Pairwise Test Case Generator

Configuration testing explodes quickly: four browsers, three operating systems, three payment methods, two user types and three languages already give 216 combinations, and every new option multiplies the total. Pairwise testing (also called all-pairs testing, a form of combinatorial interaction testing) picks a much smaller set of test cases in which every pair of values from any two parameters appears together at least once. Research on real-world failures, notably by NIST, found that most defects are triggered by a single parameter or by the interaction of two, so covering all pairs catches a large share of bugs at a fraction of the cost.

This generator uses a greedy, AETG-style algorithm. It repeatedly builds several candidate test cases, each starting from a value combination that is not yet covered and filling the remaining parameters with the values that cover the most new pairs, and keeps the best one. A seeded random number generator breaks ties, so the same model and seed always produce the same suite, and the best of several seeded runs is kept for small models. After generation the tool enumerates every valid pair again and checks it against the output, so the coverage figure you see is measured, not assumed.

Rules describe combinations that are impossible or meaningless, such as Safari on Linux or a guest checkout with a stored bank account. Pairs that no valid test can contain are reported as excluded instead of being forced into a test case. 3-wise coverage guarantees every triple of values and is useful when three settings are known to interact (for example locale, currency and tax rules); expect the suite to be several times larger.

Pairwise output works well as the Examples table of a Gherkin Scenario Outline — tidy it with the Gherkin formatter — and as a test matrix for parameterised unit tests. Fill each case with realistic inputs from the mock data generator, and convert the CSV export to JSON or YAML fixtures with the data format converter.