LUCENE-9583: Remove RandomAccessVectorValuesProducer

lucene/backward-codecs/src/java/org/apache/lucene/backward_codecs/lucene90/Lucene90HnswGraphBuilder.java

@@ -22,7 +22,6 @@
 import java.util.Objects;
 import java.util.SplittableRandom;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.util.InfoStream;
 import org.apache.lucene.util.hnsw.NeighborQueue;
@@ -72,14 +71,14 @@ public final class Lucene90HnswGraphBuilder {
    *     to ensure repeatable construction.
    */
   public Lucene90HnswGraphBuilder(
-      RandomAccessVectorValuesProducer vectors,
+      RandomAccessVectorValues vectors,
       VectorSimilarityFunction similarityFunction,
       int maxConn,
       int beamWidth,
       long seed)
       throws IOException {
-    vectorValues = vectors.randomAccess();
-    buildVectors = vectors.randomAccess();
+    vectorValues = vectors.copy();
+    buildVectors = vectors.copy();
     this.similarityFunction = Objects.requireNonNull(similarityFunction);
     if (maxConn <= 0) {
       throw new IllegalArgumentException("maxConn must be positive");

lucene/backward-codecs/src/java/org/apache/lucene/backward_codecs/lucene90/Lucene90HnswVectorsReader.java

@@ -33,7 +33,6 @@
 import org.apache.lucene.index.FieldInfos;
 import org.apache.lucene.index.IndexFileNames;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.SegmentReadState;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.index.VectorValues;
@@ -381,8 +380,7 @@ int size() {
   }
 
   /** Read the vector values from the index input. This supports both iterated and random access. */
-  static class OffHeapVectorValues extends VectorValues
-      implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+  static class OffHeapVectorValues extends VectorValues implements RandomAccessVectorValues {
 
     final int dimension;
     final int[] ordToDoc;
@@ -468,7 +466,7 @@ public long cost() {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() {
+    public RandomAccessVectorValues copy() {
       return new OffHeapVectorValues(dimension, ordToDoc, dataIn.clone());
     }
 

lucene/backward-codecs/src/java/org/apache/lucene/backward_codecs/lucene91/Lucene91HnswVectorsReader.java

@@ -33,7 +33,6 @@
 import org.apache.lucene.index.FieldInfos;
 import org.apache.lucene.index.IndexFileNames;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.SegmentReadState;
 import org.apache.lucene.index.VectorEncoding;
 import org.apache.lucene.index.VectorSimilarityFunction;
@@ -422,8 +421,7 @@ int ordToDoc(int ord) {
   }
 
   /** Read the vector values from the index input. This supports both iterated and random access. */
-  static class OffHeapVectorValues extends VectorValues
-      implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+  static class OffHeapVectorValues extends VectorValues implements RandomAccessVectorValues {
 
     private final int dimension;
     private final int size;
@@ -516,7 +514,7 @@ public long cost() {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() {
+    public RandomAccessVectorValues copy() {
       return new OffHeapVectorValues(dimension, size, ordToDoc, dataIn.clone());
     }
 

lucene/backward-codecs/src/java/org/apache/lucene/backward_codecs/lucene92/OffHeapVectorValues.java

@@ -21,7 +21,6 @@
 import java.nio.ByteBuffer;
 import org.apache.lucene.codecs.lucene90.IndexedDISI;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.VectorValues;
 import org.apache.lucene.store.IndexInput;
 import org.apache.lucene.store.RandomAccessInput;
@@ -30,8 +29,7 @@
 import org.apache.lucene.util.packed.DirectMonotonicReader;
 
 /** Read the vector values from the index input. This supports both iterated and random access. */
-abstract class OffHeapVectorValues extends VectorValues
-    implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+abstract class OffHeapVectorValues extends VectorValues implements RandomAccessVectorValues {
 
   protected final int dimension;
   protected final int size;
@@ -144,7 +142,7 @@ public int advance(int target) throws IOException {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       return new DenseOffHeapVectorValues(dimension, size, slice.clone());
     }
 
@@ -217,7 +215,7 @@ public int advance(int target) throws IOException {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       return new SparseOffHeapVectorValues(fieldEntry, dataIn, slice.clone());
     }
 
@@ -294,7 +292,7 @@ public long cost() {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       throw new UnsupportedOperationException();
     }
 

lucene/backward-codecs/src/test/org/apache/lucene/backward_codecs/lucene90/Lucene90HnswVectorsWriter.java

@@ -26,7 +26,7 @@
 import org.apache.lucene.index.BufferingKnnVectorsWriter;
 import org.apache.lucene.index.FieldInfo;
 import org.apache.lucene.index.IndexFileNames;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
+import org.apache.lucene.index.RandomAccessVectorValues;
 import org.apache.lucene.index.SegmentWriteState;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.index.VectorValues;
@@ -224,7 +224,7 @@ private void writeGraphOffsets(IndexOutput out, long[] offsets) throws IOExcepti
 
   private void writeGraph(
       IndexOutput graphData,
-      RandomAccessVectorValuesProducer vectorValues,
+      RandomAccessVectorValues vectorValues,
       VectorSimilarityFunction similarityFunction,
       long graphDataOffset,
       long[] offsets,
@@ -239,7 +239,7 @@ private void writeGraph(
             beamWidth,
             Lucene90HnswGraphBuilder.randSeed);
     hnswGraphBuilder.setInfoStream(segmentWriteState.infoStream);
-    Lucene90OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.randomAccess());
+    Lucene90OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.copy());
 
     for (int ord = 0; ord < offsets.length; ord++) {
       // write graph

lucene/backward-codecs/src/test/org/apache/lucene/backward_codecs/lucene91/Lucene91HnswGraphBuilder.java

@@ -24,7 +24,6 @@
 import java.util.Objects;
 import java.util.SplittableRandom;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.VectorEncoding;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.util.FixedBitSet;
@@ -79,14 +78,14 @@ public final class Lucene91HnswGraphBuilder {
    *     to ensure repeatable construction.
    */
   public Lucene91HnswGraphBuilder(
-      RandomAccessVectorValuesProducer vectors,
+      RandomAccessVectorValues vectors,
       VectorSimilarityFunction similarityFunction,
       int maxConn,
       int beamWidth,
       long seed)
       throws IOException {
-    vectorValues = vectors.randomAccess();
-    buildVectors = vectors.randomAccess();
+    vectorValues = vectors.copy();
+    buildVectors = vectors.copy();
     this.similarityFunction = Objects.requireNonNull(similarityFunction);
     if (maxConn <= 0) {
       throw new IllegalArgumentException("maxConn must be positive");

lucene/backward-codecs/src/test/org/apache/lucene/backward_codecs/lucene91/Lucene91HnswVectorsWriter.java

@@ -27,7 +27,7 @@
 import org.apache.lucene.index.DocsWithFieldSet;
 import org.apache.lucene.index.FieldInfo;
 import org.apache.lucene.index.IndexFileNames;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
+import org.apache.lucene.index.RandomAccessVectorValues;
 import org.apache.lucene.index.SegmentWriteState;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.index.VectorValues;
@@ -233,7 +233,7 @@ private void writeMeta(
   }
 
   private Lucene91OnHeapHnswGraph writeGraph(
-      RandomAccessVectorValuesProducer vectorValues, VectorSimilarityFunction similarityFunction)
+      RandomAccessVectorValues vectorValues, VectorSimilarityFunction similarityFunction)
       throws IOException {
 
     // build graph
@@ -245,7 +245,7 @@ private Lucene91OnHeapHnswGraph writeGraph(
             beamWidth,
             Lucene91HnswGraphBuilder.randSeed);
     hnswGraphBuilder.setInfoStream(segmentWriteState.infoStream);
-    Lucene91OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.randomAccess());
+    Lucene91OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.copy());
 
     // write vectors' neighbours on each level into the vectorIndex file
     int countOnLevel0 = graph.size();

lucene/backward-codecs/src/test/org/apache/lucene/backward_codecs/lucene92/Lucene92HnswVectorsWriter.java

@@ -29,7 +29,7 @@
 import org.apache.lucene.index.DocsWithFieldSet;
 import org.apache.lucene.index.FieldInfo;
 import org.apache.lucene.index.IndexFileNames;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
+import org.apache.lucene.index.RandomAccessVectorValues;
 import org.apache.lucene.index.SegmentWriteState;
 import org.apache.lucene.index.VectorEncoding;
 import org.apache.lucene.index.VectorSimilarityFunction;
@@ -268,7 +268,7 @@ private void writeMeta(
   }
 
   private OnHeapHnswGraph writeGraph(
-      RandomAccessVectorValuesProducer vectorValues,
+      RandomAccessVectorValues vectorValues,
       VectorEncoding vectorEncoding,
       VectorSimilarityFunction similarityFunction)
       throws IOException {
@@ -283,7 +283,7 @@ private OnHeapHnswGraph writeGraph(
             beamWidth,
             HnswGraphBuilder.randSeed);
     hnswGraphBuilder.setInfoStream(segmentWriteState.infoStream);
-    OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.randomAccess());
+    OnHeapHnswGraph graph = hnswGraphBuilder.build(vectorValues.copy());
 
     // write vectors' neighbours on each level into the vectorIndex file
     int countOnLevel0 = graph.size();

lucene/codecs/src/java/org/apache/lucene/codecs/simpletext/SimpleTextKnnVectorsReader.java

@@ -29,7 +29,6 @@
 import org.apache.lucene.index.FieldInfo;
 import org.apache.lucene.index.IndexFileNames;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.SegmentReadState;
 import org.apache.lucene.index.VectorSimilarityFunction;
 import org.apache.lucene.index.VectorValues;
@@ -268,7 +267,7 @@ int size() {
   }
 
   private static class SimpleTextVectorValues extends VectorValues
-      implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+      implements RandomAccessVectorValues {
 
     private final BytesRefBuilder scratch = new BytesRefBuilder();
     private final FieldEntry entry;
@@ -310,7 +309,7 @@ public BytesRef binaryValue() {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() {
+    public RandomAccessVectorValues copy() {
       return this;
     }
 

lucene/core/src/java/org/apache/lucene/codecs/lucene94/Lucene94HnswVectorsWriter.java

@@ -422,7 +422,7 @@ public void mergeOneField(FieldInfo fieldInfo, MergeState mergeState) throws IOE
                 beamWidth,
                 HnswGraphBuilder.randSeed);
         hnswGraphBuilder.setInfoStream(segmentWriteState.infoStream);
-        graph = hnswGraphBuilder.build(offHeapVectors.randomAccess());
+        graph = hnswGraphBuilder.build(offHeapVectors.copy());
         writeGraph(graph);
       }
       long vectorIndexLength = vectorIndex.getFilePointer() - vectorIndexOffset;
@@ -617,7 +617,7 @@ public float[] copyValue(float[] value) {
       hnswGraphBuilder =
           (HnswGraphBuilder<T>)
               HnswGraphBuilder.create(
-                  () -> raVectorValues,
+                  raVectorValues,
                   fieldInfo.getVectorEncoding(),
                   fieldInfo.getVectorSimilarityFunction(),
                   M,
@@ -694,5 +694,10 @@ public float[] vectorValue(int targetOrd) throws IOException {
     public BytesRef binaryValue(int targetOrd) throws IOException {
       return (BytesRef) vectors.get(targetOrd);
     }
+
+    @Override
+    public RandomAccessVectorValues copy() throws IOException {
+      return this;
+    }
   }
 }

lucene/core/src/java/org/apache/lucene/codecs/lucene94/OffHeapVectorValues.java

@@ -21,7 +21,6 @@
 import java.nio.ByteBuffer;
 import org.apache.lucene.codecs.lucene90.IndexedDISI;
 import org.apache.lucene.index.RandomAccessVectorValues;
-import org.apache.lucene.index.RandomAccessVectorValuesProducer;
 import org.apache.lucene.index.VectorValues;
 import org.apache.lucene.store.IndexInput;
 import org.apache.lucene.store.RandomAccessInput;
@@ -30,8 +29,7 @@
 import org.apache.lucene.util.packed.DirectMonotonicReader;
 
 /** Read the vector values from the index input. This supports both iterated and random access. */
-abstract class OffHeapVectorValues extends VectorValues
-    implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+abstract class OffHeapVectorValues extends VectorValues implements RandomAccessVectorValues {
 
   protected final int dimension;
   protected final int size;
@@ -150,7 +148,7 @@ public int advance(int target) throws IOException {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       return new DenseOffHeapVectorValues(dimension, size, slice.clone(), byteSize);
     }
 
@@ -226,7 +224,7 @@ public int advance(int target) throws IOException {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       return new SparseOffHeapVectorValues(fieldEntry, dataIn, slice.clone(), byteSize);
     }
 
@@ -303,7 +301,7 @@ public long cost() {
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() throws IOException {
+    public RandomAccessVectorValues copy() throws IOException {
       throw new UnsupportedOperationException();
     }
 

lucene/core/src/java/org/apache/lucene/index/BufferingKnnVectorsWriter.java

@@ -208,7 +208,7 @@ public long ramBytesUsed() {
   }
 
   private static class BufferedVectorValues extends VectorValues
-      implements RandomAccessVectorValues, RandomAccessVectorValuesProducer {
+      implements RandomAccessVectorValues {
 
     final DocsWithFieldSet docsWithField;
 
@@ -236,7 +236,7 @@ private static class BufferedVectorValues extends VectorValues
     }
 
     @Override
-    public RandomAccessVectorValues randomAccess() {
+    public RandomAccessVectorValues copy() {
       return new BufferedVectorValues(docsWithField, vectors, dimension);
     }
 

lucene/core/src/java/org/apache/lucene/index/RandomAccessVectorValues.java

@@ -49,4 +49,11 @@ public interface RandomAccessVectorValues {
    * @param targetOrd a valid ordinal, ≥ 0 and < {@link #size()}.
    */
   BytesRef binaryValue(int targetOrd) throws IOException;
+
+  /**
+   * Creates a new copy of this {@link RandomAccessVectorValues}. This is helpful when you need to
+   * access different values at once, to avoid overwriting the underlying float vector returned by
+   * {@link RandomAccessVectorValues#vectorValue}.
+   */
+  RandomAccessVectorValues copy() throws IOException;
 }