pytest-localserver

.. image:: https://img.shields.io/pypi/v/pytest-localserver.svg?style=flat :alt: PyPI Version :target: https://pypi.python.org/pypi/pytest-localserver

.. image:: https://img.shields.io/pypi/pyversions/pytest-localserver.svg :alt: Supported Python versions :target: https://pypi.python.org/pypi/pytest-localserver

.. image:: https://img.shields.io/badge/code%20style-black-000000.svg :target: https://github.com/psf/black

.. image:: https://results.pre-commit.ci/badge/github/pytest-dev/pytest-localserver/master.svg :target: https://results.pre-commit.ci/latest/github/pytest-dev/pytest-localserver/master :alt: pre-commit.ci status

================== pytest-localserver

pytest-localserver is a plugin for the pytest_ testing framework which enables you to test server connections locally.

Sometimes monkeypatching_ urllib2.urlopen() just does not cut it, for instance if you work with urllib2.Request, define your own openers/handlers or work with httplib. In these cases it may come in handy to have an HTTP server running locally which behaves just like the real thing [1]_. Well, look no further!

Quickstart

Let's say you have a function to scrape HTML which only required to be pointed at a URL ::

import requests
def scrape(url):
    html = requests.get(url).text
    # some parsing happens here
    # ...
    return result

You want to test this function in its entirety without having to rely on a remote server whose content you cannot control, neither do you want to waste time setting up a complex mechanism to mock or patch the underlying Python modules dealing with the actual HTTP request (of which there are more than one BTW). So what do you do?

You simply use pytest's funcargs feature_ and simulate an entire server locally! ::

def test_retrieve_some_content(httpserver):
    httpserver.serve_content(open('cached-content.html').read())
    assert scrape(httpserver.url) == 'Found it!'

What happened here is that for the duration of your tests an HTTP server is started on a random port on localhost which will serve the content you tell it to and behaves just like the real thing.

The added bonus is that you can test whether your code behaves gracefully if there is a network problem::

def test_content_retrieval_fails_graciously(httpserver):
    httpserver.serve_content('File not found!', 404)
    pytest.raises(ContentNotFoundException, scrape, httpserver.url)

The same thing works for SMTP servers, too::

def test_sending_some_message(smtpserver):
    mailer = MyMailer(host=smtpserver.addr[0], port=smtpserver.addr[1])
    mailer.send(to='[e-mail]', from_='[e-mail]',
        subject='MyMailer v1.0', body='Check out my mailer!')
    assert len(smtpserver.outbox)==1

Here an SMTP server is started which accepts e-mails being sent to it. The nice feature here is that you can actually check if the message was received and what was sent by looking into the smtpserver's outbox.

It is really that easy!

Available funcargs

Here is a short overview of the available funcargs. For more details I suggest poking around in the code itself.

httpserver provides a threaded HTTP server instance running on localhost. It has the following attributes:

* ``code`` - HTTP response code (int)
* ``content`` - content of next response (str, bytes, or iterable of either)
* ``headers`` - response headers (dict)
* ``chunked`` - whether to chunk-encode the response (enumeration)
* ``store_request_data`` - whether to store request data for later use

Once these attributes are set, all subsequent requests will be answered with
these values until they are changed or the server is stopped. A more
convenient way to change these is ::

    httpserver.serve_content(content=None, code=200, headers=None, chunked=pytest_localserver.http.Chunked.NO, store_request_data=True)

The ``chunked`` attribute or parameter can be set to

* ``Chunked.YES``, telling the server to always apply chunk encoding
* ``Chunked.NO``, telling the server to never apply chunk encoding
* ``Chunked.AUTO``, telling the server to apply chunk encoding only if
  the ``Transfer-Encoding`` header includes ``chunked``

If chunk encoding is applied, each str or bytes in ``content`` becomes one
chunk in the response.

You can use ``store_request_data=False`` to disable loading the request data into
memory. This will make it impossible to check the request data using
``httpserver.requests[index].data`` but may make sense when posting a larger amount of
data and you don't need to check this.

The server address can be found in property

* ``url``

which is the string representation of tuple ``server_address`` (host as str,
port as int).

If you want to check which form fields have been POSTed, Try ::

    httpserver.serve_content(..., show_post_vars=True)

which will display them as parsable text.

If you need to inspect the requests sent to the server, a list of all
received requests can be found in property

* ``requests``

which is a list of ``werkzeug.wrappers.Request`` objects.

httpsserver is the same as httpserver only with SSL encryption.

smtpserver provides a threaded SMTP server, with an API similar to smtpd.SMTPServer, (the deprecated class from the Python standard library) running on localhost. It has the following attributes:

* ``addr`` - server address as tuple (host as str, port as int)
* ``outbox`` - list of ``email.message.Message`` instances received.

Using your a WSGI application as test server

As of version 0.3 you can now use a WSGI application_ to run on the test server ::

from pytest_localserver.http import WSGIServer

def simple_app(environ, start_response):
    """Simplest possible WSGI application"""
    status = '200 OK'
    response_headers = [('Content-type', 'text/plain')]
    start_response(status, response_headers)
    return ['Hello world!\n']

@pytest.fixture
def testserver(request):
    """Defines the testserver funcarg"""
    server = WSGIServer(application=simple_app)
    server.start()
    request.addfinalizer(server.stop)
    return server

def test_retrieve_some_content(testserver):
    assert scrape(testserver.url) == 'Hello world!\n'

Have a look at the following page for more information on WSGI: http://wsgi.readthedocs.org/en/latest/learn.html

Download and Installation

You can install the plugin by running ::

pip install pytest-localserver

Alternatively, get the latest stable version from PyPI_ or the latest bleeding-edge_ from Github.

License and Credits

This plugin is released under the MIT license. You can find the full text of the license in the LICENSE file.

Copyright (C) 2010-2022 Sebastian Rahlf and others (see AUTHORS).

Some parts of this package is based on ideas or code from other people:

• I borrowed some implementation ideas for the httpserver from linkchecker_.
• The implementation for the SMTP server is based on the Mailsink recipe_ by Adam Feuer, Matt Branthwaite and Troy Frever.
• The HTTPS implementation is based on work by Sebastien Martini_.

Thanks guys!

Development and future plans

Feel free to clone the repository and add your own changes. Pull requests are always welcome!::

git clone https://github.com/pytest-dev/pytest-localserver

If you find any bugs, please file a report_.

Test can be run with tox.

I already have a couple of ideas for future versions:

• support for FTP, SSH (maybe base all on twisted?)
• making the SMTP outbox as convenient to use as django.core.mail.outbox
• add your own here!

Preparing a release

For package maintainers, here is how we release a new version:

#. Ensure that the CHANGES file is up to date with the latest changes. #. Make sure that all tests pass on the version you want to release. #. Use the new release form on Github_ (or some other equivalent method) to create a new release, following the pattern of previous releases.

• Each release has to be based on a tag. You can either create the tag first (e.g. using git tag) and then make a release from that tag, or you can have Github create the tag as part of the process of making a release; either way works.
• The tag name must be the PEP 440_-compliant version number prefixed by v, making sure to include at least three version number components (e.g. v0.6.0).
• The "Auto-generate release notes" button will be useful in summarizing the changes since the last release.

#. Using either the release workflows page_ or the link in the email you received about a "Deployment review", go to the workflow run created for the new release and click "Review deployments", then either approve or reject the two deployments, one to Test PyPI and one to real PyPI. (It should not be necessary to reject a deployment unless something really weird happens.) Once the deployment is approved, Github will automatically upload the files.

---

.. [1] The idea for this project was born when I needed to check that a piece of software_ behaved itself when receiving HTTP error codes 404 and 500. Having unsuccessfully tried to mock a server, I stumbled across linkchecker_ which uses a the same idea to test its internals.

.. _monkeypatching: http://pytest.org/latest/monkeypatch.html .. _pytest: http://pytest.org/ .. _funcargs feature: http://pytest.org/latest/funcargs.html .. _linkchecker: http://linkchecker.sourceforge.net/ .. _WSGI application: http://www.python.org/dev/peps/pep-0333/ .. _PyPI: http://pypi.python.org/pypi/pytest-localserver/ .. _bleeding-edge: https://github.com/pytest-dev/pytest-localserver .. _report: https://github.com/pytest-dev/pytest-localserver/issues/ .. _tox: http://testrun.org/tox/ .. _a piece of software: http://pypi.python.org/pypi/python-amazon-product-api/ .. _Mailsink recipe: http://code.activestate.com/recipes/440690/ .. _Sebastien Martini: http://code.activestate.com/recipes/442473/ .. _PEP 440: https://peps.python.org/pep-0440/ .. _build: https://pypa-build.readthedocs.io/en/latest/ .. _twine: https://twine.readthedocs.io/en/stable/ .. _new release form on Github: https://github.com/pytest-dev/pytest-localserver/releases/new .. _release workflows page: https://github.com/pytest-dev/pytest-localserver/actions/workflows/release.yml