Why

I needed a way to use ontologies and SPARQL to query a SQL database through HTTP.

How

Using Ontop, Kotlin and ktor. It works really with OpenJDK 8 and 11. The example docker uses version 11.

Principle

This project has an integrated webserver (port configurable) with the following endpoints:

/sparql

Takes a POST query and a data parameter (not a form parameter) that contains a SPARQL query.

Return:

• if successful: a SPARQL json of the bindings
• if not successful: HTTP 1.1/422 (Unprocessable entity) Impossible to process that query.

Example of usage with the integrated docker-compose file

There is an example of usage in the docker directory. The example would connect to a postgresql database with two tables: project and user.

Each project can have another project has its parent and a user as its owner

Any JDBC compatible database is usable. For now postgresql is integrated in the fat-jar.

The files:

• db.obda: Contains the queries and their mappings to the ontology
• db.owl: The ontology
• db.properties: The database parameters (type of DB, )
• ktor.config: The webserver configuration (port is the main thing you may want to change here)

The Dockerfile will use a build in build/libs/ontopadapter.jar by default.

I will try to keep things up to date, but you way have to check the version in build.gradle.kts.

You can use:

cd docker
cp ../build/lib/ontopadapter.jar .
docker-compose up --build

That will build a full testing environment and connect to it using:

curl -XPOST --data 'select * { ?s ?p ?o.}' http://127.0.0.1:8080

And you will get the SPARQL json output for that example query.

What is left to do

• Handle errors better
• Handle the update of parameters
• Add metrics
• Find a way to automate tests and CI
• Allow customization of CORS
• Restart the Ontop repository (in case of updates or bugs)