handy-remote-server 1.1.0 → 1.2.0

package/README.md

@@ -10,30 +10,71 @@ The easiest way to run the external inference server is using `npx`:
 npx handy-remote-server
 ```
 
-_(You must have Node.js and npm installed)_
+_(You must have Node.js and Rust/Cargo installed)_
 
 ## Usage
 
-When you run the server for the first time, it will automatically download the **GigaAM v3** model (Russian-only fast architecture model) if it's not present.
+When you run the server for the first time, it will:
 
-It will also generate a unique Bearer API Token for your active session:
+1. **Download the GigaAM v3 model** (~100 MB) with a progress bar:
 
-```bash
-Your API KEY is: xxxxx-xxxxx-xxxxx-xxxxx
+```
+📥 Downloading model...
+   URL:  https://blob.handy.computer/giga-am-v3.int8.onnx
+   Dest: /path/to/models/gigaam.onnx
+
+   ████████████████████░░░░░░░░░░░░░░░░░░░░  52.3%  52.10 MB / 99.60 MB  12.5 MB/s
+
+✅ Download complete in 8.2s
+```
+
+2. **Generate a persistent API key** saved to `~/.handy/api_key`:
+
+```
+======================================================
+Generated a new API KEY (saved to /Users/you/.handy/api_key)
+Your API KEY is: xxxxx...xxxxx
+======================================================
+```
+
+The key persists across restarts. On the next launch, it will be loaded automatically.
+
+3. **Start the server** and log every request in detail:
+
+```
 Handy Remote Server is running on port 3000
+
+[2026-03-07T12:00:00.000Z] ── REQUEST #1 ──────────────────────
+  Method:  POST /transcribe
+  From:    192.168.1.5
+  Auth:    OK
+  [#1]  Audio received: 156.3 KB
+  [#1]  Queued for inference (queue length: 0)
+[2026-03-07T12:00:01.234Z] ── RESPONSE #1 ─────────────────────
+  Status:   200
+  Duration: 1.23s
+  Result:   "Привет, как дела?"
 ```
 
+### Connecting from Handy
+
 1. Open **Handy** on your client machine.
-2. Go to **Settings > General**, select the `Remote` engine.
-3. Provide the Server URL: `http://<your-server-ip>:3000`
-4. Provide the generated Token.
-5. All audio chunks will now be transcribed by the server!
+2. Go to **Settings > Models**, select **Remote Server**.
+3. Go to **Settings > General**, fill in:
+   - **Remote Server URL**: `http://<your-server-ip>:3000`
+   - **API Token**: the generated token
+4. All transcriptions will now be processed by the server!
 
-## How it works
+## Environment Variables
 
-The `handy-remote-server` spins up a tiny Express server alongside a heavily optimized Rust CLI (`rust-infer`) powered by `transcribe-rs`. Audio files are dispatched sequentially from the Node server directly into the Rust engine.
+| Variable         | Default                                     | Description                     |
+| ---------------- | ------------------------------------------- | ------------------------------- |
+| `PORT`           | `3000`                                      | Server port                     |
+| `API_KEY`        | auto-generated, saved to `~/.handy/api_key` | Bearer token for authentication |
+| `INFER_CLI_PATH` | auto-detected                               | Path to the `rust-infer` binary |
 
-### Environment variables
+## How It Works
+
+The `handy-remote-server` spins up a tiny Express server alongside a heavily optimized Rust CLI (`rust-infer`) powered by `transcribe-rs`. Audio files are dispatched sequentially from the Node server directly into the Rust engine.
 
-- `PORT` - defaults to `3000`
-- `API_KEY` - defaults to an auto-generated token in development. Set this to a permanent token for production.
+Currently the server uses the **GigaAM v3** model (Russian-language, fast inference, ~100 MB).

package/dist/index.js

@@ -135,14 +135,46 @@ const gigaamModelPath = path_1.default.join(modelsDir, 'gigaam.onnx');
 async function downloadFile(url, dest) {
     if (fs_1.default.existsSync(dest))
         return;
-    console.log(`Downloading ${url} to ${dest}...`);
+    console.log(`\n📥 Downloading model...`);
+    console.log(`   URL:  ${url}`);
+    console.log(`   Dest: ${dest}\n`);
     fs_1.default.mkdirSync(path_1.default.dirname(dest), { recursive: true });
     const response = await fetch(url);
     if (!response.ok)
         throw new Error(`Failed to fetch ${url}: ${response.statusText}`);
-    const arrBuffer = await response.arrayBuffer();
-    fs_1.default.writeFileSync(dest, Buffer.from(arrBuffer));
-    console.log(`Downloaded ${dest}`);
+    const totalBytes = parseInt(response.headers.get('content-length') || '0', 10);
+    let downloadedBytes = 0;
+    const startTime = Date.now();
+    const fileStream = fs_1.default.createWriteStream(dest);
+    const reader = response.body?.getReader();
+    if (!reader)
+        throw new Error('Response body is not readable');
+    const barWidth = 40;
+    while (true) {
+        const { done, value } = await reader.read();
+        if (done)
+            break;
+        fileStream.write(Buffer.from(value));
+        downloadedBytes += value.length;
+        // Draw progress bar
+        const percent = totalBytes > 0 ? downloadedBytes / totalBytes : 0;
+        const filled = Math.round(barWidth * percent);
+        const empty = barWidth - filled;
+        const bar = '█'.repeat(filled) + '░'.repeat(empty);
+        const pct = (percent * 100).toFixed(1).padStart(5);
+        const dl = formatBytes(downloadedBytes);
+        const tot = totalBytes > 0 ? formatBytes(totalBytes) : '?';
+        const elapsed = (Date.now() - startTime) / 1000;
+        const speed = elapsed > 0 ? formatBytes(downloadedBytes / elapsed) + '/s' : '';
+        process.stdout.write(`\r   ${bar} ${pct}%  ${dl} / ${tot}  ${speed}   `);
+    }
+    await new Promise((resolve, reject) => {
+        fileStream.end(() => resolve());
+        fileStream.on('error', reject);
+    });
+    const totalTime = ((Date.now() - startTime) / 1000).toFixed(1);
+    process.stdout.write('\n');
+    console.log(`\n✅ Download complete in ${totalTime}s\n`);
 }
 async function ensureModels() {
     await downloadFile(GIGAAM_MODEL_URL, gigaamModelPath);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "handy-remote-server",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Remote Transcription Server for Handy",
   "main": "dist/index.js",
   "bin": {