Python unittest test suite for Robot Framework

This is an experimental package for wrapping Robot Framework test suites into Python unittest suites to make it possible to run Robot Framework tests as plone.testing's layered test suites:

import unittest

from plone.testing import layered
from robotsuite import RobotTestSuite

from my_package.testing import ACCEPTANCE_TESTING


def test_suite():
    suite = unittest.TestSuite()
    suite.addTests([
        layered(RobotTestSuite('mysuite.txt'),
                layer=ACCEPTANCE_TESTING),
    ])
    return suite

RobotTestSuite splits Robot Framework test suites into separate unittest test cases so that Robot will be run once for every test case in every test suite parsed from the given Robot Framework test suite. Because of that, each Robot will generate a separate test report for each test. Each report will have it's own folder, which are created recursively reflecting the structure of the given test suite.

RobotTestSuite's way of wrapping tests into unittest's test suite is similar to how doctest-module's DocTestSuite does its wrappings. See the documentation of DocTestSuite for possible common parameters (e.g. for how to pass a test suite from a different package).

The main motivation behind this package is to make Robot Framework support existing test fixtures and test isolation when testing Plone. Yet, this should help anyone wanting to use Robot Framework with zope.testrunner or other Python unittest compatible test runner.

If this works for you, please contribute at: http://github.com/collective/robotsuite/

Setting robot variables from environment variables

Robot Framework supports overriding test variables from command-line, which is not-available when running tests as robotsuite-wrapped with other test runners. That's why robotsuite supports settings variables as environment variables so that every ROBOT_-prefixed environment variable will be mapped into corresponding test variable without the ROBOT_-prefix.

Declaring tests non-critical by given set of tags

Robot Framework supports declaring tests with given tags as non-critical to prevent their failing to fail the complete build on CI. This is supported as keyword argument for RobotTestSuite as follows:

def test_suite():
    suite = unittest.TestSuite()
    suite.addTests([
        layered(RobotTestSuite('mysuite.txt',
                               noncritical=['non-critical-tag']),
                layer=ACCEPTANCE_TESTING),
    ])
    return suite

Setting zope.testrunner-level

zope.testrunner supports annotating test suites with levels to avoid slow test being run unless wanted:

def test_suite():
    suite = unittest.TestSuite()
    suite.addTests([
        layered(RobotTestSuite('mysuite.txt'),
                layer=ACCEPTANCE_TESTING),
    ])
    suite.level = 10
    return suite

Retry failing tests

You can retry a failed test. This can be useful for flaky robot browser tests. Warning: this may not be good for all types of test. For example any changes that were done in the test until the first failure, may persist.

You can enable retries in two ways:

• Set an environment variable ROBOTSUITE_RETRY_COUNT=X.
• Override this by passing retry_count=X to a RobotTestSuite call.

The default is zero: no retries. The retry count excludes the original try.

def test_suite():
    suite = unittest.TestSuite()
    suite.addTests([
        robotsuite.RobotTestSuite('test_example.robot', retry_count=3),
        robotsuite.RobotTestSuite('test_variables.robot'),
        robotsuite.RobotTestSuite('test_setups', retry_count=2)
    ])
    return suite

Appending test results to existing test report

When running Robot Framework through robotsuite, its test reports are created into the current working directory with filenames robot_output.xml, robot_log.html and robot_report.html. The default behavior is to override the existing robot_output.xml (and also the other report files generated from that).

To merge test results from separate test runs into the same test report, set environment variable ROBOTSUITE_APPEND_OUTPUT_XML=1 to prevent robotsuite from overriding the existing test results, but to always append to the existing robot_output.xml.

Filtering test execution errors

Set environment variable ROBOTSUITE_LOGLEVEL=ERROR to filter all top level Test Execution Errors below the given log level (e.g. ERROR) from the merged test report. This is useful when unnecessary warnings are leaking from the tested code into Robot Framework logs.

Including or skipping all RobotTestSuite-wrapped tests

Robot Framework is often used with Selenium2Library to write acceptance test using the Selenium-framework. Yet, because those test may be slow to run, one might want sometimes (e.g. on CI) to run everything except the robotsuite wrapped tests, and later only the robotsuite wrapped tests.

This can be achieved for sure, with injecting a custom string into the names of robotsuite-wrapped tests with ROBOTSUITE_PREFIX-environment variable and then filter the test with that string.

E.g. run everything except the robotsuite wrapped tests with:

$ ROBOTSUITE_PREFIX=ROBOTSUITE bin/test --all -t \!ROBOTSUITE

and the other way around with:

$ ROBOTSUITE_PREFIX=ROBOTSUITE bin/test --all -t ROBOTSUITE

Re-using test suites from other packages

Sometime it could be useful to re-use acceptance test from some upstream package to test your slightly tailored package (e.g. with a custom theme). This can be done with by defining the test lookup location with package-keyword argment for RobotTestSuite:

def test_suite():
    suite = unittest.TestSuite()
    suite.addTests([
        layered(leveled(
            robotsuite.RobotTestSuite('robot',
                                      package='Products.CMFPlone.tests'),
        ), layer=PLONE_APP_MOSAIC_NO_PAC_ROBOT),
    ])
    return suite