Difference between revisions of "Java Examples"

From Info216
Jump to: navigation, search
(Lecture 3: SPARQL Query and Update)
Line 488: Line 488:
 
table.forEachRemaining(row -> System.out.println(row.toString());</nowiki>
 
table.forEachRemaining(row -> System.out.println(row.toString());</nowiki>
 
Whenever Java 8 or later sees code like this, it behaves as if we had explicitly written a RowPrinter or similar Consumer-class like the one above.
 
Whenever Java 8 or later sees code like this, it behaves as if we had explicitly written a RowPrinter or similar Consumer-class like the one above.
 +
 +
==Lecture 4: TDB and Fuseki==
 +
 +
===Creating a dataset===
 +
<nowiki>
 +
        Dataset dataset = TDBFactory.createDataset();
 +
        Model defaultModel = dataset.getDefaultModel();
 +
 +
        ...
 +
 +
        dataset.close()</nowiki>
 +
 +
This creates an in-memory dataset, which is not persistent.
 +
 +
===Creating/loading a TDB-backed dataset===
 +
<nowiki>
 +
        Dataset dataset = TDBFactory.createDataset("TDBTest");
 +
        Model defaultModel = dataset.getDefaultModel();
 +
 +
        ...
 +
 +
        dataset.close()</nowiki>
 +
 +
The first time it is run, this creates a persistent dataset, backed by a TDB triple store located in the directory "TDBTest" inside your Eclipse project. Refresh the project to see it (or F5).
 +
 +
When re-run later, this loads the dataset from the TDB store.
 +
 +
It is important to close a TDB-backed dataset before the program terminates. Otherwise, you need to go into the database folder (for example the "TDBTest" directory inside your Eclipse project) and delete the file named "tdb.lock". (Do a refresh with F5 if you do not see it in Eclipse.)
 +
 +
===Fuseki===
 +
When you get started, it is easiest to run Fuseki from the directory where you unpacked it along with the other Jena downloads, for example:
 +
 +
  cd C:\Programs\Jena\apache-jena-fuseki-3.6.0
 +
or
 +
  cd /opt/Jena/apache-jena-fuseki-3.6.0
 +
 +
Start the Fuseki server with this command on Windows:
 +
  fuseki-server --localhost --loc=C:\...\your\Eclipse\workspace\INFO216\TDBTest /tdb
 +
 +
On Linux:
 +
  sh fuseki-server --localhost --loc=C:\...\your\Eclipse\workspace\INFO216\TDBTest /tdb
 +
 +
Here, TDBTest is the name of the triple store, INFO216 is the name of the Eclipse project, located inside your Eclipse workspace. Use the '''--help''' option to see what the other options do.
 +
 +
Open a web browser and go to '''localhost:3030''' to run queries/updates and otherwise explore and use the TDB-backed dataset.
 +
 +
You can also start Fuseki without the "--loc" option:
 +
  fuseki-server --localhost
 +
or
 +
  sh fuseki-server --localhost
 +
 +
When you go to '''localhost:3030''' in your web browser, Fuseki will now appear empty at first, but you can create new datasets and load triples into them from files or from the web. If you choose to create datasets that are persistent, they will not end up in the Eclipse project folder, but in a subfolder of the Fuseki-installation folder, for example:
 +
  C:\Programs\Jena\apache-jena-fuseki-3.6.0\runatabases\TDBTest
  
 
&nbsp;
 
&nbsp;
  
 
<div class="credits" style="text-align: right; direction: ltr; margin-left: 1em;">''INFO216, UiB, Spring 2017-2018, Andreas L. Opdahl (c). All code examples are [https://creativecommons.org/choose/zero/ CC0].'' </div>
 
<div class="credits" style="text-align: right; direction: ltr; margin-left: 1em;">''INFO216, UiB, Spring 2017-2018, Andreas L. Opdahl (c). All code examples are [https://creativecommons.org/choose/zero/ CC0].'' </div>

Revision as of 14:40, 15 February 2018

Here are the code examples we have used in the live sessions during the lectures - along with a few additional ones.

(More will appear as the course progresses.)

Lecture 1: Java, Jena, and Eclipse

Hello Jena

package no.uib.sinoa.info216;

import org.apache.jena.rdf.model.Model;
import org.apache.jena.rdf.model.ModelFactory;
import org.apache.jena.rdf.model.Resource;
import org.apache.jena.vocabulary.FOAF;

public class HelloJena {

    public static void main(String[] args) {
       
        Model model = ModelFactory.createDefaultModel();
       
	Resource andreas = model.createResource(base + "Andreas");
	Resource info216 = model.createResource(base + "INFO216");
	Property teaches = model.createProperty(base + "teaches");
	andreas.addProperty(teaches, info216);

	andreas.addLiteral(FOAF.name, "Andreas L Opdahl");

        model.write(System.out, "TURTLE");
    }   
}

Lecture 2: RDF

Resource objects

package no.uib.infomedia.info216;

...

public class HelloJena {
    public static void main(String[] args) {
        String iriBase = "http://no.uib.infomedia.info216/";
        String iriDbpedia = "http://dbpedia.org/resource/";
       
        Model model = ModelFactory.createDefaultModel();
       
        Resource resCadeTracy = model.createResource(iriBase + "Cade_Tracy");
        resCadeTracy.addLiteral(FOAF.name, "Cade Tracy");
       
        Resource resCanada = model.createResource(iriDbpedia + "Canada");
        Resource resFrance = model.createResource(iriDbpedia + "France");
        Property propVisited = model.createProperty(iriBase + "visited");
        resCadeTracy.addProperty(propVisited, resCanada);
        resCadeTracy.addProperty(propVisited, resFrance);

        model.write(System.out, "TURTLE");
    }
}

Language-tagged literals

        resFrance.addProperty(RDFS.label, "Frankrike", "no");
        resFrance.addProperty(RDFS.label, "France", "en");
        resFrance.addProperty(RDFS.label, "Francia", "es");

Typed literals

        Property propPopEst = model.createProperty(iriDbpedia + "ontology/populationEstimate");
        resFrance.addProperty(propPopEst, "66644000", XSDDatatype.XSDinteger);

Looping through statements

        for (Statement stmt : model.listStatements().toList()) {
            System.out.println(stmt.toString());
        }

Selecting statements

        for (Statement stmt : model
                .listStatements((Resource)null, RDFS.label, (RDFNode)null)
                .toList()) {
            System.out.println("Subject:   " + stmt.getSubject().toString());
            System.out.println("Predicate: " + stmt.getPredicate().toString());
            System.out.println("Object:    " + stmt.getObject().toString());
        }

Using a selector

        for (Statement stmt : model
                .listStatements(new SimpleSelector() {
                    public boolean test(Statement s) {
                        return (s.getPredicate().equals(FOAF.name));
                    }
                })
                .toList()) {
            System.out.println(stmt.getObject().toString());
        }

Writing to file

        try {
            model.write(new FileOutputStream("test.ttl"), "TURTLE");
        } catch (Exception e) {
            // TODO: handle exception
        }

Contents of test.ttl

<http://no.uib.infomedia.info216/Cade_Tracy>
        <http://no.uib.infomedia.info216/visited>
                <http://dbpedia.org/resource/France> , <http://dbpedia.org/resource/Canada> ;
        <http://xmlns.com/foaf/0.1/name>
                "Cade Tracy" .

<http://dbpedia.org/resource/France>
        <http://www.w3.org/2000/01/rdf-schema#label>
                "Francia"@es , "France"@en , "Frankrike"@no ;
        <http://dbpedia.org/resource/ontology/populationEstimate>
                66644000 .

Reading from file

package no.uib.infomedia.sinoa.info216;

import java.io.FileInputStream;

import org.apache.jena.rdf.model.Model;
import org.apache.jena.rdf.model.ModelFactory;

public class ReadJena {

    public static void main(String[] args) {
        Model model = ModelFactory.createDefaultModel();
       
        try {
            model.read(new FileInputStream("test.ttl"), "http://ex.org/", "TURTLE");
        } catch (Exception e) {
            // TODO: handle exception
        }
       
        model.write(System.out);
    }
}

Reading from web resource

package no.uib.infomedia.sinoa.info216;

import java.io.InputStream;
import java.net.HttpURLConnection;
import java.net.URL;

import org.apache.jena.rdf.model.Model;
import org.apache.jena.rdf.model.ModelFactory;

public class HttpTest {

    public static void main(String[] args) {
        Model model = ModelFactory.createDefaultModel();
       
        try {
            URL url = new URL("http://people.uib.no/sinoa/test.ttl");  
            HttpURLConnection urlConnection = (HttpURLConnection) url.openConnection(); 
            InputStream is = urlConnection.getInputStream();
            model.read(is, "http://ex.org/", "TURTLE");
        } catch (Exception e) {
            // TODO: handle exception
        }
       
        model.write(System.out);
    }
}

(There are more advanced ways to download web resources from Java, but HttpUrlConnection is a straightforward and built-in way to get started.)

Lecture 3: SPARQL Query and Update

Basic INSERT DATA update

		Dataset dataset = DatasetFactory.create();
		
		UpdateAction.parseExecute(""
				+ "PREFIX info216: <http://ex.org/teaching#>"
				+ "INSERT DATA {"
				+ "    info216:cade info216:teaches info216:ECO001 . "
				+ "    GRAPH <http://ex.org/personal#Graph> {"
				+ "        info216:cade info216:age '29' . "
				+ "    }"
				+ "}", dataset);

		RDFDataMgr.write(System.out, dataset, Lang.TRIG);

To output only the default graph use:

		dataset.getDefaultModel().write(System.out, "TURTLE");

The method dataset.getNamedModel("http://ex.org/personal#Graph"); lets you output a named model instead.

Basic SELECT query

    ResultSet resultSet = QueryExecutionFactory
        .create(""
            + "SELECT ?s ?p ?o WHERE {"
            + "    ?s ?p ?o ."
            + "}", dataset)
        .execSelect();

    resultSet.forEachRemaining(qsol -> System.out.println(qsol.toString()));

Convert the ResultSet into a JSON object

    List<Map> jsonList = new Vector<Map>();
    while (resultSet.hasNext()) {
        QuerySolution qsol = resultSet.nextSolution();
        Iterator<String> varNames = qsol.varNames();
        Map<String, Object> jsonMap = new HashMap<String, Object>();
        while (varNames.hasNext()) {
            String varName = varNames.next();
            jsonMap.put(varName, qsol.get(varName).toString());
        }
        jsonList.add(jsonMap);
    }
    System.out.println(JsonUtils.toPrettyString(jsonList));

SELECT query with Query object

    // select example
    Query query = QueryFactory.create(""
            + "SELECT ?s ?p ?o WHERE {"
            + "    ?s ?p ?o ."
            + "}");
    QueryExecution queryExecution = QueryExecutionFactory.create(query, dataset);   
    ResultSet resultSet = queryExecution.execSelect();

SELECT query from SPARQL endpoint

If a named graph in your dataset has a triple with a subject that is a DBpedia IRI:

    ResultSet resultSet = QueryExecutionFactory.create(""
                    + "SELECT ?s ?p ?o WHERE {"
                    + "    GRAPH ?g { ?s ?p2 ?o2 . } "
                    + "       SERVICE <http://dbpedia.org/sparql> {"
                    + "        ?s ?p ?o ."
                    + "    }"
                    + "}", dataset).execSelect();
       
    while (resultSet.hasNext()) {
        QuerySolution qs = resultSet.nextSolution();
        System.out.println(qs.toString());
    }

Basic ASK query

    QueryExecution queryExecution = QueryExecutionFactory.create(""
            + "ASK { GRAPH ?g { ?s ?p ?o } }"
            + "", dataset);
    boolean res = queryExecution.execAsk();
    System.out.println("The result is " + res);

ASK query from IRL

    QueryExecution queryExecution = QueryExecutionFactory.create(""
            + "ASK "
            + "    FROM <http://people.uib.no/sinoa/european-populations.ttl> { "
            + "       <" + iriDbpedia + "Vatican_City> ?p ?o . "
            + "    }"
            + "");
    boolean res = queryExecution.execAsk();
    System.out.println("The result is " + res);

Basic DESCRIBE query

    Model franceModel = QueryExecutionFactory.create(""
            + "DESCRIBE <" + iriDbpedia + "France>"
            + "", dataset).execDescribe();
    franceModel.write(System.out, "TURTLE");

Basic CONSTRUCT query

    Model franceModel = QueryExecutionFactory.create(""
            + "CONSTRUCT { ?s ?p ?o . } WHERE { "
            + "    GRAPH ?g { ?s ?p ?o . } "
            + "}", dataset).execConstruct();
    franceModel.write(System.out, "TURTLE");

CONSTRUCT query from IRL

    Model franceModel = QueryExecutionFactory.create(""
            + "CONSTRUCT { ?s ?p ?o . } "
            + "    FROM <http://people.uib.no/sinoa/european-populations.ttl> "
            + "WHERE { "
            + "    ?s ?p ?o . "
            + "}").execConstruct();
    franceModel.write(System.out, "TURTLE");

Complex SPARQL predicates (Fuseki)

In the apache-jena-fuseki-version folder:

fuseki-server --localhost --update --mem /mem

In your web browser, goto http://localhost:3030/ . Use SPARQL ENDPOINT http://localhost:3030/mem/update for the INSERT updates and http://localhost:3030/mem/query for the SELECT queries below.

PREFIX x: <http://example.org/myex#>

INSERT DATA {
    x:IngridAlexandra x:father x:HaakonMagnus ;
                      x:mother x:MetteMarit .
    x:HaakonMagnus x:sister x:MarthaLouise .
}

Keep the PREFIX line in all the following queries:

SELECT ?s ?o WHERE {
    ?s (x:father | x:mother) ?o .
}
SELECT ?s ?o WHERE {
    ?s (x:father / x:sister) ?o .
}
SELECT ?s ?o WHERE {
    ?s ^(x:father / x:sister) ?o .
}
SELECT ?s ?o WHERE {
    ?s (^x:sister / ^x:father) ?o .
}
SELECT ?s ?o WHERE {
    ?s !x:sister ?o .
}

Add some mother-triples:

INSERT DATA {
    x:HaakonMagnus x:father x:Harald .
    x:Harald x:father x:Olav .
    x:Olav x:father x:Haakon .
}
SELECT ?o WHERE
{
    x:IngridAlexandra x:father+ ?o .
}
SELECT ?o WHERE
{
    x:IngridAlexandra x:father* ?o .
}
SELECT ?o WHERE
{
    x:IngridAlexandra x:father? ?o .
}
SELECT ?o WHERE
{
    x:IngridAlexandra x:father{2} ?o .
}
SELECT ?o WHERE
{
    x:IngridAlexandra x:father{2,4} ?o .
}

SPARQL SELECT VALUES (and services)

The code below retreives DBpedia descriptions of the three Scandinavian capitals, using VALUES:

		Dataset dataset = DatasetFactory.create();
		
		ResultSet table = QueryExecutionFactory.create(""
				+ "SELECT * WHERE {"
				+ "    VALUES ?city {"
				+ "	       <http://dbpedia.org/resource/Copenhagen>"
				+ "	       <http://dbpedia.org/resource/Oslo>"
				+ "	       <http://dbpedia.org/resource/Stockholm>"
				+ "    }"
				+ "    SERVICE <http://dbpedia.org/sparql> {"
				+ "		   ?city <" + RDFS.comment + "> ?comment ."
				+ "    }"
				+ "}", dataset).execSelect();
		
		table.forEachRemaining(row -> System.out.println(row));

To retrieve only English-language descriptions, you can add a FILTER inside the SERVICE call:

		...
				+ "    SERVICE <http://dbpedia.org/sparql> {"
				+ "		   ?city <" + RDFS.comment + "> ?comment ."
				+ "        FILTER( lang(?comment) = 'en' )"
				+ "    }"
		...

Language-tagged literals (and functions, and services...)

This works because we use the language-tagged literal 'Copenhagen'@en in the INSERT DATA update (and, as a result, it outputs DBpedia-triples about Copenhagen):

		Dataset dataset = DatasetFactory.create();
		
		String prefixes = ""
				+ "PREFIX rex: <http://ex.org#>"
				+ "PREFIX dbpedia: <http://dbpedia.org/resource/>";
		UpdateAction.parseExecute(prefixes
				+ "INSERT DATA {"
				+ "    rex:Margrethe <" + FOAF.based_near + "> 'Copenhagen'@en ."
				+ "}", dataset);

		ResultSet table = QueryExecutionFactory.create(prefixes
				+ "SELECT * WHERE {"
				+ "    ?person <" + FOAF.based_near + "> ?label ."
				+ "    SERVICE <http://dbpedia.org/sparql> {"
				+ "        ?city <" + RDFS.label + "> ?label ."
				+ "    }"
				+ "}", dataset).execSelect();
		
		table.forEachRemaining(row -> System.out.println(row.toString()));

If we do INSERT DATA without the language tag @en, we get no result (because the city labels in DBpedia are language-tagged):

		UpdateAction.parseExecute(prefixes
				+ "INSERT DATA {"
				+ "    rex:Margrethe <" + FOAF.based_near + "> 'Copenhagen'@en ."
				+ "}", dataset);

We can, however, rewrite the query to use the strlang function that adds language tags to labels. So this works with the previous INSERT DATA:

		ResultSet table = QueryExecutionFactory.create(prefixes
				+ "SELECT * WHERE {"
				+ "    ?person <" + FOAF.based_near + "> ?label ."
				+ "    BIND(strlang(?label, 'en') AS ?taglabel)"				
				+ "    SERVICE <http://dbpedia.org/sparql> {"
				+ "        ?city <" + RDFS.label + "> ?taglabel ."
				+ "    }"
				+ "}", dataset).execSelect();

To go the other way (if our local labels were language-tagged and DBpedia's not, we could do a similar thing, but use the reverse str-function instead of strlang.

Explanation of table.forEachRemaining(...)

The lambda-syntax in this code line may be new for you:

	table.forEachRemaining(row -> System.out.println(row.toString()));

The straightforward way to write it (that doesn't work), would be just:

	table.forEachRemaining(System.out.println(row.toString()));

But this doesn't work because we cannot send a method call as a parameter in Java, and our code introduces a row variable that is not declared anywhere.

Instead, we could try to define a new method somewhere else inside our class:

	void printRow(QuerySolution row) {
		System.out.println(row.toString());
	};

and then pass that function to forEachRemaining (doesn't work either):

	table.forEachRemaining(printRow);

But this doesn't work because you cannot pass functions as arguments like that in Java.

Finally, we could try to define a whole new class - a subclass of a Consumer class:

class RowPrinter implements Consumer<QuerySolution> {
	public void accept(QuerySolution row) {
		System.out.println(row.toString());
	}
}

Here, we define the class RowPrinter as a Consumer of QuerySolutions, because Jena defines each row in a ResultSet-table to be a QuerySolution. In other words, we can view a ResultSet (table) as a stream of QuerySolutions (rows) that the RowPrinter consumes one by one. The method name accept is defined by the Consumer-class. We cannot chose another name (and, actually, Consumer is a Java interface, not a class).

We can pass an object of this class to forEachRemaining - this is exactly what it expects (this works!):

	table.forEachRemaining(new RowPrinter());

But it is pretty cumbersome to write a new class like RowPrinter every time we want to do a forEachRemaining-call - or some other streaming call. Therefore Java 8 has introduced a shorthand, a lambda expression:

	table.forEachRemaining(row -> System.out.println(row.toString());

Whenever Java 8 or later sees code like this, it behaves as if we had explicitly written a RowPrinter or similar Consumer-class like the one above.

Lecture 4: TDB and Fuseki

Creating a dataset

        Dataset dataset = TDBFactory.createDataset();
        Model defaultModel = dataset.getDefaultModel();

        ...

        dataset.close()

This creates an in-memory dataset, which is not persistent.

Creating/loading a TDB-backed dataset

        Dataset dataset = TDBFactory.createDataset("TDBTest");
        Model defaultModel = dataset.getDefaultModel();

        ...

        dataset.close()

The first time it is run, this creates a persistent dataset, backed by a TDB triple store located in the directory "TDBTest" inside your Eclipse project. Refresh the project to see it (or F5).

When re-run later, this loads the dataset from the TDB store.

It is important to close a TDB-backed dataset before the program terminates. Otherwise, you need to go into the database folder (for example the "TDBTest" directory inside your Eclipse project) and delete the file named "tdb.lock". (Do a refresh with F5 if you do not see it in Eclipse.)

Fuseki

When you get started, it is easiest to run Fuseki from the directory where you unpacked it along with the other Jena downloads, for example:

 cd C:\Programs\Jena\apache-jena-fuseki-3.6.0

or

 cd /opt/Jena/apache-jena-fuseki-3.6.0

Start the Fuseki server with this command on Windows:

 fuseki-server --localhost --loc=C:\...\your\Eclipse\workspace\INFO216\TDBTest /tdb

On Linux:

 sh fuseki-server --localhost --loc=C:\...\your\Eclipse\workspace\INFO216\TDBTest /tdb

Here, TDBTest is the name of the triple store, INFO216 is the name of the Eclipse project, located inside your Eclipse workspace. Use the --help option to see what the other options do.

Open a web browser and go to localhost:3030 to run queries/updates and otherwise explore and use the TDB-backed dataset.

You can also start Fuseki without the "--loc" option:

 fuseki-server --localhost 

or

 sh fuseki-server --localhost 

When you go to localhost:3030 in your web browser, Fuseki will now appear empty at first, but you can create new datasets and load triples into them from files or from the web. If you choose to create datasets that are persistent, they will not end up in the Eclipse project folder, but in a subfolder of the Fuseki-installation folder, for example:

 C:\Programs\Jena\apache-jena-fuseki-3.6.0\runatabases\TDBTest

 

INFO216, UiB, Spring 2017-2018, Andreas L. Opdahl (c). All code examples are CC0.