feat: add sourceProperties to text-based vectorizers

Author: bevzzz

src/main/java/io/weaviate/client6/v1/api/collections/vectorizers/Text2VecContextionaryVectorizer.java

@@ -1,5 +1,9 @@
 package io.weaviate.client6.v1.api.collections.vectorizers;
 
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
 import java.util.function.Function;
 
 import com.google.gson.annotations.SerializedName;
@@ -11,13 +15,16 @@
 public record Text2VecContextionaryVectorizer(
     /**
      * Weaviate defaults to {@code true} if the value is not provided.
-     * Because text2vec-contextionary cannot handle understores in collection names,
+     * Because text2vec-contextionary cannot handle underscores in collection names,
      * this quickly becomes inconvenient.
      *
      * To avoid that we send "vectorizeClassName": false all the time
      * and make it impossible to enable this feature, as it is deprecated.
      */
     @Deprecated @SerializedName("vectorizeClassName") boolean vectorizeCollectionName,
+    /** Properties included in the embedding. */
+    @SerializedName("sourceProperties") List<String> sourceProperties,
+    /** Vector index configuration. */
     VectorIndex vectorIndex) implements Vectorizer {
 
   @Override
@@ -42,20 +49,41 @@ public static Text2VecContextionaryVectorizer of(
   /**
    * Canonical constructor always sets {@link #vectorizeCollectionName} to false.
    */
-  public Text2VecContextionaryVectorizer(boolean vectorizeCollectionName, VectorIndex vectorIndex) {
+  public Text2VecContextionaryVectorizer(boolean vectorizeCollectionName, List<String> sourceProperties,
+      VectorIndex vectorIndex) {
     this.vectorizeCollectionName = false;
     this.vectorIndex = vectorIndex;
+    this.sourceProperties = Collections.emptyList();
   }
 
   public Text2VecContextionaryVectorizer(Builder builder) {
-    this(builder.vectorizeCollectionName, builder.vectorIndex);
+    this(builder.vectorizeCollectionName, builder.sourceProperties, builder.vectorIndex);
   }
 
   public static class Builder implements ObjectBuilder<Text2VecContextionaryVectorizer> {
     private final boolean vectorizeCollectionName = false;
 
+    private List<String> sourceProperties = new ArrayList<>();
     private VectorIndex vectorIndex = VectorIndex.DEFAULT_VECTOR_INDEX;
 
+    /** Add properties to include in the embedding. */
+    public Builder sourceProperties(String... properties) {
+      return sourceProperties(Arrays.asList(properties));
+    }
+
+    /** Add properties to include in the embedding. */
+    public Builder sourceProperties(List<String> properties) {
+      this.sourceProperties.addAll(properties);
+      return this;
+    }
+
+    /**
+     * Override default vector index configuration.
+     *
+     * <a href=
+     * "https://docs.weaviate.io/weaviate/config-refs/indexing/vector-index#hnsw-index-parameters">HNSW</a>
+     * is the default vector index.
+     */
     public Builder vectorIndex(VectorIndex vectorIndex) {
       this.vectorIndex = vectorIndex;
       return this;

src/main/java/io/weaviate/client6/v1/api/collections/vectorizers/Text2VecWeaviateVectorizer.java

@@ -1,5 +1,8 @@
 package io.weaviate.client6.v1.api.collections.vectorizers;
 
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
 import java.util.function.Function;
 
 import com.google.gson.annotations.SerializedName;
@@ -9,9 +12,15 @@
 import io.weaviate.client6.v1.internal.ObjectBuilder;
 
 public record Text2VecWeaviateVectorizer(
+    /** Weaviate Embeddings Service base URL. */
     @SerializedName("baseUrl") String inferenceUrl,
+    /** Dimensionality of the generated vectors. */
     @SerializedName("dimensions") Integer dimensions,
+    /** Embedding model. */
     @SerializedName("model") String model,
+    /** Properties included in the embedding. */
+    @SerializedName("sourceProperties") List<String> sourceProperties,
+    /** Vector index configuration. */
     VectorIndex vectorIndex) implements Vectorizer {
 
   @Override
@@ -37,33 +46,65 @@ public Text2VecWeaviateVectorizer(Builder builder) {
         builder.inferenceUrl,
         builder.dimensions,
         builder.model,
+        builder.sourceProperties,
         builder.vectorIndex);
   }
 
-  public static final String SNOWFLAKE_ARCTIC_EMBED_L_20 = "Snowflake/snowflake-arctic-embed-l-v2.0";
   public static final String SNOWFLAKE_ARCTIC_EMBED_M_15 = "Snowflake/snowflake-arctic-embed-m-v1.5";
+  public static final String SNOWFLAKE_ARCTIC_EMBED_L_20 = "Snowflake/snowflake-arctic-embed-l-v2.0";
 
   public static class Builder implements ObjectBuilder<Text2VecWeaviateVectorizer> {
     private VectorIndex vectorIndex = VectorIndex.DEFAULT_VECTOR_INDEX;
     private String inferenceUrl;
     private Integer dimensions;
     private String model;
+    private List<String> sourceProperties = new ArrayList<>();
 
+    /**
+     * Base URL for Weaviate Embeddings Service. This can be omitted when connecting
+     * to a Weaviate Cloud instance: the client will automatically set the necessary
+     * headers.
+     */
     public Builder inferenceUrl(String inferenceUrl) {
       this.inferenceUrl = inferenceUrl;
       return this;
     }
 
+    /** Set target dimensionality for generated embeddings. */
     public Builder dimensions(int dimensions) {
       this.dimensions = dimensions;
       return this;
     }
 
+    /**
+     * Select the embedding model.
+     *
+     * @see {@link Text2VecWeaviateVectorizer#SNOWFLAKE_ARCTIC_EMBED_M_15},
+     *      {@link Text2VecWeaviateVectorizer#SNOWFLAKE_ARCTIC_EMBED_L_20}
+     */
     public Builder model(String model) {
       this.model = model;
       return this;
     }
 
+    /** Add properties to include in the embedding. */
+    public Builder sourceProperties(String... properties) {
+      return sourceProperties(Arrays.asList(properties));
+    }
+
+    /** Add properties to include in the embedding. */
+    public Builder sourceProperties(List<String> properties) {
+      this.sourceProperties.addAll(properties);
+      return this;
+    }
+
+    /**
+     * Override default vector index configuration.
+     *
+     * <a href=
+     * "https://docs.weaviate.io/weaviate/config-refs/indexing/vector-index#hnsw-index-parameters">HNSW</a>
+     * is the default vector index.
+     */
     public Builder vectorIndex(VectorIndex vectorIndex) {
       this.vectorIndex = vectorIndex;
       return this;

src/test/java/io/weaviate/client6/v1/internal/json/JSONTest.java

@@ -101,7 +101,8 @@ public static Object[][] testCases() {
                   "vectorIndexConfig": {},
                   "vectorizer": {
                     "text2vec-contextionary": {
-                      "vectorizeClassName": false
+                      "vectorizeClassName": false,
+                      "sourceProperties": []
                     }
                   }
                 }
@@ -121,7 +122,9 @@ public static Object[][] testCases() {
                     "text2vec-weaviate": {
                       "baseUrl": "http://example.com",
                       "dimensions": 4,
-                      "model": "very-good-model"
+                      "model": "very-good-model",
+                      "sourceProperties": []
+
                     }
                   }
                 }