26.3. unittest — Unit testing framework

(If you are already familiar with the basic concepts of testing, you might want to skip to the list of assert methods.)

The unittest unit testing framework was originally inspired by JUnit and has a similar flavor as major unit testing frameworks in other languages. It supports test automation, sharing of setup and shutdown code for tests, aggregation of tests into collections, and independence of the tests from the reporting framework.

To achieve this, unittest supports some important concepts in an object-oriented way:

test fixture
A test fixture represents the preparation needed to perform one or more tests, and any associate cleanup actions. This may involve, for example, creating temporary or proxy databases, directories, or starting a server process.
test case
A test case is the individual unit of testing. It checks for a specific response to a particular set of inputs. unittest provides a base class, TestCase, which may be used to create new test cases.
test suite
A test suite is a collection of test cases, test suites, or both. It is used to aggregate tests that should be executed together.
test runner
A test runner is a component which orchestrates the execution of tests and provides the outcome to the user. The runner may use a graphical interface, a textual interface, or return a special value to indicate the results of executing the tests.

See also

Module doctest
Another test-support module with a very different flavor.
Simple Smalltalk Testing: With Patterns
Kent Beck’s original paper on testing frameworks using the pattern shared by unittest.
Nose and py.test
Third-party unittest frameworks with a lighter-weight syntax for writing tests. For example, assert func(10) == 42.
The Python Testing Tools Taxonomy
An extensive list of Python testing tools including functional testing frameworks and mock object libraries.
Testing in Python Mailing List
A special-interest-group for discussion of testing, and testing tools, in Python.

The script Tools/unittestgui/unittestgui.py in the Python source distribution is a GUI tool for test discovery and execution. This is intended largely for ease of use for those new to unit testing. For production environments it is recommended that tests be driven by a continuous integration system such as Buildbot, Jenkins or Hudson.

26.3.1. Basic example

The unittest module provides a rich set of tools for constructing and running tests. This section demonstrates that a small subset of the tools suffice to meet the needs of most users.

Here is a short script to test three functions from the random module:

import random
import unittest

class TestSequenceFunctions(unittest.TestCase):

    def setUp(self):
        self.seq = list(range(10))

    def test_shuffle(self):
        # make sure the shuffled sequence does not lose any elements
        self.assertEqual(self.seq, list(range(10)))

        # should raise an exception for an immutable sequence
        self.assertRaises(TypeError, random.shuffle, (1,2,3))

    def test_choice(self):
        element = random.choice(self.seq)
        self.assertTrue(element in self.seq)

    def test_sample(self):
        with self.assertRaises(ValueError):
            random.sample(self.seq, 20)
        for element in random.sample(self.seq, 5):
            self.assertTrue(element in self.seq)

if __name__ == '__main__':

A testcase is created by subclassing unittest.TestCase. The three individual tests are defined with methods whose names start with the letters test. This naming convention informs the test runner about which methods represent tests.

The crux of each test is a call to assertEqual() to check for an expected result; assertTrue() to verify a condition; or assertRaises() to verify that an expected exception gets raised. These methods are used instead of the assert statement so the test runner can accumulate all test results and produce a report.

When a setUp() method is defined, the test runner will run that method prior to each test. Likewise, if a tearDown() method is defined, the test runner will invoke that method after each test. In the example, setUp() was used to create a fresh sequence for each test.

The final block shows a simple way to run the tests. unittest.main() provides a command-line interface to the test script. When run from the command line, the above script produces an output that looks like this:

Ran 3 tests in 0.000s


Passing the -v option to your test script will instruct unittest.main() to enable a higher level of verbosity, and produce the following output:

test_choice (__main__.TestSequenceFunctions) ... ok
test_sample (__main__.TestSequenceFunctions) ... ok
test_shuffle (__main__.TestSequenceFunctions) ... ok

Ran 3 tests in 0.110s


The above examples show the most commonly used unittest features which are sufficient to meet many everyday testing needs. The remainder of the documentation explores the full feature set from first principles.

26.3.2. Command-Line Interface

The unittest module can be used from the command line to run tests from modules, classes or even individual test methods:

python -m unittest test_module1 test_module2
python -m unittest test_module.TestClass
python -m unittest test_module.TestClass.test_method

You can pass in a list with any combination of module names, and fully qualified class or method names.

Test modules can be specified by file path as well:

python -m unittest tests/test_something.py

This allows you to use the shell filename completion to specify the test module. The file specified must still be importable as a module. The path is converted to a module name by removing the ‘.py’ and converting path separators into ‘.’. If you want to execute a test file that isn’t importable as a module you should execute the file directly instead.

You can run tests with more detail (higher verbosity) by passing in the -v flag:

python -m unittest -v test_module

When executed without arguments Test Discovery is started:

python -m unittest

For a list of all the command-line options:

python -m unittest -h

Changed in version 3.2: In earlier versions it was only possible to run individual test methods and not modules or classes. Command-line options

unittest supports these command-line options:

-b, --buffer

The standard output and standard error streams are buffered during the test run. Output during a passing test is discarded. Output is echoed normally on test fail or error and is added to the failure messages.

-c, --catch

Control-C during the test run waits for the current test to end and then reports all the results so far. A second control-C raises the normal KeyboardInterrupt exception.

See Signal Handling for the functions that provide this functionality.

-f, --failfast

Stop the test run on the first error or failure.

New in version 3.2: The command-line options -b, -c and -f were added.

The command line can also be used for test discovery, for running all of the tests in a project or just a subset.

26.3.3. Test Discovery

New in version 3.2.

Unittest supports simple test discovery. In order to be compatible with test discovery, all of the test files must be modules or packages importable from the top-level directory of the project (this means that their filenames must be valid identifiers).

Test discovery is implemented in TestLoader.discover(), but can also be used from the command line. The basic command-line usage is:

cd project_directory
python -m unittest discover


As a shortcut, python -m unittest is the equivalent of python -m unittest discover. If you want to pass arguments to test discovery the discover sub-command must be used explicitly.

The discover sub-command has the following options:

-v, --verbose

Verbose output

-s, --start-directory directory

Directory to start discovery (. default)

-p, --pattern pattern

Pattern to match test files (test*.py default)

-t, --top-level-directory directory

Top level directory of project (defaults to start directory)

The -s, -p, and -t options can be passed in as positional arguments in that order. The following two command lines are equivalent:

python -m unittest discover -s project_directory -p '*_test.py'
python -m unittest discover project_directory '*_test.py'

As well as being a path it is possible to pass a package name, for example myproject.subpackage.test, as the start directory. The package name you supply will then be imported and its location on the filesystem will be used as the start directory.


Test discovery loads tests by importing them. Once test discovery has found all the test files from the start directory you specify it turns the paths into package names to import. For example foo/bar/baz.py will be imported as foo.bar.baz.

If you have a package installed globally and attempt test discovery on a different copy of the package then the import could happen from the wrong place. If this happens test discovery will warn you and exit.

If you supply the start directory as a package name rather than a path to a directory then discover assumes that whichever location it imports from is the location you intended, so you will not get the warning.

Test modules and packages can customize test loading and discovery by through the load_tests protocol.

26.3.4. Organizing test code

The basic building blocks of unit testing are test cases — single scenarios that must be set up and checked for correctness. In unittest, test cases are represented by unittest.TestCase instances. To make your own test cases you must write subclasses of TestCase or use FunctionTestCase.

The testing code of a TestCase instance should be entirely self contained, such that it can be run either in isolation or in arbitrary combination with any number of other test cases.

The simplest TestCase subclass will simply implement a test method (i.e. a method whose name starts with test) in order to perform specific testing code:

import unittest

class DefaultWidgetSizeTestCase(unittest.TestCase):
    def test_default_widget_size(self):
        widget = Widget('The widget')
        self.assertEqual(widget.size(), (50, 50))

Note that in order to test something, we use one of the assert*() methods provided by the TestCase base class. If the test fails, an exception will be raised, and unittest will identify the test case as a failure. Any other exceptions will be treated as errors.

Tests can be numerous, and their set-up can be repetitive. Luckily, we can factor out set-up code by implementing a method called setUp(), which the testing framework will automatically call for every single test we run:

import unittest

class SimpleWidgetTestCase(unittest.TestCase):
    def setUp(self):
        self.widget = Widget('The widget')

    def test_default_widget_size(self):
        self.assertEqual(self.widget.size(), (50,50),
                         'incorrect default size')

    def test_widget_resize(self):
        self.assertEqual(self.widget.size(), (100,150),
                         'wrong size after resize')


The order in which the various tests will be run is determined by sorting the test method names with respect to the built-in ordering for strings.

If the setUp() method raises an exception while the test is running, the framework will consider the test to have suffered an error, and the test method will not be executed.

Similarly, we can provide a tearDown() method that tidies up after the test method has been run:

import unittest

class SimpleWidgetTestCase(unittest.TestCase):
    def setUp(self):
        self.widget = Widget('The widget')

    def tearDown(self):

If setUp() succeeded, tearDown() will be run whether the test method succeeded or not.

Such a working environment for the testing code is called a fixture.

Test case instances are grouped together according to the features they test. unittest provides a mechanism for this: the test suite, represented by unittest‘s TestSuite class. In most cases, calling unittest.main() will do the right thing and collect all the module’s test cases for you, and then execute them.

However, should you want to customize the building of your test suite, you can do it yourself:

def suite():
    suite = unittest.TestSuite()
    return suite

You can place the definitions of test cases and test suites in the same modules as the code they are to test (such as widget.py), but there are several advantages to placing the test code in a separate module, such as test_widget.py:

  • The test module can be run standalone from the command line.
  • The test code can more easily be separated from shipped code.
  • There is less temptation to change test code to fit the code it tests without a good reason.
  • Test code should be modified much less frequently than the code it tests.
  • Tested code can be refactored more easily.
  • Tests for modules written in C must be in separate modules anyway, so why not be consistent?
  • If the testing strategy changes, there is no need to change the source code.

26.3.5. Re-using old test code

Some users will find that they have existing test code that they would like to run from unittest, without converting every old test function to a TestCase subclass.

For this reason, unittest provides a FunctionTestCase class. This subclass of TestCase can be used to wrap an existing test function. Set-up and tear-down functions can also be provided.

Given the following test function:

def testSomething():
    something = makeSomething()
    assert something.name is not None
    # ...

one can create an equivalent test case instance as follows, with optional set-up and tear-down methods:

testcase = unittest.FunctionTestCase(testSomething,


Even though FunctionTestCase can be used to quickly convert an existing test base over to a unittest-based system, this approach is not recommended. Taking the time to set up proper TestCase subclasses will make future test refactorings infinitely easier.

In some cases, the existing tests may have been written using the doctest module. If so, doctest provides a DocTestSuite class that can automatically build unittest.TestSuite instances from the existing doctest-based tests.

26.3.6. Skipping tests and expected failures

New in version 3.1.

Unittest supports skipping individual test methods and even whole classes of tests. In addition, it supports marking a test as a “expected failure,” a test that is broken and will fail, but shouldn’t be counted as a failure on a TestResult.

Skipping a test is simply a matter of using the skip() decorator or one of its conditional variants.

Basic skipping looks like this:

class MyTestCase(unittest.TestCase):

    @unittest.skip("demonstrating skipping")
    def test_nothing(self):
        self.fail("shouldn't happen")

    @unittest.skipIf(mylib.__version__ < (1, 3),
                     "not supported in this library version")
    def test_format(self):
        # Tests that work for only a certain version of the library.

    @unittest.skipUnless(sys.platform.startswith("win"), "requires Windows")
    def test_windows_support(self):
        # windows specific testing code

This is the output of running the example above in verbose mode:

test_format (__main__.MyTestCase) ... skipped 'not supported in this library version'
test_nothing (__main__.MyTestCase) ... skipped 'demonstrating skipping'
test_windows_support (__main__.MyTestCase) ... skipped 'requires Windows'

Ran 3 tests in 0.005s

OK (skipped=3)

Classes can be skipped just like methods:

@unittest.skip("showing class skipping")
class MySkippedTestCase(unittest.TestCase):
    def test_not_run(self):

TestCase.setUp() can also skip the test. This is useful when a resource that needs to be set up is not available.

Expected failures use the expectedFailure() decorator.

class ExpectedFailureTestCase(unittest.TestCase):
    def test_fail(self):
        self.assertEqual(1, 0, "broken")

It’s easy to roll your own skipping decorators by making a decorator that calls skip() on the test when it wants it to be skipped. This decorator skips the test unless the passed object has a certain attribute:

def skipUnlessHasattr(obj, attr):
    if hasattr(obj, attr):
        return lambda func: func
    return unittest.skip("{!r} doesn't have {!r}".format(obj, attr))

The following decorators implement test skipping and expected failures:


Unconditionally skip the decorated test. reason should describe why the test is being skipped.

@unittest.skipIf(condition, reason)

Skip the decorated test if condition is true.

@unittest.skipUnless(condition, reason)

Skip the decorated test unless condition is true.


Mark the test as an expected failure. If the test fails when run, the test is not counted as a failure.

exception unittest.SkipTest(reason)

This exception is raised to skip a test.

Usually you can use TestCase.skipTest() or one of the skipping decorators instead of raising this directly.

Skipped tests will not have setUp() or tearDown() run around them. Skipped classes will not have setUpClass() or tearDownClass() run.

26.3.7. Classes and functions

This section describes in depth the API of unittest. Test cases

class unittest.TestCase(methodName='runTest')

Instances of the TestCase class represent the logical test units in the unittest universe. This class is intended to be used as a base class, with specific tests being implemented by concrete subclasses. This class implements the interface needed by the test runner to allow it to drive the tests, and methods that the test code can use to check for and report various kinds of failure.

Each instance of TestCase will run a single base method: the method named methodName. However, the standard implementation of the default methodName, runTest(), will run every method starting with test as an individual test, and count successes and failures accordingly. Therefore, in most uses of TestCase, you will neither change the methodName nor reimplement the default runTest() method.

Changed in version 3.2: TestCase can be instantiated successfully without providing a methodName. This makes it easier to experiment with TestCase from the interactive interpreter.

TestCase instances provide three groups of methods: one group used to run the test, another used by the test implementation to check conditions and report failures, and some inquiry methods allowing information about the test itself to be gathered.

Methods in the first group (running the test) are:


Method called to prepare the test fixture. This is called immediately before calling the test method; any exception raised by this method will be considered an error rather than a test failure. The default implementation does nothing.


Method called immediately after the test method has been called and the result recorded. This is called even if the test method raised an exception, so the implementation in subclasses may need to be particularly careful about checking internal state. Any exception raised by this method will be considered an error rather than a test failure. This method will only be called if the setUp() succeeds, regardless of the outcome of the test method. The default implementation does nothing.


A class method called before tests in an individual class run. setUpClass is called with the class as the only argument and must be decorated as a classmethod():

def setUpClass(cls):

See Class and Module Fixtures for more details.

New in version 3.2.


A class method called after tests in an individual class have run. tearDownClass is called with the class as the only argument and must be decorated as a classmethod():

def tearDownClass(cls):

See Class and Module Fixtures for more details.

New in version 3.2.


Run the test, collecting the result into the TestResult object passed as result. If result is omitted or None, a temporary result object is created (by calling the defaultTestResult() method) and used. The result object is returned to run()‘s caller.

The same effect may be had by simply calling the TestCase instance.

Changed in version 3.3: Previous versions of run did not return the result. Neither did calling an instance.


Calling this during a test method or setUp() skips the current test. See Skipping tests and expected failures for more information.

New in version 3.1.


Run the test without collecting the result. This allows exceptions raised by the test to be propagated to the caller, and can be used to support running tests under a debugger.

The TestCase class provides a number of methods to check for and report failures, such as:

Method Checks that New in
assertEqual(a, b) a == b  
assertNotEqual(a, b) a != b  
assertTrue(x) bool(x) is True  
assertFalse(x) bool(x) is False  
assertIs(a, b) a is b 3.1
assertIsNot(a, b) a is not b 3.1
assertIsNone(x) x is None 3.1
assertIsNotNone(x) x is not None 3.1
assertIn(a, b) a in b 3.1
assertNotIn(a, b) a not in b 3.1
assertIsInstance(a, b) isinstance(a, b) 3.2
assertNotIsInstance(a, b) not isinstance(a, b) 3.2

All the assert methods accept a msg argument that, if specified, is used as the error message on failure (see also longMessage). Note that the msg keyword argument can be passed to assertRaises(), assertRaisesRegex(), assertWarns(), assertWarnsRegex() only when they are used as a context manager.

assertEqual(first, second, msg=None)

Test that first and second are equal. If the values do not compare equal, the test will fail.

In addition, if first and second are the exact same type and one of list, tuple, dict, set, frozenset or str or any type that a subclass registers with addTypeEqualityFunc() the type-specific equality function will be called in order to generate a more useful default error message (see also the list of type-specific methods).

Changed in version 3.1: Added the automatic calling of type-specific equality function.

Changed in version 3.2: assertMultiLineEqual() added as the default type equality function for comparing strings.

assertNotEqual(first, second, msg=None)

Test that first and second are not equal. If the values do compare equal, the test will fail.

assertTrue(expr, msg=None)
assertFalse(expr, msg=None)

Test that expr is true (or false).

Note that this is equivalent to bool(expr) is True and not to expr is True (use assertIs(expr, True) for the latter). This method should also be avoided when more specific methods are available (e.g. assertEqual(a, b) instead of assertTrue(a == b)), because they provide a better error message in case of failure.

assertIs(first, second, msg=None)
assertIsNot(first, second, msg=None)

Test that first and second evaluate (or don’t evaluate) to the same object.

New in version 3.1.

assertIsNone(expr, msg=None)
assertIsNotNone(expr, msg=None)

Test that expr is (or is not) None.

New in version 3.1.

assertIn(first, second, msg=None)
assertNotIn(first, second, msg=None)

Test that first is (or is not) in second.

New in version 3.1.

assertIsInstance(obj, cls, msg=None)
assertNotIsInstance(obj, cls, msg=None)

Test that obj is (or is not) an instance of cls (which can be a class or a tuple of classes, as supported by isinstance()). To check for the exact type, use assertIs(type(obj), cls).

New in version 3.2.

It is also possible to check that exceptions and warnings are raised using the following methods:

Method Checks that New in
assertRaises(exc, fun, *args, **kwds) fun(*args, **kwds) raises exc  
assertRaisesRegex(exc, r, fun, *args, **kwds) fun(*args, **kwds) raises exc and the message matches regex r 3.1
assertWarns(warn, fun, *args, **kwds) fun(*args, **kwds) raises warn 3.2
assertWarnsRegex(warn, r, fun, *args, **kwds) fun(*args, **kwds) raises warn and the message matches regex r 3.2
assertRaises(exception, callable, *args, **kwds)
assertRaises(exception, msg=None)

Test that an exception is raised when callable is called with any positional or keyword arguments that are also passed to assertRaises(). The test passes if exception is raised, is an error if another exception is raised, or fails if no exception is raised. To catch any of a group of exceptions, a tuple containing the exception classes may be passed as exception.

If only the exception and possibly the msg arguments are given, return a context manager so that the code under test can be written inline rather than as a function:

with self.assertRaises(SomeException):

When used as a context manager, assertRaises() accepts the additional keyword argument msg.

The context manager will store the caught exception object in its exception attribute. This can be useful if the intention is to perform additional checks on the exception raised:

with self.assertRaises(SomeException) as cm:

the_exception = cm.exception
self.assertEqual(the_exception.error_code, 3)

Changed in version 3.1: Added the ability to use assertRaises() as a context manager.

Changed in version 3.2: Added the exception attribute.

Changed in version 3.3: Added the msg keyword argument when used as a context manager.

assertRaisesRegex(exception, regex, callable, *args, **kwds)
assertRaisesRegex(exception, regex, msg=None)

Like assertRaises() but also tests that regex matches on the string representation of the raised exception. regex may be a regular expression object or a string containing a regular expression suitable for use by re.search(). Examples:

self.assertRaisesRegex(ValueError, "invalid literal for.*XYZ'$",
                       int, 'XYZ')


with self.assertRaisesRegex(ValueError, 'literal'):

New in version 3.1: under the name assertRaisesRegexp.

Changed in version 3.2: Renamed to assertRaisesRegex().

Changed in version 3.3: Added the msg keyword argument when used as a context manager.

assertWarns(warning, callable, *args, **kwds)
assertWarns(warning, msg=None)

Test that a warning is triggered when callable is called with any positional or keyword arguments that are also passed to assertWarns(). The test passes if warning is triggered and fails if it isn’t. Any exception is an error. To catch any of a group of warnings, a tuple containing the warning classes may be passed as warnings.

If only the warning and possibly the msg arguments are given, return a context manager so that the code under test can be written inline rather than as a function:

with self.assertWarns(SomeWarning):

When used as a context manager, assertWarns() accepts the additional keyword argument msg.

The context manager will store the caught warning object in its warning attribute, and the source line which triggered the warnings in the filename and lineno attributes. This can be useful if the intention is to perform additional checks on the warning caught:

with self.assertWarns(SomeWarning) as cm:

self.assertIn('myfile.py', cm.filename)
self.assertEqual(320, cm.lineno)

This method works regardless of the warning filters in place when it is called.

New in version 3.2.

Changed in version 3.3: Added the msg keyword argument when used as a context manager.

assertWarnsRegex(warning, regex, callable, *args, **kwds)
assertWarnsRegex(warning, regex, msg=None)

Like assertWarns() but also tests that regex matches on the message of the triggered warning. regex may be a regular expression object or a string containing a regular expression suitable for use by re.search(). Example:

                      r'legacy_function\(\) is deprecated',
                      legacy_function, 'XYZ')


with self.assertWarnsRegex(RuntimeWarning, 'unsafe frobnicating'):

New in version 3.2.

Changed in version 3.3: Added the msg keyword argument when used as a context manager.

There are also other methods used to perform more specific checks, such as:

Method Checks that New in
assertAlmostEqual(a, b) round(a-b, 7) == 0  
assertNotAlmostEqual(a, b) round(a-b, 7) != 0  
assertGreater(a, b) a > b 3.1
assertGreaterEqual(a, b) a >= b 3.1
assertLess(a, b) a < b 3.1
assertLessEqual(a, b) a <= b 3.1
assertRegex(s, r) r.search(s) 3.1
assertNotRegex(s, r) not r.search(s) 3.2
assertCountEqual(a, b) a and b have the same elements in the same number, regardless of their order 3.2
assertAlmostEqual(first, second, places=7, msg=None, delta=None)
assertNotAlmostEqual(first, second, places=7, msg=None, delta=None)

Test that first and second are approximately (or not approximately) equal by computing the difference, rounding to the given number of decimal places (default 7), and comparing to zero. Note that these methods round the values to the given number of decimal places (i.e. like the round() function) and not significant digits.

If delta is supplied instead of places then the difference between first and second must be less or equal to (or greater than) delta.

Supplying both delta and places raises a TypeError.

Changed in version 3.2: assertAlmostEqual() automatically considers almost equal objects that compare equal. assertNotAlmostEqual() automatically fails if the objects compare equal. Added the delta keyword argument.

assertGreater(first, second, msg=None)
assertGreaterEqual(first, second, msg=None)
assertLess(first, second, msg=None)
assertLessEqual(first, second, msg=None)

Test that first is respectively >, >=, < or <= than second depending on the method name. If not, the test will fail:

>>> self.assertGreaterEqual(3, 4)
AssertionError: "3" unexpectedly not greater than or equal to "4"

New in version 3.1.

assertRegex(text, regex, msg=None)
assertNotRegex(text, regex, msg=None)

Test that a regex search matches (or does not match) text. In case of failure, the error message will include the pattern and the text (or the pattern and the part of text that unexpectedly matched). regex may be a regular expression object or a string containing a regular expression suitable for use by re.search().

New in version 3.1: under the name assertRegexpMatches.

Changed in version 3.2: The method assertRegexpMatches() has been renamed to assertRegex().

New in version 3.2: assertNotRegex().

assertCountEqual(first, second, msg=None)

Test that sequence first contains the same elements as second, regardless of their order. When they don’t, an error message listing the differences between the sequences will be generated.

Duplicate elements are not ignored when comparing first and second. It verifies whether each element has the same count in both sequences. Equivalent to: assertEqual(Counter(list(first)), Counter(list(second))) but works with sequences of unhashable objects as well.

New in version 3.2.

The assertEqual() method dispatches the equality check for objects of the same type to different type-specific methods. These methods are already implemented for most of the built-in types, but it’s also possible to register new methods using addTypeEqualityFunc():

addTypeEqualityFunc(typeobj, function)

Registers a type-specific method called by assertEqual() to check if two objects of exactly the same typeobj (not subclasses) compare equal. function must take two positional arguments and a third msg=None keyword argument just as assertEqual() does. It must raise self.failureException(msg) when inequality between the first two parameters is detected – possibly providing useful information and explaining the inequalities in details in the error message.

New in version 3.1.

The list of type-specific methods automatically used by assertEqual() are summarized in the following table. Note that it’s usually not necessary to invoke these methods directly.

Method Used to compare New in
assertMultiLineEqual(a, b) strings 3.1
assertSequenceEqual(a, b) sequences 3.1
assertListEqual(a, b) lists 3.1
assertTupleEqual(a, b) tuples 3.1
assertSetEqual(a, b) sets or frozensets 3.1
assertDictEqual(a, b) dicts 3.1
assertMultiLineEqual(first, second, msg=None)

Test that the multiline string first is equal to the string second. When not equal a diff of the two strings highlighting the differences will be included in the error message. This method is used by default when comparing strings with assertEqual().

New in version 3.1.

assertSequenceEqual(first, second, msg=None, seq_type=None)

Tests that two sequences are equal. If a seq_type is supplied, both first and second must be instances of seq_type or a failure will be raised. If the sequences are different an error message is constructed that shows the difference between the two.

This method is not called directly by assertEqual(), but it’s used to implement assertListEqual() and assertTupleEqual().

New in version 3.1.

assertListEqual(first, second, msg=None)
assertTupleEqual(first, second, msg=None)

Tests that two lists or tuples are equal. If not, an error message is constructed that shows only the differences between the two. An error is also raised if either of the parameters are of the wrong type. These methods are used by default when comparing lists or tuples with assertEqual().

New in version 3.1.

assertSetEqual(first, second, msg=None)

Tests that two sets are equal. If not, an error message is constructed that lists the differences between the sets. This method is used by default when comparing sets or frozensets with assertEqual().

Fails if either of first or second does not have a set.difference() method.

New in version 3.1.

assertDictEqual(first, second, msg=None)

Test that two dictionaries are equal. If not, an error message is constructed that shows the differences in the dictionaries. This method will be used by default to compare dictionaries in calls to assertEqual().

New in version 3.1.

Finally the TestCase provides the following methods and attributes:


Signals a test failure unconditionally, with msg or None for the error message.


This class attribute gives the exception raised by the test method. If a test framework needs to use a specialized exception, possibly to carry additional information, it must subclass this exception in order to “play fair” with the framework. The initial value of this attribute is AssertionError.


If set to True then any explicit failure message you pass in to the assert methods will be appended to the end of the normal failure message. The normal messages contain useful information about the objects involved, for example the message from assertEqual shows you the repr of the two unequal objects. Setting this attribute to True allows you to have a custom error message in addition to the normal one.

This attribute defaults to True. If set to False then a custom message passed to an assert method will silence the normal message.

The class setting can be overridden in individual tests by assigning an instance attribute to True or False before calling the assert methods.

New in version 3.1.


This attribute controls the maximum length of diffs output by assert methods that report diffs on failure. It defaults to 80*8 characters. Assert methods affected by this attribute are assertSequenceEqual() (including all the sequence comparison methods that delegate to it), assertDictEqual() and assertMultiLineEqual().

Setting maxDiff to None means that there is no maximum length of diffs.

New in version 3.2.

Testing frameworks can use the following methods to collect information on the test:


Return the number of tests represented by this test object. For TestCase instances, this will always be 1.


Return an instance of the test result class that should be used for this test case class (if no other result instance is provided to the run() method).

For TestCase instances, this will always be an instance of TestResult; subclasses of TestCase should override this as necessary.


Return a string identifying the specific test case. This is usually the full name of the test method, including the module and class name.


Returns a description of the test, or None if no description has been provided. The default implementation of this method returns the first line of the test method’s docstring, if available, or None.

Changed in version 3.1: In 3.1 this was changed to add the test name to the short description even in the presence of a docstring. This caused compatibility issues with unittest extensions and adding the test name was moved to the TextTestResult in Python 3.2.

addCleanup(function, *args, **kwargs)

Add a function to be called after tearDown() to cleanup resources used during the test. Functions will be called in reverse order to the order they are added (LIFO). They are called with any arguments and keyword arguments passed into addCleanup() when they are added.

If setUp() fails, meaning that tearDown() is not called, then any cleanup functions added will still be called.

New in version 3.1.


This method is called unconditionally after tearDown(), or after setUp() if setUp() raises an exception.

It is responsible for calling all the cleanup functions added by addCleanup(). If you need cleanup functions to be called prior to tearDown() then you can call doCleanups() yourself.

doCleanups() pops methods off the stack of cleanup functions one at a time, so it can be called at any time.

New in version 3.1.

class unittest.FunctionTestCase(testFunc, setUp=None, tearDown=None, description=None)

This class implements the portion of the TestCase interface which allows the test runner to drive the test, but does not provide the methods which test code can use to check and report errors. This is used to create test cases using legacy test code, allowing it to be integrated into a unittest-based test framework. Deprecated aliases

For historical reasons, some of the TestCase methods had one or more aliases that are now deprecated. The following table lists the correct names along with their deprecated aliases:

Method Name Deprecated alias Deprecated alias
assertEqual() failUnlessEqual assertEquals
assertNotEqual() failIfEqual assertNotEquals
assertTrue() failUnless assert_
assertFalse() failIf  
assertRaises() failUnlessRaises  
assertAlmostEqual() failUnlessAlmostEqual assertAlmostEquals
assertNotAlmostEqual() failIfAlmostEqual assertNotAlmostEquals
assertRegex()   assertRegexpMatches
assertRaisesRegex()   assertRaisesRegexp

Deprecated since version 3.1: the fail* aliases listed in the second column.

Deprecated since version 3.2: the assert* aliases listed in the third column.

Deprecated since version 3.2: assertRegexpMatches and assertRaisesRegexp have been renamed to assertRegex() and assertRaisesRegex() Grouping tests

class unittest.TestSuite(tests=())

This class represents an aggregation of individual tests cases and test suites. The class presents the interface needed by the test runner to allow it to be run as any other test case. Running a TestSuite instance is the same as iterating over the suite, running each test individually.

If tests is given, it must be an iterable of individual test cases or other test suites that will be used to build the suite initially. Additional methods are provided to add test cases and suites to the collection later on.

TestSuite objects behave much like TestCase objects, except they do not actually implement a test. Instead, they are used to aggregate tests into groups of tests that should be run together. Some additional methods are available to add tests to TestSuite instances:


Add a TestCase or TestSuite to the suite.


Add all the tests from an iterable of TestCase and TestSuite instances to this test suite.

This is equivalent to iterating over tests, calling addTest() for each element.

TestSuite shares the following methods with TestCase:


Run the tests associated with this suite, collecting the result into the test result object passed as result. Note that unlike TestCase.run(), TestSuite.run() requires the result object to be passed in.


Run the tests associated with this suite without collecting the result. This allows exceptions raised by the test to be propagated to the caller and can be used to support running tests under a debugger.


Return the number of tests represented by this test object, including all individual tests and sub-suites.


Tests grouped by a TestSuite are always accessed by iteration. Subclasses can lazily provide tests by overriding __iter__(). Note that this method maybe called several times on a single suite (for example when counting tests or comparing for equality) so the tests returned must be the same for repeated iterations.

Changed in version 3.2: In earlier versions the TestSuite accessed tests directly rather than through iteration, so overriding __iter__() wasn’t sufficient for providing tests.

In the typical usage of a TestSuite object, the run() method is invoked by a TestRunner rather than by the end-user test harness. Loading and running tests

class unittest.TestLoader

The TestLoader class is used to create test suites from classes and modules. Normally, there is no need to create an instance of this class; the unittest module provides an instance that can be shared as unittest.defaultTestLoader. Using a subclass or instance, however, allows customization of some configurable properties.

TestLoader objects have the following methods:


Return a suite of all tests cases contained in the TestCase-derived testCaseClass.


Return a suite of all tests cases contained in the given module. This method searches module for classes derived from TestCase and creates an instance of the class for each test method defined for the class.


While using a hierarchy of TestCase-derived classes can be convenient in sharing fixtures and helper functions, defining test methods on base classes that are not intended to be instantiated directly does not play well with this method. Doing so, however, can be useful when the fixtures are different and defined in subclasses.

If a module provides a load_tests function it will be called to load the tests. This allows modules to customize test loading. This is the load_tests protocol.

Changed in version 3.2: Support for load_tests added.

loadTestsFromName(name, module=None)

Return a suite of all tests cases given a string specifier.

The specifier name is a “dotted name” that may resolve either to a module, a test case class, a test method within a test case class, a TestSuite instance, or a callable object which returns a TestCase or TestSuite instance. These checks are applied in the order listed here; that is, a method on a possible test case class will be picked up as “a test method within a test case class”, rather than “a callable object”.

For example, if you have a module SampleTests containing a TestCase-derived class SampleTestCase with three test methods (test_one(), test_two(), and test_three()), the specifier 'SampleTests.SampleTestCase' would cause this method to return a suite which will run all three test methods. Using the specifier 'SampleTests.SampleTestCase.test_two' would cause it to return a test suite which will run only the test_two() test method. The specifier can refer to modules and packages which have not been imported; they will be imported as a side-effect.

The method optionally resolves name relative to the given module.

loadTestsFromNames(names, module=None)

Similar to loadTestsFromName(), but takes a sequence of names rather than a single name. The return value is a test suite which supports all the tests defined for each name.


Return a sorted sequence of method names found within testCaseClass; this should be a subclass of TestCase.

discover(start_dir, pattern='test*.py', top_level_dir=None)

Find all the test modules by recursing into subdirectories from the specified start directory, and return a TestSuite object containing them. Only test files that match pattern will be loaded. (Using shell style pattern matching.) Only module names that are importable (i.e. are valid Python identifiers) will be loaded.

All test modules must be importable from the top level of the project. If the start directory is not the top level directory then the top level directory must be specified separately.

If importing a module fails, for example due to a syntax error, then this will be recorded as a single error and discovery will continue.

If a test package name (directory with __init__.py) matches the pattern then the package will be checked for a load_tests function. If this exists then it will be called with loader, tests, pattern.

If load_tests exists then discovery does not recurse into the package, load_tests is responsible for loading all tests in the package.

The pattern is deliberately not stored as a loader attribute so that packages can continue discovery themselves. top_level_dir is stored so load_tests does not need to pass this argument in to loader.discover().

start_dir can be a dotted module name as well as a directory.

New in version 3.2.

The following attributes of a TestLoader can be configured either by subclassing or assignment on an instance:


String giving the prefix of method names which will be interpreted as test methods. The default value is 'test'.

This affects getTestCaseNames() and all the loadTestsFrom*() methods.


Function to be used to compare method names when sorting them in getTestCaseNames() and all the loadTestsFrom*() methods.


Callable object that constructs a test suite from a list of tests. No methods on the resulting object are needed. The default value is the TestSuite class.

This affects all the loadTestsFrom*() methods.

class unittest.TestResult

This class is used to compile information about which tests have succeeded and which have failed.

A TestResult object stores the results of a set of tests. The TestCase and TestSuite classes ensure that results are properly recorded; test authors do not need to worry about recording the outcome of tests.

Testing frameworks built on top of unittest may want access to the TestResult object generated by running a set of tests for reporting purposes; a TestResult instance is returned by the TestRunner.run() method for this purpose.

TestResult instances have the following attributes that will be of interest when inspecting the results of running a set of tests:


A list containing 2-tuples of TestCase instances and strings holding formatted tracebacks. Each tuple represents a test which raised an unexpected exception.


A list containing 2-tuples of TestCase instances and strings holding formatted tracebacks. Each tuple represents a test where a failure was explicitly signalled using the TestCase.assert*() methods.


A list containing 2-tuples of TestCase instances and strings holding the reason for skipping the test.

New in version 3.1.


A list containing 2-tuples of TestCase instances and strings holding formatted tracebacks. Each tuple represents an expected failure of the test case.


A list containing TestCase instances that were marked as expected failures, but succeeded.


Set to True when the execution of tests should stop by stop().


The total number of tests run so far.


If set to true, sys.stdout and sys.stderr will be buffered in between startTest() and stopTest() being called. Collected output will only be echoed onto the real sys.stdout and sys.stderr if the test fails or errors. Any output is also attached to the failure / error message.

New in version 3.2.


If set to true stop() will be called on the first failure or error, halting the test run.

New in version 3.2.


Return True if all tests run so far have passed, otherwise returns False.


This method can be called to signal that the set of tests being run should be aborted by setting the shouldStop attribute to True. TestRunner objects should respect this flag and return without running any additional tests.

For example, this feature is used by the TextTestRunner class to stop the test framework when the user signals an interrupt from the keyboard. Interactive tools which provide TestRunner implementations can use this in a similar manner.

The following methods of the TestResult class are used to maintain the internal data structures, and may be extended in subclasses to support additional reporting requirements. This is particularly useful in building tools which support interactive reporting while tests are being run.


Called when the test case test is about to be run.


Called after the test case test has been executed, regardless of the outcome.


Called once before any tests are executed.

New in version 3.1.


Called once after all tests are executed.

New in version 3.1.

addError(test, err)

Called when the test case test raises an unexpected exception. err is a tuple of the form returned by sys.exc_info(): (type, value, traceback).

The default implementation appends a tuple (test, formatted_err) to the instance’s errors attribute, where formatted_err is a formatted traceback derived from err.

addFailure(test, err)

Called when the test case test signals a failure. err is a tuple of the form returned by sys.exc_info(): (type, value, traceback).

The default implementation appends a tuple (test, formatted_err) to the instance’s failures attribute, where formatted_err is a formatted traceback derived from err.


Called when the test case test succeeds.

The default implementation does nothing.

addSkip(test, reason)

Called when the test case test is skipped. reason is the reason the test gave for skipping.

The default implementation appends a tuple (test, reason) to the instance’s skipped attribute.

addExpectedFailure(test, err)

Called when the test case test fails, but was marked with the expectedFailure() decorator.

The default implementation appends a tuple (test, formatted_err) to the instance’s expectedFailures attribute, where formatted_err is a formatted traceback derived from err.


Called when the test case test was marked with the expectedFailure() decorator, but succeeded.

The default implementation appends the test to the instance’s unexpectedSuccesses attribute.

class unittest.TextTestResult(stream, descriptions, verbosity)

A concrete implementation of TestResult used by the TextTestRunner.

New in version 3.2: This class was previously named _TextTestResult. The old name still exists as an alias but is deprecated.


Instance of the TestLoader class intended to be shared. If no customization of the TestLoader is needed, this instance can be used instead of repeatedly creating new instances.

class unittest.TextTestRunner(stream=None, descriptions=True, verbosity=1, failfast=False, buffer=False, resultclass=None, warnings=None)

A basic test runner implementation that outputs results to a stream. If stream is None, the default, sys.stderr is used as the output stream. This class has a few configurable parameters, but is essentially very simple. Graphical applications which run test suites should provide alternate implementations.

By default this runner shows DeprecationWarning, PendingDeprecationWarning, ResourceWarning and ImportWarning even if they are ignored by default. Deprecation warnings caused by deprecated unittest methods are also special-cased and, when the warning filters are 'default' or 'always', they will appear only once per-module, in order to avoid too many warning messages. This behavior can be overridden using the -Wd or -Wa options and leaving warnings to None.

Changed in version 3.2: Added the warnings argument.

Changed in version 3.2: The default stream is set to sys.stderr at instantiation time rather than import time.


This method returns the instance of TestResult used by run(). It is not intended to be called directly, but can be overridden in subclasses to provide a custom TestResult.

_makeResult() instantiates the class or callable passed in the TextTestRunner constructor as the resultclass argument. It defaults to TextTestResult if no resultclass is provided. The result class is instantiated with the following arguments:

stream, descriptions, verbosity
unittest.main(module='__main__', defaultTest=None, argv=None, testRunner=None, testLoader=unittest.defaultTestLoader, exit=True, verbosity=1, failfast=None, catchbreak=None, buffer=None, warnings=None)

A command-line program that loads a set of tests from module and runs them; this is primarily for making test modules conveniently executable. The simplest use for this function is to include the following line at the end of a test script:

if __name__ == '__main__':

You can run tests with more detailed information by passing in the verbosity argument:

if __name__ == '__main__':

The defaultTest argument is the name of the test to run if no test names are specified via argv. If not specified or None and no test names are provided via argv, all tests found in module are run.

The argv argument can be a list of options passed to the program, with the first element being the program name. If not specified or None, the values of sys.argv are used.

The testRunner argument can either be a test runner class or an already created instance of it. By default main calls sys.exit() with an exit code indicating success or failure of the tests run.

The testLoader argument has to be a TestLoader instance, and defaults to defaultTestLoader.

main supports being used from the interactive interpreter by passing in the argument exit=False. This displays the result on standard output without calling sys.exit():

>>> from unittest import main
>>> main(module='test_module', exit=False)

The failfast, catchbreak and buffer parameters have the same effect as the same-name command-line options.

The warning argument specifies the warning filter that should be used while running the tests. If it’s not specified, it will remain None if a -W option is passed to python, otherwise it will be set to 'default'.

Calling main actually returns an instance of the TestProgram class. This stores the result of the tests run as the result attribute.

Changed in version 3.1: The exit parameter was added.

Changed in version 3.2: The verbosity, failfast, catchbreak, buffer and warnings parameters were added. load_tests Protocol

New in version 3.2.

Modules or packages can customize how tests are loaded from them during normal test runs or test discovery by implementing a function called load_tests.

If a test module defines load_tests it will be called by TestLoader.loadTestsFromModule() with the following arguments:

load_tests(loader, standard_tests, None)

It should return a TestSuite.

loader is the instance of TestLoader doing the loading. standard_tests are the tests that would be loaded by default from the module. It is common for test modules to only want to add or remove tests from the standard set of tests. The third argument is used when loading packages as part of test discovery.

A typical load_tests function that loads tests from a specific set of TestCase classes may look like:

test_cases = (TestCase1, TestCase2, TestCase3)

def load_tests(loader, tests, pattern):
    suite = TestSuite()
    for test_class in test_cases:
        tests = loader.loadTestsFromTestCase(test_class)
    return suite

If discovery is started, either from the command line or by calling TestLoader.discover(), with a pattern that matches a package name then the package __init__.py will be checked for load_tests.


The default pattern is 'test*.py'. This matches all Python files that start with 'test' but won’t match any test directories.

A pattern like 'test*' will match test packages as well as modules.

If the package __init__.py defines load_tests then it will be called and discovery not continued into the package. load_tests is called with the following arguments:

load_tests(loader, standard_tests, pattern)

This should return a TestSuite representing all the tests from the package. (standard_tests will only contain tests collected from __init__.py.)

Because the pattern is passed into load_tests the package is free to continue (and potentially modify) test discovery. A ‘do nothing’ load_tests function for a test package would look like:

def load_tests(loader, standard_tests, pattern):
    # top level directory cached on loader instance
    this_dir = os.path.dirname(__file__)
    package_tests = loader.discover(start_dir=this_dir, pattern=pattern)
    return standard_tests

26.3.8. Class and Module Fixtures

Class and module level fixtures are implemented in TestSuite. When the test suite encounters a test from a new class then tearDownClass() from the previous class (if there is one) is called, followed by setUpClass() from the new class.

Similarly if a test is from a different module from the previous test then tearDownModule from the previous module is run, followed by setUpModule from the new module.

After all the tests have run the final tearDownClass and tearDownModule are run.

Note that shared fixtures do not play well with [potential] features like test parallelization and they break test isolation. They should be used with care.

The default ordering of tests created by the unittest test loaders is to group all tests from the same modules and classes together. This will lead to setUpClass / setUpModule (etc) being called exactly once per class and module. If you randomize the order, so that tests from different modules and classes are adjacent to each other, then these shared fixture functions may be called multiple times in a single test run.

Shared fixtures are not intended to work with suites with non-standard ordering. A BaseTestSuite still exists for frameworks that don’t want to support shared fixtures.

If there are any exceptions raised during one of the shared fixture functions the test is reported as an error. Because there is no corresponding test instance an _ErrorHolder object (that has the same interface as a TestCase) is created to represent the error. If you are just using the standard unittest test runner then this detail doesn’t matter, but if you are a framework author it may be relevant. setUpClass and tearDownClass

These must be implemented as class methods:

import unittest

class Test(unittest.TestCase):
    def setUpClass(cls):
        cls._connection = createExpensiveConnectionObject()

    def tearDownClass(cls):

If you want the setUpClass and tearDownClass on base classes called then you must call up to them yourself. The implementations in TestCase are empty.

If an exception is raised during a setUpClass then the tests in the class are not run and the tearDownClass is not run. Skipped classes will not have setUpClass or tearDownClass run. If the exception is a SkipTest exception then the class will be reported as having been skipped instead of as an error. setUpModule and tearDownModule

These should be implemented as functions:

def setUpModule():

def tearDownModule():

If an exception is raised in a setUpModule then none of the tests in the module will be run and the tearDownModule will not be run. If the exception is a SkipTest exception then the module will be reported as having been skipped instead of as an error.

26.3.9. Signal Handling

New in version 3.2.

The -c/--catch command-line option to unittest, along with the catchbreak parameter to unittest.main(), provide more friendly handling of control-C during a test run. With catch break behavior enabled control-C will allow the currently running test to complete, and the test run will then end and report all the results so far. A second control-c will raise a KeyboardInterrupt in the usual way.

The control-c handling signal handler attempts to remain compatible with code or tests that install their own signal.SIGINT handler. If the unittest handler is called but isn’t the installed signal.SIGINT handler, i.e. it has been replaced by the system under test and delegated to, then it calls the default handler. This will normally be the expected behavior by code that replaces an installed handler and delegates to it. For individual tests that need unittest control-c handling disabled the removeHandler() decorator can be used.

There are a few utility functions for framework authors to enable control-c handling functionality within test frameworks.


Install the control-c handler. When a signal.SIGINT is received (usually in response to the user pressing control-c) all registered results have stop() called.


Register a TestResult object for control-c handling. Registering a result stores a weak reference to it, so it doesn’t prevent the result from being garbage collected.

Registering a TestResult object has no side-effects if control-c handling is not enabled, so test frameworks can unconditionally register all results they create independently of whether or not handling is enabled.


Remove a registered result. Once a result has been removed then stop() will no longer be called on that result object in response to a control-c.


When called without arguments this function removes the control-c handler if it has been installed. This function can also be used as a test decorator to temporarily remove the handler whilst the test is being executed:

def test_signal_handling(self):