Merge pull request #200 from ate47/dev_quads

Fix some quad issues with some tests
rdfhdt · Sep 15, 2023 · aedace3 · aedace3
2 parents 8a73fcd + 2cce19c
commit aedace3
Show file tree

Hide file tree

Showing 9 changed files with 308 additions and 21 deletions.
diff --git a/hdt-api/src/main/java/org/rdfhdt/hdt/quad/QuadString.java b/hdt-api/src/main/java/org/rdfhdt/hdt/quad/QuadString.java
@@ -30,6 +30,10 @@ public void clear() {
 	@Override
 	public boolean equals(Object other) {
 		if (!(other instanceof QuadString)) {
+			if (context.length() == 0) {
+				// not a quad string, maybe it is a TripleString
+				return super.equals(other);
+			}
 			return false;
 		}
 		QuadString qs = (QuadString) other;
@@ -116,6 +120,9 @@ public QuadString tripleToString() {
 
 	@Override
 	public String toString() {
+		if (context.length() == 0) {
+			return super.toString();
+		}
 		return super.toString() + " " + context;
 	}
 }
diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/dictionary/impl/HashQuadDictionary.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/dictionary/impl/HashQuadDictionary.java
@@ -123,14 +123,21 @@ public void reorganize(TempTriples triples) {
 			mapPred.setNewID(j, this.stringToId(mapPred.getString(j), TripleComponentRole.PREDICATE));
 		}
 
-		for(long j=0;j<mapGraph.size();j++) {
-			mapGraph.setNewID(j, this.stringToId(mapGraph.getString(j), TripleComponentRole.GRAPH));
-		}
-
 		for(long j=0;j<mapObj.size();j++) {
 			mapObj.setNewID(j, this.stringToId(mapObj.getString(j), TripleComponentRole.OBJECT));
 		}
 
+		for(long j=0;j<mapGraph.size();j++) {
+			CharSequence str = mapGraph.getString(j);
+			// check that because stringToId returns 0 for empty strings and if a string is empty its id is
+			// 0 no matter what
+			if (str.length() == 0) {
+				mapGraph.setNewID(j, 1);
+			} else {
+				mapGraph.setNewID(j, this.stringToId(str, TripleComponentRole.GRAPH));
+			}
+		}
+
 		// Replace old IDs with news
 		triples.replaceAllIds(mapSubj, mapPred, mapObj, mapGraph);
 

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/TempHDTImporterOnePass.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/TempHDTImporterOnePass.java
@@ -68,11 +68,12 @@ public void processTriple(TripleString triple, long pos) {
 			if (dict.supportGraphs()) {
 				long g = dict.insert(triple.getGraph(), TripleComponentRole.GRAPH);
 				triples.insert(s, p, o, g);
+				size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+triple.getGraph().length()+5;
 			} else {
 				triples.insert(s, p, o);
+				size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+triple.getGraph().length()+4;  // Spaces and final dot
 			}
 			num++;
-			size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+triple.getGraph().length()+4;  // Spaces and final dot
 			ListenerUtil.notifyCond(listener, "Loaded "+num+" triples", num, 0, 100);
 		}
 	}
@@ -122,13 +123,23 @@ public TempHDT loadFromTriples(HDTOptions specs, Iterator<TripleString> iterator
         long size=0;
         while(iterator.hasNext()) {
         	TripleString triple = iterator.next();
-        	triples.insert(
-        			dictionary.insert(triple.getSubject(), TripleComponentRole.SUBJECT),
-        			dictionary.insert(triple.getPredicate(), TripleComponentRole.PREDICATE),
-        			dictionary.insert(triple.getObject(), TripleComponentRole.OBJECT)
-        			);
+			if (dictionary.supportGraphs()) {
+				triples.insert(
+						dictionary.insert(triple.getSubject(), TripleComponentRole.SUBJECT),
+						dictionary.insert(triple.getPredicate(), TripleComponentRole.PREDICATE),
+						dictionary.insert(triple.getObject(), TripleComponentRole.OBJECT),
+						dictionary.insert(triple.getGraph(), TripleComponentRole.GRAPH)
+				);
+				size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+triple.getGraph().length()+5;  // Spaces and final dot
+			} else {
+				triples.insert(
+						dictionary.insert(triple.getSubject(), TripleComponentRole.SUBJECT),
+						dictionary.insert(triple.getPredicate(), TripleComponentRole.PREDICATE),
+						dictionary.insert(triple.getObject(), TripleComponentRole.OBJECT)
+				);
+				size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+4;  // Spaces and final dot
+			}
         	num++;
-			size+=triple.getSubject().length()+triple.getPredicate().length()+triple.getObject().length()+4;  // Spaces and final dot
         	ListenerUtil.notifyCond(listener, "Loaded "+num+" triples", num, 0, 100);
         }
         dictionary.endProcessing();

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/quads/impl/BitmapQuadsIteratorG.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/quads/impl/BitmapQuadsIteratorG.java
@@ -10,7 +10,7 @@ public class BitmapQuadsIteratorG extends BitmapTriplesIterator {
 
 	// resolves ???G, S??G, SP?G, SPOG queries
 
-	private Bitmap bitmapGraph; // the graph bitmap for the search
+	private final Bitmap bitmapGraph; // the graph bitmap for the search
 
 	public BitmapQuadsIteratorG(BitmapTriples triples, TripleID pattern) {
 		super();

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/rdf/parsers/JenaNodeFormatter.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/rdf/parsers/JenaNodeFormatter.java
@@ -26,6 +26,7 @@
 import org.apache.jena.datatypes.xsd.impl.RDFLangString;
 import org.apache.jena.graph.Node;
 import org.apache.jena.rdf.model.RDFNode;
+import org.apache.jena.sparql.core.Quad;
 
 /**
  * Converts a Jena {@link Node} to a String format that will round trip back to the same Node via
@@ -42,7 +43,7 @@ public static String format(RDFNode n) {
     }
 
     public static String format(Node node) {
-        if (node == null) {
+        if (node == null || Quad.isDefaultGraph(node)) {
             return "";
         }
         if (node.isURI()) {

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/TriplesList.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/TriplesList.java
@@ -536,7 +536,7 @@ public void replaceAllIds(
 					(int)mapGraph.getNewID(triple.getGraph()    -1)
 				);
 			} else {
-				throw new RuntimeException("You must call the replaceAllIds method without a DictionaryIDMapping for graphs if the triples are not quads.");
+				throw new IllegalArgumentException("You must call the replaceAllIds method without a DictionaryIDMapping for graphs if the triples are not quads.");
 			}
 		}
 	}
@@ -550,7 +550,7 @@ public void replaceAllIds(
 		sorted=false;
 		for(TripleIDInt triple : arrayOfTriples) {
 			if (triple.isQuad()) {
-				throw new RuntimeException("You must call the replaceAllIds  method with a DictionaryIDMapping for graphs if the triples are quads.");
+				throw new IllegalArgumentException("You must call the replaceAllIds  method with a DictionaryIDMapping for graphs if the triples are quads.");
 			} else {
 				triple.setAll(
 					(int)mapSubj.getNewID(triple.getSubject()  -1),