@grinev/opencode-telegram-bot 0.18.0 → 0.19.0

package/dist/tts/client.js

@@ -1,22 +1,80 @@
 import { config } from "../config.js";
 import { logger } from "../utils/logger.js";
+import textToSpeech from "@google-cloud/text-to-speech";
 const TTS_REQUEST_TIMEOUT_MS = 60_000;
 export function isTtsConfigured() {
+    if (config.tts.provider === "google") {
+        return Boolean(process.env.GOOGLE_APPLICATION_CREDENTIALS);
+    }
     return Boolean(config.tts.apiUrl && config.tts.apiKey);
 }
-export async function synthesizeSpeech(text) {
-    if (!isTtsConfigured()) {
-        throw new Error("TTS is not configured: set TTS API credentials");
+/**
+ * Removes markdown syntax that TTS engines would read aloud
+ * (asterisks, backticks, heading markers, etc.).
+ */
+export function stripMarkdownForSpeech(text) {
+    let clean = text;
+    // fenced code blocks → inline content
+    clean = clean.replace(/```[\s\S]*?```/g, (match) => {
+        const inner = match.replace(/^```\w*\n?/, "").replace(/\n?```$/, "");
+        return inner.replace(/\n/g, " ");
+    });
+    clean = clean.replace(/`([^`]+)`/g, "$1");
+    clean = clean.replace(/\*\*\*(.+?)\*\*\*/g, "$1");
+    clean = clean.replace(/\*\*(.+?)\*\*/g, "$1");
+    clean = clean.replace(/(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)/g, "$1");
+    clean = clean.replace(/~~(.+?)~~/g, "$1");
+    clean = clean.replace(/\[([^\]]+)\]\([^)]+\)/g, "$1");
+    clean = clean.replace(/^#{1,6}\s+/gm, "");
+    clean = clean.replace(/^>\s?/gm, "");
+    clean = clean.replace(/^[-*]\s+/gm, "");
+    clean = clean.replace(/^\d+\.\s+/gm, "");
+    clean = clean.replace(/^[-*_]{3,}\s*$/gm, "");
+    clean = clean.replace(/<\/?[A-Za-z][^>]*>/g, "");
+    clean = clean.replace(/[ \t]+/g, " ");
+    clean = clean.replace(/\n{3,}/g, "\n\n");
+    return clean.trim();
+}
+/** Extracts "ll-CC" from Google voice names like "de-DE-Neural2-B". */
+export function extractLanguageCode(voiceName) {
+    const match = voiceName.match(/^([a-z]{2,3}-[A-Z]{2})/);
+    return match ? match[1] : "en-US";
+}
+// --- Provider implementations ---
+let googleClient = null;
+function getGoogleClient() {
+    if (!googleClient) {
+        googleClient = new textToSpeech.TextToSpeechClient();
     }
-    const input = text.trim();
-    if (!input) {
-        throw new Error("TTS input text is empty");
+    return googleClient;
+}
+/** @internal Reset Google client singleton (for tests only). */
+export function _resetGoogleClient() {
+    googleClient = null;
+}
+async function synthesizeWithGoogle(text) {
+    const client = getGoogleClient();
+    const voiceName = config.tts.voice || "en-US-Studio-O";
+    const languageCode = extractLanguageCode(voiceName);
+    logger.debug(`[TTS] Google Cloud TTS: voice=${voiceName}, languageCode=${languageCode}, chars=${text.length}`);
+    const [response] = await client.synthesizeSpeech({
+        input: { text },
+        voice: { languageCode, name: voiceName },
+        audioConfig: { audioEncoding: "MP3" },
+    }, { timeout: TTS_REQUEST_TIMEOUT_MS });
+    const raw = response.audioContent;
+    const buffer = Buffer.isBuffer(raw) ? raw : Buffer.from(raw);
+    if (buffer.length === 0) {
+        throw new Error("Google TTS API returned an empty audio response");
     }
+    return { buffer, filename: "assistant-reply.mp3", mimeType: "audio/mpeg" };
+}
+async function synthesizeWithOpenAi(text) {
+    const url = `${config.tts.apiUrl}/audio/speech`;
+    logger.debug(`[TTS] OpenAI-compatible: url=${url}, model=${config.tts.model}, voice=${config.tts.voice}, chars=${text.length}`);
     const controller = new AbortController();
     const timeout = setTimeout(() => controller.abort(), TTS_REQUEST_TIMEOUT_MS);
     try {
-        const url = `${config.tts.apiUrl}/audio/speech`;
-        logger.debug(`[TTS] Sending speech synthesis request: url=${url}, model=${config.tts.model}, voice=${config.tts.voice}, chars=${input.length}`);
         const response = await fetch(url, {
             method: "POST",
             headers: {
@@ -26,7 +84,7 @@ export async function synthesizeSpeech(text) {
             body: JSON.stringify({
                 model: config.tts.model,
                 voice: config.tts.voice,
-                input,
+                input: text,
                 response_format: "mp3",
             }),
             signal: controller.signal,
@@ -35,17 +93,37 @@ export async function synthesizeSpeech(text) {
             const errorBody = await response.text().catch(() => "");
             throw new Error(`TTS API returned HTTP ${response.status}: ${errorBody || response.statusText}`);
         }
-        const arrayBuffer = await response.arrayBuffer();
-        const buffer = Buffer.from(arrayBuffer);
+        const buffer = Buffer.from(await response.arrayBuffer());
         if (buffer.length === 0) {
             throw new Error("TTS API returned an empty audio response");
         }
         logger.debug(`[TTS] Generated speech audio: ${buffer.length} bytes`);
-        return {
-            buffer,
-            filename: "assistant-reply.mp3",
-            mimeType: "audio/mpeg",
-        };
+        return { buffer, filename: "assistant-reply.mp3", mimeType: "audio/mpeg" };
+    }
+    finally {
+        clearTimeout(timeout);
+    }
+}
+// --- Public API ---
+function getNotConfiguredMessage() {
+    return config.tts.provider === "google"
+        ? "TTS is not configured: set GOOGLE_APPLICATION_CREDENTIALS for Google Cloud TTS"
+        : "TTS is not configured: set TTS_API_URL and TTS_API_KEY";
+}
+export async function synthesizeSpeech(text) {
+    if (!isTtsConfigured()) {
+        throw new Error(getNotConfiguredMessage());
+    }
+    const raw = text.trim();
+    if (!raw) {
+        throw new Error("TTS input text is empty");
+    }
+    const input = stripMarkdownForSpeech(raw);
+    try {
+        if (config.tts.provider === "google") {
+            return await synthesizeWithGoogle(input);
+        }
+        return await synthesizeWithOpenAi(input);
     }
     catch (err) {
         if (err instanceof DOMException && err.name === "AbortError") {
@@ -53,7 +131,4 @@ export async function synthesizeSpeech(text) {
         }
         throw err;
     }
-    finally {
-        clearTimeout(timeout);
-    }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@grinev/opencode-telegram-bot",
-  "version": "0.18.0",
+  "version": "0.19.0",
   "description": "Telegram bot client for OpenCode to run and monitor coding tasks from chat.",
   "type": "module",
   "main": "./dist/index.js",
@@ -51,6 +51,7 @@
   },
   "homepage": "https://github.com/grinev/opencode-telegram-bot#readme",
   "dependencies": {
+    "@google-cloud/text-to-speech": "^6.4.0",
     "@grammyjs/menu": "^1.3.1",
     "@opencode-ai/sdk": "^1.1.21",
     "better-sqlite3": "^12.6.2",
@@ -61,7 +62,6 @@
     "remark-gfm": "^4.0.1",
     "remark-parse": "^11.0.0",
     "socks-proxy-agent": "^8.0.5",
-    "telegram-markdown-v2": "^0.0.4",
     "unified": "^11.0.5"
   },
   "devDependencies": {