Skip to content

A pytest plugin to rerun tests dynamically based off of test outcome and output.

License

Notifications You must be signed in to change notification settings

gnikonorov/pytest-dynamicrerun

Repository files navigation

pytest-dynamicrerun

PyPI version See Build Status on Travis CI

pytest-dynamicrerun is a pytest plugin to rerun tests dynamically based off of test outcome and output.

Supported versions

This plugin is tested against the following Python and pytest versions. Each Python version is tested against all pytest versions. Please file an issue to request additional targets.

Python Versions:python 3.5, python 3.6, python 3.7, python 3.8, pypy3
Pytest Versions:5.4.0, 5.4.1, 5.4.2, 5.4.3, 6.0.0, 6.0.1

Installation

Install this plugin from PyPI by running the following:

$ pip install pytest-dynamicrerun

Usage

Specifying how many times to rerun

By default, one rerun attempt is made. You can set the amount of times to attempt a rerun by passing the --dynamic-rerun-attempts flag when invoking pytest or including the dynamic_rerun_attempts INI key.

To pass the flag:

python3 -m pytest --dynamic-rerun-attempts=3

To set the INI key add the following to your config file's [pytest] section:

[pytest]
dynamic_rerun_attempts = 123

Passing a non positive integer value will set the number of rerun attempts to the default.

Specifying what to rerun on

By default, all failed tests are rerun. You can change this behavior by either passing the --dynamic-rerun-triggers flag when invoking pytest or including the dynamic_rerun_triggers INI key. Note that regular expressions are allowed.

To pass the flag:

python3 -m pytest --dynamic-rerun-triggers="a triggering trace"

To set the INI key add the following to your config file's [pytest] section:

[pytest]
dynamic_rerun_triggers = a triggering trace

You can accumulate values by either providing the flag multiple times or appending to the INI key. For example, the below two snippets would cause this plugin to trigger on both foo and bar:

python3 -m pytest --dynamic-rerun-triggers="foo" --dynamic-rerun-triggers="bar"

[pytest]
dynamic_rerun_triggers = foo
    bar

Note that at this time only stdout, stderr, and exceptions are checked.

Specifying a rerun interval

You can specify an interval to rerun tests on by either passing the --dynamic-rerun-schedule flag to python when invoking pytest or including the dyanmic_rerun_schedule INI key.

Internally, this plugin uses croniter to schedule wait times. Because of this, we are able to schedule wait times with second level granularity. Visit the croniter repository README to find out more information on this.

To pass the flag:

python3 -m pytest --dynamic-rerun-schedule="* * * * * *"

To set the INI key add the following to your config file's [pytest] section:

[pytest]
dynamic_rerun_schedule = * * * * * *

Note that any valid cron schedule is accepted. If this flag is not passed or set in the INI file, this plugin will not take effect. Passing an invalid value will force the interval to default to * * * * * * ( every second ).

Ignoring this plugin

You can ignore this plugin by passing the --dynamic-rerun-disabled flag to python when invoking pytest or including the dynamic_rerun_disabled INI key.

To pass the flag:

python3 -m pytest --dynamic-rerun-disabled="True"

To set the INI key add the following to your config file's [pytest] section:

[pytest]
dynamic_rerun_disabled = True

Note that if this flag is omitted, we do not disable the plugin ( so it is equivalent to passing --dynamic-rerun-disabled=False )

Using markers to rerun tests

We can achieve the above functionality through markers as well. This plugin defines the dynamicrerun mark, which can be used as follows:

@pytest.mark.dynamicrerun(attempts=10, disabled=False, schedule="* * * * * *", triggers="foo")
def test_print_foo():
    print("foo")

Mark arguments correspond to INI keys as follows:

  • attempts corresponds to dynamic_rerun_attempts
  • disabled corresponds to dynanic_rerun_disabled
  • schedule corresponds to dynamic_rerun_schedule
  • triggers corresponds to dynamic_rerun_triggers

To pass multiple values to the triggers argument, provide a list as so:

@pytest.mark.dynamicrerun(attempts=10, disabled=False, schedule="* * * * * *", triggers=["foo", "bar", "baz"])
def test_print_foo():
    print("foo")

In the above example, reruns will be triggered on foo, bar, and baz.

Argument precedence

Note that first we check for arguments in markers, then command line switches, and only then do we check in INI files. Values found at lower levels ( those checked first ) take precedence over values defined at a higher level ( those checked later ).

For example, if we define the number of rerun attempts as 10 when invoking pytest from the command line, but later in a marker define the number of rerun attempts as 3, 3 would take precedence over 10 since we check markers before we check command line flags.

Developing against this plugin

This plugin exposes the following attributes on the item object:

  • dynamic_rerun_run_times ( list ): The list of times this item was run by the plugin. Note this includes the original non dynamically rerun run.
  • dynamic_rerun_schedule(string): The schedule to rerun this item on. See the section Specifying a rerun interval above for more details.
  • dynamic_rerun_sleep_times (list): A list of timedelta objects representing the time slept in between reruns for the item
  • dynamic_rerun_triggers (list): The rerun triggers for this specific item. See the section Specifying what to rerun on above for more details.
  • max_allowed_dynamic_rerun_attempts(int): The maximum amount of times we are allowed to rerun this item. See the section Specifying how many times to rerun above for more details.
  • num_dynamic_reruns_kicked_off (int): The amount of reruns launched at the moment of inspection for this item.

This plugin exposes the following attributes on the session object:

  • dynamic_rerun_items (list): The list of items that are set to be dynamically rerun on the next iteration

Contributing

Contributions are always welcome. Tests can be run with tox.

Please remember to add a changelog entry when adding a non-trivial feature.

pre-commit is used to ensure basic checks pass.

License

Distributed under the terms of the MIT license, "pytest-dynamicrerun" is free and open source software

Issues

If you encounter any problems, please file an issue along with a detailed description.

About

A pytest plugin to rerun tests dynamically based off of test outcome and output.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages