Release 0.6.15 (#1295 )

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
fix: lazy load openai (#1294 )
2026-07-01 22:14:03 -04:00 · 2024-10-03 19:44:55 -07:00 · 2024-10-03 17:12:33 -07:00 · 2024-10-03 17:07:50 -07:00 · 2024-10-03 16:35:54 -07:00 · 2024-10-03 11:52:26 -07:00
54 changed files with 386 additions and 238 deletions
@@ -1,5 +1,18 @@
 # docs

+## 0.0.84
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.83
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.0.82

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "docs",
-  "version": "0.0.82",
+  "version": "0.0.84",
  "private": true,
  "scripts": {
    "docusaurus": "docusaurus",
@@ -13,7 +13,7 @@ import { FunctionTool, OpenAI, ToolCallOptions } from "llamaindex";
  }
 })();

-async function callLLM(init: Partial<OpenAI>) {
+async function callLLM(init: { model: string }) {
  const csvData =
    "Country,Average Height (cm)\nNetherlands,156\nDenmark,158\nNorway,160";

@@ -1,5 +1,18 @@
 # @llamaindex/autotool

+## 3.0.15
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 3.0.14
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 3.0.13

 ### Patch Changes
@@ -1,5 +1,20 @@
 # @llamaindex/autotool-01-node-example

+## 0.0.24
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+  - @llamaindex/autotool@3.0.15
+
+## 0.0.23
+
+### Patch Changes
+
+- llamaindex@0.6.14
+- @llamaindex/autotool@3.0.14
+
 ## 0.0.22

 ### Patch Changes
@@ -13,5 +13,5 @@
  "scripts": {
    "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
  },
-  "version": "0.0.22"
+  "version": "0.0.24"
 }
@@ -1,5 +1,20 @@
 # @llamaindex/autotool-02-next-example

+## 0.1.68
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+  - @llamaindex/autotool@3.0.15
+
+## 0.1.67
+
+### Patch Changes
+
+- llamaindex@0.6.14
+- @llamaindex/autotool@3.0.14
+
 ## 0.1.66

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/autotool-02-next-example",
  "private": true,
-  "version": "0.1.66",
+  "version": "0.1.68",
  "scripts": {
    "dev": "next dev",
    "build": "next build",
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/autotool",
  "type": "module",
-  "version": "3.0.13",
+  "version": "3.0.15",
  "description": "auto transpile your JS function to LLM Agent compatible",
  "files": [
    "dist",
@@ -1,5 +1,11 @@
 # @llamaindex/cloud

+## 0.2.11
+
+### Patch Changes
+
+- 0b20ff9: fix: package.json format
+
 ## 0.2.10

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloud",
-  "version": "0.2.10",
+  "version": "0.2.11",
  "type": "module",
  "license": "MIT",
  "scripts": {
@@ -51,13 +51,13 @@
  "devDependencies": {
    "@hey-api/client-fetch": "^0.2.4",
    "@hey-api/openapi-ts": "^0.53.0",
-    "@llamaindex/core": "workspace:^0.2.8",
-    "@llamaindex/env": "workspace:^0.1.13",
+    "@llamaindex/core": "workspace:*",
+    "@llamaindex/env": "workspace:*",
    "bunchee": "5.3.2"
  },
  "peerDependencies": {
-    "@llamaindex/core": "workspace:^0.2.8",
-    "@llamaindex/env": "workspace:^0.1.13"
+    "@llamaindex/core": "workspace:*",
+    "@llamaindex/env": "workspace:*"
  },
  "dependencies": {
    "magic-bytes.js": "^1.10.0"
@@ -1,5 +1,12 @@
 # @llamaindex/community

+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+
 ## 0.0.43

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/community",
  "description": "Community package for LlamaIndexTS",
-  "version": "0.0.43",
+  "version": "0.0.44",
  "type": "module",
  "types": "dist/type/index.d.ts",
  "main": "dist/cjs/index.js",
@@ -1,5 +1,11 @@
 # @llamaindex/core

+## 0.2.9
+
+### Patch Changes
+
+- b17d439: Fix #1278: resolved issue where the id\_ was not correctly passed as the id when creating a TextNode. As a result, the upsert operation to the vector database was using a generated ID instead of the provided document ID, if available.
+
 ## 0.2.8

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/core",
  "type": "module",
-  "version": "0.2.8",
+  "version": "0.2.9",
  "description": "LlamaIndex Core Module",
  "exports": {
    "./agent": {
@@ -479,7 +479,7 @@ export function buildNodeFromSplits(
    ) {
      const imageDoc = doc as ImageNode;
      const imageNode = new ImageNode({
-        id_: idGenerator(i, imageDoc),
+        id_: imageDoc.id_ ?? idGenerator(i, imageDoc),
        text: textChunk,
        image: imageDoc.image,
        embedding: imageDoc.embedding,
@@ -496,7 +496,7 @@ export function buildNodeFromSplits(
    ) {
      const textDoc = doc as TextNode;
      const node = new TextNode({
-        id_: idGenerator(i, textDoc),
+        id_: textDoc.id_ ?? idGenerator(i, textDoc),
        text: textChunk,
        embedding: textDoc.embedding,
        excludedEmbedMetadataKeys: [...textDoc.excludedEmbedMetadataKeys],
@@ -1,5 +1,18 @@
 # @llamaindex/experimental

+## 0.0.93
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.92
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.0.91

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/experimental",
  "description": "Experimental package for LlamaIndexTS",
-  "version": "0.0.91",
+  "version": "0.0.93",
  "type": "module",
  "types": "dist/type/index.d.ts",
  "main": "dist/cjs/index.js",
@@ -1,5 +1,26 @@
 # llamaindex

+## 0.6.15
+
+### Patch Changes
+
+- 2a82413: fix(core): set `Settings.llm` to OpenAI by default and support lazy load openai
+- Updated dependencies [2a82413]
+- Updated dependencies [0b20ff9]
+  - @llamaindex/groq@0.0.11
+  - @llamaindex/openai@0.1.12
+  - @llamaindex/cloud@0.2.11
+
+## 0.6.14
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+  - @llamaindex/ollama@0.0.4
+  - @llamaindex/openai@0.1.11
+  - @llamaindex/groq@0.0.10
+
 ## 0.6.13

 ### Patch Changes
@@ -1,5 +1,18 @@
 # @llamaindex/cloudflare-worker-agent-test

+## 0.0.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.0.75

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.75",
+  "version": "0.0.77",
  "type": "module",
  "private": true,
  "scripts": {
@@ -1,5 +1,12 @@
 # @llamaindex/llama-parse-browser-test

+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [0b20ff9]
+  - @llamaindex/cloud@0.2.11
+
 ## 0.0.6

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/llama-parse-browser-test",
  "private": true,
-  "version": "0.0.6",
+  "version": "0.0.7",
  "type": "module",
  "scripts": {
    "dev": "vite",
@@ -1,5 +1,18 @@
 # @llamaindex/next-agent-test

+## 0.1.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.1.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.1.75

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-agent-test",
-  "version": "0.1.75",
+  "version": "0.1.77",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,18 @@
 # test-edge-runtime

+## 0.1.76
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.1.75
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.1.74

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.74",
+  "version": "0.1.76",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,18 @@
 # @llamaindex/next-node-runtime

+## 0.0.58
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.57
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.0.56

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.0.56",
+  "version": "0.0.58",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,18 @@
 # @llamaindex/waku-query-engine-test

+## 0.0.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
 ## 0.0.75

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.75",
+  "version": "0.0.77",
  "type": "module",
  "private": true,
  "scripts": {
@@ -1,6 +1,6 @@
 {
  "name": "llamaindex",
-  "version": "0.6.13",
+  "version": "0.6.15",
  "license": "MIT",
  "type": "module",
  "keywords": [
@@ -2,7 +2,6 @@ import {
  type CallbackManager,
  Settings as CoreSettings,
 } from "@llamaindex/core/global";
-import { OpenAI } from "@llamaindex/openai";

 import { PromptHelper } from "@llamaindex/core/indices";

@@ -61,12 +60,6 @@ class GlobalSettings implements Config {
  }

  get llm(): LLM {
-    // fixme: we might need check internal error instead of try-catch here
-    try {
-      CoreSettings.llm;
-    } catch (error) {
-      CoreSettings.llm = new OpenAI();
-    }
    return CoreSettings.llm;
  }

@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "@llamaindex/openai";

 export class FireworksEmbedding extends OpenAIEmbedding {
-  constructor(init?: Partial<OpenAIEmbedding>) {
+  constructor(init?: Omit<Partial<OpenAIEmbedding>, "session">) {
    const {
      apiKey = getEnv("FIREWORKS_API_KEY"),
      additionalSessionOptions = {},
@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "@llamaindex/openai";

 export class TogetherEmbedding extends OpenAIEmbedding {
-  constructor(init?: Partial<OpenAIEmbedding>) {
+  constructor(init?: Omit<Partial<OpenAIEmbedding>, "session">) {
    const {
      apiKey = getEnv("TOGETHER_API_KEY"),
      additionalSessionOptions = {},
@@ -1,3 +1,15 @@
+//#region initial setup for OpenAI
+import { OpenAI } from "@llamaindex/openai";
+import { Settings } from "./Settings.js";
+
+try {
+  Settings.llm;
+} catch {
+  Settings.llm = new OpenAI();
+}
+
+//#endregion
+
 export {
  LlamaParseReader,
  type Language,
@@ -28,12 +40,12 @@ export type {
  JSONArray,
  JSONObject,
  JSONValue,
+  LlamaIndexEventMaps,
  LLMEndEvent,
  LLMStartEvent,
  LLMStreamEvent,
  LLMToolCallEvent,
  LLMToolResultEvent,
-  LlamaIndexEventMaps,
 } from "@llamaindex/core/global";
 export * from "@llamaindex/core/indices";
 export * from "@llamaindex/core/llms";
@@ -61,7 +73,7 @@ export * from "./postprocessors/index.js";
 export * from "./QuestionGenerator.js";
 export * from "./selectors/index.js";
 export * from "./ServiceContext.js";
-export { Settings } from "./Settings.js";
 export * from "./storage/StorageContext.js";
 export * from "./tools/index.js";
 export * from "./types.js";
+export { Settings };
@@ -6,7 +6,7 @@ const DEFAULT_MODEL = "mistralai/Mixtral-8x22B-Instruct-v0.1";
 const BASE_URL = "https://api.deepinfra.com/v1/openai";

 export class DeepInfra extends OpenAI {
-  constructor(init?: Partial<OpenAI>) {
+  constructor(init?: Omit<Partial<OpenAI>, "session">) {
    const {
      apiKey = getEnv(ENV_VARIABLE_NAME),
      additionalSessionOptions = {},
@@ -10,7 +10,9 @@ type DeepSeekModelName = keyof typeof DEEPSEEK_MODELS;
 const DEFAULT_MODEL: DeepSeekModelName = "deepseek-coder";

 export class DeepSeekLLM extends OpenAI {
-  constructor(init?: Partial<OpenAI> & { model?: DeepSeekModelName }) {
+  constructor(
+    init?: Omit<Partial<OpenAI>, "session"> & { model?: DeepSeekModelName },
+  ) {
    const {
      apiKey = getEnv("DEEPSEEK_API_KEY"),
      additionalSessionOptions = {},
@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAI } from "@llamaindex/openai";

 export class FireworksLLM extends OpenAI {
-  constructor(init?: Partial<OpenAI>) {
+  constructor(init?: Omit<Partial<OpenAI>, "session">) {
    const {
      apiKey = getEnv("FIREWORKS_API_KEY"),
      additionalSessionOptions = {},
@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAI } from "@llamaindex/openai";

 export class TogetherLLM extends OpenAI {
-  constructor(init?: Partial<OpenAI>) {
+  constructor(init?: Omit<Partial<OpenAI>, "session">) {
    const {
      apiKey = getEnv("TOGETHER_API_KEY"),
      additionalSessionOptions = {},
@@ -27,7 +27,7 @@ describe("VectorStoreIndex", () => {
      runs: number = 2,
    ): Promise<Array<number>> => {
      const documents = [new Document({ text: "lorem ipsem", id_: "1" })];
-      const entries = [];
+      const entries: number[] = [];
      for (let i = 0; i < runs; i++) {
        await VectorStoreIndex.fromDocuments(documents, {
          serviceContext,
@@ -43,7 +43,7 @@ describe("VectorStoreIndex", () => {

  test("fromDocuments stores duplicates without a doc store strategy", async () => {
    const entries = await testStrategy(DocStoreStrategy.NONE);
-    expect(entries[0]! + 1).toBe(entries[1]);
+    expect(entries[0]).toBe(entries[1]);
  });

  test("fromDocuments ignores duplicates with upserts doc store strategy", async () => {
@@ -0,0 +1,7 @@
+import { expect, test, vi } from "vitest";
+
+test("init without error", async () => {
+  vi.stubEnv("OPENAI_API_KEY", undefined);
+  const { Settings } = await import("llamaindex");
+  expect(Settings.llm).toBeDefined();
+});
@@ -1,5 +1,19 @@
 # @llamaindex/groq

+## 0.0.11
+
+### Patch Changes
+
+- 2a82413: fix(core): set `Settings.llm` to OpenAI by default and support lazy load openai
+- Updated dependencies [2a82413]
+  - @llamaindex/openai@0.1.12
+
+## 0.0.10
+
+### Patch Changes
+
+- @llamaindex/openai@0.1.11
+
 ## 0.0.9

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/groq",
  "description": "Groq Adapter for LlamaIndex",
-  "version": "0.0.9",
+  "version": "0.0.11",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -4,7 +4,7 @@ import GroqSDK, { type ClientOptions } from "groq-sdk";

 export class Groq extends OpenAI {
  constructor(
-    init?: Partial<OpenAI> & {
+    init?: Omit<Partial<OpenAI>, "session"> & {
      additionalSessionOptions?: ClientOptions;
    },
  ) {
@@ -22,9 +22,10 @@ export class Groq extends OpenAI {
      ...rest,
    });

-    this.session.openai = new GroqSDK({
-      apiKey,
-      ...init?.additionalSessionOptions,
-    }) as any;
+    this.lazySession = async () =>
+      new GroqSDK({
+        apiKey,
+        ...init?.additionalSessionOptions,
+      }) as any;
  }
 }
@@ -1,5 +1,12 @@
 # @llamaindex/ollama

+## 0.0.4
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+
 ## 0.0.3

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/ollama",
  "description": "Ollama Adapter for LlamaIndex",
-  "version": "0.0.3",
+  "version": "0.0.4",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -1,5 +1,18 @@
 # @llamaindex/openai

+## 0.1.12
+
+### Patch Changes
+
+- 2a82413: fix(core): set `Settings.llm` to OpenAI by default and support lazy load openai
+
+## 0.1.11
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+
 ## 0.1.10

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/openai",
  "description": "OpenAI Adapter for LlamaIndex",
-  "version": "0.1.10",
+  "version": "0.1.12",
  "type": "module",
  "main": "./dist/index.cjs",
  "module": "./dist/index.js",
@@ -2,11 +2,6 @@ import { getEnv } from "@llamaindex/env";

 import type { AzureClientOptions } from "openai";

-export interface AzureOpenAIConfig extends AzureClientOptions {
-  /** @deprecated use "deployment" instead */
-  deploymentName?: string | undefined;
-}
-
 // NOTE we're not supporting the legacy models as they're not available for new deployments
 // https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/legacy-models
 // If you have a need for them, please open an issue on GitHub
@@ -85,14 +80,15 @@ const DEFAULT_API_VERSION = "2023-05-15";
 //^ NOTE: this will change over time, if you want to pin it, use a specific version

 export function getAzureConfigFromEnv(
-  init?: Partial<AzureOpenAIConfig> & { model?: string },
-): AzureOpenAIConfig {
+  init?: Partial<AzureClientOptions> & { model?: string },
+): AzureClientOptions {
  const deployment =
-    init?.deploymentName ??
-    init?.deployment ??
-    getEnv("AZURE_OPENAI_DEPLOYMENT") ?? // From Azure docs
-    getEnv("AZURE_OPENAI_API_DEPLOYMENT_NAME") ?? // LCJS compatible
-    init?.model; // Fall back to model name, Python compatible
+    init && "deploymentName" in init && typeof init.deploymentName === "string"
+      ? init?.deploymentName
+      : (init?.deployment ??
+        getEnv("AZURE_OPENAI_DEPLOYMENT") ?? // From Azure docs
+        getEnv("AZURE_OPENAI_API_DEPLOYMENT_NAME") ?? // LCJS compatible
+        init?.model); // Fall back to model name, Python compatible
  return {
    apiKey:
      init?.apiKey ??
@@ -110,15 +106,10 @@ export function getAzureConfigFromEnv(
      getEnv("OPENAI_API_VERSION") ?? // Python compatible
      getEnv("AZURE_OPENAI_API_VERSION") ?? // LCJS compatible
      DEFAULT_API_VERSION,
-    deploymentName: deployment, // LCJS compatible
    deployment, // For Azure OpenAI
  };
 }

-export function getAzureBaseUrl(config: AzureOpenAIConfig): string {
-  return `${config.endpoint}/openai/deployments/${config.deploymentName}`;
-}
-
 export function getAzureModel(openAIModel: string) {
  for (const [key, value] of Object.entries(
    ALL_AZURE_OPENAI_EMBEDDING_MODELS,
@@ -1,14 +1,16 @@
 import { BaseEmbedding } from "@llamaindex/core/embeddings";
-import { Tokenizers } from "@llamaindex/env";
-import type { ClientOptions as OpenAIClientOptions } from "openai";
-import type { AzureOpenAIConfig } from "./azure.js";
+import { getEnv, Tokenizers } from "@llamaindex/env";
+import type {
+  AzureClientOptions,
+  AzureOpenAI as AzureOpenAILLM,
+  ClientOptions as OpenAIClientOptions,
+  OpenAI as OpenAILLM,
+} from "openai";
 import {
  getAzureConfigFromEnv,
  getAzureModel,
  shouldUseAzure,
 } from "./azure.js";
-import type { OpenAISession } from "./llm.js";
-import { getOpenAISession } from "./llm.js";

 export const ALL_OPENAI_EMBEDDING_MODELS = {
  "text-embedding-ada-002": {
@@ -32,6 +34,8 @@ export const ALL_OPENAI_EMBEDDING_MODELS = {

 type ModelKeys = keyof typeof ALL_OPENAI_EMBEDDING_MODELS;

+type LLMInstance = Pick<AzureOpenAILLM | OpenAILLM, "embeddings" | "apiKey">;
+
 export class OpenAIEmbedding extends BaseEmbedding {
  /** embeddding model. defaults to "text-embedding-ada-002" */
  model: string;
@@ -51,14 +55,26 @@ export class OpenAIEmbedding extends BaseEmbedding {
    | Omit<Partial<OpenAIClientOptions>, "apiKey" | "maxRetries" | "timeout">
    | undefined;

-  /** session object */
-  session: OpenAISession;
+  // use lazy here to avoid check OPENAI_API_KEY immediately
+  lazySession: () => Promise<LLMInstance>;
+  #session: Promise<LLMInstance> | null = null;
+  get session() {
+    if (!this.#session) {
+      this.#session = this.lazySession();
+    }
+    return this.#session;
+  }

  /**
   * OpenAI Embedding
   * @param init - initial parameters
   */
-  constructor(init?: Partial<OpenAIEmbedding> & { azure?: AzureOpenAIConfig }) {
+  constructor(
+    init?: Omit<Partial<OpenAIEmbedding>, "lazySession"> & {
+      session?: LLMInstance | undefined;
+      azure?: AzureClientOptions;
+    },
+  ) {
    super();

    this.model = init?.model ?? "text-embedding-ada-002";
@@ -77,7 +93,6 @@ export class OpenAIEmbedding extends BaseEmbedding {
    if (key) {
      this.embedInfo = ALL_OPENAI_EMBEDDING_MODELS[key];
    }
-
    if (init?.azure || shouldUseAzure()) {
      const azureConfig = {
        ...getAzureConfigFromEnv({
@@ -85,26 +100,32 @@ export class OpenAIEmbedding extends BaseEmbedding {
        }),
        ...init?.azure,
      };
-
-      this.apiKey = azureConfig.apiKey;
-      this.session =
-        init?.session ??
-        getOpenAISession({
-          azure: true,
-          maxRetries: this.maxRetries,
-          timeout: this.timeout,
-          ...this.additionalSessionOptions,
-          ...azureConfig,
-        });
+      this.apiKey =
+        init?.session?.apiKey ?? azureConfig.apiKey ?? getEnv("OPENAI_API_KEY");
+      this.lazySession = async () =>
+        import("openai").then(
+          async ({ AzureOpenAI }) =>
+            init?.session ??
+            new AzureOpenAI({
+              maxRetries: this.maxRetries,
+              timeout: this.timeout!,
+              ...this.additionalSessionOptions,
+              ...azureConfig,
+            }),
+        );
    } else {
-      this.apiKey = init?.apiKey ?? undefined;
-      this.session =
-        init?.session ??
-        getOpenAISession({
-          apiKey: this.apiKey,
-          maxRetries: this.maxRetries,
-          timeout: this.timeout,
-          ...this.additionalSessionOptions,
+      this.apiKey = init?.session?.apiKey ?? getEnv("OPENAI_API_KEY");
+      this.lazySession = async () =>
+        import("openai").then(({ OpenAI }) => {
+          return (
+            init?.session ??
+            new OpenAI({
+              apiKey: this.apiKey,
+              maxRetries: this.maxRetries,
+              timeout: this.timeout!,
+              ...this.additionalSessionOptions,
+            })
+          );
        });
    }
  }
@@ -118,7 +139,9 @@ export class OpenAIEmbedding extends BaseEmbedding {
    // TODO: ensure this for every sub class by calling it in the base class
    input = this.truncateMaxTokens(input);

-    const { data } = await this.session.openai.embeddings.create(
+    const { data } = await (
+      await this.session
+    ).embeddings.create(
      this.dimensions
        ? {
            model: this.model,
@@ -10,9 +10,6 @@ export {
  GPT4_MODELS,
  O1_MODELS,
  OpenAI,
-  OpenAISession,
  type OpenAIAdditionalChatOptions,
  type OpenAIAdditionalMetadata,
 } from "./llm";
-
-export { type AzureOpenAIConfig } from "./azure";
@@ -1,12 +1,11 @@
 import { getEnv } from "@llamaindex/env";
-import type OpenAILLM from "openai";
 import type {
-  ClientOptions,
+  AzureClientOptions,
+  AzureOpenAI as AzureOpenAILLM,
  ClientOptions as OpenAIClientOptions,
+  OpenAI as OpenAILLM,
 } from "openai";
-import { AzureOpenAI, OpenAI as OrigOpenAI } from "openai";
 import type { ChatModel } from "openai/resources/chat/chat";
-import { isDeepEqual } from "remeda";

 import { wrapEventCaller, wrapLLMEvent } from "@llamaindex/core/decorator";
 import {
@@ -35,64 +34,12 @@ import type {
  ChatCompletionUserMessageParam,
 } from "openai/resources/chat/completions";
 import type { ChatCompletionMessageParam } from "openai/resources/index.js";
-import type { AzureOpenAIConfig } from "./azure.js";
 import {
  getAzureConfigFromEnv,
  getAzureModel,
  shouldUseAzure,
 } from "./azure.js";

-export class OpenAISession {
-  openai: Pick<OrigOpenAI, "chat" | "embeddings">;
-
-  constructor(options: ClientOptions & { azure?: boolean } = {}) {
-    if (options.azure) {
-      this.openai = new AzureOpenAI(options as AzureOpenAIConfig);
-    } else {
-      if (!options.apiKey) {
-        options.apiKey = getEnv("OPENAI_API_KEY");
-      }
-
-      if (!options.apiKey) {
-        throw new Error("Set OpenAI Key in OPENAI_API_KEY env variable"); // Overriding OpenAI package's error message
-      }
-
-      this.openai = new OrigOpenAI({
-        ...options,
-      });
-    }
-  }
-}
-
-// I'm not 100% sure this is necessary vs. just starting a new session
-// every time we make a call. They say they try to reuse connections
-// so in theory this is more efficient, but we should test it in the future.
-const defaultOpenAISession: {
-  session: OpenAISession;
-  options: ClientOptions;
-}[] = [];
-
-/**
- * Get a session for the OpenAI API. If one already exists with the same options,
- * it will be returned. Otherwise, a new session will be created.
- * @param options
- * @returns
- */
-export function getOpenAISession(
-  options: ClientOptions & { azure?: boolean } = {},
-) {
-  let session = defaultOpenAISession.find((session) => {
-    return isDeepEqual(session.options, options);
-  })?.session;
-
-  if (!session) {
-    session = new OpenAISession(options);
-    defaultOpenAISession.push({ session, options });
-  }
-
-  return session;
-}
-
 export const GPT4_MODELS = {
  "chatgpt-4o-latest": {
    contextWindow: 128000,
@@ -182,6 +129,8 @@ export type OpenAIAdditionalChatOptions = Omit<
  | "toolChoice"
 >;

+type LLMInstance = Pick<AzureOpenAILLM | OpenAILLM, "chat" | "apiKey">;
+
 export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
  model:
    | ChatModel
@@ -196,14 +145,24 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
  apiKey?: string | undefined = undefined;
  maxRetries: number;
  timeout?: number;
-  session: OpenAISession;
  additionalSessionOptions?:
    | undefined
    | Omit<Partial<OpenAIClientOptions>, "apiKey" | "maxRetries" | "timeout">;

+  // use lazy here to avoid check OPENAI_API_KEY immediately
+  lazySession: () => Promise<LLMInstance>;
+  #session: Promise<LLMInstance> | null = null;
+  get session() {
+    if (!this.#session) {
+      this.#session = this.lazySession();
+    }
+    return this.#session;
+  }
+
  constructor(
-    init?: Partial<OpenAI> & {
-      azure?: AzureOpenAIConfig;
+    init?: Omit<Partial<OpenAI>, "session"> & {
+      session?: LLMInstance | undefined;
+      azure?: AzureClientOptions;
    },
  ) {
    super();
@@ -216,6 +175,8 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
    this.timeout = init?.timeout ?? 60 * 1000; // Default is 60 seconds
    this.additionalChatOptions = init?.additionalChatOptions;
    this.additionalSessionOptions = init?.additionalSessionOptions;
+    this.apiKey =
+      init?.session?.apiKey ?? init?.apiKey ?? getEnv("OPENAI_API_KEY");

    if (init?.azure || shouldUseAzure()) {
      const azureConfig = {
@@ -225,25 +186,26 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
        ...init?.azure,
      };

-      this.apiKey = azureConfig.apiKey;
-      this.session =
+      this.lazySession = async () =>
        init?.session ??
-        getOpenAISession({
-          azure: true,
-          maxRetries: this.maxRetries,
-          timeout: this.timeout,
-          ...this.additionalSessionOptions,
-          ...azureConfig,
+        import("openai").then(({ AzureOpenAI }) => {
+          return new AzureOpenAI({
+            maxRetries: this.maxRetries,
+            timeout: this.timeout!,
+            ...this.additionalSessionOptions,
+            ...azureConfig,
+          });
        });
    } else {
-      this.apiKey = init?.apiKey ?? undefined;
-      this.session =
+      this.lazySession = async () =>
        init?.session ??
-        getOpenAISession({
-          apiKey: this.apiKey,
-          maxRetries: this.maxRetries,
-          timeout: this.timeout,
-          ...this.additionalSessionOptions,
+        import("openai").then(({ OpenAI }) => {
+          return new OpenAI({
+            apiKey: this.apiKey,
+            maxRetries: this.maxRetries,
+            timeout: this.timeout!,
+            ...this.additionalSessionOptions,
+          });
        });
    }
  }
@@ -382,7 +344,9 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
    }

    // Non-streaming
-    const response = await this.session.openai.chat.completions.create({
+    const response = await (
+      await this.session
+    ).chat.completions.create({
      ...baseRequestParams,
      stream: false,
    });
@@ -414,11 +378,12 @@ export class OpenAI extends ToolCallLLM<OpenAIAdditionalChatOptions> {
  protected async *streamChat(
    baseRequestParams: OpenAILLM.Chat.ChatCompletionCreateParams,
  ): AsyncIterable<ChatResponseChunk<ToolCallLLMMessageOptions>> {
-    const stream: AsyncIterable<OpenAILLM.Chat.ChatCompletionChunk> =
-      await this.session.openai.chat.completions.create({
-        ...baseRequestParams,
-        stream: true,
-      });
+    const stream: AsyncIterable<OpenAILLM.Chat.ChatCompletionChunk> = await (
+      await this.session
+    ).chat.completions.create({
+      ...baseRequestParams,
+      stream: true,
+    });

    // TODO: add callback to streamConverter and use streamConverter here
    // this will be used to keep track of the current tool call, make sure input are valid json object.
@@ -360,10 +360,10 @@ importers:
        specifier: ^0.53.0
        version: 0.53.0(typescript@5.6.2)
      '@llamaindex/core':
-        specifier: workspace:^0.2.8
+        specifier: workspace:*
        version: link:../core
      '@llamaindex/env':
-        specifier: workspace:^0.1.13
+        specifier: workspace:*
        version: link:../env
      bunchee:
        specifier: 5.3.2
@@ -19518,8 +19518,8 @@ snapshots:
      '@typescript-eslint/parser': 7.2.0(eslint@8.57.0)(typescript@5.6.2)
      eslint: 8.57.0
      eslint-import-resolver-node: 0.3.9
-      eslint-import-resolver-typescript: 3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0)
-      eslint-plugin-import: 2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0)
+      eslint-import-resolver-typescript: 3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1)(eslint@8.57.0)
+      eslint-plugin-import: 2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-typescript@3.6.3)(eslint@8.57.0)
      eslint-plugin-jsx-a11y: 6.9.0(eslint@8.57.0)
      eslint-plugin-react: 7.35.0(eslint@8.57.0)
      eslint-plugin-react-hooks: 4.6.2(eslint@8.57.0)
@@ -19566,25 +19566,6 @@ snapshots:
    transitivePeerDependencies:
      - supports-color

-  eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0):
-    dependencies:
-      '@nolyfill/is-core-module': 1.0.39
-      debug: 4.3.7
-      enhanced-resolve: 5.17.1
-      eslint: 8.57.0
-      eslint-module-utils: 2.8.2(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0)
-      fast-glob: 3.3.2
-      get-tsconfig: 4.8.0
-      is-bun-module: 1.1.0
-      is-glob: 4.0.3
-    optionalDependencies:
-      eslint-plugin-import: 2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0)
-    transitivePeerDependencies:
-      - '@typescript-eslint/parser'
-      - eslint-import-resolver-node
-      - eslint-import-resolver-webpack
-      - supports-color
-
  eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1)(eslint@8.57.0):
    dependencies:
      '@nolyfill/is-core-module': 1.0.39
@@ -19604,17 +19585,6 @@ snapshots:
      - eslint-import-resolver-webpack
      - supports-color

-  eslint-module-utils@2.8.2(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0):
-    dependencies:
-      debug: 3.2.7
-    optionalDependencies:
-      '@typescript-eslint/parser': 7.2.0(eslint@8.57.0)(typescript@5.6.2)
-      eslint: 8.57.0
-      eslint-import-resolver-node: 0.3.9
-      eslint-import-resolver-typescript: 3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0)
-    transitivePeerDependencies:
-      - supports-color
-
  eslint-module-utils@2.8.2(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1)(eslint@8.57.0))(eslint@8.57.0):
    dependencies:
      debug: 3.2.7
@@ -19626,33 +19596,6 @@ snapshots:
    transitivePeerDependencies:
      - supports-color

-  eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0):
-    dependencies:
-      array-includes: 3.1.8
-      array.prototype.findlastindex: 1.2.5
-      array.prototype.flat: 1.3.2
-      array.prototype.flatmap: 1.3.2
-      debug: 3.2.7
-      doctrine: 2.1.0
-      eslint: 8.57.0
-      eslint-import-resolver-node: 0.3.9
-      eslint-module-utils: 2.8.2(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-import-resolver-typescript@3.6.3(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-node@0.3.9)(eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0))(eslint@8.57.0))(eslint@8.57.0)
-      hasown: 2.0.2
-      is-core-module: 2.15.1
-      is-glob: 4.0.3
-      minimatch: 3.1.2
-      object.fromentries: 2.0.8
-      object.groupby: 1.0.3
-      object.values: 1.2.0
-      semver: 6.3.1
-      tsconfig-paths: 3.15.0
-    optionalDependencies:
-      '@typescript-eslint/parser': 7.2.0(eslint@8.57.0)(typescript@5.6.2)
-    transitivePeerDependencies:
-      - eslint-import-resolver-typescript
-      - eslint-import-resolver-webpack
-      - supports-color
-
  eslint-plugin-import@2.29.1(@typescript-eslint/parser@7.2.0(eslint@8.57.0)(typescript@5.6.2))(eslint-import-resolver-typescript@3.6.3)(eslint@8.57.0):
    dependencies:
      array-includes: 3.1.8
@@ -19679,7 +19622,6 @@ snapshots:
      - eslint-import-resolver-typescript
      - eslint-import-resolver-webpack
      - supports-color
-    optional: true

  eslint-plugin-import@2.29.1(@typescript-eslint/parser@8.5.0(eslint@8.57.0)(typescript@5.6.2))(eslint@8.57.0):
    dependencies:
Author	SHA1	Message	Date
github-actions[bot]	b7382b0d24	Release 0.6.15 (#1295 ) Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>	2024-10-03 19:44:55 -07:00
Alex Yang	2a8241328d	fix: lazy load openai (#1294 )	2024-10-03 17:12:33 -07:00
Alex Yang	0b20ff9f17	fix(cloud): package.json format (#1291 )	2024-10-03 17:07:50 -07:00
github-actions[bot]	1fc26046e3	Release 0.6.14 (#1290 ) Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>	2024-10-03 16:35:54 -07:00
LAWG	b17d439d6d	fix: ensure `id_` is correctly passed during creation (#1282 ) Co-authored-by: lawrencegb <lawrence@3api.com> Co-authored-by: Alex Yang <himself65@outlook.com>	2024-10-03 11:52:26 -07:00