bergel / SmallSuiteGenerator

SmallSuiteGenerator

SmallSuiteGenerator is an Smalltalk tool to generate test cases automatically

MIT Licensed.

SmallSuiteGenerator can be used to discover bugs in the code, because it generates automatically test cases

Installation

You can load SmallSuiteGenerator into Pharo image executing the following code in the playground:

Metacello new
 baseline:'SmallSuiteGenerator';
 repository: 'github://OBJECTSEMANTICS/SmallSuiteGenerator:master/src';
 load.

Configuration

The first step is to define the code block that will be instrumented to get the typeInfo. Define package's regular expression pattern also.

| typeinfo |
typeInfo := SSTypeCollector profile:[ 
		(SSTeacher name: 'Ann' with: 50)
		nickname;
		canRegister: ((SConference price: 50) offerPrice: 50);
		idTeacher;
		yearsWorkExperience ] onPackagesMatching: 'SmallSuiteGenerator-Scenario'.

The second step is to configure the class STestCaseFactory.

STestCaseFactory new
    typeInfo: typeInfo;
    fitness: SStatementCoverage new;
    targetClassName: #SSTeacher;
    targetPackagesRegex:'SmallSuiteGenerator-Scenario';
    outputPackageName:'SmallSuiteGenerator-Tests-Generated';
    numberOfIterations: 20;
    numberOfStatements: 3;
    createTestCases;
    visualize;
    yourself.

Some other options to configure are:

• fitness: it will be used in genetic algorithm to improve its value either increasing or reducing, depending on the function. In this case we use: SStatementCoverage.
• targetClassName: It represents the target class name that will be analyzed in the package.
• targetPackagesRegex: Regular expression of the package where the class is found.
• outputPackageName: Package name where the tests cases will be created.

To watch the evolution of fitness function call visualize method.

After to execute this script the testCases will be generated with the higher fitness of the population. In the output package name specified you can find the generated testCases. Usually the last enumerated testCases have the highests fitness.

Additionally to the described configurations, you can configure some parameters before to generate test cases, e.g:

populationSize: 30.