Skip to main content

A pytest plugin for managing processes across test runs.

Project description

Maintenance GitHub last commit GitHub closed pull requests GitHub closed issues PyPI - Downloads Code size Release License supported python versions Issues build status style

pytest plugin for managing processes across test runs.

Setting Up

To use pytest-xprocess you just need to install it via:

pip install pytest-xprocess

and that’s all!

This will provide a xprocess fixture which can be used to ensure that external processes on which your application depends are up and running during testing. You can also use it to start and pre-configure test-specific databases (i.e. Postgres, Couchdb).

Additionally, there are two new command line options:

--xkill  # terminates all external processes
--xshow  # shows currently running processes and log files

xprocess fixture usage

You typically define a project-specific fixture which uses xprocess internally. Following are two examples:

Minimal reference fixture

# content of conftest.py

import pytest
from xprocess import ProcessStarter

@pytest.fixture
def myserver(xprocess):
    class Starter(ProcessStarter):
        # startup pattern
        pattern = "PATTERN"

        # command to start process
        args = ['command', 'arg1', 'arg2']

    # ensure process is running and return its logfile
    logfile = xprocess.ensure("myserver", Starter)

    conn = # create a connection or url/port info to the server
    yield conn

    # clean up whole process tree afterwards
    xprocess.getinfo("myserver").terminate()

Complete reference fixture

# content of conftest.py

import pytest
from xprocess import ProcessStarter

@pytest.fixture
def myserver(xprocess):
    class Starter(ProcessStarter):
        # startup pattern
        pattern = "PATTERN"

        # command to start process
        args = ['command', 'arg1', 'arg2']

        # max startup waiting time
        # optional, defaults to 120 seconds
        timeout = 45

        # max lines read from stdout when matching pattern
        # optional, defaults to 50 lines
        max_read_lines = 100

        def startup_check(self):
            """
            Optional callback used to check process responsiveness
            after the provided pattern has been matched. Returned
            value must be a boolean, where:

            True: Process has been sucessfuly started and is ready
                  to answer queries.

            False: Callback failed during process startup.

            This method will be called multiple times to check if the
            process is ready to answer queries. A 'TimeoutError' exception
            will be raised if the provied 'startup_check' does not
            return 'True' before 'timeout' seconds.
            """
            sock = socket.socket()
            sock.connect(("localhost", 6777))
            sock.sendall(b"testing connection\n")
            return sock.recv(1) == "connection ok!"

    # ensure process is running and return its logfile
    logfile = xprocess.ensure("myserver", Starter)

    conn = # create a connection or url/port info to the server
    yield conn

    # clean up whole process tree afterwards
    xprocess.getinfo("myserver").terminate()

The xprocess.ensure method takes the name of an external process and will make sure it is running during your testing phase. Also, you are not restricted to having a single external process at a time, xprocess can be used to handle multiple diferent processes or several instances of the same process.

Starter Class

Your Starter must be a subclass of ProcessStarter where the required information to start a process instance will be provided:

  • pattern is waited for in the logfile before returning. It should thus match a state of your server where it is ready to answer queries.

  • args is a list of arguments, used to invoke a new subprocess.

  • timeout may be used to specify the maximum time in seconds to wait for process startup.

  • max_read_lines may be be used to extend the number of lines searched for pattern prior to considering the external process dead. By default, the first 50 lines of stdout are redirected to a logfile, which is returned pointing to the line right after the pattern match.

  • startup_check when provided will be called upon to check process responsiveness after ProcessStarter.pattern is matched. By default, XProcess.ensure will attempt to match ProcessStarter.pattern when starting a process, if matched, xprocess will consider the process as ready to answer queries. If startup_check is provided though, its return value will also be considered to determine if the process has been properly started. If startup_check returns True after ProcessStarter.pattern has been matched, XProcess.ensure will return sucessfully. In contrast, if startup_check does not return True before timing out, XProcess.ensure will raise a TimeoutError exception.

  • Adicionally, env may be defined to customize the environment in which the new subprocess is invoked. To inherit the main test process environment, leave env set to the default (None).

If the process is already running, simply the logfile is returned.

Overriding Wait Behavior

To override the wait behavior, override ProcessStarter.wait. See the xprocess.ProcessStarter interface for more details. Note that the plugin uses a subdirectory in .pytest_cache to persist the process ID and logfile information.

An Important Note Regarding Stream Buffering

There have been reports of issues with test suites hanging when users attempt to start external python processes with xprocess.ensure method. The reason for this is that pytest-xprocess relies on matching predefined string patterns written to your environment standard output streams to detect when processes start and python’s sys.stdout/sys.stderr buffering ends up getting in the way of that. A possible solution for this problem is making both streams unbuffered by passing the -u command-line option to your process start command or setting the PYTHONUNBUFFERED environment variable.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pytest-xprocess-0.17.0.tar.gz (19.2 kB view details)

Uploaded Source

Built Distribution

pytest_xprocess-0.17.0-py2.py3-none-any.whl (9.4 kB view details)

Uploaded Python 2 Python 3

File details

Details for the file pytest-xprocess-0.17.0.tar.gz.

File metadata

  • Download URL: pytest-xprocess-0.17.0.tar.gz
  • Upload date:
  • Size: 19.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/50.3.2 requests-toolbelt/0.9.1 tqdm/4.53.0 CPython/3.8.6

File hashes

Hashes for pytest-xprocess-0.17.0.tar.gz
Algorithm Hash digest
SHA256 2506d637c4f54c65dd195c1d094abdeedacc9bf0689131d847a324ad0fc951c0
MD5 e1dae2defe63edada325c5c6ff50ddaf
BLAKE2b-256 a810b5beb8f8d83791bb9df557a9913d354af7104d0937520cbcef3016810772

See more details on using hashes here.

File details

Details for the file pytest_xprocess-0.17.0-py2.py3-none-any.whl.

File metadata

  • Download URL: pytest_xprocess-0.17.0-py2.py3-none-any.whl
  • Upload date:
  • Size: 9.4 kB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/50.3.2 requests-toolbelt/0.9.1 tqdm/4.53.0 CPython/3.8.6

File hashes

Hashes for pytest_xprocess-0.17.0-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 b90052ebb13840a4716c035fb90b69c5fbc07bd04f0bbccc9dba452ed299ccf2
MD5 bdb8bc2064d4a60099838a40d3946c40
BLAKE2b-256 b203f4bc4f008ffa730315e2866e01e29bbd26bf952b7bb5d51120496344593f

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page