A simple testing library for your python code.
Key features:
Via pip:
pip install checking
from checking import *
def my_function_to_test(a,b):
return a + b
@test
def any_name_you_like():
# Assert 1+1=2
equals(2, my_function_to_test(1,1))
if __name__ == '__main__':
# Run all tests in current module
start()
Only functions wrapped with the @test decorator will be marked as tests to execute. You can name your tests however you like, just put that @test decorator.
You can use standard Python asserts if you want, but it is recommended to use simple and readable asserts this library provides.
By default, even if you will use simple python assert like
assert 1==2
checking will give you a readable message, for example
Objects are not equals (1 != 2)
@test
def checks_basic_asserts():
is_true('1', 'Error message') # checks, if the first arg equals to True
is_false([], 'Error message') # checks, if the first arg equals to False
equals(1, 1, 'Error message') # checks, if two args are equal (==)
not_equals(1, 2, 'Error message') # checks, if two args are NOT equal (!=)
is_none(None, 'Error message') # checks, if the first arg is None
is_not_none('1', 'Error message') # checks, if the first arg is not None
contains(1, [1, 2, 3], 'Error message') # checks, if the second arg contains the first arg
not_contains(4, [1, 2, 3], 'Error message') # checks, if the second arg does not contain the first arg
is_zero(0, 'Error message') # checks, if the first arg is equal to 0 (assumed to be int or float)
is_positive(1, 'Error message') # checks, if the first arg is greater than 0 (for int or float), or len of the arg is non-zero (for Sequence)
is_negative(-1, 'Error message') # checks, if the first arg is less than 0 (assumed to be int or float)
is_empty([], 'Error message') # checks, if the length of the first arg (Sized type) equals 0, e.g. a collection is empty
is_not_empty([1,2], 'Error message') # checks, if the length of first arg (Sized type) is greater than 0, e.g. a collection is NOT empty
Messages in all asserts are optional, but it's strongly recommended using them.
If you need to check if an exception is raised or the message it contains, you can use the provided context manager:
@test
def check_with_exception():
with should_raise(ZeroDivisionError) as e:
x = 1 / 0 # Force ZeroDivisionError
assert e.message == 'division by zero' # Note, that the exception is wrapped into a library provided convenience class
The test will fail if no exception is raised or the exception is of another type. Please note, that you have to check the exception message after the context manager exits, not within it's scope.
The library forbids the usage of the BaseException type here, and it is strongly recommended not to use the Exception type as well. Use concrete exception types.
In some cases you only need to run the code and make sure no exception is raised. There is a special way to do that:
@test
def no_exceptions_bad():
do_something() # Wrong: no asserts here, this is not a proper test
@test
def check_no_exception_raises():
with no_exception_expected(): # Correct: explicitly state that we do not expect any exceptions
do_something()
The test fails if any exception is raised.
Sometimes, you need to fail, skip or break a test during runtime due to some reason (wrong OS, wrong parameters, etc.)
@test
def must_fail_on_condition():
if some_condition():
test_fail('Expected to fail.')
@test
def must_be_broken():
if some_condition():
test_break('Expected to break.')
@test
def must_be_ignored():
if some_condition():
test_skip('Expected to be ignored.')
The standard testing procedure compels the "fail fast" workflow: the whole test should halt if a single check fails. But sometimes you need to check a bunch of conditions and only fail at the end of the test if needed, collecting all the information on the executed checks. Soft Assert is a simple and convenient tool designed for this purpose.
For example, you need to check all the fields in a JSON object, collecting the info on which fields are correct and which are not:
@test
def check_all_json_fields():
my_json = fetch_json()
soft_assert = SoftAssert()
soft_assert.check(lambda : equals(1, my_json['field1'], 'message')) # Check a field, the test will not stop executing on failure
soft_assert.check(lambda : equals('text', my_json['field2'], 'message'))
soft_assert.contains(1, my_json['list'])
soft_assert.assert_all() # The test will fail here if some of the checks failed earlier
Attention! You must use assert_all() at the end of the test to actually raise an assertion exception if something went wrong.
Fluent assert is just syntactic sugar to make a series of checks for an object simpler and more readable. Fluent assert is not a soft assert, if one of the checks fails -- the whole chain halts. Fluent assert interface has methods analogous to the basic asserts as well as exclusive ones:
@test
def check_fluents():
my_list = fetch_list()
# Check if 'my_list' is not None, is instance of 'list', contains 2 and is sorted
verify(my_list).is_not_none().AND.type_is(list).AND.contains(2).AND.is_sorted()
some_object = SomeClass()
same_object = some_object
other_object = SomeClass()
# Check if objects are the same, and are not same with another object
verify(some_object).same_as(same_object).not_same_as(other_object)
# Check if 1 is in list [1, 2], not is in set {3, 4, 5} and is greater than 0
verify(1).is_in([1, 2]).is_not_in({3, 4, 5}).greater_than(0)
There are special "switches" to check some property of an object. Please note, that after invoking a switch, you cannot return to the original object you started with.
@test
def switch_to_length():
# Check if the length of the given list is positive, equals to 2 and is greater than the length of [1]
# the 'size' method is a switch -- all of the following checks will be executed against
# the int object (length of the starting object), NOT against the [1, 2] list
verify([1,2]).size.is_positive().AND.equal(2).AND.greater_than_length_of([1])
@test
def switch_to_attribute():
class Example:
pass
ex = Example()
ex.x = 100
# Check if the object has attribute 'x' and whether it is equal to 100
# the 'attribute' method is a switch -- all of the following checks will be executed against
# the int object ('x' attribute of 'ex'), and NOT against 'ex' object itself
verify(ex).has_attribute('x').AND.attribute('x').equal(100)
Often you need to run the same test with different data, there is @provider annotation for that target. Mark function you want with @provider annotation and you can use it in your tests. The function for data-provider should not have arguments, and it has to return iterable, sequence or generator.
Important! Name of the provider has to be unique, you can specify it in parameter
@provider(name='provider')
or it takes the function name by default. It is not necessary to have data-provider with the test (in the same module)
The data provider takes sets of values from the iterable one by one and pushes them onto your test.
# Create data-provider
@provider
def pairs():
return [(1, 1, 2), (2, 2, 4)] # Each tuple in a list is a unit set of data a test will be run against
@test(data_provider='pairs') # Specify which provider to use
def check_sum(it): # The test function must take one argument, the data provider will supply the test data via this argument
equals(it[0] + it[1], it[2]) # Run an assert on the data unit
You can specify mapping function to map values from the provider to some format, this string representation will be shown in test parameter, by default it use str(value) result. Pay attention - mapping function just change parameter representation in logs, console or report, but not the values itself!
from checking import *
class Cat:
def show(self):
return f'Cat from {id(self)}'
@provider(map_to_str=Cat.show) # uses show() of Cat to show values
def cats():
return (Cat(), Cat())
@test(data_provider='cats')
def check_cat(it):
assert isinstance(it, Cat) # assert 'it' is a Cat object
if __name__ == '__main__':
start(3)
In test logs you will see following information
Test "__main__.check_cat" [Cat from 140288585438160] SUCCESS!
----------
Test "__main__.check_cat" [Cat from 140288585437776] SUCCESS!
If you want to use a text file as a data source, you can use DATA_FILE helper function to skip the file handling boilerplate code:
from checking import *
DATA_FILE('files/data.txt', name='provider') # Use the file located at <module folder>/files/data.txt
@test(data_provider='provider')
def try_prov(it):
print(it)
is_true(it)
The helper lazy-loads specified data file line by line. Raises FileNotFoundError if the file is not found. Also, you can transform all the lines before feeding them into the test, for example delete trailing newlines at the end of each line:
from checking import *
DATA_FILE('files/data.txt', name='provider', map_function=str.rstrip) # Feed each line through str.rstrip()
@test(data_provider='provider')
def try_prov(it):
is_true(it)
If you don't specify provider_name for the DATA_FILE helper, file_path will be used:
from checking import *
DATA_FILE('data.txt') # Use text file located at the module folder. Note, that no provider_name is specified.
@test(data_provider='data.txt') # Use the specified file_name parameter for provider lookup
def try_prov(it):
is_true(it)
If your test suite uses a data provider more than once, you might want to avoid the IO overhead,
if this provider fetches the data from some external source (database, file system, http request etc.).
You can use the cached parameter to force the provider to fetch the data only once and store it into memory.
Please, be varied of the memory consumption, because the cache persists until the whole suite is done running.
Also, be careful when using the cache when running tests in parallel.
DATA_FILE helper can use this parameter too.
from checking import *
DATA_FILE('data.csv', name='csv', cached=True) # Enable caching
@test(data_provider='csv') # First provider use -- data is fetched from the file and stored into memory
def check_one(it):
not_none(it)
@test(data_provider='csv') # Second use -- no file reads, cached data is used
def check_two(it):
not_none(it)
if __name__ == '__main__':
start(0)
If your provider is a simple one-liner (string, list comprehension, generator expression, etc.), you can use the CONTAINER helper function to avoid full function definition boilerplate:
from checking import *
CONTAINER([e for e in range(10)], name='range') # Provide data from a listcomps, set provider name to 'range'
@test(data_provider='range')
def try_container(it):
is_true(it in range(10))
'name' parameter is optional, 'container' is used by default, but it's strongly recommended using a unique name:
from checking import *
CONTAINER((e for e in range(10))) # Provide data from a genexps
@test(data_provider='container')
def try_container(it):
is_true(it in range(10))
Important! You must define DATA_FILE or CONTAINER providers at the module scope, not in the fixtures and tests.
You can manage the test execution mode by passing a number of parameters to the @test decorator:
enabled (bool) - if set to False, the test will be skipped, all other parameters are ignored. By default, set to True.
name (str) - the name of the test. Is bound to the decorated function name if not specified.
description (str) - test description. If absent, the test function docstring is used. If both description and docstring are present, description takes precedence.
data_provider (str) - the name of the data provider to use with the test. If specified, the test function must take one argument to be fed with the data from the provider. Raises UnknownProviderName if no providers with the specified name found.
retries (int) - the number of times to run the failing test. If test does not fail, no more runs attempted. By default, set to 1.
groups (Tuple[str]) - a tuple of strings, representing the test group names a test is a part of. All tests belong to some test group, the default group holds all tests from the current module and is named after the module. Use this parameter to manage test execution groups.
priority (int) - test priority. The higher the value the later the test will be executed. Use this parameter to fine tune test run order. By default, set to 0.
timeout (int) - amount of time to wait for the test to end. If the time runs out, the thread running the test is terminated and the test is marked as "broken". Use sparingly due to potential memory leaks.
only_if (Callable[None, bool]) - boolean predicate, which is evaluated before the test execution. The test will be executed only if the predicate evaluates to True. Use this parameter for conditional test execution e.g. run only if the OS is Linux, etc.
Each test group or all test-suite can have preconditions and post-actions. For example, open DB connection before test starts and close it after that. You can easily make it with before/after fixtures. The function that marked with before/after should be without arguments.
@before - run function before EACH test in group, by default group is current module, but you can specify it with parameter
@after - run function after EACH test in group, by default group is current module, but you can specify it with parameter. This function will not be run if there is @before and it failed!
@before(group_name='api')
def my_func():
do_some_precondition()
@after(group_name='api')
def another_func():
do_post_actions()
@before_group - function run once before running test in group, by default group is current module, but you can specify it with parameter.
@after_group - function run once after running all test in group, by default group is current module, but you can specify it with parameter. This function will not be run if there is @before_group and it failed, except using parameter always_run = True
@before_group(name='api')
def my_func():
do_some_precondition_for_whole_group()
@after_group(name='api', always_run =True)
def another_func():
do_post_actions_for_whole_group()
@before_suite - function runs once before any group at start of the test-suite
@after_suite - function run once after all groups, at the end of the test-suite. This function will not be run if there is @before_suite, and it failed, except using parameter 'always_run = True'
@before_suite
def my_func():
print('start suite!')
@after_suite(always_run=True)
def another_func():
print('will be printed, even if before_suite failed!')
For testing purposes you sometimes need to fake some behaviour or to isolate your application from any other classes/libraries etc.
If you need your test to use fake object, without doing any real calls, you can use mocks:
1. Fake one of the builtin function.
Let say you need to test function which is using standard input() inside.
But you cannot wait for real user input during the test, so fake it with mock object.
def our_weird_function_with_input_inside():
text = input()
return text.upper()
@test
def mock_builtins_input():
with mock_builtins('input', lambda : 'test'): # Now input() just returns 'test', it does not wait for user input.
result_text = our_weird_function_with_input_inside()
equals('TEST', result_text)
More convenient way is to use mock_input or mock_print for simple and most common cases. From code above we can test our_weird_function this way
@test
def check_input():
with mock_input(['test']): # Now input() just returns 'test', it does not wait for user input.
result_text = our_weird_function_with_input_inside()
equals('TEST', result_text)
Now let's say we have simple function with print inside and need to test it:
def my_print(x):
print(x)
@test
def check_print():
with mock_print([]) as result: # now print just collects all to list result
my_print(1)
my_print('1')
equals([(1,), ('1',)], result) # checks all args are in result list
and more complicated case, when our function works forever, printing all inputs, until gets 'exit':
def use_both():
while True:
word = input('text>>>')
if word == 'exit':
break
print(word)
@test
def check_print_and_input():
# you can see inputs will get 'a','b' and 'exit' to break cycle, all args will
# be collected to result list
with mock_input(['a', 'b', 'exit']), mock_print([]) as result:
use_both()
equals([('a',), ('b',)], result)
2. Fake function of the 3-d party library
For working with other modules and libraries in test module, you need to import this module and to mock it function. For example, you need to test function, which is using requests.get inside, but you do not want to make real http request. Let it mock
some_module_to_test.py
import requests
def func_with_get_inside(url):
response = requests.get(url)
return response.text
our_tests.py
import requests # need to import it for mock!
from some_module_to_test import func_with_get_inside
@test
def mock_requests_get():
stub = Stub(text='test') # create simple stub, with attribute text equals to 'test'
with mock(requests, 'get', lambda x: stub): # Mock real requests with stub object
equals('test', func_with_get_inside('https://yandex.ru')) # Now no real requests be performed!
3. Mock read/write to file
If you need to mock open function, push data to read from file and gets back with write to file, you can use mock_open context-manager
def my_open():
# We read from one file, uppercase results and write to another file
with open('my_file.txt', encoding='utf-8') as f, open('another.txt', 'wt') as f2:
f2.write(f.readline().upper())
@test
def mock_open_both():
# Here we specify what we must "read from file" ('test') and where we want to get all writes(result)
with mock_open(on_read_text='test') as result:
my_open()
equals(['TEST'], result) # checks we get test uppercase
4. Spy object
Spy is the object which has all attributes of original, but spy not performed any action, all methods return None (if not specified what to return). Therefore, spy log all actions and arguments. It can be useful if your code has inner object, and you need to test what functions were called.
def function_with_str_inside(value):
# Suppose we need to check upper was called here inside
return value.upper()
@test
def spy_for_str():
spy = Spy('it is a string') # Spy, which is like str, but it is not str!
function_with_str_inside(spy) # Send our spy instead a str
is_true(spy.upper.was_called()) # Verify upper was called
You can even specify what to return when some function of the spy will be called!
def function_with_str_inside(value):
# Suppose we need to check upper was called here inside
return value.upper()
@test
def spy_with_return():
spy = Spy('string')
spy.upper.returns('test') # Tells what to return, when upper will be call
result = function_with_str_inside(spy)
is_true(spy.upper.was_called())
equals('test', result) # verify our spy returns 'test'
Spy object can be created without original inner object and can be call itself, it can be useful when you need some dumb object to know it was called.
@test
def check_spy():
spy = Spy() # Create "empty" spy
spy() # Call it
is_true(spy.was_called()) # Checks spy was called
5. TestDouble object
Test-Double object is like the Spy, but it saves original object behaviour, so its methods returns real object methods results if not specified otherwise.
@test
def check_double():
spy = TestDouble("string") # Create str double-object
equals(6, len(spy)) # Len returns 6 - the real length of original object ("string")
spy.len.returns(100) # Fake len result
equals(100, len(spy)) # Len now returns 100
Important! Both spy and TestDouble override isinstance, so they emulate type of the original object. It can be useful for testing functions, which has isinstance check inside. For example:
def function_that_checks_class(obj):
if isinstance(obj, str): # check for argument type (string)
return "OK"
return "Not OK"
@test
def isinstance_check():
spy = Spy("fake string") # fake the real string
result = function_that_checks_class(spy) # get "OK" here, cause function thinks it's a string, not Spy
equals("OK", result)
6. Stub object
Stub object is just a helper for testing, its purpose not to check or assert something, but to give data and perform some simple action, when application under test need it. Unlike spy or double, Stub is not remember calls, it just a simple replacement for some object with minimum or no logic inside.
Let's say we have a function which gets some object, take its attribute, calculates something and return result. We wish to isolate our testing from real objects, just test important behaviour, besides this data-object can be hard to create or complicated.
from checking import *
# Our function to test, it get some object and use it attribute and method, but we just
# need to test how it works!
def function(some_object)->int:
initial_value = some_object.value
result = 2 + some_object.complicate_function()*initial_value # Some calculation we need to test
return result
@test
def check_with_stub():
stub = Stub(value=2) # Creates stub with attribute value=2
stub.complicate_function.returns(2) # Says, when complicate_function will be called returns 2
equals(6, function(stub)) # Asserts 6 == 2+(2*2)
Pay attention - when you look for some attribute in stub - it always has it! But it will be a wrapper to use with
expression like stub.any_attribute.returns('test').
So, if you need to have some attribute (not method) on stub, you just use stub.attr=10, but for methods just use expression above.
You can execute all test at current module using function start(). For example:
from checking import *
@test
def some_check():
equals(4, 2+2)
if __name__ == '__main__':
start(3) # Here we run our test function some_check
There are parameters to run your tests in different ways:
suite_name - name of the test-suite, to use in reports or in logs
listener - object of Listener class, test listener, is the way to work with test results and execution DefaultListener is used by default. If set, then the verbose parameter is ignored (the one in the listener is used).
verbose is the report detail, 0 - briefly (only dots and 1 letter), 1 - detail, indicating only failed tests, 2 - detail, indicating successful and fallen tests, 3 - detail and at the end, a list of fallen and broken ones If verbose is not between 0 and 3, then 0 is accepted
Example (name and verbose)
from checking import *
@test
def some_check():
equals(4, 2 + 2)
@test
def some_check_two():
equals(2, 1 + 1)
@test
def failed():
equals(5, 2 + 2) # Will fail
@test
def broken():
int('a') # Will be broken
if __name__ == '__main__':
start(suite_name='My Suite', verbose=0)
This code will gave output (mention dots and chars!):
Starting suite "My Suite"
..FB
==============================
Total tests: 4, success tests: 2, failed tests: 1, broken tests: 1, ignored tests: 0
Time elapsed: 0.00 seconds.
Success percent: 50.00%
==============================
If you use parameter verbose=3 in example above, result will be:
Starting suite "My Suite"
----------
Test "__main__.some_check" SUCCESS!
----------
Test "__main__.some_check_two" SUCCESS!
----------
Test "__main__.failed" FAILED!
File "\checking\runner.py", line 285, in _run_test
--> test.run()
File "your_module_with_test_path", line 16, in failed
--> equals(5, 2 + 2) # Will fail
Objects are not equal!
Expected:"5" <int>
Actual :"4" <int>!
----------
Test "__main__.broken" BROKEN!
File "\checking\runner.py", line 285, in _run_test
--> test.run()
File "your_module_with_test_path", line 21, in broken
--> int('a') # Will be broken
invalid literal for int() with base 10: 'a'
==============================
Total tests: 4, success tests: 2, failed tests: 1, broken tests: 1, ignored tests: 0
Time elapsed: 0.00 seconds.
Success percent: 50.00%
Failed tests are:
__main__.failed
Broken tests are:
__main__.broken
==============================
groups is the list of test-group names to run. Only tests with that group will be run.
from checking import *
@test(groups=('api',))
def api_check():
equals(4, 2 + 2)
@test(groups=('ui',))
def ui_check():
equals(2, 1 + 1)
if __name__ == '__main__':
start(verbose=3, groups=['api'])
When you run this example, only function api_check will be executed, because we specify groups to run.
params is the dictionary of parameters available in all tests (general run parameters)
from checking import *
@test
def api_check():
equals(4, 2 + common['value']) # Here we use common - object available from all tests
if __name__ == '__main__':
start(verbose=3, params={'value': 2}) # Here we adds a parameter to common object
You can also mark any function as common, for example if you need to use it in different tests in
different modules.
from checking import *
@common_function
def sum_of_two(a, b):
return a + b
@test
def test_common_function():
equals(4, common.sum_of_two(2, 2)) # here we use dot notation to get common attribute
threads is the number of threads to run tests, by default is 1. Each group can run in a separate thread if necessary. This is an experimental feature, and it can be useful only for tests NOT performing any complex calculations (CPU bound). It is best to use this parameter (more than 1) for tests related to the use of I / O operations - disk work, network requests. Obey the GIL!
dry_run if True runs test-suite with fake function except of real tests and fixtures, can be useful to find out order, number of tests, params of provider etc. No real tests or fixtures will be executed!
filter_by_name if specified - runs only tests with name containing this parameter
from checking import *
@test
def api_check():
equals(4, 2 + 2)
@test
def ui_check():
equals(2, 1 + 1)
if __name__ == '__main__':
start(verbose=3, filter_by_name='ui')
In example above only function ui_name will be runs, because name of the function (ui_check) contains 'ui'.
random_order if True - runs tests inside each group in random order. Can be useful to make sure your tests really independent as they should be.
max_fail if greater than 0, then suite will stop, when failed tests count reach that number. For example, if you specify max_fail=1, then suite will stop after first failure. Pay attention that real failed test count can be bigger than max_fail if you use parallel execution, so parallel test will not interrupt until ends, even if count is reached.
generate_report if True - creates html report with the results in test folder. Experimental!
You can run all your tests in current folder and all sub-folders with your terminal:
python -m checking
There are few parameters for run in terminal:
-o options.json tells looking at the file options.json for test-suite parameters. If specified, -d and -f options will be ignored!
-g just generates default .json file for your options! If specified, all other options will be ignored!
-a arg or -a key=value add argument to common parameters of the suite to use later at tests
-d dry-run mode, no real tests will be executed, just collects and counts.
-f name filters test name, only test whose name contains filter will be executed
-r runs all tests in random order (priority will be ignored)
-m int_argument runs all tests till not reach specified number of failed tests
-R generate html-report with results of the tests
Options file is a way to manage your suite(s), you can have a few of them with different names to use in any case you need, for example 1 file for run all tests, another to run only api tests etc. File must have .json extension and contains valid json. For your convenience you can generate one with
python -m checking -g
In current working folder a file will appear with content like:
{
"suite_name": "Default Test Suite",
"verbose": 0,
"groups": [],
"params": {},
"listener": "",
"modules": [],
"threads": 1,
"dry_run": false,
"filter_by_name": "",
"random_order": false,
"max_fail": 0,
"generate_report": false
}
Changing these parameters you can manage your suites and test - for example specify what listener to use, or what group to run only. Some rules for parameters:
All types must be as in example, so you can't put string to "verbose" it must be int, etc.
If groups not empty ("groups":["api"]) then only group with that name will run. If no such group found, no tests will executed
Listener must be specified with module, like "listener": "my_module.MyListener". It is not necessary to specify whole path, just module name and class name. If not specified, default listener will be used. You can use default listener names here, like "listener":"DefaultFileListener"
Modules must be specified without ".py"! If modules parameter is empty, then all found modules with tests will be imported. If modules specified("modules":["my_package.my_module"]) only that modules be imported, and tests wil be collected from it. You can specify just module names or package.module (no need to specify full path)
FAQs
A small library for unit-testing
We found that checking demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
PyPI confirms no security flaws were exploited in the Ultralytics supply chain attack and highlights improvements for safer package publishing.
Security News
Research
The Socket Research Team breaks down a malicious wrapper package that uses obfuscation to harvest credentials and exfiltrate sensitive data.
Research
Security News
Attackers used a malicious npm package typosquatting a popular ESLint plugin to steal sensitive data, execute commands, and exploit developer systems.