More REST compliant services have the same URL for different operations and the difference between them is the HTTP method. This change allows multiple functions to mock the same URL but for different HTTP methods i.e. GET vs POST can be mocked separately for the same URL.

It maintains previous behaviour so if a function doesn't specify the method param and multiple functions match the same URL the first one to return a value that's not None is taken.

>>> import requests
>>> requests.get('http://fake.fake')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/site-packages/requests/api.py", line 55, in get
    return request('get', url, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/requests/api.py", line 44, in request
    return session.request(method=method, url=url, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/requests/sessions.py", line 383, in request
    resp = self.send(prep, **send_kwargs)
  File "/usr/local/lib/python2.7/site-packages/requests/sessions.py", line 486, in send
    r = adapter.send(request, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/requests/adapters.py", line 378, in send
    raise ConnectionError(e)
requests.exceptions.ConnectionError: HTTPConnectionPool(host='fake.fake', port=80): Max retries exceeded with url: / (Caused by <class 'socket.gaierror'>: [Errno 8] nodename nor servname provided, or not known)

I want to throw the ConnectionError above.

Providing a Request instead of a PreparedRequest would make complicated mocking much easier: Request basically holds the kwargs to requests.method, while PreparedRequest only stores encoded versions of the kwargs.

For example, if you'd like to replay request params (maybe when testing an OAuth flow), PreparedRequest requires reparsing request.path_url, while Request would provide request.params directly.

See https://github.com/kennethreitz/requests/blob/master/requests/adapters.py#L180-L189.

I'm using httmock to test a new API wrapper and checking the internals of the request object to ensure that my wrapper is setting up the calls correctly. In https://github.com/kennethreitz/requests/blob/master/requests/adapters.py#L180-L189 , requests adds both the request and request.url as attributes on the response. This does the same.

I need to assert the payload sent to an external host, so I added the request list in order to ensure my data was properly formed.

The scenario it's a functional tests with no integration servers. I need to test a call to an api, although it's my code I consider it as a blackbox. this blackbox must to do a call to another api so I have not direct way to validate the third call. Using the request history I can validate the call. Something like this:

def test_store_several_requests(self):
    with HTTMock(google_mock_store_requests):
        payload = {"query": "foo"}
        requests.post('http://google.com', data=payload)

    self.assertTrue(google_mock_store_requests.call['called'])
    self.assertEqual(google_mock_store_requests.call['count'], 1)
    request = google_mock_store_requests.call['requests'][0]
    self.assertEqual(request.body, 'query=foo')

As issue #16 points out, the PreparedRequest object is not always very useful. So, provide a original attribute so you can access the originating Request object.

This adds support for:

• mocking out a redirect
• redirecting to a mock

Basically, I just copy/pasted the redirect-handling part of session.send into the fake send code. I had to fake Response.close as well.

Without the encoding, requests will bomb out in Python 3 with:

Traceback (most recent call last):
    self.assertEqual(self.content, response.json())
  File "/home/travis/virtualenv/python3.2/lib/python3.2/site-packages/requests/models.py", line 687, in json
    encoding = guess_json_utf(self.content)
  File "/home/travis/virtualenv/python3.2/lib/python3.2/site-packages/requests/utils.py", line 518, in guess_json_utf
    nullcount = sample.count(_null)
TypeError: Can't convert 'bytes' object to str implicitly

I am trying to mock a response for requests but httmock breaks iter_lines from requests.

With

import requests
from httmock import urlmatch, HTTMock


@urlmatch(netloc=r'uw.edu')
def local_mock(url, request):
    return {'status_code': 200, 'content': 'Hello, World'}


def main():
    with HTTMock(local_mock):
        print list(requests.get('http://uw.edu').iter_lines())

if __name__ == '__main__':
    main()

I get the following error

Traceback (most recent call last):
  File "mock_bug.py", line 15, in <module>
    main()
  File "mock_bug.py", line 12, in main
    print list(requests.get('http://uw.edu').iter_lines())
  File ".../python2.7/site-packages/requests/models.py", line 648, in iter_lines
    decode_unicode=decode_unicode):
  File ".../python2.7/site-packages/requests/models.py", line 625, in generate
    chunk = self.raw.read(chunk_size)
AttributeError: 'NoneType' object has no attribute 'read'

Running the same code without the mock works fine.

Perhaps I am mistaken, but it looks like your library does not support mocking by method, e.g. how would you distinguish between a GET and a POST with the same request url?

The 1.1.0 release no longer works when used as the return value for a mock object.

Failing test case:

class PutTest(unittest.TestCase):
    def test_put(self):
        res = httmock.response(status_code=204)
        put_mock = mock.Mock()
        put_mock.return_value = res
        response = put_mock('http://example.org', data={'foo', 'bar'})

Fails with an exception:

======================================================================
ERROR: test_put (main.PutTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/jimrollenhagen/code/httmock-bug/main.py", line 13, in test_put
    res = httmock.response(status_code=204)
  File "/Users/jimrollenhagen/virtualenvs/httmockbug/lib/python2.7/site-packages/httmock.py", line 43, in response
    res.url = request.url
AttributeError: 'NoneType' object has no attribute 'url'

----------------------------------------------------------------------

Includes #64.

Travis CI has a new pricing model making it harder to test open source projects, and it's also stopped running.

This PR add testing on GitHub Actions.

Pros

• 20 parallel jobs, not 5 like Travis
• Can test on Windows and macOS, not just Ubuntu
• Very flexible, lots of extra actions (aka add-ons) available
• The new popular CI for Python projects
• No confusing pricing model

Cons

• Does not have ppc64le architecture
• Does not have EOL Python 3.4 (I would recommend dropping EOL 2.7 and 3.4-3.6)

If ppc64le isn't important, I'd recommend deleting .travis.yml outright.

Demo

https://github.com/hugovk/httmock/actions/runs/1677534178

To use:

python -m pip install tox
tox               # to test all versions
tox -s            # --skip-missing-interpreters, if you don't have a version installed
tox -e py310      # to test a single version
tox -e py27,py310 # to test more than one
tox -p auto       # run in parallel

For example:

This is useful for downstream users, including OS packagers to QA/test if the package works correctly without requiring fetching the sources from the repository.

Identified by: FreeBSD Ports Bug 242694

Closes: #55

This is useful for downstream distributions to test if the package works correctly with their build recipe. Also, it provides a way to test if the dependencies as they are packaged by the distribution are compatible.

I'm in the process of adding httmock to the FreeBSD ports tree. One standard for most python port is to include tests.

Looking at the httmock-1.2.6.tar.gz from https://pypi.python.org/pypi/httmock this tarball does not include tests.py.

Would it be possible to include the tests.py file in the pypi package? If so I can add do-tests to the FreeBSD port.