Menu
▾
▴
[Bigdata-commit] SF.net SVN: bigdata:[4312] branches/QUADS_QUERY_BRANCH/bigdata
|
From: <mrp...@us...> - 2011-03-17 20:55:19
|
Revision: 4312
http://bigdata.svn.sourceforge.net/bigdata/?rev=4312&view=rev
Author: mrpersonick
Date: 2011-03-17 20:55:10 +0000 (Thu, 17 Mar 2011)
Log Message:
-----------
added matchAll capability to free text search
Modified Paths:
--------------
branches/QUADS_QUERY_BRANCH/bigdata/src/java/com/bigdata/search/FullTextIndex.java
branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/lexicon/ITextIndexer.java
branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/store/BD.java
branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/test/com/bigdata/rdf/lexicon/TestFullTextIndex.java
branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl3.java
branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java
branches/QUADS_QUERY_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java
Added Paths:
-----------
branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/RunningQueryCloseableIterator.java
Modified: branches/QUADS_QUERY_BRANCH/bigdata/src/java/com/bigdata/search/FullTextIndex.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata/src/java/com/bigdata/search/FullTextIndex.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata/src/java/com/bigdata/search/FullTextIndex.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -37,6 +37,7 @@
import java.util.Iterator;
import java.util.List;
import java.util.Locale;
+import java.util.Map;
import java.util.Properties;
import java.util.Set;
import java.util.UUID;
@@ -901,9 +902,6 @@
}
- /*
- * FIXME [prefixMatch] has not been implemented yet.
- */
public Hiterator search(final String query, final String languageCode,
final boolean prefixMatch) {
@@ -913,6 +911,7 @@
prefixMatch,//
.4, // minCosine
10000, // maxRank
+ false, // matchAllTerms
this.timeout,//
TimeUnit.MILLISECONDS//
);
@@ -942,7 +941,7 @@
final double minCosine, final int maxRank) {
return search(query, languageCode, false/* prefixMatch */, minCosine,
- maxRank, this.timeout, TimeUnit.MILLISECONDS);
+ maxRank, false, this.timeout, TimeUnit.MILLISECONDS);
}
@@ -990,6 +989,8 @@
* <code>free</code> will be an exact match on <code>free</code>
* but a partial match on <code>freedom</code>. When
* <code>false</code>, only exact matches will be made.
+ * @param matchAllTerms
+ * if true, return only hits that match all search terms
* @param timeout
* The timeout -or- ZERO (0) for NO timeout (this is equivalent
* to using {@link Long#MAX_VALUE}).
@@ -1014,7 +1015,8 @@
*/
public Hiterator<Hit> search(final String query, final String languageCode,
final boolean prefixMatch, final double minCosine,
- final int maxRank, long timeout, final TimeUnit unit) {
+ final int maxRank, final boolean matchAllTerms,
+ long timeout, final TimeUnit unit) {
final long begin = System.currentTimeMillis();
@@ -1039,6 +1041,7 @@
if (log.isInfoEnabled())
log.info("languageCode=[" + languageCode + "], text=[" + query
+ "], minCosine=" + minCosine + ", maxRank=" + maxRank
+ + ", matchAllTerms=" + matchAllTerms
+ ", timeout=" + timeout + ", unit=" + unit);
if (timeout == 0L) {
@@ -1126,6 +1129,22 @@
}
}
+
+ if (matchAllTerms) {
+ final int nterms = qdata.terms.size();
+
+ if (log.isInfoEnabled())
+ log.info("nterms: " + nterms);
+
+ final Iterator<Map.Entry<Long,Hit>> it = hits.entrySet().iterator();
+ while (it.hasNext()) {
+ final Hit hit = it.next().getValue();
+ if (log.isInfoEnabled())
+ log.info("hit terms: " + hit.getTermCount());
+ if (hit.getTermCount() != nterms)
+ it.remove();
+ }
+ }
// #of hits.
final int nhits = hits.size();
@@ -1151,7 +1170,7 @@
if (log.isInfoEnabled())
log.info("Rank ordering "+nhits+" hits by relevance");
- final Hit[] a = hits.values().toArray(new Hit[0]);
+ final Hit[] a = hits.values().toArray(new Hit[nhits]);
Arrays.sort(a);
Modified: branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/lexicon/ITextIndexer.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/lexicon/ITextIndexer.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/lexicon/ITextIndexer.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -100,6 +100,8 @@
* The minimum cosine that will be returned.
* @param maxRank
* The upper bound on the #of hits in the result set.
+ * @param matchAllTerms
+ * if true, return only hits that match all search terms
* @param timeout
* The timeout -or- ZERO (0) for NO timeout (this is equivalent
* to using {@link Long#MAX_VALUE}).
@@ -110,6 +112,7 @@
*/
public Hiterator<A> search(final String query, final String languageCode,
final boolean prefixMatch, final double minCosine,
- final int maxRank, long timeout, final TimeUnit unit);
+ final int maxRank, final boolean matchAllTerms,
+ long timeout, final TimeUnit unit);
}
Modified: branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/store/BD.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/store/BD.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/java/com/bigdata/rdf/store/BD.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -158,7 +158,7 @@
* <p>
* <pre>
*
- * select ?s ?rank
+ * select ?s
* where {
* ?s bd:search "scale-out RDF triplestore" .
* ?s bd:maxHits "5"^^xsd:int .
@@ -174,7 +174,7 @@
* <p>
* <pre>
*
- * select ?s ?matched
+ * select ?s
* where {
* ?s bd:search "scale-out RDF triplestore" .
* ?s bd:minRelevance "0.5"^^xsd:double .
@@ -185,6 +185,22 @@
final URI MIN_RELEVANCE = new URIImpl(SEARCH_NAMESPACE+"minRelevance");
/**
+ * Magic predicate used to query for free text search metadata. Use
+ * in conjunction with {@link #SEARCH} as follows:
+ * <p>
+ * <pre>
+ *
+ * select ?s
+ * where {
+ * ?s bd:search "scale-out RDF triplestore" .
+ * ?s bd:matchAllTerms "true" .
+ * }
+ *
+ * </pre>
+ */
+ final URI MATCH_ALL_TERMS = new URIImpl(SEARCH_NAMESPACE+"matchAllTerms");
+
+ /**
* Sesame has the notion of a "null" graph. Any time you insert a statement
* into a quad store and the context position is not specified, it is
* actually inserted into this "null" graph. If SPARQL <code>DATASET</code>
Modified: branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/test/com/bigdata/rdf/lexicon/TestFullTextIndex.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/test/com/bigdata/rdf/lexicon/TestFullTextIndex.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-rdf/src/test/com/bigdata/rdf/lexicon/TestFullTextIndex.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -133,7 +133,9 @@
final Hiterator hitr = store.getLexiconRelation().getSearchEngine()
.search(query, languageCode, false/* prefixMatch */, minCosine,
- Integer.MAX_VALUE/* maxRank */, 2L/* timeout */,
+ Integer.MAX_VALUE/* maxRank */,
+ false/* matchAllTerms */,
+ 2L/* timeout */,
TimeUnit.SECONDS);
// assertEquals("#hits", (long) expected.length, itr.size());
Modified: branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl3.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl3.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl3.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -22,6 +22,7 @@
import org.openrdf.model.URI;
import org.openrdf.model.Value;
import org.openrdf.model.impl.BooleanLiteralImpl;
+import org.openrdf.query.Binding;
import org.openrdf.query.BindingSet;
import org.openrdf.query.Dataset;
import org.openrdf.query.QueryEvaluationException;
@@ -60,6 +61,7 @@
import org.openrdf.query.algebra.evaluation.impl.EvaluationStrategyImpl;
import org.openrdf.query.algebra.evaluation.iterator.FilterIterator;
import org.openrdf.query.algebra.helpers.QueryModelVisitorBase;
+import org.openrdf.query.impl.MapBindingSet;
import com.bigdata.bop.BOp;
import com.bigdata.bop.BOpUtility;
@@ -75,6 +77,7 @@
import com.bigdata.bop.NV;
import com.bigdata.bop.PipelineOp;
import com.bigdata.bop.ap.Predicate;
+import com.bigdata.bop.bindingSet.ListBindingSet;
import com.bigdata.bop.constraint.INBinarySearch;
import com.bigdata.bop.engine.IRunningQuery;
import com.bigdata.bop.engine.QueryEngine;
@@ -98,7 +101,6 @@
import com.bigdata.rdf.internal.constraints.RangeBOp;
import com.bigdata.rdf.internal.constraints.SPARQLConstraint;
import com.bigdata.rdf.internal.constraints.SameTermBOp;
-import com.bigdata.rdf.internal.constraints.XSDBooleanIVValueExpression;
import com.bigdata.rdf.lexicon.LexiconRelation;
import com.bigdata.rdf.model.BigdataValue;
import com.bigdata.rdf.sail.BigdataSail.Options;
@@ -133,6 +135,7 @@
import com.bigdata.striterator.Dechunkerator;
import com.bigdata.striterator.DistinctFilter;
import com.bigdata.striterator.IChunkedOrderedIterator;
+import com.bigdata.striterator.ICloseableIterator;
/**
* Extended to rewrite Sesame {@link TupleExpr}s onto native {@link Rule}s and
@@ -662,7 +665,8 @@
final Value p = sp.getPredicateVar().getValue();
if (s == null && p != null &&
(BD.RELEVANCE.equals(p) || BD.MAX_HITS.equals(p) ||
- BD.MIN_RELEVANCE.equals(p))) {
+ BD.MIN_RELEVANCE.equals(p) ||
+ BD.MATCH_ALL_TERMS.equals(p))) {
final Var sVar = sp.getSubjectVar();
Set<StatementPattern> metadata = searchMetadata.get(sVar);
if (metadata != null) {
@@ -887,13 +891,9 @@
}
- /*
- * FIXME What is [bs]? It is not being used within this context.
- */
CloseableIteration<BindingSet, QueryEvaluationException>
doEvaluateNatively(final PipelineOp query, final BindingSet bs,
final QueryEngine queryEngine, final IVariable[] required
-// , final Collection<Filter> sesameFilters
)
throws QueryEvaluationException {
@@ -901,13 +901,16 @@
try {
// Submit query for evaluation.
- runningQuery = queryEngine.eval(query);
+ if (bs != null)
+ runningQuery = queryEngine.eval(query, toBindingSet(bs));
+ else
+ runningQuery = queryEngine.eval(query);
/*
* Wrap up the native bigdata query solution iterator as Sesame
* compatible iteration with materialized RDF Values.
*/
- return wrapQuery(runningQuery, required);//, sesameFilters);
+ return iterator(runningQuery, database, required);
} catch (UnsupportedOperatorException t) {
if (runningQuery != null) {
@@ -928,7 +931,35 @@
}
}
+
+ private static IBindingSet toBindingSet(final BindingSet src) {
+ if (src == null)
+ throw new IllegalArgumentException();
+
+ final ListBindingSet bindingSet = new ListBindingSet();
+
+ final Iterator<Binding> itr = src.iterator();
+
+ while(itr.hasNext()) {
+
+ final Binding binding = itr.next();
+
+ final IVariable var = com.bigdata.bop.Var.var(binding.getName());
+
+ final IV iv = ((BigdataValue) binding.getValue()).getIV();
+
+ final IConstant val = new Constant<IV>(iv);
+
+ bindingSet.set(var, val);
+
+ }
+
+ return bindingSet;
+
+ }
+
+
/**
* Wrap the {@link IRunningQuery#iterator()}, returning a Sesame compatible
* iteration which will visit Sesame binding sets having materialized RDF
@@ -952,22 +983,62 @@
// De-chunk the IBindingSet[] visited by that iterator.
final IChunkedOrderedIterator<IBindingSet> it2 =
new ChunkedWrappedIterator<IBindingSet>(
- new Dechunkerator<IBindingSet>(it1));
+ // Monitor IRunningQuery and cancel if Sesame iterator is closed.
+ new RunningQueryCloseableIterator<IBindingSet>(runningQuery,
+ new Dechunkerator<IBindingSet>(it1)));
// Materialize IVs as RDF Values.
final CloseableIteration<BindingSet, QueryEvaluationException> result =
- // Monitor IRunningQuery and cancel if Sesame iterator is closed.
- new RunningQueryCloseableIteration<BindingSet, QueryEvaluationException>(runningQuery,
// Convert bigdata binding sets to Sesame binding sets.
new Bigdata2Sesame2BindingSetIterator<QueryEvaluationException>(
// Materialize IVs as RDF Values.
new BigdataBindingSetResolverator(database, it2, required).start(
- database.getExecutorService())));
+ database.getExecutorService()));
return result;
}
+
+ public static ICloseableIterator<IBindingSet> iterator(
+ final IRunningQuery runningQuery) {
+
+ // The iterator draining the query solutions.
+ final IAsynchronousIterator<IBindingSet[]> it1 =
+ runningQuery.iterator();
+
+ // Dechunkify the original iterator
+ final ICloseableIterator<IBindingSet> it2 =
+ new Dechunkerator<IBindingSet>(it1);
+
+ // Monitor IRunningQuery and cancel if Sesame iterator is closed.
+ final ICloseableIterator<IBindingSet> it3 =
+ new RunningQueryCloseableIterator<IBindingSet>(runningQuery, it2);
+
+ return it3;
+
+ }
+
+ public static CloseableIteration<BindingSet, QueryEvaluationException>
+ iterator(final IRunningQuery runningQuery, final AbstractTripleStore db,
+ final IVariable[] required) {
+ final ICloseableIterator<IBindingSet> it1 = iterator(runningQuery);
+
+ // Wrap in an IChunkedOrderedIterator
+ final IChunkedOrderedIterator<IBindingSet> it2 =
+ new ChunkedWrappedIterator<IBindingSet>(it1);
+
+ // Materialize IVs as RDF Values.
+ final CloseableIteration<BindingSet, QueryEvaluationException> it3 =
+ // Convert bigdata binding sets to Sesame binding sets.
+ new Bigdata2Sesame2BindingSetIterator<QueryEvaluationException>(
+ // Materialize IVs as RDF Values.
+ new BigdataBindingSetResolverator(db, it2, required).start(
+ db.getExecutorService()));
+
+ return it3;
+
+ }
private void attachNamedGraphsFilterToSearches(final SOpTree sopTree) {
@@ -1582,6 +1653,7 @@
IVariableOrConstant<IV> relevance = new Constant(DummyIV.INSTANCE);
Literal maxHits = null;
Literal minRelevance = null;
+ boolean matchAllTerms = false;
for (StatementPattern meta : metadata) {
if (!meta.getSubjectVar().equals(subjVar)) {
@@ -1608,12 +1680,17 @@
throw new IllegalArgumentException("illegal metadata: " + meta);
}
minRelevance = (Literal) oVal;
+ } else if (BD.MATCH_ALL_TERMS.equals(pVal)) {
+ if (oVal == null || !(oVal instanceof Literal)) {
+ throw new IllegalArgumentException("illegal metadata: " + meta);
+ }
+ matchAllTerms = ((Literal) oVal).booleanValue();
}
}
final IAccessPathExpander expander =
new FreeTextSearchExpander(database, (Literal) objValue,
- maxHits, minRelevance);
+ maxHits, minRelevance, matchAllTerms);
// Decide on the correct arity for the predicate.
final BOp[] vars = new BOp[] {
@@ -1988,8 +2065,11 @@
final Iterator<IHit> itr = (Iterator)database.getLexiconRelation()
.getSearchEngine().search(label, languageCode,
- false/* prefixMatch */, 0d/* minCosine */,
- 10000/* maxRank */, 1000L/* timeout */,
+ false/* prefixMatch */,
+ 0d/* minCosine */,
+ 10000/* maxRank */,
+ false/* matchAllTerms */,
+ 0L/* timeout */,
TimeUnit.MILLISECONDS);
// ensure that named graphs are handled correctly for quads
Modified: branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -61,19 +61,21 @@
private final Literal query, maxHits, minRelevance;
+ private final boolean matchAllTerms;
+
/** Note: volatile for visibility (or use AtomicReference). */
private volatile Set<URI> graphs;
public FreeTextSearchExpander(final AbstractTripleStore database,
final Literal query) {
- this(database, query, null, null);
+ this(database, query, null, null, false);
}
public FreeTextSearchExpander(final AbstractTripleStore database,
final Literal query, final Literal maxHits,
- final Literal minRelevance) {
+ final Literal minRelevance, final boolean matchAllTerms) {
if (database == null)
throw new IllegalArgumentException();
@@ -89,6 +91,8 @@
this.minRelevance = minRelevance;
+ this.matchAllTerms = matchAllTerms;
+
}
public boolean backchain() {
@@ -176,7 +180,8 @@
prefixMatch,
minRelevance == null ? 0d : minRelevance.doubleValue()/* minCosine */,
maxHits == null ? 10000 : maxHits.intValue()+1/* maxRank */,
- 1000L/* timeout */, TimeUnit.MILLISECONDS);
+ matchAllTerms,
+ 0L/* timeout */, TimeUnit.MILLISECONDS);
}
return hiterator;
Added: branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/RunningQueryCloseableIterator.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/RunningQueryCloseableIterator.java (rev 0)
+++ branches/QUADS_QUERY_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/RunningQueryCloseableIterator.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -0,0 +1,106 @@
+package com.bigdata.rdf.sail;
+
+import java.util.NoSuchElementException;
+
+import com.bigdata.bop.IBindingSet;
+import com.bigdata.bop.engine.IRunningQuery;
+import com.bigdata.striterator.ICloseableIterator;
+
+/**
+ * Iteration construct wraps an {@link IRunningQuery} with logic to (a) verify
+ * that the {@link IRunningQuery} has not encountered an error; and (b) to cancel
+ * the {@link IRunningQuery} when the iteration is {@link #close() closed}.
+ */
+public class RunningQueryCloseableIterator<E extends IBindingSet>
+ implements ICloseableIterator<E> {
+
+ private final IRunningQuery runningQuery;
+ private final ICloseableIterator<E> src;
+ private boolean checkedFuture = false;
+ /**
+ * The next element is buffered so we can always return it if the
+ * {@link #runningQuery} was not aborted at the time that {@link #hasNext()}
+ * return <code>true</code>.
+ */
+ private E current = null;
+
+ public RunningQueryCloseableIterator(final IRunningQuery runningQuery,
+ final ICloseableIterator<E> src) {
+
+ this.runningQuery = runningQuery;
+ this.src = src;
+
+ }
+
+ public void close() {
+ runningQuery.cancel(true/* mayInterruptIfRunning */);
+ src.close();
+ }
+
+ public boolean hasNext() {
+
+ if (current != null) {
+ // Already buffered.
+ return true;
+ }
+
+ if (!src.hasNext()) {
+ // Source is exhausted.
+ return false;
+ }
+
+ // buffer the next element.
+ current = src.next();
+
+ // test for abnormal completion of the runningQuery.
+ if (!checkedFuture && runningQuery.isDone()) {
+ try {
+ runningQuery.get();
+ } catch (InterruptedException e) {
+ /*
+ * Interrupted while waiting on the Future (should not happen
+ * since the Future is already done).
+ */
+ throw new RuntimeException(e);
+ } catch (Throwable e) {
+ /*
+ * Exception thrown by the runningQuery.
+ */
+ if (runningQuery.getCause() != null) {
+ // abnormal termination - wrap and rethrow.
+ throw new RuntimeException(e);
+ }
+ // otherwise this is normal termination.
+ }
+ checkedFuture = true;
+ }
+
+ // the next element is now buffered.
+ return true;
+
+ }
+
+ public E next() {
+
+ if (!hasNext())
+ throw new NoSuchElementException();
+
+ final E tmp = current;
+
+ current = null;
+
+ return tmp;
+
+ }
+
+ /**
+ * Operation is not supported.
+ */
+ public void remove() {
+
+ // Not supported since we are buffering ahead.
+ throw new UnsupportedOperationException();
+
+ }
+
+}
Modified: branches/QUADS_QUERY_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java
===================================================================
--- branches/QUADS_QUERY_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java 2011-03-17 19:17:15 UTC (rev 4311)
+++ branches/QUADS_QUERY_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java 2011-03-17 20:55:10 UTC (rev 4312)
@@ -798,6 +798,7 @@
false, // prefixMatch
0d, // minCosine
10000, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -859,6 +860,7 @@
false, // prefixMatch
0d, // minCosine
maxHits+1, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -920,6 +922,7 @@
false, // prefixMatch
minRelevance, // minCosine
10000, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -985,6 +988,7 @@
false, // prefixMatch
minRelevance, // minCosine
10000, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -1052,6 +1056,7 @@
true, // prefixMatch
minRelevance, // minCosine
10000, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -1117,6 +1122,7 @@
true, // prefixMatch
minRelevance, // minCosine
10000, // maxRank (=maxResults + 1)
+ false, // matchAllTerms
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -1139,6 +1145,66 @@
}
+ { // match all terms
+
+ final String searchQuery = "how now brown cow";
+ final double minRelevance = 0.0d;
+
+ final String query =
+ "select ?s ?o " +
+ "where " +
+ "{ " +
+ " ?s <"+RDFS.LABEL+"> ?o . " +
+ " ?o <"+BD.SEARCH+"> \""+searchQuery+"\" . " +
+ " ?o <"+BD.MATCH_ALL_TERMS+"> \"true\" . " +
+ "}";
+
+ log.info("\n"+query);
+
+ final TupleQuery tupleQuery =
+ cxn.prepareTupleQuery(QueryLanguage.SPARQL, query);
+ tupleQuery.setIncludeInferred(true /* includeInferred */);
+ TupleQueryResult result = tupleQuery.evaluate();
+
+ int i = 0;
+ while (result.hasNext()) {
+ log.info(i++ + ": " + result.next().toString());
+ }
+// assertTrue("wrong # of results: " + i, i == 1);
+
+ result = tupleQuery.evaluate();
+
+ Collection<BindingSet> answer = new LinkedList<BindingSet>();
+
+ final ITextIndexer search =
+ sail.getDatabase().getLexiconRelation().getSearchEngine();
+ final Hiterator<IHit> hits =
+ search.search(searchQuery,
+ null, // languageCode
+ true, // prefixMatch
+ minRelevance, // minCosine
+ 10000, // maxRank (=maxResults + 1)
+ true, // matchAllTerms
+ 1000L, // timeout
+ TimeUnit.MILLISECONDS // unit
+ );
+
+ while (hits.hasNext()) {
+ final IHit hit = hits.next();
+ final IV id = new TermId(VTE.LITERAL, hit.getDocId());
+ final URI s = uris.get(id);
+ final Literal o = literals.get(id);
+ final BindingSet bs = createBindingSet(
+ new BindingImpl("s", s),
+ new BindingImpl("o", o));
+ log.info(bs);
+ answer.add(bs);
+ }
+
+ compare(result, answer);
+
+ }
+
} finally {
cxn.close();
}
@@ -1148,4 +1214,61 @@
}
+ /*
+
+prefix BIGDATA_QUERY_HINTS: <http://www.bigdata.com/queryHints#com.bigdata.rdf.sail.QueryHints.optimizer=None>
+prefix xsd: <http://www.w3.org/2001/XMLSchema#>
+select distinct ?target0
+where {
+ ?obj0 <http://www.bigdata.com/rdf/search#search> "stainless" .
+ ?target0 ?p0 ?obj0 .
+ ?obj1 <http://www.bigdata.com/rdf/search#search> "innovations" .
+ ?target1 ?p1 ?obj1 .
+ ?obj2 <http://www.bigdata.com/rdf/search#search> "cabin" .
+ ?target2 ?p2 ?obj2 .
+ filter(?target0 = ?target1 && ?target1 = ?target2) .
+}
+
+ FILTER (?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlock> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockAttribute> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockReference> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADFile> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADTable> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADTitleBlock> || ?category = <http://www.ms2w.com/ontologies/file/Directory> || ?category = <http://www.ms2w.com/ontologies/file/File> || ?category = <http://www.ms2w.com/ontologies/pdffile/PdfAnnotation> || ?category = <http://www.ms2w.com/ontologies/pdffile/PdfFile> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentAssembly> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeBOM> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/Component> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/Configuration> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentDrawing> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeIndented> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypePartsOnly> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeRevision> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxCopiedPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxStandardPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeTopLevelOnly> ) .
+
+prefix BIGDATA_QUERY_HINTS: <http://www.bigdata.com/queryHints#com.bigdata.rdf.sail.QueryHints.optimizer=None>
+prefix xsd: <http://www.w3.org/2001/XMLSchema#>
+select distinct ?target
+where {
+ ?obj0 <http://www.bigdata.com/rdf/search#search> "stainless" .
+ ?obj0 <http://www.bigdata.com/rdf/search#relevance> ?score0 .
+ ?target ?p0 ?obj0 .
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category0 .
+ {
+ ?obj1 <http://www.bigdata.com/rdf/search#search> "innovations" .
+ ?obj1 <http://www.bigdata.com/rdf/search#relevance> ?score1 .
+ ?target ?p1 ?obj1 .
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category1 .
+ }
+ {
+ ?obj2 <http://www.bigdata.com/rdf/search#search> "cabin" .
+ ?obj2 <http://www.bigdata.com/rdf/search#relevance> ?score2 .
+ ?target ?p2 ?obj2 .
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category2 .
+ }
+}
+ORDER BY DESC(?score2) DESC(?score1) DESC(?score0)
+LIMIT 10 OFFSET 0
+
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category0 .
+
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category1 .
+
+ ?target <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?category2 .
+
+ FILTER (?category0 = ... ) .
+ FILTER (?category1 = ... ) .
+ FILTER (?category2 = ... ) .
+
+ FILTER (?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlock> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockAttribute> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockReference> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADFile> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADTable> || ?category = <http://www.ms2w.com/ontologies/autocad/AutoCADTitleBlock> || ?category = <http://www.ms2w.com/ontologies/file/Directory> || ?category = <http://www.ms2w.com/ontologies/file/File> || ?category = <http://www.ms2w.com/ontologies/pdffile/PdfAnnotation> || ?category = <http://www.ms2w.com/ontologies/pdffile/PdfFile> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentAssembly> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeBOM> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/Component> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/Configuration> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentDrawing> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeIndented> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypePartsOnly> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeRevision> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxCopiedPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxStandardPart> || ?category = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeTopLevelOnly> ) .
+ FILTER (?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlock> || ?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockAttribute> || ?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockReference> || ?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADFile> || ?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADTable> || ?category0 = <http://www.ms2w.com/ontologies/autocad/AutoCADTitleBlock> || ?category0 = <http://www.ms2w.com/ontologies/file/Directory> || ?category0 = <http://www.ms2w.com/ontologies/file/File> || ?category0 = <http://www.ms2w.com/ontologies/pdffile/PdfAnnotation> || ?category0 = <http://www.ms2w.com/ontologies/pdffile/PdfFile> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentAssembly> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeBOM> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/Component> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/Configuration> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentDrawing> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeIndented> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentPart> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypePartsOnly> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeRevision> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxCopiedPart> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxStandardPart> || ?category0 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeTopLevelOnly> ) .
+ FILTER (?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlock> || ?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockAttribute> || ?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockReference> || ?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADFile> || ?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADTable> || ?category1 = <http://www.ms2w.com/ontologies/autocad/AutoCADTitleBlock> || ?category1 = <http://www.ms2w.com/ontologies/file/Directory> || ?category1 = <http://www.ms2w.com/ontologies/file/File> || ?category1 = <http://www.ms2w.com/ontologies/pdffile/PdfAnnotation> || ?category1 = <http://www.ms2w.com/ontologies/pdffile/PdfFile> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentAssembly> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeBOM> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/Component> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/Configuration> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentDrawing> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeIndented> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentPart> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypePartsOnly> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeRevision> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxCopiedPart> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxStandardPart> || ?category1 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeTopLevelOnly> ) .
+ FILTER (?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlock> || ?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockAttribute> || ?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADBlockReference> || ?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADFile> || ?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADTable> || ?category2 = <http://www.ms2w.com/ontologies/autocad/AutoCADTitleBlock> || ?category2 = <http://www.ms2w.com/ontologies/file/Directory> || ?category2 = <http://www.ms2w.com/ontologies/file/File> || ?category2 = <http://www.ms2w.com/ontologies/pdffile/PdfAnnotation> || ?category2 = <http://www.ms2w.com/ontologies/pdffile/PdfFile> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentAssembly> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeBOM> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/Component> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/Configuration> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentDrawing> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeIndented> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmDocumentPart> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypePartsOnly> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmTableTypeRevision> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxCopiedPart> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmToolboxStandardPart> || ?category2 = <http://www.ms2w.com/ontologies/solidworks/20091023/swDmBOMTableTypeTopLevelOnly> ) .
+ */
+
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
