From d5f8853fda8f488d73aed09d376e1c1c3c9ea85f Mon Sep 17 00:00:00 2001
From: Benjamin Trent <ben.w.trent@gmail.com>
Date: Wed, 20 Mar 2024 11:49:30 -0400
Subject: [PATCH] Fix NPE for when fields are missing in `searchNearestVectors`
 (#13195)

Related to: https://github.com/apache/lucene/pull/13162

Since this is unreleased, no changelog entry is necessary.
---
 .../org/apache/lucene/index/LeafReader.java   |  6 ++++
 .../TestPerFieldKnnVectorsFormat.java         | 34 +++++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/lucene/core/src/java/org/apache/lucene/index/LeafReader.java b/lucene/core/src/java/org/apache/lucene/index/LeafReader.java
index 4856feedca3c..5c3c9f8da5e1 100644
--- a/lucene/core/src/java/org/apache/lucene/index/LeafReader.java
+++ b/lucene/core/src/java/org/apache/lucene/index/LeafReader.java
@@ -246,6 +246,9 @@ public final PostingsEnum postings(Term term) throws IOException {
   public final TopDocs searchNearestVectors(
       String field, float[] target, int k, Bits acceptDocs, int visitedLimit) throws IOException {
     FieldInfo fi = getFieldInfos().fieldInfo(field);
+    if (fi == null || fi.getVectorDimension() == 0) {
+      return TopDocsCollector.EMPTY_TOPDOCS;
+    }
     FloatVectorValues floatVectorValues = getFloatVectorValues(fi.name);
     if (floatVectorValues == null) {
       return TopDocsCollector.EMPTY_TOPDOCS;
@@ -287,6 +290,9 @@ public final TopDocs searchNearestVectors(
   public final TopDocs searchNearestVectors(
       String field, byte[] target, int k, Bits acceptDocs, int visitedLimit) throws IOException {
     FieldInfo fi = getFieldInfos().fieldInfo(field);
+    if (fi == null || fi.getVectorDimension() == 0) {
+      return TopDocsCollector.EMPTY_TOPDOCS;
+    }
     ByteVectorValues byteVectorValues = getByteVectorValues(fi.name);
     if (byteVectorValues == null) {
       return TopDocsCollector.EMPTY_TOPDOCS;
diff --git a/lucene/core/src/test/org/apache/lucene/codecs/perfield/TestPerFieldKnnVectorsFormat.java b/lucene/core/src/test/org/apache/lucene/codecs/perfield/TestPerFieldKnnVectorsFormat.java
index c51f51f5f113..45814144d10a 100644
--- a/lucene/core/src/test/org/apache/lucene/codecs/perfield/TestPerFieldKnnVectorsFormat.java
+++ b/lucene/core/src/test/org/apache/lucene/codecs/perfield/TestPerFieldKnnVectorsFormat.java
@@ -71,6 +71,40 @@ protected Codec getCodec() {
     return codec;
   }
 
+  public void testMissingFieldReturnsNoResults() throws IOException {
+    try (Directory directory = newDirectory()) {
+      IndexWriterConfig iwc = newIndexWriterConfig(new MockAnalyzer(random()));
+      iwc.setCodec(
+          new AssertingCodec() {
+            @Override
+            public KnnVectorsFormat getKnnVectorsFormatForField(String field) {
+              return TestUtil.getDefaultKnnVectorsFormat();
+            }
+          });
+      try (IndexWriter iwriter = new IndexWriter(directory, iwc)) {
+        Document doc = new Document();
+        doc.add(newTextField("id", "1", Field.Store.YES));
+        iwriter.addDocument(doc);
+      }
+
+      try (IndexReader ireader = DirectoryReader.open(directory)) {
+        LeafReader reader = ireader.leaves().get(0).reader();
+        TopDocs hits =
+            reader.searchNearestVectors(
+                "missing_field",
+                new float[] {1, 2, 3},
+                10,
+                reader.getLiveDocs(),
+                Integer.MAX_VALUE);
+        assertEquals(0, hits.scoreDocs.length);
+        hits =
+            reader.searchNearestVectors(
+                "id", new float[] {1, 2, 3}, 10, reader.getLiveDocs(), Integer.MAX_VALUE);
+        assertEquals(0, hits.scoreDocs.length);
+      }
+    }
+  }
+
   public void testTwoFieldsTwoFormats() throws IOException {
     try (Directory directory = newDirectory()) {
       // we don't use RandomIndexWriter because it might add more values than we expect !!!!1