macroclaw 0.11.0 → 0.12.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "macroclaw",
-  "version": "0.11.0",
+  "version": "0.12.0",
   "description": "Telegram-to-Claude-Code bridge",
   "license": "MIT",
   "type": "module",

package/src/app.test.ts

@@ -3,18 +3,13 @@ import { existsSync, rmSync } from "node:fs";
 import { App, type AppConfig } from "./app";
 import { type Claude, QueryProcessError, type QueryResult, type RunningQuery } from "./claude";
 import { saveSessions } from "./sessions";
+import type { SpeechToText } from "./speech-to-text";
 
-const mockOpenAICreate = mock(async () => ({ text: "transcribed text" }));
+const mockTranscribe = mock(async (_filePath: string) => "transcribed text");
 
-mock.module("openai", () => ({
-  default: class MockOpenAI {
-    audio = {
-      transcriptions: {
-        create: mockOpenAICreate,
-      },
-    };
-  },
-}));
+function mockStt(): SpeechToText {
+  return { transcribe: mockTranscribe } as unknown as SpeechToText;
+}
 
 // Mock Grammy Bot
 mock.module("grammy", () => ({
@@ -59,17 +54,14 @@ mock.module("grammy", () => ({
 
 const tmpSettingsDir = "/tmp/macroclaw-test-settings";
 
-const savedOpenAIKey = process.env.OPENAI_API_KEY;
-
 beforeEach(() => {
-  process.env.OPENAI_API_KEY = "test-key";
+  mockTranscribe.mockReset();
+  mockTranscribe.mockImplementation(async () => "transcribed text");
   if (existsSync(tmpSettingsDir)) rmSync(tmpSettingsDir, { recursive: true });
   saveSessions({ mainSessionId: "test-session" }, tmpSettingsDir);
 });
 
 afterEach(() => {
-  if (savedOpenAIKey) process.env.OPENAI_API_KEY = savedOpenAIKey;
-  else delete process.env.OPENAI_API_KEY;
   if (existsSync(tmpSettingsDir)) rmSync(tmpSettingsDir, { recursive: true });
 });
 
@@ -128,6 +120,7 @@ function makeConfig(overrides?: Partial<AppConfig>): AppConfig {
     workspace: "/tmp/macroclaw-test-workspace",
     settingsDir: tmpSettingsDir,
     claude: defaultMockClaude(),
+    stt: mockStt(),
     ...overrides,
   };
 }
@@ -350,7 +343,7 @@ describe("App", () => {
       globalThis.fetch = mock(() =>
         Promise.resolve(new Response("fake-audio", { status: 200 })),
       ) as any;
-      mockOpenAICreate.mockImplementationOnce(async () => ({ text: "hello from voice" }));
+      mockTranscribe.mockImplementationOnce(async () => "hello from voice");
 
       const config = makeConfig();
       const app = new App(config);
@@ -380,7 +373,7 @@ describe("App", () => {
       globalThis.fetch = mock(() =>
         Promise.resolve(new Response("fake-audio", { status: 200 })),
       ) as any;
-      mockOpenAICreate.mockImplementationOnce(async () => { throw new Error("API error"); });
+      mockTranscribe.mockImplementationOnce(async () => { throw new Error("API error"); });
 
       const config = makeConfig();
       const app = new App(config);
@@ -406,7 +399,7 @@ describe("App", () => {
       globalThis.fetch = mock(() =>
         Promise.resolve(new Response("fake-audio", { status: 200 })),
       ) as any;
-      mockOpenAICreate.mockImplementationOnce(async () => ({ text: "  " }));
+      mockTranscribe.mockImplementationOnce(async () => "  ");
 
       const config = makeConfig();
       const app = new App(config);
@@ -442,9 +435,8 @@ describe("App", () => {
       expect((config.claude as Claude & { calls: CallInfo[] }).calls).toHaveLength(0);
     });
 
-    it("responds with unavailable message when OPENAI_API_KEY is not set", async () => {
-      delete process.env.OPENAI_API_KEY;
-      const config = makeConfig();
+    it("responds with unavailable message when stt is not configured", async () => {
+      const config = makeConfig({ stt: undefined });
       const app = new App(config);
       const bot = app.bot as any;
       const handler = bot.filterHandlers.get("message:voice")![0];
@@ -455,7 +447,7 @@ describe("App", () => {
       });
 
       const sendCalls = (bot.api.sendMessage as any).mock.calls;
-      const call = sendCalls.find((c: any) => c[1].includes("OPENAI_API_KEY"));
+      const call = sendCalls.find((c: any) => c[1].includes("openaiApiKey"));
       expect(call).toBeDefined();
       expect((config.claude as Claude & { calls: CallInfo[] }).calls).toHaveLength(0);
     });

package/src/app.ts

@@ -2,7 +2,7 @@ import type { Bot } from "grammy";
 import { CronScheduler } from "./cron";
 import { createLogger } from "./logger";
 import { type Claude, Orchestrator, type OrchestratorResponse } from "./orchestrator";
-import { isAvailable as isSttAvailable, transcribe } from "./stt";
+import type { SpeechToText } from "./speech-to-text";
 import { createBot, downloadFile, sendFile, sendResponse } from "./telegram";
 
 const log = createLogger("app");
@@ -14,6 +14,7 @@ export interface AppConfig {
   model?: string;
   settingsDir?: string;
   claude?: Claude;
+  stt?: SpeechToText;
 }
 
 export class App {
@@ -118,13 +119,13 @@ export class App {
 
     this.#bot.on("message:voice", async (ctx) => {
       if (ctx.chat.id.toString() !== this.#config.authorizedChatId) return;
-      if (!isSttAvailable()) {
-        await sendResponse(this.#bot, this.#config.authorizedChatId, "[Voice messages not available — set OPENAI_API_KEY to enable]");
+      if (!this.#config.stt) {
+        await sendResponse(this.#bot, this.#config.authorizedChatId, "[Voice messages not available — set openaiApiKey in settings to enable]");
         return;
       }
       try {
         const path = await downloadFile(this.#bot, ctx.message.voice.file_id, this.#config.botToken, "voice.ogg");
-        const text = await transcribe(path);
+        const text = await this.#config.stt.transcribe(path);
         if (!text.trim()) {
           await sendResponse(this.#bot, this.#config.authorizedChatId, "[Could not understand audio]");
           return;

package/src/index.ts

@@ -3,6 +3,7 @@ import { dirname, join, resolve } from "node:path";
 import { App, type AppConfig } from "./app";
 import { createLogger, initLogger } from "./logger";
 import { applyEnvOverrides, loadSettings, printSettings } from "./settings";
+import { SpeechToText } from "./speech-to-text";
 
 export async function start(): Promise<void> {
   const log = createLogger("index");
@@ -42,6 +43,7 @@ export async function start(): Promise<void> {
     authorizedChatId: resolved.chatId,
     workspace,
     model: resolved.model,
+    stt: resolved.openaiApiKey ? new SpeechToText(resolved.openaiApiKey) : undefined,
   };
 
   new App(config).start();

package/src/speech-to-text.ts

@@ -0,0 +1,28 @@
+import { readFile } from "node:fs/promises";
+import { basename } from "node:path";
+import OpenAI from "openai";
+import { createLogger } from "./logger";
+
+const log = createLogger("speech-to-text");
+
+export class SpeechToText {
+  #client: OpenAI;
+
+  constructor(apiKey: string) {
+    this.#client = new OpenAI({ apiKey });
+  }
+
+  async transcribe(filePath: string): Promise<string> {
+    const buffer = await readFile(filePath);
+    const file = new File([buffer], basename(filePath), { type: "audio/ogg" });
+
+    log.debug({ filePath }, "Transcribing audio");
+    const result = await this.#client.audio.transcriptions.create({
+      model: "whisper-1",
+      file,
+    });
+
+    log.debug({ text: result.text }, "Transcription complete");
+    return result.text;
+  }
+}

package/src/stt.ts

@@ -1,31 +0,0 @@
-import { readFile } from "node:fs/promises";
-import { basename } from "node:path";
-import OpenAI from "openai";
-import { createLogger } from "./logger";
-
-const log = createLogger("stt");
-
-let client: OpenAI | undefined;
-
-function getClient(): OpenAI {
-  if (!client) client = new OpenAI();
-  return client;
-}
-
-export function isAvailable(): boolean {
-  return !!process.env.OPENAI_API_KEY;
-}
-
-export async function transcribe(filePath: string): Promise<string> {
-  const buffer = await readFile(filePath);
-  const file = new File([buffer], basename(filePath), { type: "audio/ogg" });
-
-  log.debug({ filePath }, "Transcribing audio");
-  const result = await getClient().audio.transcriptions.create({
-    model: "whisper-1",
-    file,
-  });
-
-  log.debug({ text: result.text }, "Transcription complete");
-  return result.text;
-}