chromadb 3.4.2 → 3.5.0

package/dist/chromadb.d.ts

@@ -55,8 +55,21 @@ type FloatListValueType$1 = {
 type FloatValueType$1 = {
     float_inverted_index?: null | FloatInvertedIndexType$1;
 };
+/**
+ * Full-text search index algorithm.
+ *
+ * Controls which index format and query pipeline are used for
+ * document substring search within a collection.
+ */
+type FtsAlgorithm = 'trigram' | 'token_bitmap';
 type FtsIndexConfig$1 = {
-    [key: string]: never;
+    /**
+     * FTS index algorithm.
+     * Omitted from JSON when set to the default (Trigram) so that old
+     * servers/clients that do not know about this field can still
+     * deserialize the schema.
+     */
+    algorithm?: FtsAlgorithm;
 };
 type FtsIndexType$1 = {
     config: FtsIndexConfig$1;
@@ -209,10 +222,6 @@ type Schema$1 = {
     keys: {
         [key: string]: ValueTypes$1;
     };
-    /**
-     * ID of the attached function that created this output collection (if applicable)
-     */
-    source_attached_function_id?: string | null;
 };
 type SearchPayload = {
     filter?: {
@@ -284,6 +293,13 @@ type SpannIndexConfig = {
     write_rng_epsilon?: number | null;
     write_rng_factor?: number | null;
 };
+/**
+ * Sparse vector index algorithm.
+ *
+ * Controls which posting list format and query engine are used for
+ * sparse vector search within a collection.
+ */
+type SparseIndexAlgorithm = 'wand' | 'max_score';
 /**
  * Represents a sparse vector using parallel arrays for indices and values.
  *
@@ -307,6 +323,13 @@ type SparseVector = {
     values: Array<number>;
 };
 type SparseVectorIndexConfig$1 = {
+    /**
+     * Sparse index algorithm (cloud-only, tenant-gated).
+     * Omitted from JSON when set to the default (Wand) so that old
+     * servers/clients that do not know about this field can still
+     * deserialize the schema.
+     */
+    algorithm?: SparseIndexAlgorithm;
     /**
      * Whether this embedding is BM25
      */
@@ -396,10 +419,13 @@ type UserIdentity = GetUserIdentityResponse;
  *   All committed writes will be visible. This is the default.
  * - INDEX_ONLY: Read only from the compacted index, skipping the WAL.
  *   Recent writes that haven't been compacted may not be visible, but queries are faster.
+ * - INDEX_AND_BOUNDED_WAL: Read from the index and up to a server-configured number of
+ *   WAL entries for bounded query latency.
  */
 declare const ReadLevel: {
     readonly INDEX_AND_WAL: "index_and_wal";
     readonly INDEX_ONLY: "index_only";
+    readonly INDEX_AND_BOUNDED_WAL: "index_and_bounded_wal";
 };
 type ReadLevel = (typeof ReadLevel)[keyof typeof ReadLevel];
 
@@ -1290,7 +1316,6 @@ declare class Schema {
     private setIndexForKey;
     private enableAllIndexesForKey;
     private disableAllIndexesForKey;
-    private validateSingleSparseVectorIndex;
     private validateSparseVectorConfig;
     private initializeDefaults;
     private initializeKeys;
@@ -1492,6 +1517,13 @@ declare class ChromaClient {
      * @throws Error if the collection does not exist
      */
     getCollectionByCrn(crn: string): Promise<Collection>;
+    /**
+     * Retrieves an existing collection by its ID.
+     * @param id - The UUID of the collection to retrieve
+     * @returns Promise resolving to the Collection instance
+     * @throws Error if the collection does not exist
+     */
+    getCollectionById(id: string): Promise<Collection>;
     /**
      * Retrieves multiple collections by name.
      * @param items - Array of collection names or objects with name and optional embedding function (should match the ones used to create the collections)
@@ -1594,6 +1626,7 @@ interface Collection {
          * Controls whether to read from the write-ahead log.
          * - ReadLevel.INDEX_AND_WAL: Read from both index and WAL (default)
          * - ReadLevel.INDEX_ONLY: Read only from index, faster but recent writes may not be visible
+         * - ReadLevel.INDEX_AND_BOUNDED_WAL: Read up to a server-configured number of WAL entries
          */
         readLevel?: ReadLevel;
     }): Promise<number>;
@@ -1747,6 +1780,7 @@ interface Collection {
          * Controls whether to read from the write-ahead log.
          * - ReadLevel.INDEX_AND_WAL: Read from both index and WAL (default)
          * - ReadLevel.INDEX_ONLY: Read only from index, faster but recent writes may not be visible
+         * - ReadLevel.INDEX_AND_BOUNDED_WAL: Read up to a server-configured number of WAL entries
          */
         readLevel?: ReadLevel;
     }): Promise<SearchResult>;

package/dist/chromadb.legacy-esm.js

@@ -26,7 +26,8 @@ if (typeof globalThis.Deno !== "undefined") {
 // src/types.ts
 var ReadLevel = {
   INDEX_AND_WAL: "index_and_wal",
-  INDEX_ONLY: "index_only"
+  INDEX_ONLY: "index_only",
+  INDEX_AND_BOUNDED_WAL: "index_and_bounded_wal"
 };
 var baseRecordSetFields = [
   "ids",
@@ -436,6 +437,22 @@ var CollectionService = class {
       }
     });
   }
+  /**
+   * Get collection by ID
+   * Returns a collection by its UUID within a specific tenant and database.
+   */
+  static getCollectionById(options) {
+    return (options.client ?? client).get({
+      security: [
+        {
+          name: "x-chroma-token",
+          type: "apiKey"
+        }
+      ],
+      url: "/api/v2/tenants/{tenant}/databases/{database}/collections/by-id/{collection_id}",
+      ...options
+    });
+  }
   /**
    * Delete collection
    * Deletes a collection in a database.
@@ -3301,9 +3318,7 @@ var Schema = class _Schema {
       );
     }
     if (keyProvided && key && key === EMBEDDING_KEY) {
-      throw new Error(
-        "Cannot modify #embedding. Currently not supported"
-      );
+      throw new Error("Cannot modify #embedding. Currently not supported");
     }
     if (keyProvided && key === DOCUMENT_KEY && !(config instanceof FtsIndexConfig)) {
       throw new Error(
@@ -3441,7 +3456,6 @@ var Schema = class _Schema {
   }
   setIndexForKey(key, config, enabled) {
     if (config instanceof SparseVectorIndexConfig && enabled) {
-      this.validateSingleSparseVectorIndex(key);
       this.validateSparseVectorConfig(config);
     }
     const current = this.keys[key] = ensureValueTypes(this.keys[key]);
@@ -3528,17 +3542,6 @@ var Schema = class _Schema {
       new BoolInvertedIndexType(false, new BoolInvertedIndexConfig())
     );
   }
-  validateSingleSparseVectorIndex(targetKey) {
-    for (const [existingKey, valueTypes] of Object.entries(this.keys)) {
-      if (existingKey === targetKey) continue;
-      const sparseIndex = valueTypes.sparseVector?.sparseVectorIndex;
-      if (sparseIndex?.enabled) {
-        throw new Error(
-          `Cannot enable sparse vector index on key '${targetKey}'. A sparse vector index is already enabled on key '${existingKey}'. Only one sparse vector index is allowed per collection.`
-        );
-      }
-    }
-  }
   validateSparseVectorConfig(config) {
     if (config.sourceKey !== null && config.sourceKey !== void 0 && !config.embeddingFunction) {
       throw new Error(
@@ -5109,6 +5112,36 @@ var ChromaClient = class {
       schema
     });
   }
+  /**
+   * Retrieves an existing collection by its ID.
+   * @param id - The UUID of the collection to retrieve
+   * @returns Promise resolving to the Collection instance
+   * @throws Error if the collection does not exist
+   */
+  async getCollectionById(id) {
+    const { data } = await CollectionService.getCollectionById({
+      client: this.apiClient,
+      path: { ...await this._path(), collection_id: id }
+    });
+    const schema = await Schema.deserializeFromJSON(data.schema ?? null, this);
+    const schemaEmbeddingFunction = resolveSchemaEmbeddingFunction(schema);
+    const resolvedEmbeddingFunction = await getEmbeddingFunction({
+      efConfig: data.configuration_json.embedding_function ?? void 0,
+      client: this
+    }) ?? schemaEmbeddingFunction;
+    return new CollectionImpl({
+      chromaClient: this,
+      apiClient: this.apiClient,
+      name: data.name,
+      tenant: data.tenant,
+      database: data.database,
+      configuration: data.configuration_json,
+      metadata: deserializeMetadata(data.metadata ?? void 0) ?? void 0,
+      embeddingFunction: resolvedEmbeddingFunction,
+      id: data.id,
+      schema
+    });
+  }
   /**
    * Retrieves multiple collections by name.
    * @param items - Array of collection names or objects with name and optional embedding function (should match the ones used to create the collections)

package/dist/chromadb.mjs

@@ -26,7 +26,8 @@ if (typeof globalThis.Deno !== "undefined") {
 // src/types.ts
 var ReadLevel = {
   INDEX_AND_WAL: "index_and_wal",
-  INDEX_ONLY: "index_only"
+  INDEX_ONLY: "index_only",
+  INDEX_AND_BOUNDED_WAL: "index_and_bounded_wal"
 };
 var baseRecordSetFields = [
   "ids",
@@ -436,6 +437,22 @@ var CollectionService = class {
       }
     });
   }
+  /**
+   * Get collection by ID
+   * Returns a collection by its UUID within a specific tenant and database.
+   */
+  static getCollectionById(options) {
+    return (options.client ?? client).get({
+      security: [
+        {
+          name: "x-chroma-token",
+          type: "apiKey"
+        }
+      ],
+      url: "/api/v2/tenants/{tenant}/databases/{database}/collections/by-id/{collection_id}",
+      ...options
+    });
+  }
   /**
    * Delete collection
    * Deletes a collection in a database.
@@ -3301,9 +3318,7 @@ var Schema = class _Schema {
       );
     }
     if (keyProvided && key && key === EMBEDDING_KEY) {
-      throw new Error(
-        "Cannot modify #embedding. Currently not supported"
-      );
+      throw new Error("Cannot modify #embedding. Currently not supported");
     }
     if (keyProvided && key === DOCUMENT_KEY && !(config instanceof FtsIndexConfig)) {
       throw new Error(
@@ -3441,7 +3456,6 @@ var Schema = class _Schema {
   }
   setIndexForKey(key, config, enabled) {
     if (config instanceof SparseVectorIndexConfig && enabled) {
-      this.validateSingleSparseVectorIndex(key);
       this.validateSparseVectorConfig(config);
     }
     const current = this.keys[key] = ensureValueTypes(this.keys[key]);
@@ -3528,17 +3542,6 @@ var Schema = class _Schema {
       new BoolInvertedIndexType(false, new BoolInvertedIndexConfig())
     );
   }
-  validateSingleSparseVectorIndex(targetKey) {
-    for (const [existingKey, valueTypes] of Object.entries(this.keys)) {
-      if (existingKey === targetKey) continue;
-      const sparseIndex = valueTypes.sparseVector?.sparseVectorIndex;
-      if (sparseIndex?.enabled) {
-        throw new Error(
-          `Cannot enable sparse vector index on key '${targetKey}'. A sparse vector index is already enabled on key '${existingKey}'. Only one sparse vector index is allowed per collection.`
-        );
-      }
-    }
-  }
   validateSparseVectorConfig(config) {
     if (config.sourceKey !== null && config.sourceKey !== void 0 && !config.embeddingFunction) {
       throw new Error(
@@ -5109,6 +5112,36 @@ var ChromaClient = class {
       schema
     });
   }
+  /**
+   * Retrieves an existing collection by its ID.
+   * @param id - The UUID of the collection to retrieve
+   * @returns Promise resolving to the Collection instance
+   * @throws Error if the collection does not exist
+   */
+  async getCollectionById(id) {
+    const { data } = await CollectionService.getCollectionById({
+      client: this.apiClient,
+      path: { ...await this._path(), collection_id: id }
+    });
+    const schema = await Schema.deserializeFromJSON(data.schema ?? null, this);
+    const schemaEmbeddingFunction = resolveSchemaEmbeddingFunction(schema);
+    const resolvedEmbeddingFunction = await getEmbeddingFunction({
+      efConfig: data.configuration_json.embedding_function ?? void 0,
+      client: this
+    }) ?? schemaEmbeddingFunction;
+    return new CollectionImpl({
+      chromaClient: this,
+      apiClient: this.apiClient,
+      name: data.name,
+      tenant: data.tenant,
+      database: data.database,
+      configuration: data.configuration_json,
+      metadata: deserializeMetadata(data.metadata ?? void 0) ?? void 0,
+      embeddingFunction: resolvedEmbeddingFunction,
+      id: data.id,
+      schema
+    });
+  }
   /**
    * Retrieves multiple collections by name.
    * @param items - Array of collection names or objects with name and optional embedding function (should match the ones used to create the collections)