mirror of https://github.com/apache/lucene.git
LUCENE-2858: fix queries module
git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/branches/lucene2858@1237320 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
parent
6d4484cb4a
commit
45c377007f
|
@ -22,8 +22,8 @@ import java.util.ArrayList;
|
|||
import java.util.List;
|
||||
import java.util.Iterator;
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.search.BitsFilteredDocIdSet;
|
||||
import org.apache.lucene.search.BooleanClause.Occur;
|
||||
import org.apache.lucene.search.DocIdSet;
|
||||
|
@ -52,7 +52,7 @@ public class BooleanFilter extends Filter implements Iterable<FilterClause> {
|
|||
@Override
|
||||
public DocIdSet getDocIdSet(AtomicReaderContext context, Bits acceptDocs) throws IOException {
|
||||
FixedBitSet res = null;
|
||||
final IndexReader reader = context.reader;
|
||||
final AtomicIndexReader reader = context.reader();
|
||||
|
||||
boolean hasShouldClauses = false;
|
||||
for (final FilterClause fc : clauses) {
|
||||
|
|
|
@ -17,8 +17,8 @@ package org.apache.lucene.queries;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.search.BitsFilteredDocIdSet;
|
||||
import org.apache.lucene.search.DocIdSet;
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
|
@ -129,7 +129,7 @@ public class ChainedFilter extends Filter {
|
|||
|
||||
private OpenBitSetDISI initialResult(AtomicReaderContext context, int logic, int[] index)
|
||||
throws IOException {
|
||||
IndexReader reader = context.reader;
|
||||
AtomicIndexReader reader = context.reader();
|
||||
OpenBitSetDISI result;
|
||||
/**
|
||||
* First AND operation takes place against a completely false
|
||||
|
|
|
@ -19,7 +19,7 @@ package org.apache.lucene.queries;
|
|||
|
||||
import java.io.IOException;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.IndexReader; // for javadocs
|
||||
import org.apache.lucene.search.Explanation;
|
||||
import org.apache.lucene.search.FieldCache; // for javadocs
|
||||
|
|
|
@ -22,7 +22,7 @@ import java.util.Set;
|
|||
import java.util.Arrays;
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.ComplexExplanation;
|
||||
import org.apache.lucene.search.Explanation;
|
||||
|
|
|
@ -18,7 +18,7 @@ package org.apache.lucene.queries;
|
|||
*/
|
||||
|
||||
import org.apache.lucene.index.*;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.search.DocIdSet;
|
||||
import org.apache.lucene.search.Filter;
|
||||
import org.apache.lucene.util.Bits;
|
||||
|
@ -55,7 +55,7 @@ public class TermsFilter extends Filter {
|
|||
|
||||
@Override
|
||||
public DocIdSet getDocIdSet(AtomicReaderContext context, Bits acceptDocs) throws IOException {
|
||||
IndexReader reader = context.reader;
|
||||
AtomicIndexReader reader = context.reader();
|
||||
FixedBitSet result = new FixedBitSet(reader.maxDoc());
|
||||
Fields fields = reader.fields();
|
||||
|
||||
|
|
|
@ -19,7 +19,7 @@ package org.apache.lucene.queries.function;
|
|||
|
||||
import org.apache.lucene.search.*;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.util.Bits;
|
||||
import org.apache.lucene.util.ToStringUtils;
|
||||
|
||||
|
|
|
@ -18,7 +18,7 @@ package org.apache.lucene.queries.function;
|
|||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.search.*;
|
||||
import org.apache.lucene.index.MultiFields;
|
||||
import org.apache.lucene.util.Bits;
|
||||
|
@ -96,7 +96,7 @@ public class FunctionQuery extends Query {
|
|||
|
||||
@Override
|
||||
public Explanation explain(AtomicReaderContext context, int doc) throws IOException {
|
||||
return ((AllScorer)scorer(context, true, true, context.reader.getLiveDocs())).explain(doc);
|
||||
return ((AllScorer)scorer(context, true, true, context.reader().getLiveDocs())).explain(doc);
|
||||
}
|
||||
}
|
||||
|
||||
|
@ -113,7 +113,7 @@ public class FunctionQuery extends Query {
|
|||
super(w);
|
||||
this.weight = w;
|
||||
this.qWeight = qWeight;
|
||||
this.reader = context.reader;
|
||||
this.reader = context.reader();
|
||||
this.maxDoc = reader.maxDoc();
|
||||
this.liveDocs = acceptDocs;
|
||||
vals = func.getValues(weight.context, context);
|
||||
|
|
|
@ -18,7 +18,7 @@ package org.apache.lucene.queries.function;
|
|||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.search.FieldComparator;
|
||||
import org.apache.lucene.search.FieldComparatorSource;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
|
|
|
@ -22,7 +22,7 @@ import org.apache.lucene.queries.function.ValueSource;
|
|||
import org.apache.lucene.queries.function.ValueSourceScorer;
|
||||
import org.apache.lucene.search.FieldCache;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.util.BytesRef;
|
||||
import org.apache.lucene.util.CharsRef;
|
||||
import org.apache.lucene.util.UnicodeUtil;
|
||||
|
@ -43,7 +43,7 @@ public abstract class StringIndexDocValues extends FunctionValues {
|
|||
|
||||
public StringIndexDocValues(ValueSource vs, AtomicReaderContext context, String field) throws IOException {
|
||||
try {
|
||||
termsIndex = FieldCache.DEFAULT.getTermsIndex(context.reader, field);
|
||||
termsIndex = FieldCache.DEFAULT.getTermsIndex(context.reader(), field);
|
||||
} catch (RuntimeException e) {
|
||||
throw new StringIndexException(field, e);
|
||||
}
|
||||
|
|
|
@ -19,7 +19,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.search.FieldCache;
|
||||
|
||||
|
@ -51,7 +51,7 @@ public class ByteFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final byte[] arr = cache.getBytes(readerContext.reader, field, parser, false);
|
||||
final byte[] arr = cache.getBytes(readerContext.reader(), field, parser, false);
|
||||
|
||||
return new FunctionValues() {
|
||||
@Override
|
||||
|
|
|
@ -17,7 +17,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.StringIndexDocValues;
|
||||
import org.apache.lucene.queries.function.ValueSource; //javadoc
|
||||
|
@ -35,7 +35,7 @@ public class BytesRefFieldSource extends FieldCacheSource {
|
|||
}
|
||||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, IndexReader.AtomicReaderContext readerContext) throws IOException {
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
return new StringIndexDocValues(this, readerContext, field) {
|
||||
|
||||
@Override
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
||||
|
|
|
@ -16,7 +16,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.DoubleDocValues;
|
||||
|
||||
|
|
|
@ -20,7 +20,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSourceScorer;
|
||||
|
@ -57,8 +57,8 @@ public class DoubleFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final double[] arr = cache.getDoubles(readerContext.reader, field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader, field);
|
||||
final double[] arr = cache.getDoubles(readerContext.reader(), field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader(), field);
|
||||
return new DoubleDocValues(this) {
|
||||
@Override
|
||||
public double doubleVal(int doc) {
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -20,7 +20,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
import org.apache.lucene.search.FieldCache;
|
||||
|
@ -55,8 +55,8 @@ public class FloatFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final float[] arr = cache.getFloats(readerContext.reader, field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader, field);
|
||||
final float[] arr = cache.getFloats(readerContext.reader(), field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader(), field);
|
||||
|
||||
return new FloatDocValues(this) {
|
||||
@Override
|
||||
|
|
|
@ -18,7 +18,7 @@
|
|||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.*;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
import org.apache.lucene.search.similarities.Similarity;
|
||||
|
|
|
@ -18,7 +18,7 @@
|
|||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.search.Explanation;
|
||||
|
|
|
@ -20,7 +20,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSourceScorer;
|
||||
|
@ -57,8 +57,8 @@ public class IntFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final int[] arr = cache.getInts(readerContext.reader, field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader, field);
|
||||
final int[] arr = cache.getInts(readerContext.reader(), field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader(), field);
|
||||
|
||||
return new IntDocValues(this) {
|
||||
final MutableValueInt val = new MutableValueInt();
|
||||
|
|
|
@ -21,7 +21,7 @@ import java.io.IOException;
|
|||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.IntDocValues;
|
||||
import org.apache.lucene.search.FieldCache.DocTerms;
|
||||
|
@ -52,8 +52,8 @@ public class JoinDocFreqValueSource extends FieldCacheSource {
|
|||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException
|
||||
{
|
||||
final DocTerms terms = cache.getTerms(readerContext.reader, field, true );
|
||||
final IndexReader top = ReaderUtil.getTopLevelContext(readerContext).reader;
|
||||
final DocTerms terms = cache.getTerms(readerContext.reader(), field, true );
|
||||
final IndexReader top = ReaderUtil.getTopLevelContext(readerContext).reader();
|
||||
|
||||
return new IntDocValues(this) {
|
||||
BytesRef ref = new BytesRef();
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -16,7 +16,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.StrDocValues;
|
||||
|
|
|
@ -20,7 +20,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSourceScorer;
|
||||
|
@ -66,8 +66,8 @@ public class LongFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final long[] arr = cache.getLongs(readerContext.reader, field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader, field);
|
||||
final long[] arr = cache.getLongs(readerContext.reader(), field, parser, true);
|
||||
final Bits valid = cache.getDocsWithField(readerContext.reader(), field);
|
||||
|
||||
return new LongDocValues(this) {
|
||||
@Override
|
||||
|
|
|
@ -16,7 +16,7 @@
|
|||
*/
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.BoolDocValues;
|
||||
|
|
|
@ -16,7 +16,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -16,7 +16,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
|
|
|
@ -18,7 +18,7 @@
|
|||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.DocValues;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
@ -57,7 +57,7 @@ public class NormValueSource extends ValueSource {
|
|||
throw new UnsupportedOperationException("requires a TFIDFSimilarity (such as DefaultSimilarity)");
|
||||
}
|
||||
final TFIDFSimilarity similarity = (TFIDFSimilarity) sim;
|
||||
DocValues dv = readerContext.reader.normValues(field);
|
||||
DocValues dv = readerContext.reader().normValues(field);
|
||||
|
||||
if (dv == null) {
|
||||
return new ConstDoubleDocValues(0.0, this);
|
||||
|
|
|
@ -16,7 +16,7 @@
|
|||
*/
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.util.ReaderUtil;
|
||||
|
@ -37,7 +37,7 @@ public class NumDocsValueSource extends ValueSource {
|
|||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
// Searcher has no numdocs so we must use the reader instead
|
||||
return new ConstIntDocValues(ReaderUtil.getTopLevelContext(readerContext).reader.numDocs(), this);
|
||||
return new ConstIntDocValues(ReaderUtil.getTopLevelContext(readerContext).reader().numDocs(), this);
|
||||
}
|
||||
|
||||
@Override
|
||||
|
|
|
@ -21,7 +21,7 @@ import java.util.Map;
|
|||
|
||||
import org.apache.lucene.index.DocValues.Source;
|
||||
import org.apache.lucene.index.DocValues.Type;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
|
||||
|
@ -43,7 +43,7 @@ public class NumericIndexDocValueSource extends ValueSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final Source source = readerContext.reader.docValues(field)
|
||||
final Source source = readerContext.reader().docValues(field)
|
||||
.getSource();
|
||||
Type type = source.type();
|
||||
switch (type) {
|
||||
|
|
|
@ -17,9 +17,11 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.AtomicIndexReader;
|
||||
import org.apache.lucene.index.CompositeIndexReader;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.SlowMultiReaderWrapper;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.SlowCompositeReaderWrapper;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.IntDocValues;
|
||||
|
@ -66,8 +68,11 @@ public class OrdFieldSource extends ValueSource {
|
|||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final int off = readerContext.docBase;
|
||||
final IndexReader topReader = ReaderUtil.getTopLevelContext(readerContext).reader;
|
||||
final FieldCache.DocTermsIndex sindex = FieldCache.DEFAULT.getTermsIndex(new SlowMultiReaderWrapper(topReader), field);
|
||||
final IndexReader topReader = ReaderUtil.getTopLevelContext(readerContext).reader();
|
||||
final AtomicIndexReader r = topReader instanceof CompositeIndexReader
|
||||
? new SlowCompositeReaderWrapper((CompositeIndexReader)topReader)
|
||||
: (AtomicIndexReader) topReader;
|
||||
final FieldCache.DocTermsIndex sindex = FieldCache.DEFAULT.getTermsIndex(r, field);
|
||||
return new IntDocValues(this) {
|
||||
protected String toTerm(String readableValue) {
|
||||
return readableValue;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
@ -96,7 +96,7 @@ class QueryDocValues extends FloatDocValues {
|
|||
super(vs);
|
||||
|
||||
this.readerContext = readerContext;
|
||||
this.acceptDocs = readerContext.reader.getLiveDocs();
|
||||
this.acceptDocs = readerContext.reader().getLiveDocs();
|
||||
this.defVal = vs.defVal;
|
||||
this.q = vs.q;
|
||||
this.fcontext = fcontext;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -17,9 +17,11 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.AtomicIndexReader;
|
||||
import org.apache.lucene.index.CompositeIndexReader;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.SlowMultiReaderWrapper;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.SlowCompositeReaderWrapper;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.IntDocValues;
|
||||
|
@ -65,10 +67,13 @@ public class ReverseOrdFieldSource extends ValueSource {
|
|||
// TODO: this is trappy? perhaps this query instead should make you pass a slow reader yourself?
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final IndexReader topReader = ReaderUtil.getTopLevelContext(readerContext).reader;
|
||||
final IndexReader topReader = ReaderUtil.getTopLevelContext(readerContext).reader();
|
||||
final AtomicIndexReader r = topReader instanceof CompositeIndexReader
|
||||
? new SlowCompositeReaderWrapper((CompositeIndexReader)topReader)
|
||||
: (AtomicIndexReader) topReader;
|
||||
final int off = readerContext.docBase;
|
||||
|
||||
final FieldCache.DocTermsIndex sindex = FieldCache.DEFAULT.getTermsIndex(new SlowMultiReaderWrapper(topReader), field);
|
||||
final FieldCache.DocTermsIndex sindex = FieldCache.DEFAULT.getTermsIndex(r, field);
|
||||
final int end = sindex.numOrd();
|
||||
|
||||
return new IntDocValues(this) {
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
@ -66,7 +66,7 @@ public class ScaleFloatFunction extends ValueSource {
|
|||
float maxVal = Float.NEGATIVE_INFINITY;
|
||||
|
||||
for (AtomicReaderContext leaf : leaves) {
|
||||
int maxDoc = leaf.reader.maxDoc();
|
||||
int maxDoc = leaf.reader().maxDoc();
|
||||
FunctionValues vals = source.getValues(context, leaf);
|
||||
for (int i=0; i<maxDoc; i++) {
|
||||
|
||||
|
|
|
@ -19,7 +19,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.search.FieldCache;
|
||||
|
||||
|
@ -48,7 +48,7 @@ public class ShortFieldSource extends FieldCacheSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
final short[] arr = cache.getShorts(readerContext.reader, field, parser, false);
|
||||
final short[] arr = cache.getShorts(readerContext.reader(), field, parser, false);
|
||||
|
||||
return new FunctionValues() {
|
||||
@Override
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.BoolDocValues;
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
|
|
|
@ -17,8 +17,8 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.Fields;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.Terms;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
|
@ -50,15 +50,15 @@ public class SumTotalTermFreqValueSource extends ValueSource {
|
|||
}
|
||||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, IndexReader.AtomicReaderContext readerContext) throws IOException {
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
return (FunctionValues)context.get(this);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void createWeight(Map context, IndexSearcher searcher) throws IOException {
|
||||
long sumTotalTermFreq = 0;
|
||||
for (IndexReader.AtomicReaderContext readerContext : searcher.getTopReaderContext().leaves()) {
|
||||
Fields fields = readerContext.reader.fields();
|
||||
for (AtomicReaderContext readerContext : searcher.getTopReaderContext().leaves()) {
|
||||
Fields fields = readerContext.reader().fields();
|
||||
if (fields == null) continue;
|
||||
Terms terms = fields.terms(indexedField);
|
||||
if (terms == null) continue;
|
||||
|
|
|
@ -18,7 +18,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
*/
|
||||
|
||||
import org.apache.lucene.index.*;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.FloatDocValues;
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
|
@ -42,7 +42,7 @@ public class TFValueSource extends TermFreqValueSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
Fields fields = readerContext.reader.fields();
|
||||
Fields fields = readerContext.reader().fields();
|
||||
final Terms terms = fields.terms(field);
|
||||
final Similarity sim = ((IndexSearcher)context.get("searcher")).getSimilarityProvider().get(field);
|
||||
if (!(sim instanceof TFIDFSimilarity)) {
|
||||
|
|
|
@ -18,7 +18,7 @@
|
|||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.*;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.docvalues.IntDocValues;
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
|
@ -39,7 +39,7 @@ public class TermFreqValueSource extends DocFreqValueSource {
|
|||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
Fields fields = readerContext.reader.fields();
|
||||
Fields fields = readerContext.reader().fields();
|
||||
final Terms terms = fields.terms(field);
|
||||
|
||||
return new IntDocValues(this) {
|
||||
|
|
|
@ -17,7 +17,7 @@
|
|||
|
||||
package org.apache.lucene.queries.function.valuesource;
|
||||
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.queries.function.docvalues.LongDocValues;
|
||||
|
@ -54,15 +54,15 @@ public class TotalTermFreqValueSource extends ValueSource {
|
|||
}
|
||||
|
||||
@Override
|
||||
public FunctionValues getValues(Map context, IndexReader.AtomicReaderContext readerContext) throws IOException {
|
||||
public FunctionValues getValues(Map context, AtomicReaderContext readerContext) throws IOException {
|
||||
return (FunctionValues)context.get(this);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void createWeight(Map context, IndexSearcher searcher) throws IOException {
|
||||
long totalTermFreq = 0;
|
||||
for (IndexReader.AtomicReaderContext readerContext : searcher.getTopReaderContext().leaves()) {
|
||||
totalTermFreq += readerContext.reader.totalTermFreq(indexedField, indexedBytes);
|
||||
for (AtomicReaderContext readerContext : searcher.getTopReaderContext().leaves()) {
|
||||
totalTermFreq += readerContext.reader().totalTermFreq(indexedField, indexedBytes);
|
||||
}
|
||||
final long ttf = Math.max(-1, totalTermFreq); // we may have added up -1s if not supported
|
||||
context.put(this, new LongDocValues(this) {
|
||||
|
|
|
@ -16,7 +16,7 @@ package org.apache.lucene.queries.function.valuesource;
|
|||
* limitations under the License.
|
||||
*/
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.FunctionValues;
|
||||
import org.apache.lucene.queries.function.ValueSource;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
|
|
|
@ -25,6 +25,7 @@ import org.apache.lucene.document.Document;
|
|||
import org.apache.lucene.index.Fields;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexableField;
|
||||
import org.apache.lucene.index.MultiFields;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.index.Terms;
|
||||
import org.apache.lucene.index.TermsEnum;
|
||||
|
@ -34,7 +35,6 @@ import org.apache.lucene.search.similarities.TFIDFSimilarity;
|
|||
import org.apache.lucene.util.BytesRef;
|
||||
import org.apache.lucene.util.CharsRef;
|
||||
import org.apache.lucene.util.PriorityQueue;
|
||||
import org.apache.lucene.util.ReaderUtil;
|
||||
import org.apache.lucene.util.UnicodeUtil;
|
||||
|
||||
|
||||
|
@ -570,7 +570,7 @@ public final class MoreLikeThis {
|
|||
public Query like(int docNum) throws IOException {
|
||||
if (fieldNames == null) {
|
||||
// gather list of valid fields from lucene
|
||||
Collection<String> fields = ReaderUtil.getIndexedFields(ir);
|
||||
Collection<String> fields = MultiFields.getIndexedFields(ir);
|
||||
fieldNames = fields.toArray(new String[fields.size()]);
|
||||
}
|
||||
|
||||
|
|
|
@ -21,17 +21,16 @@ import org.apache.lucene.analysis.MockAnalyzer;
|
|||
import org.apache.lucene.analysis.MockTokenizer;
|
||||
import org.apache.lucene.document.Document;
|
||||
import org.apache.lucene.document.TextField;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.RandomIndexWriter;
|
||||
import org.apache.lucene.index.SlowMultiReaderWrapper;
|
||||
import org.apache.lucene.index.SlowCompositeReaderWrapper;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.BooleanClause.Occur;
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
import org.apache.lucene.search.Filter;
|
||||
import org.apache.lucene.search.TermRangeFilter;
|
||||
import org.apache.lucene.search.DocIdSet;
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
import org.apache.lucene.search.TermQuery;
|
||||
import org.apache.lucene.search.QueryWrapperFilter;
|
||||
import org.apache.lucene.store.Directory;
|
||||
|
@ -43,7 +42,7 @@ import java.io.IOException;
|
|||
|
||||
public class BooleanFilterTest extends LuceneTestCase {
|
||||
private Directory directory;
|
||||
private IndexReader reader;
|
||||
private AtomicIndexReader reader;
|
||||
|
||||
@Override
|
||||
public void setUp() throws Exception {
|
||||
|
@ -57,7 +56,7 @@ public class BooleanFilterTest extends LuceneTestCase {
|
|||
addDoc(writer, "guest", "020", "20050101", "Y");
|
||||
addDoc(writer, "admin", "020", "20050101", "Maybe");
|
||||
addDoc(writer, "admin guest", "030", "20050101", "N");
|
||||
reader = new SlowMultiReaderWrapper(writer.getReader());
|
||||
reader = new SlowCompositeReaderWrapper(writer.getReader());
|
||||
writer.close();
|
||||
}
|
||||
|
||||
|
@ -97,7 +96,7 @@ public class BooleanFilterTest extends LuceneTestCase {
|
|||
return new Filter() {
|
||||
@Override
|
||||
public DocIdSet getDocIdSet(AtomicReaderContext context, Bits acceptDocs) {
|
||||
return new FixedBitSet(context.reader.maxDoc());
|
||||
return new FixedBitSet(context.reader().maxDoc());
|
||||
}
|
||||
};
|
||||
}
|
||||
|
|
|
@ -22,10 +22,10 @@ import java.util.HashSet;
|
|||
import org.apache.lucene.document.Document;
|
||||
import org.apache.lucene.document.StringField;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.MultiReader;
|
||||
import org.apache.lucene.index.RandomIndexWriter;
|
||||
import org.apache.lucene.index.SlowMultiReaderWrapper;
|
||||
import org.apache.lucene.index.SlowCompositeReaderWrapper;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.Filter;
|
||||
import org.apache.lucene.store.Directory;
|
||||
|
@ -62,27 +62,26 @@ public class TermsFilterTest extends LuceneTestCase {
|
|||
doc.add(newField(fieldName, "" + term, StringField.TYPE_STORED));
|
||||
w.addDocument(doc);
|
||||
}
|
||||
IndexReader reader = new SlowMultiReaderWrapper(w.getReader());
|
||||
assertTrue(reader.getTopReaderContext().isAtomic);
|
||||
IndexReader reader = new SlowCompositeReaderWrapper(w.getReader());
|
||||
assertTrue(reader.getTopReaderContext() instanceof AtomicReaderContext);
|
||||
AtomicReaderContext context = (AtomicReaderContext) reader.getTopReaderContext();
|
||||
assertTrue(context.isAtomic);
|
||||
w.close();
|
||||
|
||||
TermsFilter tf = new TermsFilter();
|
||||
tf.addTerm(new Term(fieldName, "19"));
|
||||
FixedBitSet bits = (FixedBitSet) tf.getDocIdSet(context, context.reader.getLiveDocs());
|
||||
FixedBitSet bits = (FixedBitSet) tf.getDocIdSet(context, context.reader().getLiveDocs());
|
||||
assertEquals("Must match nothing", 0, bits.cardinality());
|
||||
|
||||
tf.addTerm(new Term(fieldName, "20"));
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader.getLiveDocs());
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader().getLiveDocs());
|
||||
assertEquals("Must match 1", 1, bits.cardinality());
|
||||
|
||||
tf.addTerm(new Term(fieldName, "10"));
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader.getLiveDocs());
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader().getLiveDocs());
|
||||
assertEquals("Must match 2", 2, bits.cardinality());
|
||||
|
||||
tf.addTerm(new Term(fieldName, "00"));
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader.getLiveDocs());
|
||||
bits = (FixedBitSet) tf.getDocIdSet(context, context.reader().getLiveDocs());
|
||||
assertEquals("Must match 2", 2, bits.cardinality());
|
||||
|
||||
reader.close();
|
||||
|
@ -112,8 +111,8 @@ public class TermsFilterTest extends LuceneTestCase {
|
|||
tf.addTerm(new Term(fieldName, "content1"));
|
||||
|
||||
MultiReader multi = new MultiReader(reader1, reader2);
|
||||
for (IndexReader.AtomicReaderContext context : ReaderUtil.leaves(multi.getTopReaderContext())) {
|
||||
FixedBitSet bits = (FixedBitSet) tf.getDocIdSet(context, context.reader.getLiveDocs());
|
||||
for (AtomicReaderContext context : ReaderUtil.leaves(multi.getTopReaderContext())) {
|
||||
FixedBitSet bits = (FixedBitSet) tf.getDocIdSet(context, context.reader().getLiveDocs());
|
||||
assertTrue("Must be >= 0", bits.cardinality() >= 0);
|
||||
}
|
||||
multi.close();
|
||||
|
|
|
@ -28,7 +28,7 @@ import java.io.IOException;
|
|||
import java.util.HashMap;
|
||||
import java.util.Map;
|
||||
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.AtomicIndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.Term;
|
||||
|
||||
|
@ -174,11 +174,11 @@ public class TestCustomScoreQuery extends FunctionTestSetup {
|
|||
|
||||
@Override
|
||||
protected CustomScoreProvider getCustomScoreProvider(AtomicReaderContext context) throws IOException {
|
||||
final int[] values = FieldCache.DEFAULT.getInts(context.reader, INT_FIELD, false);
|
||||
final int[] values = FieldCache.DEFAULT.getInts(context.reader(), INT_FIELD, false);
|
||||
return new CustomScoreProvider(context) {
|
||||
@Override
|
||||
public float customScore(int doc, float subScore, float valSrcScore) throws IOException {
|
||||
assertTrue(doc <= context.reader.maxDoc());
|
||||
assertTrue(doc <= context.reader().maxDoc());
|
||||
return values[doc];
|
||||
}
|
||||
};
|
||||
|
|
|
@ -19,11 +19,9 @@ package org.apache.lucene.queries.function;
|
|||
|
||||
import org.apache.lucene.index.CorruptIndexException;
|
||||
import org.apache.lucene.index.IndexReader;
|
||||
import org.apache.lucene.index.IndexReader.AtomicReaderContext;
|
||||
import org.apache.lucene.queries.function.valuesource.OrdFieldSource;
|
||||
import org.apache.lucene.queries.function.valuesource.ReverseOrdFieldSource;
|
||||
import org.apache.lucene.search.*;
|
||||
import org.apache.lucene.util.ReaderUtil;
|
||||
import org.junit.BeforeClass;
|
||||
import org.junit.Test;
|
||||
|
||||
|
|
Loading…
Reference in New Issue