Socket Mock Framework - for all kinds of socket animals, web-clients included - with gevent/asyncio/SSL support
Reason this release was yanked:
pip cannot find it
Project description
A socket mock framework
for all kinds of socket animals, web-clients included - with gevent/asyncio/SSL support
…and then MicroPython’s urequest (mocket >= 3.9.1)
Versioning
Starting from 3.7.0, Mocket major version will follow the same numbering pattern as Python’s and therefore indicate the most recent Python version that is supported.
FYI: the last version compatible with Python 2.7 is 3.9.4, bugfixing or backporting of features introduced after that release will only be available as commercial support.
Support it
Star the project on GitHub, Buy Me a Coffee clicking the button below or, even better, contribute with patches or documentation.
Thanks to @felixonmars Mocket is available in the Arch Linux repository.
How to use it
Read these three blog posts if you want to have a big picture of what Mocket is capable of:
https://medium.com/p/mocket-is-alive-and-is-fighting-with-us-b2810d52597a
https://hackernoon.com/make-development-great-again-faab769d264e
https://hackernoon.com/httpretty-now-supports-asyncio-e310814704c6
The starting point to understand how to use Mocket to write a custom mock is the following example:
As next step, you are invited to have a look at both the implementation of the two mocks it provides:
HTTP mock (similar to HTTPretty) - https://github.com/mindflayer/python-mocket/blob/master/mocket/mockhttp.py
Redis mock (basic implementation) - https://github.com/mindflayer/python-mocket/blob/master/mocket/mockredis.py
Please also have a look at the huge test suite:
Tests module at https://github.com/mindflayer/python-mocket/tree/master/tests
Installation
Using pip:
$ pip install mocket
Speedups
Mocket uses xxhash when available instead of hashlib.md5 for creating hashes, you can install it as follows:
$ pip install mocket[speedups]
Issues
When opening an Issue, please add few lines of code as failing test, or -better- open its relative Pull request adding this test to our test suite.
Example of how to mock an HTTP[S] call
Let’s create a new virtualenv with all we need:
$ virtualenv example $ source example/bin/activate $ pip install pytest requests mocket
As second step, we create an example.py file as the following one:
import json
from mocket import mocketize
from mocket.mockhttp import Entry
import requests
import pytest
@pytest.fixture
def response():
return {
"integer": 1,
"string": "asd",
"boolean": False,
}
@mocketize # use its decorator
def test_json(response):
url_to_mock = 'https://testme.org/json'
Entry.single_register(
Entry.GET,
url_to_mock,
body=json.dumps(response),
headers={'content-type': 'application/json'}
)
mocked_response = requests.get(url_to_mock).json()
assert response == mocked_response
# OR use its context manager
from mocket import Mocketizer
def test_json_with_context_manager(response):
url_to_mock = 'https://testme.org/json'
Entry.single_register(
Entry.GET,
url_to_mock,
body=json.dumps(response),
headers={'content-type': 'application/json'}
)
with Mocketizer():
mocked_response = requests.get(url_to_mock).json()
assert response == mocked_response
Let’s fire our example test:
$ py.test example.py
Example of how to fake socket errors
It’s very important that we test non-happy paths.
@mocketize
def test_raise_exception(self):
url = "http://github.com/fluidicon.png"
Entry.single_register(Entry.GET, url, exception=socket.error())
with self.assertRaises(requests.exceptions.ConnectionError):
requests.get(url)
Example of how to record real socket traffic
You probably know what VCRpy is capable of, that’s the mocket’s way of achieving it:
@mocketize(truesocket_recording_dir=tempfile.mkdtemp())
def test_truesendall_with_recording_https():
url = 'https://httpbin.org/ip'
requests.get(url, headers={"Accept": "application/json"})
resp = requests.get(url, headers={"Accept": "application/json"})
assert resp.status_code == 200
dump_filename = os.path.join(
Mocket.get_truesocket_recording_dir(),
Mocket.get_namespace() + '.json',
)
with io.open(dump_filename) as f:
response = json.load(f)
assert len(response['httpbin.org']['443'].keys()) == 1
HTTPretty compatibility layer
Mocket HTTP mock can work as HTTPretty replacement for many different use cases. Two main features are missing:
URL entries containing regular expressions;
response body from functions (used mostly to fake errors, mocket doesn’t need to do it this way).
Two features which are against the Zen of Python, at least imho (mindflayer), but of course I am open to call it into question.
Example:
import json
import aiohttp
import asyncio
import async_timeout
from unittest import TestCase
from mocket.plugins.httpretty import httpretty, httprettified
class AioHttpEntryTestCase(TestCase):
@httprettified
def test_https_session(self):
url = 'https://httpbin.org/ip'
httpretty.register_uri(
httpretty.GET,
url,
body=json.dumps(dict(origin='127.0.0.1')),
)
async def main(l):
async with aiohttp.ClientSession(loop=l) as session:
with async_timeout.timeout(3):
async with session.get(url) as get_response:
assert get_response.status == 200
assert await get_response.text() == '{"origin": "127.0.0.1"}'
loop = asyncio.get_event_loop()
loop.set_debug(True)
loop.run_until_complete(main(loop))
What about the other socket animals?
Using Mocket with asyncio based clients:
$ pip install aiohttp
Example:
class AioHttpEntryTestCase(TestCase):
@mocketize
def test_http_session(self):
url = 'http://httpbin.org/ip'
body = "asd" * 100
Entry.single_register(Entry.GET, url, body=body, status=404)
Entry.single_register(Entry.POST, url, body=body*2, status=201)
async def main(l):
async with aiohttp.ClientSession(loop=l) as session:
with async_timeout.timeout(3):
async with session.get(url) as get_response:
assert get_response.status == 404
assert await get_response.text() == body
with async_timeout.timeout(3):
async with session.post(url, data=body * 6) as post_response:
assert post_response.status == 201
assert await post_response.text() == body * 2
loop = asyncio.get_event_loop()
loop.run_until_complete(main(loop))
# or again with a unittest.IsolatedAsyncioTestCase
from mocket.async_mocket import async_mocketize
class AioHttpEntryTestCase(IsolatedAsyncioTestCase):
@async_mocketize
async def test_http_session(self):
url = 'http://httpbin.org/ip'
body = "asd" * 100
Entry.single_register(Entry.GET, url, body=body, status=404)
Entry.single_register(Entry.POST, url, body=body * 2, status=201)
async with aiohttp.ClientSession() as session:
with async_timeout.timeout(3):
async with session.get(url) as get_response:
assert get_response.status == 404
assert await get_response.text() == body
with async_timeout.timeout(3):
async with session.post(url, data=body * 6) as post_response:
assert post_response.status == 201
assert await post_response.text() == body * 2
assert Mocket.last_request().method == 'POST'
assert Mocket.last_request().body == body * 6
Works well with others
Using Mocket as pook engine:
$ pip install mocket[pook]
Example:
import pook
from mocket.plugins.pook_mock_engine import MocketEngine
pook.set_mock_engine(MocketEngine)
pook.on()
url = 'http://twitter.com/api/1/foobar'
status = 404
response_json = {'error': 'foo'}
mock = pook.get(
url,
headers={'content-type': 'application/json'},
reply=status,
response_json=response_json,
)
mock.persist()
requests.get(url)
assert mock.calls == 1
resp = requests.get(url)
assert resp.status_code == status
assert resp.json() == response_json
assert mock.calls == 2
First appearance
EuroPython 2013, Florence
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.