Descartes includes its own output formats which are focused on pseudo
and partially-tested methods.
These formats can be used by specifying METHODS and/or ISSUES in the
PITest outputFormats parameter.
METHODS produces a JSON file with the following structure:
{
"methods": [
...
{
"name": "getOptionValue",
"description": "(Ljava/lang/String;)Ljava/lang/String;",
"class": "CommandLine",
"package": "org/apache/commons/cli",
"file-name": "CommandLine.java",
"line-number": 202,
"classification": "tested",
"detected": ["\"A\"", "null", "\"\""],
"not-detected": [],
"tests": [...],
"mutations": [
...
{
"status": "KILLED",
"mutator": "\"A\"",
"tests-run": 1,
"detected-by": "org.apache.commons.cli.bug.BugsTest.test31148(org.apache.commons.cli.bug.BugsTest)",
"tests": [...]
...
]
}
...
],
"analysis": {
"time": 15918,
"mutators": [
"void",
"null",
"empty",
"true",
"false",
"0",
"1",
"(short)0",
"(short)1",
"(byte)0",
"(byte)1",
"0L",
"1L",
"0.0",
"1.0",
"0.0f",
"1.0f",
"'\\40'",
"'A'",
"\"\"",
"\"A\""
]
}The methods list contains an entry for every method that was part of the analysis. For each method there is information about its metadata, its testing classification (tested, partially-tested, pseudo-tested, not-covered), which mutations were detected and which not, all tests covering the method and there is also a list of entries for every mutation performed with more information.
A method is said to be not-covered if it is not covered by the test suite. It is said to be pseudo-tested if it is covered by the test suite, yet no extreme mutation applied to the method was detected by any test case. A method is said to be partially-tested if it has mixed results: some mutations were detected and others were not. Finally, a method is classified as tested if all extreme transformations are detected by the test suite.
Examples of files with this format can be checked here and also here. These were obtained from actual open-source projects.
The ISSUES output format contains this same information, but it is rendered as an HTML document with human-readable descriptions. An example can be checked here.