/allpairspy

A python library for test combinations generator. The generator allows one to create a set of tests using "pairwise combinations" method, reducing a number of combinations of variables into a lesser set that covers most situations.

Primary LanguagePythonMIT LicenseMIT

allpairspy forked from bayandin/allpairs

PyPI package version Supported Python versions Linux/macOS CI status Windows CI status Test coverage

AllPairs test combinations generator

AllPairs is an open source test combinations generator written in Python, developed and maintained by MetaCommunications Engineering. The generator allows one to create a set of tests using "pairwise combinations" method, reducing a number of combinations of variables into a lesser set that covers most situations.

For more info on pairwise testing see http://www.pairwise.org.

Features

  • Produces good enough dataset.
  • Pythonic, iterator-style enumeration interface.
  • Allows to filter out "invalid" combinations during search for the next combination.
  • Goes beyond pairs! If/when required can generate n-wise combinations.

Get Started

Basic Usage

Sample Code:
from allpairspy import AllPairs

parameters = [
    ["Brand X", "Brand Y"],
    ["98", "NT", "2000", "XP"],
    ["Internal", "Modem"],
    ["Salaried", "Hourly", "Part-Time", "Contr."],
    [6, 10, 15, 30, 60],
]

print("PAIRWISE:")
for i, pairs in enumerate(AllPairs(parameters)):
    print("{:2d}: {}".format(i, pairs))
Output:
PAIRWISE:
 0: ['Brand X', '98', 'Internal', 'Salaried', 6]
 1: ['Brand Y', 'NT', 'Modem', 'Hourly', 6]
 2: ['Brand Y', '2000', 'Internal', 'Part-Time', 10]
 3: ['Brand X', 'XP', 'Modem', 'Contr.', 10]
 4: ['Brand X', '2000', 'Modem', 'Part-Time', 15]
 5: ['Brand Y', 'XP', 'Internal', 'Hourly', 15]
 6: ['Brand Y', '98', 'Modem', 'Salaried', 30]
 7: ['Brand X', 'NT', 'Internal', 'Contr.', 30]
 8: ['Brand X', '98', 'Internal', 'Hourly', 60]
 9: ['Brand Y', '2000', 'Modem', 'Contr.', 60]
10: ['Brand Y', 'NT', 'Modem', 'Salaried', 60]
11: ['Brand Y', 'XP', 'Modem', 'Part-Time', 60]
12: ['Brand Y', '2000', 'Modem', 'Hourly', 30]
13: ['Brand Y', '98', 'Modem', 'Contr.', 15]
14: ['Brand Y', 'XP', 'Modem', 'Salaried', 15]
15: ['Brand Y', 'NT', 'Modem', 'Part-Time', 15]
16: ['Brand Y', 'XP', 'Modem', 'Part-Time', 30]
17: ['Brand Y', '98', 'Modem', 'Part-Time', 6]
18: ['Brand Y', '2000', 'Modem', 'Salaried', 6]
19: ['Brand Y', '98', 'Modem', 'Salaried', 10]
20: ['Brand Y', 'XP', 'Modem', 'Contr.', 6]
21: ['Brand Y', 'NT', 'Modem', 'Hourly', 10]

Filtering

You can restrict pairs by setting filtering function to filter_func at AllPairs constructor.

Sample Code:
from allpairspy import AllPairs

def is_valid_combination(row):
    """
    This is a filtering function. Filtering functions should return True
    if combination is valid and False otherwise.

    Test row that is passed here can be incomplete.
    To prevent search for unnecessary items filtering function
    is executed with found subset of data to validate it.
    """

    n = len(row)

    if n > 1:
        # Brand Y does not support Windows 98
        if "98" == row[1] and "Brand Y" == row[0]:
            return False

        # Brand X does not work with XP
        if "XP" == row[1] and "Brand X" == row[0]:
            return False

    if n > 4:
        # Contractors are billed in 30 min increments
        if "Contr." == row[3] and row[4] < 30:
            return False

    return True

parameters = [
    ["Brand X", "Brand Y"],
    ["98", "NT", "2000", "XP"],
    ["Internal", "Modem"],
    ["Salaried", "Hourly", "Part-Time", "Contr."],
    [6, 10, 15, 30, 60]
]

print("PAIRWISE:")
for i, pairs in enumerate(AllPairs(parameters, filter_func=is_valid_combination)):
    print("{:2d}: {}".format(i, pairs))
Output:
PAIRWISE:
 0: ['Brand X', '98', 'Internal', 'Salaried', 6]
 1: ['Brand Y', 'NT', 'Modem', 'Hourly', 6]
 2: ['Brand Y', '2000', 'Internal', 'Part-Time', 10]
 3: ['Brand X', '2000', 'Modem', 'Contr.', 30]
 4: ['Brand X', 'NT', 'Internal', 'Contr.', 60]
 5: ['Brand Y', 'XP', 'Modem', 'Salaried', 60]
 6: ['Brand X', '98', 'Modem', 'Part-Time', 15]
 7: ['Brand Y', 'XP', 'Internal', 'Hourly', 15]
 8: ['Brand Y', 'NT', 'Internal', 'Part-Time', 30]
 9: ['Brand X', '2000', 'Modem', 'Hourly', 10]
10: ['Brand Y', 'XP', 'Modem', 'Contr.', 30]
11: ['Brand Y', '2000', 'Modem', 'Salaried', 15]
12: ['Brand Y', 'NT', 'Modem', 'Salaried', 10]
13: ['Brand Y', 'XP', 'Modem', 'Part-Time', 6]
14: ['Brand Y', '2000', 'Modem', 'Contr.', 60]

OrderedDict

You can use collections.OrderedDict instance as an argument for AllPairs constructor. Pairs returned as collections.namedtuple instances.

Sample Code:
from collections import OrderedDict
from allpairspy import AllPairs

parameters = OrderedDict({
    "brand": ["Brand X", "Brand Y"],
    "os": ["98", "NT", "2000", "XP"],
    "minute": [15, 30, 60],
})

print("PAIRWISE:")
for i, pairs in enumerate(AllPairs(parameters)):
    print("{:2d}: {}".format(i, pairs))
Sample Code:
PAIRWISE:
 0: Pairs(brand='Brand X', os='98', minute=15)
 1: Pairs(brand='Brand Y', os='NT', minute=15)
 2: Pairs(brand='Brand Y', os='2000', minute=30)
 3: Pairs(brand='Brand X', os='XP', minute=30)
 4: Pairs(brand='Brand X', os='2000', minute=60)
 5: Pairs(brand='Brand Y', os='XP', minute=60)
 6: Pairs(brand='Brand Y', os='98', minute=60)
 7: Pairs(brand='Brand X', os='NT', minute=60)
 8: Pairs(brand='Brand X', os='NT', minute=30)
 9: Pairs(brand='Brand X', os='98', minute=30)
10: Pairs(brand='Brand X', os='XP', minute=15)
11: Pairs(brand='Brand X', os='2000', minute=15)

Parameterized testing with pairwise by using py.test

Sample Code:
import pytest
from allpairspy import AllPairs

def function_to_be_tested(brand, operating_system, minute):
    # do something
    return True

class TestParameterized(object):

    @pytest.mark.parametrize(["brand", "operating_system", "minute"], [
        value_list for value_list in AllPairs([
            ["Brand X", "Brand Y"],
            ["98", "NT", "2000", "XP"],
            [10, 15, 30, 60]
        ])
    ])
    def test(self, brand, operating_system, minute):
        assert function_to_be_tested(brand, operating_system, minute)
Output:
$ py.test test_parameterize.py -v
============================= test session starts ==============================
...
collected 16 items

test_parameterize.py::TestParameterized::test[Brand X-98-10] PASSED      [  6%]
test_parameterize.py::TestParameterized::test[Brand Y-NT-10] PASSED      [ 12%]
test_parameterize.py::TestParameterized::test[Brand Y-2000-15] PASSED    [ 18%]
test_parameterize.py::TestParameterized::test[Brand X-XP-15] PASSED      [ 25%]
test_parameterize.py::TestParameterized::test[Brand X-2000-30] PASSED    [ 31%]
test_parameterize.py::TestParameterized::test[Brand Y-XP-30] PASSED      [ 37%]
test_parameterize.py::TestParameterized::test[Brand Y-98-60] PASSED      [ 43%]
test_parameterize.py::TestParameterized::test[Brand X-NT-60] PASSED      [ 50%]
test_parameterize.py::TestParameterized::test[Brand X-NT-30] PASSED      [ 56%]
test_parameterize.py::TestParameterized::test[Brand X-98-30] PASSED      [ 62%]
test_parameterize.py::TestParameterized::test[Brand X-XP-60] PASSED      [ 68%]
test_parameterize.py::TestParameterized::test[Brand X-2000-60] PASSED    [ 75%]
test_parameterize.py::TestParameterized::test[Brand X-2000-10] PASSED    [ 81%]
test_parameterize.py::TestParameterized::test[Brand X-XP-10] PASSED      [ 87%]
test_parameterize.py::TestParameterized::test[Brand X-98-15] PASSED      [ 93%]
test_parameterize.py::TestParameterized::test[Brand X-NT-15] PASSED      [100%]

Other Examples

Other examples could be found in examples directory.

Installation

Install from PyPI

pip install allpairspy

Install from PPA (for Ubuntu)

sudo add-apt-repository ppa:thombashi/ppa
sudo apt update
sudo apt install python3-allpairspy

Known issues

  • Not optimal - there are tools that can create smaller set covering all the pairs. However, they are missing some other important features and/or do not integrate well with Python.
  • Lousy written filtering function may lead to full permutation of parameters.
  • Version 2.0 has become slower (a side-effect of introducing ability to produce n-wise combinations).

Dependencies

Python 2.7+ or 3.5+