(requests, 'post')
def your_test_method(self, mockpost):
    mockresponse = Mock()
    mockpost.return_value = mockresponse
    mockresponse.text = 'mock return'

    #call your target method now

尝试使用响应库。以下是他们文档中的一个例子:

import responses
import requests

@responses.activate
def test_simple():
    responses.add(responses.GET, 'http://twitter.com/api/1/foobar',
                  json={'error': 'not found'}, status=404)

    resp = requests.get('http://twitter.com/api/1/foobar')

    assert resp.json() == {"error": "not found"}

    assert len(responses.calls) == 1
    assert responses.calls[0].request.url == 'http://twitter.com/api/1/foobar'
    assert responses.calls[0].response.text == '{"error": "not found"}'

相比于自己设置所有的mock，它提供了相当好的便利。

还有HTTPretty:

它不是特定于请求库，在某些方面更强大，尽管我发现它不太适合检查它拦截的请求，而响应则很容易

还有httmock。

最近，一个比古老的请求更受欢迎的新库是httpx，它增加了对异步的一等支持。httpx的模拟库是:https://github.com/lundberg/respx

解决请求的一个可能的方法是使用库betamax，它记录所有的请求，之后如果你在相同的url中使用相同的参数发出请求，betamax将使用记录的请求，我一直在用它来测试网络爬虫，它节省了我很多时间。

import os

import requests
from betamax import Betamax
from betamax_serializers import pretty_json


WORKERS_DIR = os.path.dirname(os.path.abspath(__file__))
CASSETTES_DIR = os.path.join(WORKERS_DIR, u'resources', u'cassettes')
MATCH_REQUESTS_ON = [u'method', u'uri', u'path', u'query']

Betamax.register_serializer(pretty_json.PrettyJSONSerializer)
with Betamax.configure() as config:
    config.cassette_library_dir = CASSETTES_DIR
    config.default_cassette_options[u'serialize_with'] = u'prettyjson'
    config.default_cassette_options[u'match_requests_on'] = MATCH_REQUESTS_ON
    config.default_cassette_options[u'preserve_exact_body_bytes'] = True


class WorkerCertidaoTRT2:
    session = requests.session()

    def make_request(self, input_json):
        with Betamax(self.session) as vcr:
            vcr.use_cassette(u'google')
            response = session.get('http://www.google.com')

https://betamax.readthedocs.io/en/latest/

如果使用pytest:

>>> import pytest
>>> import requests

>>> def test_url(requests_mock):
...     requests_mock.get('http://test.com', text='data')
...     assert 'data' == requests.get('http://test.com').text

摘自官方文件