Merge pull request #48 from AssemblyAI/E07417BDFEA3614F5967B1520F8B2F61

Sync from internal repo (2024/05/03)

CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## [4.4.2]
+
+### Changed
+
+- Caching is disabled for all HTTP request made by the SDK
+- Change how the WebSocket libraries are imported for better compatibility across frameworks and runtimes.
+  The library no longer relies on a internal `#ws` import, and instead compiles the imports into the dist bundles.
+  Browser builds will use the native `WebSocket`, other builds will use the `ws` package.
+
 ## [4.4.1] - 2024-04-16
 
 ### Changed

package.json

@@ -1,6 +1,6 @@
 {
   "name": "assemblyai",
-  "version": "4.4.1",
+  "version": "4.4.2",
   "description": "The AssemblyAI JavaScript SDK provides an easy-to-use interface for interacting with the AssemblyAI API, which supports async and real-time transcription, as well as the latest LeMUR models.",
   "engines": {
     "node": ">=18"
@@ -17,7 +17,7 @@
         "default": "./dist/deno.mjs"
       },
       "workerd": "./dist/index.mjs",
-      "browser": "./dist/index.mjs",
+      "browser": "./dist/browser.mjs",
       "node": {
         "types": "./dist/index.d.ts",
         "import": "./dist/node.mjs",
@@ -40,14 +40,10 @@
       "node": "./src/polyfills/streams/node.ts",
       "default": "./src/polyfills/streams/index.ts"
     },
-    "#ws": {
-      "types": "./src/polyfills/ws/index.d.ts",
-      "browser": "./src/polyfills/ws/browser.mjs",
-      "default": {
-        "types": "./src/polyfills/ws/index.d.ts",
-        "import": "./src/polyfills/ws/index.mjs",
-        "require": "./src/polyfills/ws/index.cjs"
-      }
+    "#websocket": {
+      "browser": "./src/polyfills/websocket/browser.ts",
+      "node": "./src/polyfills/websocket/default.ts",
+      "default": "./src/polyfills/websocket/default.ts"
     }
   },
   "type": "commonjs",
@@ -70,7 +66,7 @@
     "clean": "rimraf dist/* && rimraf temp/* && rimraf temp-docs/*",
     "lint": "eslint -c .eslintrc.json '{src,tests}/**/*.{js,ts}' && publint && tsc --noEmit -p tsconfig.json",
     "test": "pnpm run test:unit && pnpm run test:integration",
-    "test:unit": "jest --config jest.unit.config.js",
+    "test:unit": "jest --config jest.unit.config.js --testTimeout 1000",
     "test:integration": "jest --config jest.integration.config.js --testTimeout 360000",
     "format": "prettier '**/*' --write",
     "generate:types": "tsx ./scripts/generate-types.ts && prettier 'src/types/*.generated.ts' --write",

rollup.config.js

@@ -21,7 +21,7 @@ module.exports = [
       // so we compile to es2015 for maximum compatibility.
       ts({ compilerOptions: { target: "ES2015" } }),
     ],
-    external: ["#ws"],
+    external: ["ws"],
     output: [
       {
         file: `./dist/index.mjs`,
@@ -38,7 +38,7 @@ module.exports = [
   {
     input: "src/index.ts",
     plugins: [ts({ compilerOptions: { customConditions: ["node"] } })],
-    external: ["fs", "stream", "stream/web", "#ws"],
+    external: ["fs", "stream", "stream/web", "ws"],
     output: [
       {
         file: `./dist/node.mjs`,
@@ -55,7 +55,7 @@ module.exports = [
   {
     input: "src/index.ts",
     plugins: [ts({ compilerOptions: { customConditions: ["deno"] } })],
-    external: ["#ws"],
+    external: ["ws"],
     output: [
       {
         file: `./dist/deno.mjs`,
@@ -67,7 +67,7 @@ module.exports = [
   {
     input: "src/index.ts",
     plugins: [ts({ compilerOptions: { customConditions: ["bun"] } })],
-    external: ["#ws"],
+    external: ["ws"],
     output: [
       {
         file: `./dist/bun.mjs`,
@@ -76,6 +76,18 @@ module.exports = [
       },
     ],
   },
+  // Browser ESM build
+  {
+    input: "src/index.ts",
+    plugins: [ts({ compilerOptions: { customConditions: ["browser"] } })],
+    output: [
+      {
+        file: `./dist/browser.mjs`,
+        format: "es",
+        exports: "named",
+      },
+    ],
+  },
   // Browser UMD build to reference directly in the browser.
   {
     ...umdConfig,

src/polyfills/websocket/browser.ts

@@ -0,0 +1,18 @@
+import { PolyfillWebSocketFactory, PolyfillWebSocket } from ".";
+export { PolyfillWebSocket } from ".";
+
+const PolyfillWebSocket =
+  WebSocket ?? global?.WebSocket ?? window?.WebSocket ?? self?.WebSocket;
+
+export const factory: PolyfillWebSocketFactory = (
+  url: string,
+  params?: unknown,
+) => {
+  if (params) {
+    return new PolyfillWebSocket(
+      url,
+      params as string | string[],
+    ) as unknown as PolyfillWebSocket;
+  }
+  return new PolyfillWebSocket(url) as unknown as PolyfillWebSocket;
+};

src/polyfills/websocket/default.ts

@@ -0,0 +1,8 @@
+import ws from "ws";
+import { PolyfillWebSocket, PolyfillWebSocketFactory } from ".";
+export { PolyfillWebSocket } from ".";
+
+export const factory: PolyfillWebSocketFactory = (
+  url: string,
+  params?: unknown,
+) => new ws(url, params as ws.ClientOptions) as unknown as PolyfillWebSocket;

src/polyfills/websocket/index.ts

@@ -0,0 +1,41 @@
+import ws, { Event, ErrorEvent, CloseEvent, MessageEvent } from "ws";
+
+export type PolyfillWebSocket = {
+  OPEN: typeof ws.OPEN;
+  binaryType: string;
+  onopen: ((event: Event) => void) | null;
+  onerror: ((event: ErrorEvent) => void) | null;
+  onclose: ((event: CloseEvent) => void) | null;
+  onmessage: ((event: MessageEvent) => void) | null;
+  readonly readyState:
+    | typeof ws.CONNECTING
+    | typeof ws.OPEN
+    | typeof ws.CLOSING
+    | typeof ws.CLOSED;
+  removeAllListeners?: () => void;
+  send(
+    data:
+      | string
+      | number
+      | Buffer
+      | DataView
+      | ArrayBufferView
+      | Uint8Array
+      | ArrayBuffer
+      | SharedArrayBuffer
+      | readonly unknown[]
+      | readonly number[]
+      | { valueOf(): ArrayBuffer }
+      | { valueOf(): SharedArrayBuffer }
+      | { valueOf(): Uint8Array }
+      | { valueOf(): readonly number[] }
+      | { valueOf(): string }
+      | { [Symbol.toPrimitive](hint: string): string },
+  ): unknown;
+  close(): unknown;
+};
+
+export type PolyfillWebSocketFactory = (
+  url: string,
+  params?: unknown,
+) => PolyfillWebSocket;

src/services/base.ts

@@ -21,6 +21,7 @@ export abstract class BaseService {
       "Content-Type": "application/json",
       ...init.headers,
     };
+    init.cache = "no-store";
     if (!input.startsWith("http")) input = this.params.baseUrl + input;
 
     const response = await fetch(input, init);

src/services/files/index.ts

@@ -10,8 +10,13 @@ export class FileService extends BaseService {
    */
   async upload(input: FileUploadParams): Promise<string> {
     let fileData: FileUploadData;
-    if (typeof input === "string") fileData = await readFile(input);
-    else fileData = input;
+    if (typeof input === "string") {
+      if (input.startsWith("data:")) {
+        fileData = dataUrlToBlob(input);
+      } else {
+        fileData = await readFile(input);
+      }
+    } else fileData = input;
 
     const data = await this.fetchJson<UploadedFile>("/v2/upload", {
       method: "POST",
@@ -24,3 +29,15 @@ export class FileService extends BaseService {
     return data.upload_url;
   }
 }
+
+function dataUrlToBlob(dataUrl: string) {
+  const arr = dataUrl.split(",");
+  const mime = arr[0].match(/:(.*?);/)![1];
+  const bstr = atob(arr[1]);
+  let n = bstr.length;
+  const u8arr = new Uint8Array(n);
+  while (n--) {
+    u8arr[n] = bstr.charCodeAt(n);
+  }
+  return new Blob([u8arr], { type: mime });
+}

src/services/realtime/service.ts

@@ -1,5 +1,8 @@
 import { WritableStream } from "#streams";
-import WebSocket from "#ws";
+import {
+  PolyfillWebSocket,
+  factory as polyfillWebSocketFactory,
+} from "#websocket";
 import { ErrorEvent, MessageEvent, CloseEvent } from "ws";
 import {
   RealtimeEvents,
@@ -52,7 +55,7 @@ export class RealtimeTranscriber {
   private endUtteranceSilenceThreshold?: number;
   private disablePartialTranscripts?: boolean;
 
-  private socket?: WebSocket;
+  private socket?: PolyfillWebSocket;
   private listeners: RealtimeListeners = {};
   private sessionTerminatedResolve?: () => void;
 
@@ -136,15 +139,15 @@ export class RealtimeTranscriber {
       const url = this.connectionUrl();
 
       if (this.token) {
-        this.socket = new WebSocket(url.toString());
+        this.socket = polyfillWebSocketFactory(url.toString());
       } else {
-        this.socket = new WebSocket(url.toString(), {
+        this.socket = polyfillWebSocketFactory(url.toString(), {
           headers: { Authorization: this.apiKey },
         });
       }
-      this.socket.binaryType = "arraybuffer";
+      this.socket!.binaryType = "arraybuffer";
 
-      this.socket.onopen = () => {
+      this.socket!.onopen = () => {
         if (
           this.endUtteranceSilenceThreshold === undefined ||
           this.endUtteranceSilenceThreshold === null
@@ -156,7 +159,7 @@ export class RealtimeTranscriber {
         );
       };
 
-      this.socket.onclose = ({ code, reason }: CloseEvent) => {
+      this.socket!.onclose = ({ code, reason }: CloseEvent) => {
         if (!reason) {
           if (code in RealtimeErrorType) {
             reason = RealtimeErrorMessages[code as RealtimeErrorType];
@@ -165,12 +168,12 @@ export class RealtimeTranscriber {
         this.listeners.close?.(code, reason);
       };
 
-      this.socket.onerror = (event: ErrorEvent) => {
+      this.socket!.onerror = (event: ErrorEvent) => {
         if (event.error) this.listeners.error?.(event.error as Error);
         else this.listeners.error?.(new Error(event.message));
       };
 
-      this.socket.onmessage = ({ data }: MessageEvent) => {
+      this.socket!.onmessage = ({ data }: MessageEvent) => {
         const message = JSON.parse(data.toString()) as RealtimeMessage;
         if ("error" in message) {
           this.listeners.error?.(new RealtimeError(message.error));
@@ -242,15 +245,15 @@ export class RealtimeTranscriber {
   }
 
   private send(data: BufferLike) {
-    if (!this.socket || this.socket.readyState !== WebSocket.OPEN) {
+    if (!this.socket || this.socket.readyState !== this.socket.OPEN) {
       throw new Error("Socket is not open for communication");
     }
     this.socket.send(data);
   }
 
   async close(waitForSessionTermination = true) {
     if (this.socket) {
-      if (this.socket.readyState === WebSocket.OPEN) {
+      if (this.socket.readyState === this.socket.OPEN) {
         if (waitForSessionTermination) {
           const sessionTerminatedPromise = new Promise<void>((resolve) => {
             this.sessionTerminatedResolve = resolve;
@@ -261,7 +264,7 @@ export class RealtimeTranscriber {
           this.socket.send(terminateSessionMessage);
         }
       }
-      if ("removeAllListeners" in this.socket) this.socket.removeAllListeners();
+      if (this.socket?.removeAllListeners) this.socket.removeAllListeners();
       this.socket.close();
     }
 

src/services/transcripts/index.ts

@@ -60,8 +60,12 @@ export class TranscriptService extends BaseService {
           // audio is local path, upload local file
           audioUrl = await this.files.upload(path);
         } else {
-          // audio is not a local path, assume it's a URL
-          audioUrl = audio;
+          if (audio.startsWith("data:")) {
+            audioUrl = await this.files.upload(audio);
+          } else {
+            // audio is not a local path, and not a data-URI, assume it's a normal URL
+            audioUrl = audio;
+          }
         }
       } else {
         // audio is of uploadable type