search wip

hdt-api/src/main/java/org/rdfhdt/hdt/rdf/RDFAccess.java

@@ -52,4 +52,28 @@ public interface RDFAccess {
 	 * @throws NotFoundException when the triple cannot be found
 	 */
 	IteratorTripleString search(CharSequence subject, CharSequence predicate, CharSequence object) throws NotFoundException;
+
+	/**
+	 * Iterate over the triples of an RDF Set that match the specified pattern.
+	 * null and empty strings act as a wildcard. 
+	 * (e.g. search(null, null, null, null) iterates over all elements)
+	 * 
+	 * @param subject
+	 *            The subject to search
+	 * @param predicate
+	 *            The predicate to search
+	 * @param object
+	 *            The object to search
+	 * @param graph
+	 *            The graph to search
+	 * 
+	 * @return Iterator of TripleStrings
+	 * @throws NotFoundException when the triple cannot be found
+	 */
+	IteratorTripleString search(
+		CharSequence subject,
+		CharSequence predicate,
+		CharSequence object,
+		CharSequence graph
+	) throws NotFoundException;
 }

hdt-java-cli/src/main/java/org/rdfhdt/hdt/tools/HdtSearch.java

@@ -95,12 +95,18 @@ protected static void iterate(
 		subject = subject.length()==1 && subject.charAt(0)=='?' ? "" : subject;
 		predicate = predicate.length()==1 && predicate.charAt(0)=='?' ? "" : predicate;
 		object = object.length()==1 && object.charAt(0)=='?' ? "" : object;
+		graph = graph.length()==1 && graph.charAt(0)=='?' ? "" : graph;
 		// Iterate over triples as Strings
-		IteratorTripleString it = hdt.search(subject,predicate,object);
+		System.out.println("Searching for: "+subject+" "+predicate+" "+object+" "+graph);
+		System.out.println("hdt class name: " + hdt.getClass().getName());
+		IteratorTripleString it =
+			isHDTQ
+			? hdt.search(subject, predicate, object, graph)
+			: hdt.search(subject, predicate, object);
 		count = 0;
 		while(it.hasNext()) {
 			TripleString triple = it.next();
-			System.out.println(triple + " (graph: " + triple.getGraph() + ")");
+			System.out.println(triple);
 			count++;
 		}
 

hdt-java-core/src/main/java/org/rdfhdt/hdt/dictionary/impl/DictionaryPFCOptimizedExtractor.java

@@ -5,7 +5,12 @@
 import org.rdfhdt.hdt.enums.TripleComponentRole;
 
 public class DictionaryPFCOptimizedExtractor implements OptimizedExtractor{
-	private final PFCOptimizedExtractor shared, subjects, predicates, objects;
+	private final PFCOptimizedExtractor
+		shared,
+		subjects,
+		predicates,
+		objects,
+		graphs;
 	private final long numshared;
 
 	public DictionaryPFCOptimizedExtractor(FourSectionDictionary origDict) {
@@ -14,6 +19,11 @@ public DictionaryPFCOptimizedExtractor(FourSectionDictionary origDict) {
 		subjects = new PFCOptimizedExtractor((PFCDictionarySectionMap) origDict.subjects);
 		predicates = new PFCOptimizedExtractor((PFCDictionarySectionMap) origDict.predicates);
 		objects = new PFCOptimizedExtractor((PFCDictionarySectionMap) origDict.objects);
+		if (origDict.graphs == null) {
+			graphs = null;
+		} else {
+			graphs = new PFCOptimizedExtractor((PFCDictionarySectionMap) origDict.graphs);
+		}
 	}
 
 	public CharSequence idToString(long id, TripleComponentRole role) {
@@ -38,6 +48,8 @@ private PFCOptimizedExtractor getSection(long id, TripleComponentRole role) {
 			} else {
 				return objects;
 			}
+		case GRAPH:
+			return graphs;
 		}
 		throw new IllegalArgumentException();
 	}
@@ -54,6 +66,8 @@ private long getLocalId(long id, TripleComponentRole position) {
 			}
 		case PREDICATE:
 			return id;
+		case GRAPH:
+			return id;
 		}
 
 		throw new IllegalArgumentException();

hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/HDTImpl.java

@@ -55,6 +55,7 @@
 import org.rdfhdt.hdt.header.HeaderPrivate;
 import org.rdfhdt.hdt.iterator.DictionaryTranslateIterator;
 import org.rdfhdt.hdt.iterator.DictionaryTranslateIteratorBuffer;
+import org.rdfhdt.hdt.iterator.SuppliableIteratorTripleID;
 import org.rdfhdt.hdt.listener.ProgressListener;
 import org.rdfhdt.hdt.options.ControlInfo;
 import org.rdfhdt.hdt.options.ControlInformation;
@@ -320,6 +321,78 @@ public long getLastTriplePosition() {
 		}
 	}
 
+	@Override
+	public IteratorTripleString search(
+		CharSequence subject,
+		CharSequence predicate,
+		CharSequence object,
+		CharSequence graph
+	) throws NotFoundException {
+		if(isClosed) {
+			throw new IllegalStateException("Cannot search an already closed HDT");
+		}
+
+		// Conversion from TripleString to TripleID
+		TripleID triple = new TripleID(
+				dictionary.stringToId(subject, TripleComponentRole.SUBJECT),
+				dictionary.stringToId(predicate, TripleComponentRole.PREDICATE),
+				dictionary.stringToId(object, TripleComponentRole.OBJECT),
+				dictionary.stringToId(graph, TripleComponentRole.GRAPH)
+			);
+
+		if(triple.isNoMatch()) {
+			return new IteratorTripleString() {
+				@Override
+				public TripleString next() {
+					return null;
+				}
+				@Override
+				public boolean hasNext() {
+					return false;
+				}
+				@Override
+				public ResultEstimationType numResultEstimation() {
+					return ResultEstimationType.EXACT;
+				}
+				@Override
+				public void goToStart() {
+				}
+				@Override
+				public long estimatedNumResults() {
+					return 0;
+				}
+
+				@Override
+				public long getLastTriplePosition() {
+					throw new NotImplementedException();
+				}
+			};
+		}
+
+		if(isMapped) {
+			try {
+				if(dictionary instanceof MultipleSectionDictionary){
+					return new DictionaryTranslateIteratorBuffer(triples.search(triple), (MultipleSectionDictionary) dictionary, subject, predicate, object, graph);
+				}else{
+					SuppliableIteratorTripleID iterator = triples.search(triple);
+					return new DictionaryTranslateIteratorBuffer(
+						iterator,
+						(FourSectionDictionary) dictionary,
+						subject,
+						predicate,
+						object,
+						graph
+					);
+				}
+			}catch(NullPointerException e) {
+				e.printStackTrace();
+				return new DictionaryTranslateIterator(triples.search(triple), dictionary, subject, predicate, object, graph);
+			}
+		} else {
+			return new DictionaryTranslateIterator(triples.search(triple), dictionary, subject, predicate, object, graph);
+		}
+	}
+
 	public void loadFromParts(HeaderPrivate h, DictionaryPrivate d, TriplesPrivate t) {
 		this.header = h;
 		this.dictionary = d;

hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/WriteHDTImpl.java

@@ -121,4 +121,13 @@ public void close() throws IOException {
 	public IteratorTripleString search(CharSequence subject, CharSequence predicate, CharSequence object) {
 		throw new NotImplementedException();
 	}
+	@Override
+	public IteratorTripleString search(
+		CharSequence subject,
+		CharSequence predicate,
+		CharSequence object,
+		CharSequence graph
+	) {
+		throw new NotImplementedException();
+	}
 }

hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/diskimport/MapOnCallHDT.java

@@ -92,4 +92,13 @@ public void close() throws IOException {
 	public IteratorTripleString search(CharSequence subject, CharSequence predicate, CharSequence object) throws NotFoundException {
 		return mapOrGetHDT().search(subject, predicate, object);
 	}
+	@Override
+	public IteratorTripleString search(
+		CharSequence subject,
+		CharSequence predicate,
+		CharSequence object,
+		CharSequence graph
+	) throws NotFoundException {
+		return mapOrGetHDT().search(subject, predicate, object, graph);
+	}
 }

hdt-java-core/src/main/java/org/rdfhdt/hdt/header/PlainHeader.java

@@ -33,7 +33,6 @@
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.util.ArrayList;
-import java.util.Iterator;
 import java.util.List;
 
 import org.rdfhdt.hdt.enums.RDFNotation;
@@ -44,6 +43,7 @@
 import org.rdfhdt.hdt.options.ControlInfo;
 import org.rdfhdt.hdt.options.HDTOptions;
 import org.rdfhdt.hdt.options.HDTSpecification;
+import org.rdfhdt.hdt.quad.QuadString;
 import org.rdfhdt.hdt.rdf.RDFParserCallback.RDFCallback;
 import org.rdfhdt.hdt.rdf.parsers.RDFParserSimple;
 import org.rdfhdt.hdt.triples.IteratorTripleString;
@@ -163,6 +163,32 @@ public IteratorTripleString search(CharSequence subject, CharSequence predicate,
 		}
 		return new PlainHeaderIterator(this, pattern);
 	}
+	@Override
+	public IteratorTripleString search(
+		CharSequence subject,
+		CharSequence predicate,
+		CharSequence object,
+		CharSequence graph
+	) {
+		TripleString pattern;
+		String objStr = object.toString();
+		if(objStr.isEmpty() || objStr.charAt(0)=='<'|| objStr.charAt(0)=='"' || objStr.startsWith("http://")||objStr.startsWith("file://")) {
+			pattern = new QuadString(
+				HeaderUtil.cleanURI(subject),
+				HeaderUtil.cleanURI(predicate),
+				HeaderUtil.cleanURI(object),
+				HeaderUtil.cleanURI(graph)
+			);
+		} else {
+			pattern = new QuadString(
+				HeaderUtil.cleanURI(subject),
+				HeaderUtil.cleanURI(predicate),
+				'"'+objStr+'"',
+				HeaderUtil.cleanURI(graph)
+			);
+		}
+		return new PlainHeaderIterator(this, pattern);
+	}
 
 	@Override
 	public void processTriple(TripleString triple, long pos) {

hdt-java-core/src/main/java/org/rdfhdt/hdt/iterator/DictionaryTranslateIterator.java

@@ -46,7 +46,7 @@ public class DictionaryTranslateIterator implements IteratorTripleString {
 	/** The dictionary */
     final Dictionary dictionary;
 
-	CharSequence s, p, o;
+	CharSequence s, p, o, g;
 
 	long lastSid, lastPid, lastOid;
 	CharSequence lastSstr, lastPstr, lastOstr;
@@ -62,7 +62,7 @@ public class DictionaryTranslateIterator implements IteratorTripleString {
 	public DictionaryTranslateIterator(IteratorTripleID iteratorTripleID, Dictionary dictionary) {
 		this.iterator = iteratorTripleID;
 		this.dictionary = dictionary;
-		this.s = this.p = this.o = "";
+		this.s = this.p = this.o = this.g = "";
 	}
 
 	/**
@@ -79,6 +79,31 @@ public DictionaryTranslateIterator(IteratorTripleID iteratorTripleID, Dictionary
 		this.s = s==null ? "" : s;
 		this.p = p==null ? "" : p;
 		this.o = o==null ? "" : o;
+		this.g = "";
+	}
+
+	/**
+	 * Basic constructor
+	 * 
+	 * @param iteratorTripleID
+	 *            Iterator of TripleID to be used
+	 * @param dictionary
+	 *            The dictionary to be used
+	 */
+	public DictionaryTranslateIterator(
+		IteratorTripleID iteratorTripleID,
+		Dictionary dictionary,
+		CharSequence s,
+		CharSequence p,
+		CharSequence o,
+		CharSequence g
+	) {
+		this.iterator = iteratorTripleID;
+		this.dictionary = dictionary;
+		this.s = s==null ? "" : s;
+		this.p = p==null ? "" : p;
+		this.o = o==null ? "" : o;
+		this.g = g==null ? "" : g;
 	}
 
 	/*