Made it possible to change the index on runtime, however returning en…

…tities from two URIs at the same time is impossible as of now, this is related to #55 and #46. If 55 is fixed, 46 should be fixed.

src/main/java/org/aksw/agdistis/algorithm/CandidateUtil.java

@@ -28,6 +28,10 @@ public class CandidateUtil {
 
 	private static Logger log = LoggerFactory.getLogger(CandidateUtil.class);
 	private String nodeType;
+	public void setNodeType(String nodeType) {
+		this.nodeType = nodeType;
+	}
+
 	private TripleIndex index;
 	private TripleIndexContext index2;
 	private NGramDistance nGramDistance;
@@ -47,7 +51,8 @@ public CandidateUtil() throws IOException {
 		String envNodeType = System.getenv("AGDISTIS_NODE_TYPE");
 		this.nodeType = envNodeType != null ? envNodeType : prop.getProperty("nodeType");
 		String envNgramDistance = System.getenv("AGDISTIS_NGRAM_DISTANCE");
-		this.nGramDistance = new NGramDistance(Integer.valueOf(envNgramDistance != null ? envNgramDistance : prop.getProperty("ngramDistance")));
+		this.nGramDistance = new NGramDistance(
+				Integer.valueOf(envNgramDistance != null ? envNgramDistance : prop.getProperty("ngramDistance")));
 		this.index = new TripleIndex();
 		String envContext = System.getenv("AGDISTIS_CONTEXT");
 		this.context = Boolean.valueOf(envContext != null ? envContext : prop.getProperty("context"));
@@ -61,11 +66,23 @@ public CandidateUtil() throws IOException {
 		String envAcronym = System.getenv("AGDISTIS_ACRONYM");
 		this.acronym = Boolean.valueOf(envAcronym != null ? envAcronym : prop.getProperty("acronym"));
 		String envCommonEntities = System.getenv("AGDISTIS_COMMON_ENTITIES");
-		this.commonEntities = Boolean.valueOf(envCommonEntities != null ? envCommonEntities : prop.getProperty("commonEntities"));
+		this.commonEntities = Boolean
+				.valueOf(envCommonEntities != null ? envCommonEntities : prop.getProperty("commonEntities"));
 		String envAlgorithm = System.getenv("AGDISTIS_ALGORITHM");
 		this.algorithm = envAlgorithm != null ? envAlgorithm : prop.getProperty("algorithm");
 	}
 
+	public void setIndex(TripleIndex index) {
+		try {
+			this.index = index;
+			this.domainWhiteLister = new DomainWhiteLister(index);
+		} catch (IOException e) {
+			log.error("Could not set new index in Candidate Util due to DomainWhiteLister");
+			e.printStackTrace();
+		}
+
+	}
+
 	public void insertCandidatesIntoText(DirectedSparseGraph<Node, String> graph, Document document,
 			double threshholdTrigram, Boolean heuristicExpansionOn) throws IOException {
 		NamedEntitiesInText namedEntities = document.getNamedEntitiesInText();

src/main/java/org/aksw/agdistis/algorithm/NEDAlgo_HITS.java

@@ -147,5 +147,26 @@ public void setMaxDepth(int maxDepth) {
 	public void setHeuristicExpansionOn(Boolean value) {
 		this.heuristicExpansionOn = value;
 	}
+	public String getEdgeType() {
+		return edgeType;
+	}
+
+	public void setEdgeType(String edgeType) {
+		this.edgeType = edgeType;
+	}
+
+	public String getNodeType() {
+		return nodeType;
+	}
+
+	public void setNodeType(String nodeType) {
+		this.nodeType = nodeType;
+		this.cu.setNodeType(nodeType);
+	}
+
+	public void setIndex(TripleIndex index) {
+		this.index = index;
+		this.cu.setIndex(index);
+	}
 
 }

src/main/java/org/aksw/agdistis/util/TripleIndex.java

@@ -70,12 +70,19 @@ public TripleIndex() throws IOException {
 
 		cache = CacheBuilder.newBuilder().maximumSize(50000).build();
 	}
+
+	public void setIndex(String index) throws IOException {
+		directory = new MMapDirectory(new File(index));
+		ireader = DirectoryReader.open(directory);
+		isearcher = new IndexSearcher(ireader);
+	}
 
 	public List<Triple> search(String subject, String predicate, String object) {
 		return search(subject, predicate, object, defaultMaxNumberOfDocsRetrievedFromIndex);
 	}
 
 	public List<Triple> search(String subject, String predicate, String object, int maxNumberOfResults) {
+		System.out.println(predicate +" -> "+object + " : " );
 		BooleanQuery bq = new BooleanQuery();
 		List<Triple> triples = new ArrayList<Triple>();
 

src/main/java/org/aksw/agdistis/util/TripleIndexCreator.java

@@ -72,7 +72,8 @@ public static void main(String args[]) {
 			log.info("The index will be here: " + index);
 
 			String envFolderWithTtlFiles = System.getenv("AGDISTIS_FOLDER_WITH_TTL_FILES");
-			String folder = envFolderWithTtlFiles != null ? envFolderWithTtlFiles : prop.getProperty("folderWithTTLFiles");
+			String folder = envFolderWithTtlFiles != null ? envFolderWithTtlFiles
+					: prop.getProperty("folderWithTTLFiles");
 			log.info("Getting triple data from: " + folder);
 			List<File> listOfFiles = new ArrayList<File>();
 			for (File file : new File(folder).listFiles()) {
@@ -140,7 +141,11 @@ private void indexTTLFile(File file, String baseURI)
 		OnlineStatementHandler osh = new OnlineStatementHandler();
 		parser.setRDFHandler(osh);
 		parser.setStopAtFirstError(false);
-		parser.parse(new FileReader(file), baseURI);
+		if (baseURI == null) {
+			parser.parse(new FileReader(file), "");
+		} else {
+			parser.parse(new FileReader(file), baseURI);
+		}
 		log.info("Finished parsing: " + file);
 	}
 

src/test/java/TripleIndexCreatorTest.java

@@ -0,0 +1,133 @@
+
+import static org.junit.Assert.assertTrue;
+
+import java.io.File;
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Properties;
+
+import org.aksw.agdistis.algorithm.CandidateUtil;
+import org.aksw.agdistis.algorithm.NEDAlgo_HITS;
+import org.aksw.agdistis.datatypes.Document;
+import org.aksw.agdistis.datatypes.NamedEntitiesInText;
+import org.aksw.agdistis.datatypes.NamedEntityInText;
+import org.aksw.agdistis.util.Triple;
+import org.aksw.agdistis.util.TripleIndex;
+import org.aksw.agdistis.util.TripleIndexCreator;
+import org.aksw.agdistis.webapp.GetDisambiguation;
+import org.junit.After;
+import org.junit.Before;
+import org.junit.Test;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import com.google.common.collect.Lists;
+
+public class TripleIndexCreatorTest {
+
+	Logger log = LoggerFactory.getLogger(TripleIndexCreatorTest.class);
+	private TripleIndex index;
+
+	@Before
+	public void init() {
+		try {
+			index = new TripleIndex();
+
+		} catch (IOException e) {
+			log.error(
+					"Can not load index or DBpedia repository due to either wrong properties in agdistis.properties or missing index at location",
+					e);
+		}
+	}
+
+	@After
+	public void close() {
+		try {
+			index.close();
+		} catch (IOException e) {
+			log.error(
+					"Can not load index or DBpedia repository due to either wrong properties in agdistis.properties or missing index at location",
+					e);
+		}
+	}
+
+	@Test
+	/**
+	 * tests https://github.com/dice-group/AGDISTIS/issues/46 if we need to return
+	 * URIs from two different KBs AGDISTIS returns breaks
+	 * 
+	 * @throws IOException
+	 */
+	public void testMinimalOntologyExample() throws IOException {
+		// load test data into index
+		TripleIndexCreator tic = new TripleIndexCreator();
+		File file = new File("src/test/resources/test_evertec.ttl");
+		File folder = new File("src/test/resources/evertec");
+		if (folder.exists()) {
+			folder.delete();
+		}
+		tic.createIndex(Lists.newArrayList(file), folder.getAbsolutePath(), null);
+
+		// set the properties correctly
+
+		NEDAlgo_HITS agdistis = new NEDAlgo_HITS();
+		agdistis.setNodeType("http://fairhair.ai/kg/resource/");
+		agdistis.setEdgeType("http://dbpedia.org/ontology/");
+
+		// load index
+		index.setIndex(folder.getAbsolutePath());
+		agdistis.setIndex(index);
+
+		// test index
+		String taisho = "Evertec";
+		String taishoURL = "http://fairhair.ai/kg/resource/Evertec";
+		String japan = "Puerto Rico";
+		String japanURL = "http://dbpedia.org/resource/Puerto_Rico";
+
+		HashMap<String, String> correct = new HashMap<String, String>();
+		correct.put(taisho, taishoURL);
+		correct.put(japan, japanURL);
+		String preAnnotatedText = "<entity>" + taisho + "</entity> is a company in<entity>" + japan + "</entity>.";
+
+		Document d = GetDisambiguation.textToDocument(preAnnotatedText);
+		agdistis.run(d, null);
+
+		NamedEntitiesInText namedEntities = d.getNamedEntitiesInText();
+		HashMap<NamedEntityInText, String> results = new HashMap<NamedEntityInText, String>();
+		for (NamedEntityInText namedEntity : namedEntities) {
+			String disambiguatedURL = namedEntity.getNamedEntityUri();
+			System.out.println(namedEntity);
+			results.put(namedEntity, disambiguatedURL);
+		}
+		for (NamedEntityInText namedEntity : results.keySet()) {
+			String disambiguatedURL = namedEntity.getNamedEntityUri();
+			System.out.println(namedEntity.getLabel() + " -> " + results.get(namedEntity));
+			// TODO comment that line in
+			// assertTrue(correct.get(namedEntity.getLabel()).equals(disambiguatedURL));
+		}
+
+		// TODO if we change the node type, Puerto Rico gets found but not Evertec
+		agdistis.setNodeType("http://dbpedia.org/resource/");
+
+		d = GetDisambiguation.textToDocument(preAnnotatedText);
+		agdistis.run(d, null);
+
+		namedEntities = d.getNamedEntitiesInText();
+		results = new HashMap<NamedEntityInText, String>();
+		for (NamedEntityInText namedEntity : namedEntities) {
+			String disambiguatedURL = namedEntity.getNamedEntityUri();
+			System.out.println(namedEntity);
+			results.put(namedEntity, disambiguatedURL);
+		}
+		for (NamedEntityInText namedEntity : results.keySet()) {
+			String disambiguatedURL = namedEntity.getNamedEntityUri();
+			System.out.println(namedEntity.getLabel() + " -> " + results.get(namedEntity));
+			// TODO comment that line in
+			// assertTrue(correct.get(namedEntity.getLabel()).equals(disambiguatedURL));
+		}
+
+	}
+
+}