KGDD

KG Test Driven Design development. Write queries with expected results. CONSTRUCT KG. Run tests and see if queries return correct bindings.

Run tests

To run a testbed on a KG run:

kgdd -f "/path/to/KG.{ttl,nt,...}" -t "/path/to/tests.json"

Testbed

To create a testbed write a .json file like the following:

{
    "tests" : [
        {
            "description" : "Extract all the triples from a KG",
            "query" : "SELECT * WHERE {?s ?p ?o } ",
            "results" : 3,
            "rows": [{
                "s" : "artist_1"
            }            ],

            "bindings": {
                "o" : "music_1"
            },
            "skip": false
        }
        ...
    ]
}

Every item in the tests array will be used to run a test. Let's see fields:

• query: The query to be run to extract data from KG.
• results: The number of expected rows. With 0 or a value greater than 0 then the test will pass if the numbers match. With -1 the test will pass if you get a number greater than 0.
• rows: Every item in the array is an expected row. For the test to pass you need at least a row matching all the expected ones. In this example we want at least a row with "artist_1" as "s".
• bindings: At least a binding in at least a results row for all the specified variable in the bindings object should match for the test to pass. In this example At least in a "o" in one of the row should match "music_1".
• skip: The test will be skipped.

In every field you can set null as default value if you don't want the results, rows, or bindings step to run. E.g. following testbed will run only results step.

{
    "tests" : [
        {
            "description" : "Extract all the triples from a KG",
            "query" : "SELECT * WHERE {?s ?p ?o } ",
            "results" : 3,
            "rows": null,
            "bindings": null,
            "skip": false
        }
        ...
    ]
}

Usage

usage: kgdd [-h] -t TEST -f FILE [-v {0,1,2}]

A cli program to automate TDD KG development. Define some basic query and expected outputs in a test. Kgdd will take
care of run queries and see if they match expedted results.

options:
  -h, --help            show this help message and exit
  -t TEST, --test TEST  Path to testbed file. File should be in Json format
  -f FILE, --file FILE  Path to KG file
  -v {0,1,2}, --verbose {0,1,2}
                        Verbose output. Verbosity level = 0,1,2 . Default 0 no verbose. 2 is really verbose

Installation

kgdd is a python package to install it you need python 3.X installed. You can install it globally in your local machine running:

pip install kgdd

or on Windows:

py -m pip install kgdd

Alternatively if you prefere a local installation you can create a virtualenv and install the package there. The package is distributed as a cli tool, and available everywhere if the environment where is installed is activated.