feat: add Typer CLI with daemon mode and wtype keyboard injection

Replace argparse CLI with Typer-based CLI supporting `cohere on/off/status` commands. The daemon runs transcription in the background and types into the focused Wayland window via wtype. Adds wtype to flake.nix and fixes the hatchling build backend. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-05-30 21:09:32 +08:00
parent 8d517b3ea8
commit 92d8ba28d0
8 changed files with 277 additions and 43 deletions
@@ -0,0 +1,120 @@
+import os
+import subprocess
+import sys
+import time
+
+import typer
+from rich.console import Console
+
+from ..daemon import STATE_FILE, is_running, read_state, stop_daemon
+
+app = typer.Typer(help="Cohere live transcription — speaks into your keyboard.")
+console = Console()
+
+
+@app.command()
+def on(
+    language: str = typer.Option("en", "--lang", "-l", help="Language code"),
+    foreground: bool = typer.Option(False, "--fg", help="Run in foreground (don't daemonize)"),
+):
+    """Start transcribing and typing into your focused window."""
+    if is_running():
+        console.print("[yellow]Already running.[/yellow]")
+        raise typer.Exit(1)
+
+    if foreground:
+        from ..daemon import run_daemon
+        console.print("[green]Starting cohere (foreground)...[/green]")
+        run_daemon(language)
+        return
+
+    console.print("[green]Starting cohere daemon...[/green]")
+    os.makedirs(os.path.dirname(STATE_FILE), exist_ok=True)
+    subprocess.Popen(
+        [sys.executable, "-m", "cohere_transcribe.daemon_main", "--lang", language],
+        start_new_session=True,
+        stdin=subprocess.DEVNULL,
+        stdout=open(os.path.join(os.path.dirname(STATE_FILE), "daemon.log"), "a"),
+        stderr=subprocess.STDOUT,
+    )
+
+    for _ in range(50):
+        time.sleep(0.1)
+        if is_running():
+            break
+
+    if is_running():
+        console.print("[green]Cohere is on — speak and it types.[/green]")
+    else:
+        console.print("[red]Failed to start daemon. Check ~/.local/state/cohere/daemon.log[/red]")
+        raise typer.Exit(1)
+
+
+@app.command()
+def off():
+    """Stop transcribing."""
+    if not is_running():
+        console.print("[yellow]Not running.[/yellow]")
+        raise typer.Exit(0)
+
+    if stop_daemon():
+        console.print("[red]Cohere is off.[/red]")
+    else:
+        console.print("[red]Failed to stop daemon.[/red]")
+        raise typer.Exit(1)
+
+
+@app.command()
+def status():
+    """Show whether cohere is running."""
+    state = read_state()
+    running = is_running()
+
+    if running:
+        started = state.get("started_at", 0)
+        elapsed = time.time() - started
+        minutes = int(elapsed) // 60
+        console.print(f"[green]ON[/green] — running for {minutes}m")
+    else:
+        console.print("[dim]OFF[/dim]")
+
+
+@app.command()
+def transcribe(
+    audio_file: str = typer.Argument(None, help="Audio file to transcribe"),
+    mic: int = typer.Option(None, "--mic", "-m", help="Record from mic for N seconds"),
+    stream: bool = typer.Option(False, "--stream", "-s", help="Live streaming mode (prints to terminal)"),
+    language: str = typer.Option("en", "--lang", "-l", help="Language code"),
+):
+    """One-shot transcription (file, mic, or stream to terminal)."""
+    from ..model import load_model, transcribe_audio
+
+    if stream:
+        from ..stream import stream_transcribe
+        processor, model = load_model()
+        stream_transcribe(processor, model, language)
+    elif mic is not None:
+        from ..model import record_audio
+        processor, model = load_model()
+        try:
+            audio = record_audio(mic)
+            console.print("Transcribing...")
+            text = transcribe_audio(processor, model, audio, language)
+            console.print(f"\n{text}\n")
+        except OSError as e:
+            console.print(f"[red]Microphone error: {e}[/red]")
+            raise typer.Exit(1)
+    elif audio_file:
+        from transformers.audio_utils import load_audio as load_audio_file
+        from ..model import SAMPLE_RATE
+        processor, model = load_model()
+        audio = load_audio_file(audio_file, sampling_rate=SAMPLE_RATE)
+        text = transcribe_audio(processor, model, audio, language)
+        console.print(f"\n{text}\n")
+    else:
+        console.print("[yellow]Provide an audio file, --mic, or --stream[/yellow]")
+        raise typer.Exit(1)
+
+
+def main():
+    app()