v6: Create vector index Vectorizer-first style

src/it/java/io/weaviate/integration/AggregationITest.java

@@ -14,15 +14,14 @@
 import io.weaviate.ConcurrentTest;
 import io.weaviate.client6.v1.api.WeaviateClient;
 import io.weaviate.client6.v1.api.collections.Property;
+import io.weaviate.client6.v1.api.collections.Vectorizers;
 import io.weaviate.client6.v1.api.collections.Vectors;
 import io.weaviate.client6.v1.api.collections.aggregate.AggregateResponseGroup;
 import io.weaviate.client6.v1.api.collections.aggregate.AggregateResponseGrouped;
 import io.weaviate.client6.v1.api.collections.aggregate.Aggregation;
 import io.weaviate.client6.v1.api.collections.aggregate.GroupBy;
 import io.weaviate.client6.v1.api.collections.aggregate.GroupedBy;
 import io.weaviate.client6.v1.api.collections.aggregate.IntegerAggregation;
-import io.weaviate.client6.v1.api.collections.vectorindex.Hnsw;
-import io.weaviate.client6.v1.api.collections.vectorizers.NoneVectorizer;
 import io.weaviate.containers.Container;
 
 public class AggregationITest extends ConcurrentTest {
@@ -36,7 +35,7 @@ public static void beforeAll() throws IOException {
             .properties(
                 Property.text("category"),
                 Property.integer("price"))
-            .vector(Hnsw.of(new NoneVectorizer())));
+            .vectors(Vectorizers.none()));
 
     var things = client.collections.use(COLLECTION);
     for (var category : List.of("Shoes", "Hat", "Jacket")) {

src/it/java/io/weaviate/integration/CollectionsITest.java

@@ -12,7 +12,8 @@
 import io.weaviate.client6.v1.api.collections.InvertedIndex;
 import io.weaviate.client6.v1.api.collections.Property;
 import io.weaviate.client6.v1.api.collections.Replication;
-import io.weaviate.client6.v1.api.collections.VectorIndex;
+import io.weaviate.client6.v1.api.collections.Vectorizer;
+import io.weaviate.client6.v1.api.collections.Vectorizers;
 import io.weaviate.client6.v1.api.collections.config.Shard;
 import io.weaviate.client6.v1.api.collections.config.ShardStatus;
 import io.weaviate.client6.v1.api.collections.vectorindex.Hnsw;
@@ -28,18 +29,18 @@ public void testCreateGetDelete() throws IOException {
     client.collections.create(collectionName,
         col -> col
             .properties(Property.text("username"), Property.integer("age"))
-            .vector(Hnsw.of(new NoneVectorizer())));
+            .vectors(Vectorizers.none()));
 
     var thingsCollection = client.collections.getConfig(collectionName);
 
     Assertions.assertThat(thingsCollection).get()
         .hasFieldOrPropertyWithValue("collectionName", collectionName)
-        .extracting(CollectionConfig::vectors, InstanceOfAssertFactories.map(String.class, VectorIndex.class))
+        .extracting(CollectionConfig::vectors, InstanceOfAssertFactories.map(String.class, Vectorizer.class))
         .as("default vector").extractingByKey("default")
         .satisfies(defaultVector -> {
-          Assertions.assertThat(defaultVector).extracting(VectorIndex::vectorizer)
+          Assertions.assertThat(defaultVector)
               .as("has none vectorizer").isInstanceOf(NoneVectorizer.class);
-          Assertions.assertThat(defaultVector).extracting(VectorIndex::config)
+          Assertions.assertThat(defaultVector).extracting(Vectorizer::vectorIndex)
               .isInstanceOf(Hnsw.class);
         });
 

src/it/java/io/weaviate/integration/DataITest.java

@@ -11,6 +11,7 @@
 import io.weaviate.ConcurrentTest;
 import io.weaviate.client6.v1.api.WeaviateClient;
 import io.weaviate.client6.v1.api.collections.Property;
+import io.weaviate.client6.v1.api.collections.Vectorizers;
 import io.weaviate.client6.v1.api.collections.Vectors;
 import io.weaviate.client6.v1.api.collections.WeaviateObject;
 import io.weaviate.client6.v1.api.collections.data.BatchReference;
@@ -20,8 +21,6 @@
 import io.weaviate.client6.v1.api.collections.query.QueryMetadata;
 import io.weaviate.client6.v1.api.collections.query.QueryReference;
 import io.weaviate.client6.v1.api.collections.query.Where;
-import io.weaviate.client6.v1.api.collections.vectorindex.Hnsw;
-import io.weaviate.client6.v1.api.collections.vectorizers.NoneVectorizer;
 import io.weaviate.containers.Container;
 
 public class DataITest extends ConcurrentTest {
@@ -107,7 +106,7 @@ private static void createTestCollections() throws IOException {
                 Property.integer("age"))
             .references(
                 Property.reference("hasAwards", awardsGrammy, awardsOscar))
-            .vectors(named -> named.vector(VECTOR_INDEX, Hnsw.of(new NoneVectorizer()))));
+            .vectors(Vectorizers.none(VECTOR_INDEX)));
   }
 
   @Test
@@ -223,7 +222,7 @@ public void testUpdate() throws IOException {
         collection -> collection
             .properties(Property.text("title"), Property.integer("year"))
             .references(Property.reference("writtenBy", nsAuthors))
-            .vector(Hnsw.of(new NoneVectorizer())));
+            .vectors(Vectorizers.none()));
 
     var authors = client.collections.use(nsAuthors);
     var walter = authors.data.insert(Map.of("name", "walter scott"));

src/it/java/io/weaviate/integration/SearchITest.java

@@ -18,6 +18,7 @@
 import io.weaviate.ConcurrentTest;
 import io.weaviate.client6.v1.api.WeaviateClient;
 import io.weaviate.client6.v1.api.collections.Property;
+import io.weaviate.client6.v1.api.collections.Vectorizers;
 import io.weaviate.client6.v1.api.collections.Vectors;
 import io.weaviate.client6.v1.api.collections.WeaviateMetadata;
 import io.weaviate.client6.v1.api.collections.WeaviateObject;
@@ -27,10 +28,6 @@
 import io.weaviate.client6.v1.api.collections.query.QueryMetadata;
 import io.weaviate.client6.v1.api.collections.query.QueryResponseGroup;
 import io.weaviate.client6.v1.api.collections.query.Where;
-import io.weaviate.client6.v1.api.collections.vectorindex.Hnsw;
-import io.weaviate.client6.v1.api.collections.vectorizers.Img2VecNeuralVectorizer;
-import io.weaviate.client6.v1.api.collections.vectorizers.NoneVectorizer;
-import io.weaviate.client6.v1.api.collections.vectorizers.Text2VecContextionaryVectorizer;
 import io.weaviate.containers.Container;
 import io.weaviate.containers.Container.ContainerGroup;
 import io.weaviate.containers.Contextionary;
@@ -133,7 +130,7 @@ private static Map<String, Float[]> populateTest(int n) throws IOException {
   private static void createTestCollection() throws IOException {
     client.collections.create(COLLECTION, cfg -> cfg
         .properties(Property.text("category"))
-        .vector(VECTOR_INDEX, Hnsw.of(new NoneVectorizer())));
+        .vectors(Vectorizers.none(VECTOR_INDEX)));
   }
 
   @Test
@@ -142,7 +139,7 @@ public void testNearText() throws IOException {
     client.collections.create(nsSongs,
         col -> col
             .properties(Property.text("title"))
-            .vector(Hnsw.of(Text2VecContextionaryVectorizer.of())));
+            .vectors(Vectorizers.text2vecContextionary()));
 
     var songs = client.collections.use(nsSongs);
     var submarine = songs.data.insert(Map.of("title", "Yellow Submarine"));
@@ -164,13 +161,13 @@ public void testNearText() throws IOException {
 
   @Test
   public void testNearText_groupBy() throws IOException {
-    var vectorIndex = Hnsw.of(Text2VecContextionaryVectorizer.of());
+    var vectorizer = Vectorizers.text2vecContextionary();
 
     var nsArtists = ns("Artists");
     client.collections.create(nsArtists,
         col -> col
             .properties(Property.text("name"))
-            .vector(vectorIndex));
+            .vectors(vectorizer));
 
     var artists = client.collections.use(nsArtists);
     var beatles = artists.data.insert(Map.of("name", "Beatles"));
@@ -181,7 +178,7 @@ public void testNearText_groupBy() throws IOException {
         col -> col
             .properties(Property.text("title"))
             .references(Property.reference("performedBy", nsArtists))
-            .vector(vectorIndex));
+            .vectors(vectorizer));
 
     var songs = client.collections.use(nsSongs);
     songs.data.insert(Map.of("title", "Yellow Submarine"),
@@ -208,9 +205,8 @@ public void testNearImage() throws IOException {
             .properties(
                 Property.text("breed"),
                 Property.blob("img"))
-            .vector(Hnsw.of(
-                Img2VecNeuralVectorizer.of(
-                    i2v -> i2v.imageFields("img")))));
+            .vectors(Vectorizers.img2vecNeural(
+                i2v -> i2v.imageFields("img"))));
 
     var cats = client.collections.use(nsCats);
     cats.data.insert(Map.of(
@@ -325,7 +321,7 @@ public void testNearObject() throws IOException {
     client.collections.create(nsAnimals,
         collection -> collection
             .properties(Property.text("kind"))
-            .vector(Hnsw.of(Text2VecContextionaryVectorizer.of())));
+            .vectors(Vectorizers.text2vecContextionary()));
 
     var animals = client.collections.use(nsAnimals);
 
@@ -354,7 +350,7 @@ public void testHybrid() throws IOException {
     client.collections.create(nsHobbies,
         collection -> collection
             .properties(Property.text("name"), Property.text("description"))
-            .vector(Hnsw.of(Text2VecContextionaryVectorizer.of())));
+            .vectors(Vectorizers.text2vecContextionary()));
 
     var hobbies = client.collections.use(nsHobbies);
 

src/main/java/io/weaviate/client6/v1/api/collections/CollectionConfig.java

@@ -27,7 +27,7 @@ public record CollectionConfig(
     @SerializedName("description") String description,
     @SerializedName("properties") List<Property> properties,
     List<ReferenceProperty> references,
-    @SerializedName("vectorConfig") Map<String, VectorIndex> vectors,
+    @SerializedName("vectorConfig") Map<String, Vectorizer> vectors,
     @SerializedName("multiTenancyConfig") MultiTenancy multiTenancy,
     @SerializedName("shardingConfig") Sharding sharding,
     @SerializedName("replicationConfig") Replication replication,
@@ -88,7 +88,7 @@ public static class Builder implements ObjectBuilder<CollectionConfig> {
     private String description;
     private Map<String, Property> properties = new HashMap<>();
     private Map<String, ReferenceProperty> references = new HashMap<>();
-    private Map<String, VectorIndex> vectors = new HashMap<>();
+    private Map<String, Vectorizer> vectors = new HashMap<>();
     private MultiTenancy multiTenancy;
     private Sharding sharding;
     private Replication replication;
@@ -131,23 +131,14 @@ private List<ReferenceProperty> referenceList() {
       return this.references.values().stream().toList();
     }
 
-    public Builder vector(VectorIndex vector) {
-      this.vectors.put(VectorIndex.DEFAULT_VECTOR_NAME, vector);
-      return this;
-    }
-
-    public Builder vector(String name, VectorIndex vector) {
-      this.vectors.put(name, vector);
-      return this;
-    }
-
-    public Builder vectors(Map<String, VectorIndex> vectors) {
+    public final Builder vectors(Map<String, Vectorizer> vectors) {
       this.vectors.putAll(vectors);
       return this;
     }
 
-    public Builder vectors(Function<VectorsBuilder, ObjectBuilder<Map<String, VectorIndex>>> fn) {
-      this.vectors = fn.apply(new VectorsBuilder()).build();
+    @SafeVarargs
+    public final Builder vectors(Map.Entry<String, Vectorizer>... vectors) {
+      this.vectors.putAll(Map.ofEntries(vectors));
       return this;
     }
 

src/main/java/io/weaviate/client6/v1/api/collections/VectorIndex.java

@@ -19,6 +19,7 @@
 
 public interface VectorIndex {
   static final String DEFAULT_VECTOR_NAME = "default";
+  static final VectorIndex DEFAULT_VECTOR_INDEX = Hnsw.of();
 
   public enum Kind implements JsonEnum<Kind> {
     HNSW("hnsw"),
@@ -48,8 +49,6 @@ default String type() {
     return _kind().jsonValue();
   }
 
-  Vectorizer vectorizer();
-
   Object config();
 
   public static enum CustomTypeAdapterFactory implements TypeAdapterFactory {
@@ -79,7 +78,6 @@ public <T> TypeAdapter<T> create(Gson gson, TypeToken<T> type) {
         init(gson);
       }
 
-      final var vectorizerAdapter = gson.getDelegateAdapter(this, TypeToken.get(Vectorizer.class));
       final var writeAdapter = gson.getDelegateAdapter(this, TypeToken.get(rawType));
       return (TypeAdapter<T>) new TypeAdapter<VectorIndex>() {
 
@@ -89,13 +87,11 @@ public void write(JsonWriter out, VectorIndex value) throws IOException {
           out.name("vectorIndexType");
           out.value(value._kind().jsonValue());
 
-          var config = writeAdapter.toJsonTree((T) value.config());
-          config.getAsJsonObject().remove("vectorizer");
           out.name("vectorIndexConfig");
+          var config = writeAdapter.toJsonTree((T) value.config());
+          config.getAsJsonObject().remove("name");
           Streams.write(config, out);
 
-          out.name("vectorizer");
-          vectorizerAdapter.write(out, value.vectorizer());
           out.endObject();
         }
 
@@ -117,7 +113,6 @@ public VectorIndex read(JsonReader in) throws IOException {
           }
 
           var config = jsonObject.get("vectorIndexConfig").getAsJsonObject();
-          config.add("vectorizer", jsonObject.get("vectorizer"));
           return adapter.fromJsonTree(config);
         }
       }.nullSafe();

src/main/java/io/weaviate/client6/v1/api/collections/Vectorizer.java

@@ -3,20 +3,24 @@
 import java.io.IOException;
 import java.util.EnumMap;
 import java.util.Map;
+import java.util.function.Function;
 
 import com.google.gson.Gson;
+import com.google.gson.JsonObject;
+import com.google.gson.JsonParser;
 import com.google.gson.TypeAdapter;
 import com.google.gson.TypeAdapterFactory;
+import com.google.gson.internal.Streams;
 import com.google.gson.reflect.TypeToken;
 import com.google.gson.stream.JsonReader;
-import com.google.gson.stream.JsonToken;
 import com.google.gson.stream.JsonWriter;
 
 import io.weaviate.client6.v1.api.collections.vectorizers.Img2VecNeuralVectorizer;
 import io.weaviate.client6.v1.api.collections.vectorizers.Multi2VecClipVectorizer;
 import io.weaviate.client6.v1.api.collections.vectorizers.NoneVectorizer;
 import io.weaviate.client6.v1.api.collections.vectorizers.Text2VecContextionaryVectorizer;
 import io.weaviate.client6.v1.api.collections.vectorizers.Text2VecWeaviateVectorizer;
+import io.weaviate.client6.v1.internal.ObjectBuilder;
 import io.weaviate.client6.v1.internal.json.JsonEnum;
 
 public interface Vectorizer {
@@ -48,6 +52,8 @@ public static Kind valueOfJson(String jsonValue) {
 
   Object _self();
 
+  VectorIndex vectorIndex();
+
   public static enum CustomTypeAdapterFactory implements TypeAdapterFactory {
     INSTANCE;
 
@@ -84,28 +90,48 @@ public <T> TypeAdapter<T> create(Gson gson, TypeToken<T> type) {
         public void write(JsonWriter out, Vectorizer value) throws IOException {
           TypeAdapter<T> adapter = (TypeAdapter<T>) delegateAdapters.get(value._kind());
 
-          out.beginObject();
-          out.name(value._kind().jsonValue());
-          adapter.write(out, (T) value._self());
-          out.endObject();
+          // Serialize vectorizer config as { "vectorizer-kind": { ... } }
+          // and remove "vectorIndex" object which every vectorizer has.
+          var vectorizer = new JsonObject();
+          var config = adapter.toJsonTree((T) value._self());
+
+          // This will create { "vectorIndexType": "", "vectorIndexConfig": { ... } }
+          // to which we just need to add "vectorizer": { ... } key.
+          var vectorIndex = config.getAsJsonObject().remove("vectorIndex");
+
+          vectorizer.add(value._kind().jsonValue(), config);
+          vectorIndex.getAsJsonObject().add("vectorizer", vectorizer);
+
+          Streams.write(vectorIndex, out);
         }
 
         @Override
         public Vectorizer read(JsonReader in) throws IOException {
-          in.beginObject();
-          var vectorizerName = in.nextName();
+          var jsonObject = JsonParser.parseReader(in).getAsJsonObject();
+
+          // VectorIndex.CustomTypeAdapterFactory expects keys
+          // ["vectorIndexType", "vectorIndexConfig"].
+          var vectorIndex = new JsonObject();
+          vectorIndex.add("vectorIndexType", jsonObject.get("vectorIndexType"));
+          vectorIndex.add("vectorIndexConfig", jsonObject.get("vectorIndexConfig"));
+
+          var vectorizerObject = jsonObject.get("vectorizer").getAsJsonObject();
+          var vectorizerName = vectorizerObject.keySet().iterator().next();
+
+          Vectorizer.Kind kind;
           try {
-            var kind = Vectorizer.Kind.valueOfJson(vectorizerName);
-            var adapter = delegateAdapters.get(kind);
-            return adapter.read(in);
+            kind = Vectorizer.Kind.valueOfJson(vectorizerName);
           } catch (IllegalArgumentException e) {
             return null;
-          } finally {
-            if (in.peek() == JsonToken.BEGIN_OBJECT) {
-              in.beginObject();
-            }
-            in.endObject();
           }
+
+          var adapter = delegateAdapters.get(kind);
+          var concreteVectorizer = vectorizerObject.get(vectorizerName).getAsJsonObject();
+
+          // Each individual vectorizer has a `VectorIndex vectorIndex` field.
+          concreteVectorizer.add("vectorIndex", vectorIndex);
+
+          return adapter.fromJsonTree(concreteVectorizer);
         }
       }.nullSafe();
     }