Compare commits
2 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
985488bcbe | ||
|
|
524868d4f4 |
@@ -29,6 +29,8 @@ import logging
|
||||
import os
|
||||
import ssl
|
||||
import threading
|
||||
import time
|
||||
import uuid
|
||||
from http.server import BaseHTTPRequestHandler, HTTPServer
|
||||
from pathlib import Path
|
||||
from typing import Any, Callable, Dict, Optional
|
||||
@@ -441,3 +443,244 @@ class A2AMTLSClient:
|
||||
def post(self, url: str, json: Optional[Dict[str, Any]] = None, **kwargs: Any) -> Dict[str, Any]:
|
||||
data = (__import__("json").dumps(json).encode() if json is not None else None)
|
||||
return self._request("POST", url, data=data, **kwargs)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Structured A2A task delegation over mTLS
|
||||
# ---------------------------------------------------------------------------
|
||||
|
||||
_TERMINAL_TASK_STATES = {"completed", "failed", "canceled", "rejected"}
|
||||
|
||||
|
||||
def _iso_now() -> str:
|
||||
return time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime())
|
||||
|
||||
|
||||
def _task_status(state: str, message: str) -> Dict[str, Any]:
|
||||
return {
|
||||
"state": state,
|
||||
"message": message,
|
||||
"timestamp": _iso_now(),
|
||||
}
|
||||
|
||||
|
||||
def _coerce_artifact(result: Any) -> Dict[str, Any]:
|
||||
if isinstance(result, dict):
|
||||
if "text" in result:
|
||||
return result
|
||||
if "artifact" in result and isinstance(result["artifact"], dict):
|
||||
return result["artifact"]
|
||||
return {"text": str(result)}
|
||||
|
||||
|
||||
def _build_task_record(task_id: str, task: str, requester: Optional[str], metadata: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
|
||||
return {
|
||||
"taskId": task_id,
|
||||
"task": task,
|
||||
"requester": requester,
|
||||
"metadata": metadata or {},
|
||||
"artifacts": [],
|
||||
"status": _task_status("submitted", "Task submitted"),
|
||||
}
|
||||
|
||||
|
||||
def _default_agent_card(host: str, port: int) -> Dict[str, Any]:
|
||||
base_url = f"https://{host}:{port}"
|
||||
try:
|
||||
from agent.agent_card import build_agent_card
|
||||
from dataclasses import asdict
|
||||
|
||||
card = asdict(build_agent_card())
|
||||
except Exception as exc: # pragma: no cover - fallback only exercised when card build breaks
|
||||
logger.warning("Falling back to minimal agent card: %s", exc)
|
||||
card = {
|
||||
"name": os.environ.get("HERMES_AGENT_NAME", "hermes"),
|
||||
"description": "Hermes A2A task server",
|
||||
"version": "unknown",
|
||||
}
|
||||
card["url"] = base_url
|
||||
card["a2aTaskEndpoint"] = f"{base_url}/a2a/rpc"
|
||||
return card
|
||||
|
||||
|
||||
def _default_local_hermes_executor(task_payload: Dict[str, Any]) -> Dict[str, Any]:
|
||||
task_text = str(task_payload.get("task", "")).strip()
|
||||
if not task_text:
|
||||
return {"text": ""}
|
||||
from run_agent import AIAgent
|
||||
|
||||
agent = AIAgent(quiet_mode=True)
|
||||
result = agent.chat(task_text)
|
||||
return {
|
||||
"text": result,
|
||||
"metadata": {"executor": "local-hermes"},
|
||||
}
|
||||
|
||||
|
||||
class A2ATaskServer:
|
||||
"""JSON-RPC A2A task server running over the routing mTLS server."""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
cert: str | Path,
|
||||
key: str | Path,
|
||||
ca: str | Path,
|
||||
host: str = "127.0.0.1",
|
||||
port: int = 9443,
|
||||
executor: Optional[Callable[[Dict[str, Any]], Dict[str, Any]]] = None,
|
||||
card_factory: Optional[Callable[[], Dict[str, Any]]] = None,
|
||||
) -> None:
|
||||
self.host = host
|
||||
self.port = port
|
||||
self._server = A2AMTLSServer(cert=cert, key=key, ca=ca, host=host, port=port)
|
||||
self._executor = executor or _default_local_hermes_executor
|
||||
self._card_factory = card_factory or (lambda: _default_agent_card(self.host, self.port))
|
||||
self._tasks: Dict[str, Dict[str, Any]] = {}
|
||||
self._lock = threading.Lock()
|
||||
self._server.add_route("/.well-known/agent-card.json", self._handle_agent_card)
|
||||
self._server.add_route("/agent-card.json", self._handle_agent_card)
|
||||
self._server.add_route("/a2a/rpc", self._handle_rpc)
|
||||
|
||||
def __enter__(self) -> "A2ATaskServer":
|
||||
self.start()
|
||||
return self
|
||||
|
||||
def __exit__(self, *_: Any) -> None:
|
||||
self.stop()
|
||||
|
||||
def start(self) -> None:
|
||||
self._server.start()
|
||||
|
||||
def stop(self) -> None:
|
||||
self._server.stop()
|
||||
|
||||
def _handle_agent_card(self, payload: Dict[str, Any], *, peer_cn: str | None = None) -> Dict[str, Any]:
|
||||
return self._card_factory()
|
||||
|
||||
def _handle_rpc(self, payload: Dict[str, Any], *, peer_cn: str | None = None) -> Dict[str, Any]:
|
||||
req_id = payload.get("id")
|
||||
if payload.get("jsonrpc") != "2.0":
|
||||
return {"jsonrpc": "2.0", "id": req_id, "error": {"code": -32600, "message": "invalid jsonrpc version"}}
|
||||
|
||||
method = payload.get("method")
|
||||
params = payload.get("params") or {}
|
||||
try:
|
||||
if method == "tasks/send":
|
||||
result = self._rpc_send_task(params, peer_cn=peer_cn)
|
||||
elif method == "tasks/get":
|
||||
result = self._rpc_get_task(params)
|
||||
else:
|
||||
return {"jsonrpc": "2.0", "id": req_id, "error": {"code": -32601, "message": f"unknown method: {method}"}}
|
||||
except Exception as exc:
|
||||
logger.exception("A2A task RPC failed: %s", exc)
|
||||
return {"jsonrpc": "2.0", "id": req_id, "error": {"code": -32000, "message": str(exc)}}
|
||||
return {"jsonrpc": "2.0", "id": req_id, "result": result}
|
||||
|
||||
def _rpc_send_task(self, params: Dict[str, Any], *, peer_cn: str | None = None) -> Dict[str, Any]:
|
||||
task_text = str(params.get("task", "")).strip()
|
||||
if not task_text:
|
||||
raise ValueError("task is required")
|
||||
task_id = params.get("taskId") or uuid.uuid4().hex
|
||||
requester = params.get("requester") or peer_cn
|
||||
metadata = dict(params.get("metadata") or {})
|
||||
if peer_cn:
|
||||
metadata.setdefault("peer_cn", peer_cn)
|
||||
record = _build_task_record(task_id, task_text, requester, metadata)
|
||||
with self._lock:
|
||||
self._tasks[task_id] = record
|
||||
worker = threading.Thread(target=self._run_task, args=(task_id,), daemon=True, name=f"a2a-task-{task_id[:8]}")
|
||||
worker.start()
|
||||
return self._copy_task(task_id)
|
||||
|
||||
def _rpc_get_task(self, params: Dict[str, Any]) -> Dict[str, Any]:
|
||||
task_id = str(params.get("taskId", "")).strip()
|
||||
if not task_id:
|
||||
raise ValueError("taskId is required")
|
||||
return self._copy_task(task_id)
|
||||
|
||||
def _copy_task(self, task_id: str) -> Dict[str, Any]:
|
||||
with self._lock:
|
||||
if task_id not in self._tasks:
|
||||
raise KeyError(f"unknown taskId: {task_id}")
|
||||
return json.loads(json.dumps(self._tasks[task_id]))
|
||||
|
||||
def _run_task(self, task_id: str) -> None:
|
||||
with self._lock:
|
||||
task = self._tasks[task_id]
|
||||
task["status"] = _task_status("working", "Task is running")
|
||||
task_payload = {
|
||||
"taskId": task["taskId"],
|
||||
"task": task["task"],
|
||||
"requester": task.get("requester"),
|
||||
"metadata": dict(task.get("metadata") or {}),
|
||||
}
|
||||
try:
|
||||
result = self._executor(task_payload)
|
||||
artifact = _coerce_artifact(result)
|
||||
with self._lock:
|
||||
task = self._tasks[task_id]
|
||||
task["artifacts"] = [artifact]
|
||||
task["status"] = _task_status("completed", "Task completed")
|
||||
except Exception as exc:
|
||||
with self._lock:
|
||||
task = self._tasks[task_id]
|
||||
task["status"] = _task_status("failed", f"Task failed: {exc}")
|
||||
|
||||
|
||||
class A2ATaskClient(A2AMTLSClient):
|
||||
"""Client helper for A2A JSON-RPC task send/get flows."""
|
||||
|
||||
def discover_card(self, base_url: str) -> Dict[str, Any]:
|
||||
return self.get(f"{base_url.rstrip('/')}/.well-known/agent-card.json")
|
||||
|
||||
def _rpc_call(self, base_url: str, method: str, params: Dict[str, Any]) -> Dict[str, Any]:
|
||||
payload = {
|
||||
"jsonrpc": "2.0",
|
||||
"id": uuid.uuid4().hex,
|
||||
"method": method,
|
||||
"params": params,
|
||||
}
|
||||
response = self.post(f"{base_url.rstrip('/')}/a2a/rpc", json=payload)
|
||||
if "error" in response:
|
||||
error = response["error"]
|
||||
raise RuntimeError(error.get("message") or str(error))
|
||||
return response.get("result", {})
|
||||
|
||||
def send_task(
|
||||
self,
|
||||
base_url: str,
|
||||
*,
|
||||
task: str,
|
||||
requester: str | None = None,
|
||||
metadata: Optional[Dict[str, Any]] = None,
|
||||
) -> Dict[str, Any]:
|
||||
return self._rpc_call(
|
||||
base_url,
|
||||
"tasks/send",
|
||||
{
|
||||
"task": task,
|
||||
"requester": requester,
|
||||
"metadata": metadata or {},
|
||||
},
|
||||
)
|
||||
|
||||
def get_task(self, base_url: str, task_id: str) -> Dict[str, Any]:
|
||||
return self._rpc_call(base_url, "tasks/get", {"taskId": task_id})
|
||||
|
||||
def wait_for_task(
|
||||
self,
|
||||
base_url: str,
|
||||
task_id: str,
|
||||
*,
|
||||
timeout: float = 30.0,
|
||||
poll_interval: float = 0.5,
|
||||
) -> Dict[str, Any]:
|
||||
deadline = time.monotonic() + timeout
|
||||
while True:
|
||||
task = self.get_task(base_url, task_id)
|
||||
state = str(((task.get("status") or {}).get("state") or "")).lower()
|
||||
if state in _TERMINAL_TASK_STATES:
|
||||
return task
|
||||
if time.monotonic() >= deadline:
|
||||
raise TimeoutError(f"Timed out waiting for task {task_id}")
|
||||
time.sleep(poll_interval)
|
||||
|
||||
132
hermes_cli/a2a_cmd.py
Normal file
132
hermes_cli/a2a_cmd.py
Normal file
@@ -0,0 +1,132 @@
|
||||
"""CLI helpers for A2A task delegation."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import json
|
||||
import os
|
||||
import re
|
||||
import sys
|
||||
import time
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
from agent.a2a_mtls import A2ATaskClient, A2ATaskServer
|
||||
from hermes_cli.config import get_hermes_home
|
||||
|
||||
|
||||
def _registry_path() -> Path:
|
||||
return get_hermes_home() / "a2a_agents.json"
|
||||
|
||||
|
||||
def _default_identity_paths() -> tuple[str, str, str]:
|
||||
hermes_home = get_hermes_home()
|
||||
agent_name = os.environ.get("HERMES_AGENT_NAME", "hermes").lower()
|
||||
cert = os.environ.get(
|
||||
"HERMES_A2A_CERT",
|
||||
str(hermes_home / "pki" / "agents" / agent_name / f"{agent_name}.crt"),
|
||||
)
|
||||
key = os.environ.get(
|
||||
"HERMES_A2A_KEY",
|
||||
str(hermes_home / "pki" / "agents" / agent_name / f"{agent_name}.key"),
|
||||
)
|
||||
ca = os.environ.get(
|
||||
"HERMES_A2A_CA",
|
||||
str(hermes_home / "pki" / "ca" / "fleet-ca.crt"),
|
||||
)
|
||||
return cert, key, ca
|
||||
|
||||
|
||||
def load_agent_registry(path: Path | None = None) -> dict[str, Any]:
|
||||
registry_path = path or _registry_path()
|
||||
if not registry_path.exists():
|
||||
return {}
|
||||
return json.loads(registry_path.read_text(encoding="utf-8"))
|
||||
|
||||
|
||||
def resolve_agent_url(agent: str, *, registry_path: Path | None = None) -> str:
|
||||
key = re.sub(r"[^A-Za-z0-9]+", "_", agent).upper()
|
||||
env_value = os.getenv(f"HERMES_A2A_{key}_URL")
|
||||
if env_value:
|
||||
return env_value
|
||||
|
||||
registry = load_agent_registry(registry_path)
|
||||
entry = registry.get(agent)
|
||||
if isinstance(entry, str) and entry:
|
||||
return entry
|
||||
if isinstance(entry, dict):
|
||||
url = entry.get("url") or entry.get("base_url") or entry.get("card_url")
|
||||
if url:
|
||||
return str(url)
|
||||
if agent.startswith("https://") or agent.startswith("http://"):
|
||||
return agent
|
||||
raise SystemExit(f"Unknown A2A agent '{agent}'. Set HERMES_A2A_{key}_URL or add it to {_registry_path()}.")
|
||||
|
||||
|
||||
def _print(data: dict[str, Any]) -> None:
|
||||
print(json.dumps(data, indent=2, ensure_ascii=False))
|
||||
|
||||
|
||||
def cmd_send(args) -> None:
|
||||
base_url = args.url or resolve_agent_url(args.agent)
|
||||
cert, key, ca = args.cert, args.key, args.ca
|
||||
if not (cert and key and ca):
|
||||
cert, key, ca = _default_identity_paths()
|
||||
client = A2ATaskClient(cert=cert, key=key, ca=ca)
|
||||
card = client.discover_card(base_url)
|
||||
task = client.send_task(
|
||||
base_url,
|
||||
task=args.task,
|
||||
requester=args.requester,
|
||||
metadata={"agent": args.agent},
|
||||
)
|
||||
if args.wait:
|
||||
task = client.wait_for_task(
|
||||
base_url,
|
||||
task["taskId"],
|
||||
timeout=args.timeout,
|
||||
poll_interval=args.poll_interval,
|
||||
)
|
||||
_print({
|
||||
"agent": args.agent,
|
||||
"url": base_url,
|
||||
"card": card,
|
||||
"task": task,
|
||||
})
|
||||
|
||||
|
||||
def cmd_status(args) -> None:
|
||||
base_url = args.url or resolve_agent_url(args.agent)
|
||||
cert, key, ca = args.cert, args.key, args.ca
|
||||
if not (cert and key and ca):
|
||||
cert, key, ca = _default_identity_paths()
|
||||
client = A2ATaskClient(cert=cert, key=key, ca=ca)
|
||||
task = client.get_task(base_url, args.task_id)
|
||||
_print({"agent": args.agent, "url": base_url, "task": task})
|
||||
|
||||
|
||||
def cmd_serve(args) -> None:
|
||||
cert, key, ca = args.cert, args.key, args.ca
|
||||
if not (cert and key and ca):
|
||||
cert, key, ca = _default_identity_paths()
|
||||
server = A2ATaskServer(cert=cert, key=key, ca=ca, host=args.host, port=args.port)
|
||||
server.start()
|
||||
print(f"A2A task server listening on https://{args.host}:{args.port}")
|
||||
try:
|
||||
while True:
|
||||
time.sleep(1)
|
||||
except KeyboardInterrupt:
|
||||
server.stop()
|
||||
|
||||
|
||||
def cmd_a2a(args) -> None:
|
||||
command = getattr(args, "a2a_command", None) or "send"
|
||||
if command == "send":
|
||||
cmd_send(args)
|
||||
return
|
||||
if command == "status":
|
||||
cmd_status(args)
|
||||
return
|
||||
if command == "serve":
|
||||
cmd_serve(args)
|
||||
return
|
||||
raise SystemExit(f"Unknown a2a command: {command}")
|
||||
@@ -523,7 +523,7 @@ DEFAULT_CONFIG = {
|
||||
|
||||
# Text-to-speech configuration
|
||||
"tts": {
|
||||
"provider": "edge", # "edge" (free) | "elevenlabs" (premium) | "openai" | "minimax" | "mistral" | "neutts" (local) | "kittentts" (local)
|
||||
"provider": "edge", # "edge" (free) | "elevenlabs" (premium) | "openai" | "minimax" | "mistral" | "neutts" (local)
|
||||
"edge": {
|
||||
"voice": "en-US-AriaNeural",
|
||||
# Popular: AriaNeural, JennyNeural, AndrewNeural, BrianNeural, SoniaNeural
|
||||
@@ -547,12 +547,6 @@ DEFAULT_CONFIG = {
|
||||
"model": "neuphonic/neutts-air-q4-gguf", # HuggingFace model repo
|
||||
"device": "cpu", # cpu, cuda, or mps
|
||||
},
|
||||
"kittentts": {
|
||||
"model": "KittenML/kitten-tts-nano-0.8-int8", # 25MB int8 default
|
||||
"voice": "Jasper", # Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo
|
||||
"speed": 1.0,
|
||||
"clean_text": True,
|
||||
},
|
||||
},
|
||||
|
||||
"stt": {
|
||||
|
||||
@@ -173,6 +173,13 @@ from hermes_constants import OPENROUTER_BASE_URL
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
def cmd_a2a(args):
|
||||
"""Dispatch A2A CLI subcommands lazily to avoid heavy imports at startup."""
|
||||
from hermes_cli.a2a_cmd import cmd_a2a as _cmd_a2a
|
||||
|
||||
return _cmd_a2a(args)
|
||||
|
||||
|
||||
def _relative_time(ts) -> str:
|
||||
"""Format a timestamp as relative time (e.g., '2h ago', 'yesterday')."""
|
||||
if not ts:
|
||||
@@ -4781,6 +4788,45 @@ For more help on a command:
|
||||
|
||||
gateway_parser.set_defaults(func=cmd_gateway)
|
||||
|
||||
# =========================================================================
|
||||
# a2a command
|
||||
# =========================================================================
|
||||
a2a_parser = subparsers.add_parser(
|
||||
"a2a",
|
||||
help="A2A task delegation over mutual TLS",
|
||||
description="Send, inspect, and serve structured A2A tasks between Hermes agents",
|
||||
)
|
||||
a2a_subparsers = a2a_parser.add_subparsers(dest="a2a_command")
|
||||
|
||||
a2a_send = a2a_subparsers.add_parser("send", help="Send an A2A task to another agent")
|
||||
a2a_send.add_argument("--agent", required=True, help="Agent alias or URL (for example: allegro)")
|
||||
a2a_send.add_argument("--task", required=True, help="Task text to delegate")
|
||||
a2a_send.add_argument("--url", help="Explicit base URL for the remote agent")
|
||||
a2a_send.add_argument("--requester", default=None, help="Requester label included in task metadata")
|
||||
a2a_send.add_argument("--wait", action="store_true", help="Poll until the task reaches a terminal state")
|
||||
a2a_send.add_argument("--timeout", type=float, default=30.0, help="Wait timeout in seconds (default: 30)")
|
||||
a2a_send.add_argument("--poll-interval", type=float, default=0.5, help="Polling interval in seconds while waiting (default: 0.5)")
|
||||
a2a_send.add_argument("--cert", default=None, help="Client certificate path (defaults from HERMES_A2A_CERT)")
|
||||
a2a_send.add_argument("--key", default=None, help="Client private key path (defaults from HERMES_A2A_KEY)")
|
||||
a2a_send.add_argument("--ca", default=None, help="Fleet CA certificate path (defaults from HERMES_A2A_CA)")
|
||||
|
||||
a2a_status = a2a_subparsers.add_parser("status", help="Fetch the current status of an A2A task")
|
||||
a2a_status.add_argument("--agent", required=True, help="Agent alias or URL (for example: allegro)")
|
||||
a2a_status.add_argument("--task-id", required=True, help="Task identifier returned by a2a send")
|
||||
a2a_status.add_argument("--url", help="Explicit base URL for the remote agent")
|
||||
a2a_status.add_argument("--cert", default=None, help="Client certificate path (defaults from HERMES_A2A_CERT)")
|
||||
a2a_status.add_argument("--key", default=None, help="Client private key path (defaults from HERMES_A2A_KEY)")
|
||||
a2a_status.add_argument("--ca", default=None, help="Fleet CA certificate path (defaults from HERMES_A2A_CA)")
|
||||
|
||||
a2a_serve = a2a_subparsers.add_parser("serve", help="Run the local A2A task server")
|
||||
a2a_serve.add_argument("--host", default=os.environ.get("HERMES_A2A_HOST", "127.0.0.1"), help="Bind host (default: HERMES_A2A_HOST or 127.0.0.1)")
|
||||
a2a_serve.add_argument("--port", type=int, default=int(os.environ.get("HERMES_A2A_PORT", "9443")), help="Bind port (default: HERMES_A2A_PORT or 9443)")
|
||||
a2a_serve.add_argument("--cert", default=None, help="Server certificate path (defaults from HERMES_A2A_CERT)")
|
||||
a2a_serve.add_argument("--key", default=None, help="Server private key path (defaults from HERMES_A2A_KEY)")
|
||||
a2a_serve.add_argument("--ca", default=None, help="Fleet CA certificate path (defaults from HERMES_A2A_CA)")
|
||||
|
||||
a2a_parser.set_defaults(func=cmd_a2a)
|
||||
|
||||
# =========================================================================
|
||||
# setup command
|
||||
# =========================================================================
|
||||
|
||||
@@ -443,16 +443,6 @@ def _print_setup_summary(config: dict, hermes_home):
|
||||
tool_status.append(("Text-to-Speech (NeuTTS local)", True, None))
|
||||
else:
|
||||
tool_status.append(("Text-to-Speech (NeuTTS — not installed)", False, "run 'hermes setup tts'"))
|
||||
elif tts_provider == "kittentts":
|
||||
try:
|
||||
import importlib.util
|
||||
kittentts_ok = importlib.util.find_spec("kittentts") is not None
|
||||
except Exception:
|
||||
kittentts_ok = False
|
||||
if kittentts_ok:
|
||||
tool_status.append(("Text-to-Speech (KittenTTS local)", True, None))
|
||||
else:
|
||||
tool_status.append(("Text-to-Speech (KittenTTS — not installed)", False, "run 'hermes setup tts'"))
|
||||
else:
|
||||
tool_status.append(("Text-to-Speech (Edge TTS)", True, None))
|
||||
|
||||
@@ -901,7 +891,6 @@ def _install_neutts_deps() -> bool:
|
||||
return False
|
||||
else:
|
||||
print_warning("espeak-ng is required for NeuTTS. Install it manually before using NeuTTS.")
|
||||
return False
|
||||
|
||||
# Install neutts Python package
|
||||
print()
|
||||
@@ -921,34 +910,8 @@ def _install_neutts_deps() -> bool:
|
||||
return False
|
||||
|
||||
|
||||
def _install_kittentts_deps() -> bool:
|
||||
"""Install KittenTTS dependencies with user approval. Returns True on success."""
|
||||
import subprocess
|
||||
import sys
|
||||
|
||||
wheel_url = (
|
||||
"https://github.com/KittenML/KittenTTS/releases/download/"
|
||||
"0.8.1/kittentts-0.8.1-py3-none-any.whl"
|
||||
)
|
||||
print()
|
||||
print_info("Installing kittentts Python package (~25-80MB model downloaded on first use)...")
|
||||
print()
|
||||
try:
|
||||
subprocess.run(
|
||||
[sys.executable, "-m", "pip", "install", "-U", wheel_url, "soundfile", "--quiet"],
|
||||
check=True, timeout=300,
|
||||
)
|
||||
print_success("kittentts installed successfully")
|
||||
return True
|
||||
except (subprocess.CalledProcessError, subprocess.TimeoutExpired) as e:
|
||||
print_error(f"Failed to install kittentts: {e}")
|
||||
print_info(f"Try manually: python -m pip install -U '{wheel_url}' soundfile")
|
||||
return False
|
||||
|
||||
|
||||
def _setup_tts_provider(config: dict):
|
||||
"""Interactive TTS provider selection with install flow for local providers."""
|
||||
|
||||
"""Interactive TTS provider selection with install flow for NeuTTS."""
|
||||
tts_config = config.get("tts", {})
|
||||
current_provider = tts_config.get("provider", "edge")
|
||||
subscription_features = get_nous_subscription_features(config)
|
||||
@@ -960,7 +923,6 @@ def _setup_tts_provider(config: dict):
|
||||
"minimax": "MiniMax TTS",
|
||||
"mistral": "Mistral Voxtral TTS",
|
||||
"neutts": "NeuTTS",
|
||||
"kittentts": "KittenTTS",
|
||||
}
|
||||
current_label = provider_labels.get(current_provider, current_provider)
|
||||
|
||||
@@ -982,10 +944,9 @@ def _setup_tts_provider(config: dict):
|
||||
"MiniMax TTS (high quality with voice cloning, needs API key)",
|
||||
"Mistral Voxtral TTS (multilingual, native Opus, needs API key)",
|
||||
"NeuTTS (local on-device, free, ~300MB model download)",
|
||||
"KittenTTS (local on-device, free, lightweight ~25-80MB ONNX)",
|
||||
]
|
||||
)
|
||||
providers.extend(["edge", "elevenlabs", "openai", "minimax", "mistral", "neutts", "kittentts"])
|
||||
providers.extend(["edge", "elevenlabs", "openai", "minimax", "mistral", "neutts"])
|
||||
choices.append(f"Keep current ({current_label})")
|
||||
keep_current_idx = len(choices) - 1
|
||||
idx = prompt_choice("Select TTS provider:", choices, keep_current_idx)
|
||||
@@ -1027,28 +988,6 @@ def _setup_tts_provider(config: dict):
|
||||
print_info("Skipping install. Set tts.provider to 'neutts' after installing manually.")
|
||||
selected = "edge"
|
||||
|
||||
elif selected == "kittentts":
|
||||
try:
|
||||
import importlib.util
|
||||
already_installed = importlib.util.find_spec("kittentts") is not None
|
||||
except Exception:
|
||||
already_installed = False
|
||||
|
||||
if already_installed:
|
||||
print_success("KittenTTS is already installed")
|
||||
else:
|
||||
print()
|
||||
print_info("KittenTTS is lightweight (~25-80MB, CPU-only, no API key required).")
|
||||
print_info("Voices: Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo")
|
||||
print()
|
||||
if prompt_yes_no("Install KittenTTS now?", True):
|
||||
if not _install_kittentts_deps():
|
||||
print_warning("KittenTTS installation incomplete. Falling back to Edge TTS.")
|
||||
selected = "edge"
|
||||
else:
|
||||
print_info("Skipping install. Set tts.provider to 'kittentts' after installing manually.")
|
||||
selected = "edge"
|
||||
|
||||
elif selected == "elevenlabs":
|
||||
existing = get_env_value("ELEVENLABS_API_KEY")
|
||||
if not existing:
|
||||
|
||||
@@ -164,14 +164,6 @@ TOOL_CATEGORIES = {
|
||||
],
|
||||
"tts_provider": "mistral",
|
||||
},
|
||||
{
|
||||
"name": "KittenTTS",
|
||||
"badge": "local · free",
|
||||
"tag": "Lightweight local ONNX TTS (~25MB), no API key",
|
||||
"env_vars": [],
|
||||
"tts_provider": "kittentts",
|
||||
"post_setup": "kittentts",
|
||||
},
|
||||
],
|
||||
},
|
||||
"web": {
|
||||
@@ -411,36 +403,6 @@ def _run_post_setup(post_setup_key: str):
|
||||
_print_warning(" Node.js not found. Install Camofox via Docker:")
|
||||
_print_info(" docker run -p 9377:9377 -e CAMOFOX_PORT=9377 jo-inc/camofox-browser")
|
||||
|
||||
elif post_setup_key == "kittentts":
|
||||
try:
|
||||
__import__("kittentts")
|
||||
_print_success(" kittentts is already installed")
|
||||
return
|
||||
except ImportError:
|
||||
pass
|
||||
import subprocess
|
||||
_print_info(" Installing kittentts (~25-80MB model, CPU-only)...")
|
||||
wheel_url = (
|
||||
"https://github.com/KittenML/KittenTTS/releases/download/"
|
||||
"0.8.1/kittentts-0.8.1-py3-none-any.whl"
|
||||
)
|
||||
try:
|
||||
result = subprocess.run(
|
||||
[sys.executable, "-m", "pip", "install", "-U", wheel_url, "soundfile", "--quiet"],
|
||||
capture_output=True, text=True, timeout=300,
|
||||
)
|
||||
if result.returncode == 0:
|
||||
_print_success(" kittentts installed")
|
||||
_print_info(" Voices: Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo")
|
||||
_print_info(" Models: KittenML/kitten-tts-nano-0.8-int8 (25MB), micro (41MB), mini (80MB)")
|
||||
else:
|
||||
_print_warning(" kittentts install failed:")
|
||||
_print_info(f" {result.stderr.strip()[:300]}")
|
||||
_print_info(f" Run manually: python -m pip install -U '{wheel_url}' soundfile")
|
||||
except subprocess.TimeoutExpired:
|
||||
_print_warning(" kittentts install timed out (>5min)")
|
||||
_print_info(f" Run manually: python -m pip install -U '{wheel_url}' soundfile")
|
||||
|
||||
elif post_setup_key == "rl_training":
|
||||
try:
|
||||
__import__("tinker_atropos")
|
||||
|
||||
@@ -572,3 +572,94 @@ class TestA2AMTLSServerAndClient:
|
||||
|
||||
assert not errors, f"Concurrent connection errors: {errors}"
|
||||
assert len(results) == 3
|
||||
|
||||
|
||||
@_requires_crypto
|
||||
class TestA2ATaskServerAndClient:
|
||||
"""Structured A2A task send/get flow over mTLS."""
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def _pki(self, tmp_path):
|
||||
ca_dir = tmp_path / "ca"
|
||||
ca_dir.mkdir()
|
||||
self.ca_crt, self.ca_key = _make_ca_keypair(ca_dir)
|
||||
agent_dir = tmp_path / "agents"
|
||||
agent_dir.mkdir()
|
||||
self.srv_crt, self.srv_key = _make_agent_keypair(
|
||||
agent_dir, "timmy", self.ca_crt, self.ca_key
|
||||
)
|
||||
self.cli_crt, self.cli_key = _make_agent_keypair(
|
||||
agent_dir, "allegro", self.ca_crt, self.ca_key
|
||||
)
|
||||
|
||||
@pytest.fixture()
|
||||
def task_server(self):
|
||||
from agent.a2a_mtls import A2ATaskServer
|
||||
|
||||
gate = threading.Event()
|
||||
|
||||
def analyze_executor(task: dict[str, object]) -> dict[str, object]:
|
||||
gate.wait(timeout=2)
|
||||
text = str(task.get("task", ""))
|
||||
return {
|
||||
"text": f"analysis:{text}",
|
||||
"metadata": {"tool": "local-hermes-stub"},
|
||||
}
|
||||
|
||||
port = _find_free_port()
|
||||
server = A2ATaskServer(
|
||||
cert=self.srv_crt,
|
||||
key=self.srv_key,
|
||||
ca=self.ca_crt,
|
||||
host="127.0.0.1",
|
||||
port=port,
|
||||
executor=analyze_executor,
|
||||
)
|
||||
with server:
|
||||
time.sleep(0.1)
|
||||
yield server, port, gate
|
||||
|
||||
def test_task_send_get_and_completion_flow(self, task_server):
|
||||
from agent.a2a_mtls import A2ATaskClient
|
||||
|
||||
server, port, gate = task_server
|
||||
client = A2ATaskClient(cert=self.cli_crt, key=self.cli_key, ca=self.ca_crt)
|
||||
base_url = f"https://127.0.0.1:{port}"
|
||||
|
||||
card = client.discover_card(base_url)
|
||||
assert card["name"]
|
||||
|
||||
submitted = client.send_task(base_url, task="Analyze README.md", requester="timmy")
|
||||
assert submitted["status"]["state"] in {"submitted", "working"}
|
||||
|
||||
in_flight = client.get_task(base_url, submitted["taskId"])
|
||||
assert in_flight["status"]["state"] in {"submitted", "working"}
|
||||
|
||||
gate.set()
|
||||
completed = client.wait_for_task(base_url, submitted["taskId"], timeout=5.0, poll_interval=0.05)
|
||||
assert completed["status"]["state"] == "completed"
|
||||
assert completed["artifacts"][0]["text"] == "analysis:Analyze README.md"
|
||||
|
||||
def test_failed_executor_marks_task_failed(self):
|
||||
from agent.a2a_mtls import A2ATaskClient, A2ATaskServer
|
||||
|
||||
def failing_executor(task: dict[str, object]) -> dict[str, object]:
|
||||
raise RuntimeError("boom")
|
||||
|
||||
port = _find_free_port()
|
||||
server = A2ATaskServer(
|
||||
cert=self.srv_crt,
|
||||
key=self.srv_key,
|
||||
ca=self.ca_crt,
|
||||
host="127.0.0.1",
|
||||
port=port,
|
||||
executor=failing_executor,
|
||||
)
|
||||
with server:
|
||||
time.sleep(0.1)
|
||||
client = A2ATaskClient(cert=self.cli_crt, key=self.cli_key, ca=self.ca_crt)
|
||||
base_url = f"https://127.0.0.1:{port}"
|
||||
submitted = client.send_task(base_url, task="explode", requester="timmy")
|
||||
failed = client.wait_for_task(base_url, submitted["taskId"], timeout=5.0, poll_interval=0.05)
|
||||
assert failed["status"]["state"] == "failed"
|
||||
assert "boom" in failed["status"]["message"]
|
||||
|
||||
95
tests/hermes_cli/test_a2a_cmd.py
Normal file
95
tests/hermes_cli/test_a2a_cmd.py
Normal file
@@ -0,0 +1,95 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import argparse
|
||||
import json
|
||||
from pathlib import Path
|
||||
from unittest.mock import patch
|
||||
|
||||
import pytest
|
||||
|
||||
|
||||
def test_cmd_send_uses_registry_and_waits_for_terminal_task(tmp_path, monkeypatch, capsys):
|
||||
hermes_home = tmp_path / ".hermes"
|
||||
hermes_home.mkdir()
|
||||
(hermes_home / "a2a_agents.json").write_text(
|
||||
json.dumps({"allegro": {"url": "https://127.0.0.1:9443"}}),
|
||||
encoding="utf-8",
|
||||
)
|
||||
monkeypatch.setenv("HERMES_HOME", str(hermes_home))
|
||||
|
||||
from hermes_cli.a2a_cmd import cmd_a2a
|
||||
|
||||
class FakeClient:
|
||||
def __init__(self, **kwargs):
|
||||
self.kwargs = kwargs
|
||||
|
||||
def discover_card(self, base_url: str):
|
||||
assert base_url == "https://127.0.0.1:9443"
|
||||
return {"name": "allegro", "url": base_url}
|
||||
|
||||
def send_task(self, base_url: str, *, task: str, requester: str | None = None, metadata=None):
|
||||
assert task == "analyze README"
|
||||
return {"taskId": "task-123", "status": {"state": "submitted"}}
|
||||
|
||||
def wait_for_task(self, base_url: str, task_id: str, *, timeout: float, poll_interval: float):
|
||||
assert task_id == "task-123"
|
||||
return {
|
||||
"taskId": task_id,
|
||||
"status": {"state": "completed"},
|
||||
"artifacts": [{"text": "README looks healthy"}],
|
||||
}
|
||||
|
||||
args = argparse.Namespace(
|
||||
a2a_command="send",
|
||||
agent="allegro",
|
||||
task="analyze README",
|
||||
url=None,
|
||||
wait=True,
|
||||
timeout=5.0,
|
||||
poll_interval=0.01,
|
||||
requester="timmy",
|
||||
cert="cert.pem",
|
||||
key="key.pem",
|
||||
ca="ca.pem",
|
||||
)
|
||||
|
||||
with patch("hermes_cli.a2a_cmd.A2ATaskClient", FakeClient):
|
||||
cmd_a2a(args)
|
||||
|
||||
result = json.loads(capsys.readouterr().out)
|
||||
assert result["agent"] == "allegro"
|
||||
assert result["card"]["name"] == "allegro"
|
||||
assert result["task"]["status"]["state"] == "completed"
|
||||
assert result["task"]["artifacts"][0]["text"] == "README looks healthy"
|
||||
|
||||
|
||||
def test_resolve_agent_url_supports_env_override(monkeypatch):
|
||||
monkeypatch.setenv("HERMES_A2A_ALLEGRO_URL", "https://fleet-allegro:9443")
|
||||
from hermes_cli.a2a_cmd import resolve_agent_url
|
||||
|
||||
assert resolve_agent_url("allegro") == "https://fleet-allegro:9443"
|
||||
|
||||
|
||||
def test_cmd_send_requires_known_agent(tmp_path, monkeypatch):
|
||||
hermes_home = tmp_path / ".hermes"
|
||||
hermes_home.mkdir()
|
||||
monkeypatch.setenv("HERMES_HOME", str(hermes_home))
|
||||
|
||||
from hermes_cli.a2a_cmd import cmd_a2a
|
||||
|
||||
args = argparse.Namespace(
|
||||
a2a_command="send",
|
||||
agent="unknown",
|
||||
task="do work",
|
||||
url=None,
|
||||
wait=False,
|
||||
timeout=5.0,
|
||||
poll_interval=0.05,
|
||||
requester=None,
|
||||
cert="cert.pem",
|
||||
key="key.pem",
|
||||
ca="ca.pem",
|
||||
)
|
||||
|
||||
with pytest.raises(SystemExit):
|
||||
cmd_a2a(args)
|
||||
@@ -1,236 +0,0 @@
|
||||
"""Tests for the KittenTTS local provider in tools/tts_tool.py."""
|
||||
|
||||
import json
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
import numpy as np
|
||||
import pytest
|
||||
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def clean_env(monkeypatch):
|
||||
for key in ("HERMES_SESSION_PLATFORM",):
|
||||
monkeypatch.delenv(key, raising=False)
|
||||
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def clear_kittentts_cache():
|
||||
"""Reset the module-level model cache between tests."""
|
||||
from tools import tts_tool as _tt
|
||||
_tt._kittentts_model_cache.clear()
|
||||
yield
|
||||
_tt._kittentts_model_cache.clear()
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def mock_kittentts_module():
|
||||
"""Inject a fake kittentts + soundfile module that return stub objects."""
|
||||
fake_model = MagicMock()
|
||||
# 24kHz float32 PCM at ~2s of silence
|
||||
fake_model.generate.return_value = np.zeros(48000, dtype=np.float32)
|
||||
fake_cls = MagicMock(return_value=fake_model)
|
||||
fake_kittentts = MagicMock()
|
||||
fake_kittentts.KittenTTS = fake_cls
|
||||
|
||||
# Stub soundfile — the real package isn't installed in CI venv, and
|
||||
# _generate_kittentts does `import soundfile as sf` at runtime.
|
||||
fake_sf = MagicMock()
|
||||
|
||||
def _fake_write(path, audio, samplerate):
|
||||
# Emulate writing a real file so downstream path checks succeed.
|
||||
import pathlib
|
||||
|
||||
pathlib.Path(path).write_bytes(b"RIFF\x00\x00\x00\x00WAVEfmt fake")
|
||||
|
||||
fake_sf.write = _fake_write
|
||||
|
||||
with patch.dict(
|
||||
"sys.modules",
|
||||
{"kittentts": fake_kittentts, "soundfile": fake_sf},
|
||||
):
|
||||
yield fake_model, fake_cls
|
||||
|
||||
|
||||
class TestGenerateKittenTts:
|
||||
def test_successful_wav_generation(self, tmp_path, mock_kittentts_module):
|
||||
from tools.tts_tool import _generate_kittentts
|
||||
|
||||
fake_model, fake_cls = mock_kittentts_module
|
||||
output_path = str(tmp_path / "test.wav")
|
||||
result = _generate_kittentts("Hello world", output_path, {})
|
||||
|
||||
assert result == output_path
|
||||
assert (tmp_path / "test.wav").exists()
|
||||
fake_cls.assert_called_once()
|
||||
fake_model.generate.assert_called_once()
|
||||
|
||||
def test_config_passes_voice_speed_cleantext(self, tmp_path, mock_kittentts_module):
|
||||
from tools.tts_tool import _generate_kittentts
|
||||
|
||||
fake_model, _ = mock_kittentts_module
|
||||
config = {
|
||||
"kittentts": {
|
||||
"model": "KittenML/kitten-tts-mini-0.8",
|
||||
"voice": "Luna",
|
||||
"speed": 1.25,
|
||||
"clean_text": False,
|
||||
}
|
||||
}
|
||||
_generate_kittentts("Hi there", str(tmp_path / "out.wav"), config)
|
||||
|
||||
call_kwargs = fake_model.generate.call_args.kwargs
|
||||
assert call_kwargs["voice"] == "Luna"
|
||||
assert call_kwargs["speed"] == 1.25
|
||||
assert call_kwargs["clean_text"] is False
|
||||
|
||||
def test_default_model_and_voice(self, tmp_path, mock_kittentts_module):
|
||||
from tools.tts_tool import (
|
||||
DEFAULT_KITTENTTS_MODEL,
|
||||
DEFAULT_KITTENTTS_VOICE,
|
||||
_generate_kittentts,
|
||||
)
|
||||
|
||||
fake_model, fake_cls = mock_kittentts_module
|
||||
_generate_kittentts("Hi", str(tmp_path / "out.wav"), {})
|
||||
|
||||
fake_cls.assert_called_once_with(DEFAULT_KITTENTTS_MODEL)
|
||||
assert fake_model.generate.call_args.kwargs["voice"] == DEFAULT_KITTENTTS_VOICE
|
||||
|
||||
def test_model_is_cached_across_calls(self, tmp_path, mock_kittentts_module):
|
||||
from tools.tts_tool import _generate_kittentts
|
||||
|
||||
_, fake_cls = mock_kittentts_module
|
||||
_generate_kittentts("One", str(tmp_path / "a.wav"), {})
|
||||
_generate_kittentts("Two", str(tmp_path / "b.wav"), {})
|
||||
|
||||
# Same model name → class instantiated exactly once
|
||||
assert fake_cls.call_count == 1
|
||||
|
||||
def test_different_models_are_cached_separately(self, tmp_path, mock_kittentts_module):
|
||||
from tools.tts_tool import _generate_kittentts
|
||||
|
||||
_, fake_cls = mock_kittentts_module
|
||||
_generate_kittentts(
|
||||
"A",
|
||||
str(tmp_path / "a.wav"),
|
||||
{"kittentts": {"model": "KittenML/kitten-tts-nano-0.8-int8"}},
|
||||
)
|
||||
_generate_kittentts(
|
||||
"B",
|
||||
str(tmp_path / "b.wav"),
|
||||
{"kittentts": {"model": "KittenML/kitten-tts-mini-0.8"}},
|
||||
)
|
||||
|
||||
assert fake_cls.call_count == 2
|
||||
|
||||
def test_non_wav_extension_triggers_ffmpeg_conversion(
|
||||
self, tmp_path, mock_kittentts_module, monkeypatch
|
||||
):
|
||||
"""Non-.wav output path causes WAV → target ffmpeg conversion."""
|
||||
from tools import tts_tool as _tt
|
||||
|
||||
calls = []
|
||||
|
||||
def fake_shutil_which(cmd):
|
||||
return "/usr/bin/ffmpeg" if cmd == "ffmpeg" else None
|
||||
|
||||
def fake_run(cmd, check=False, timeout=None, **kw):
|
||||
calls.append(cmd)
|
||||
# Emulate ffmpeg writing the output file
|
||||
import pathlib
|
||||
|
||||
out_path = cmd[-1]
|
||||
pathlib.Path(out_path).write_bytes(b"fake-mp3-data")
|
||||
return MagicMock(returncode=0)
|
||||
|
||||
monkeypatch.setattr(_tt.shutil, "which", fake_shutil_which)
|
||||
monkeypatch.setattr(_tt.subprocess, "run", fake_run)
|
||||
|
||||
output_path = str(tmp_path / "test.mp3")
|
||||
result = _tt._generate_kittentts("Hi", output_path, {})
|
||||
|
||||
assert result == output_path
|
||||
assert len(calls) == 1
|
||||
assert calls[0][0] == "/usr/bin/ffmpeg"
|
||||
|
||||
def test_missing_kittentts_raises_import_error(self, tmp_path, monkeypatch):
|
||||
"""When kittentts package is not installed, _import_kittentts raises."""
|
||||
import sys
|
||||
|
||||
monkeypatch.setitem(sys.modules, "kittentts", None)
|
||||
from tools.tts_tool import _generate_kittentts
|
||||
|
||||
with pytest.raises((ImportError, TypeError)):
|
||||
_generate_kittentts("Hi", str(tmp_path / "out.wav"), {})
|
||||
|
||||
|
||||
class TestCheckKittenttsAvailable:
|
||||
def test_reports_available_when_package_present(self, monkeypatch):
|
||||
import importlib.util
|
||||
from tools.tts_tool import _check_kittentts_available
|
||||
|
||||
fake_spec = MagicMock()
|
||||
monkeypatch.setattr(
|
||||
importlib.util,
|
||||
"find_spec",
|
||||
lambda name: fake_spec if name == "kittentts" else None,
|
||||
)
|
||||
assert _check_kittentts_available() is True
|
||||
|
||||
def test_reports_unavailable_when_package_missing(self, monkeypatch):
|
||||
import importlib.util
|
||||
from tools.tts_tool import _check_kittentts_available
|
||||
|
||||
monkeypatch.setattr(importlib.util, "find_spec", lambda name: None)
|
||||
assert _check_kittentts_available() is False
|
||||
|
||||
|
||||
class TestDispatcherBranch:
|
||||
def test_kittentts_not_installed_returns_helpful_error(self, monkeypatch, tmp_path):
|
||||
"""When provider=kittentts but package missing, return JSON error with setup hint."""
|
||||
import sys
|
||||
|
||||
monkeypatch.setitem(sys.modules, "kittentts", None)
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
|
||||
|
||||
from tools.tts_tool import text_to_speech_tool
|
||||
|
||||
# Write a config telling it to use kittentts
|
||||
import yaml
|
||||
|
||||
(tmp_path / "config.yaml").write_text(
|
||||
yaml.safe_dump({"tts": {"provider": "kittentts"}})
|
||||
)
|
||||
|
||||
result = json.loads(text_to_speech_tool(text="Hello"))
|
||||
assert result["success"] is False
|
||||
assert "kittentts" in result["error"].lower()
|
||||
assert "hermes setup tts" in result["error"].lower()
|
||||
|
||||
def test_non_telegram_explicit_wav_path_is_preserved(
|
||||
self, monkeypatch, tmp_path, mock_kittentts_module
|
||||
):
|
||||
"""Explicit WAV outputs should stay WAV outside Telegram sessions."""
|
||||
import yaml
|
||||
from tools import tts_tool as _tt
|
||||
|
||||
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
|
||||
(tmp_path / "config.yaml").write_text(
|
||||
yaml.safe_dump({"tts": {"provider": "kittentts"}})
|
||||
)
|
||||
|
||||
def fail_convert(_path):
|
||||
raise AssertionError("_convert_to_opus should not run outside Telegram")
|
||||
|
||||
monkeypatch.setattr(_tt, "_convert_to_opus", fail_convert)
|
||||
|
||||
result = json.loads(
|
||||
_tt.text_to_speech_tool(
|
||||
text="Hello from KittenTTS",
|
||||
output_path=str(tmp_path / "out.wav"),
|
||||
)
|
||||
)
|
||||
|
||||
assert result["success"] is True
|
||||
assert result["file_path"] == str(tmp_path / "out.wav")
|
||||
assert (tmp_path / "out.wav").exists()
|
||||
@@ -2,14 +2,13 @@
|
||||
"""
|
||||
Text-to-Speech Tool Module
|
||||
|
||||
Supports seven TTS providers:
|
||||
Supports six TTS providers:
|
||||
- Edge TTS (default, free, no API key): Microsoft Edge neural voices
|
||||
- ElevenLabs (premium): High-quality voices, needs ELEVENLABS_API_KEY
|
||||
- OpenAI TTS: Good quality, needs OPENAI_API_KEY
|
||||
- MiniMax TTS: High-quality with voice cloning, needs MINIMAX_API_KEY
|
||||
- Mistral (Voxtral TTS): Multilingual, native Opus, needs MISTRAL_API_KEY
|
||||
- NeuTTS (local, free, no API key): On-device TTS via neutts_cli, needs neutts installed
|
||||
- KittenTTS (local, free, no API key): Lightweight on-device ONNX TTS via kittentts
|
||||
|
||||
Output formats:
|
||||
- Opus (.ogg) for Telegram voice bubbles (requires ffmpeg for Edge TTS)
|
||||
@@ -78,12 +77,6 @@ def _import_sounddevice():
|
||||
return sd
|
||||
|
||||
|
||||
def _import_kittentts():
|
||||
"""Lazy import KittenTTS. Returns the class or raises ImportError."""
|
||||
from kittentts import KittenTTS
|
||||
return KittenTTS
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# Defaults
|
||||
# ===========================================================================
|
||||
@@ -93,8 +86,6 @@ DEFAULT_ELEVENLABS_VOICE_ID = "pNInz6obpgDQGcFmaJgB" # Adam
|
||||
DEFAULT_ELEVENLABS_MODEL_ID = "eleven_multilingual_v2"
|
||||
DEFAULT_ELEVENLABS_STREAMING_MODEL_ID = "eleven_flash_v2_5"
|
||||
DEFAULT_OPENAI_MODEL = "gpt-4o-mini-tts"
|
||||
DEFAULT_KITTENTTS_MODEL = "KittenML/kitten-tts-nano-0.8-int8" # 25MB
|
||||
DEFAULT_KITTENTTS_VOICE = "Jasper"
|
||||
DEFAULT_OPENAI_VOICE = "alloy"
|
||||
DEFAULT_OPENAI_BASE_URL = "https://api.openai.com/v1"
|
||||
DEFAULT_MINIMAX_MODEL = "speech-2.8-hd"
|
||||
@@ -457,15 +448,6 @@ def _check_neutts_available() -> bool:
|
||||
return False
|
||||
|
||||
|
||||
def _check_kittentts_available() -> bool:
|
||||
"""Check if the kittentts engine is importable (installed locally)."""
|
||||
try:
|
||||
import importlib.util
|
||||
return importlib.util.find_spec("kittentts") is not None
|
||||
except Exception:
|
||||
return False
|
||||
|
||||
|
||||
def _default_neutts_ref_audio() -> str:
|
||||
"""Return path to the bundled default voice reference audio."""
|
||||
return str(Path(__file__).parent / "neutts_samples" / "jo.wav")
|
||||
@@ -529,51 +511,6 @@ def _generate_neutts(text: str, output_path: str, tts_config: Dict[str, Any]) ->
|
||||
return output_path
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# Provider: KittenTTS (local, lightweight)
|
||||
# ===========================================================================
|
||||
|
||||
# Module-level cache for KittenTTS model instances
|
||||
_kittentts_model_cache: Dict[str, Any] = {}
|
||||
|
||||
|
||||
def _generate_kittentts(text: str, output_path: str, tts_config: Dict[str, Any]) -> str:
|
||||
"""Generate speech using the local KittenTTS ONNX model."""
|
||||
KittenTTS = _import_kittentts()
|
||||
kt_config = tts_config.get("kittentts", {})
|
||||
model_name = kt_config.get("model", DEFAULT_KITTENTTS_MODEL)
|
||||
voice = kt_config.get("voice", DEFAULT_KITTENTTS_VOICE)
|
||||
speed = kt_config.get("speed", 1.0)
|
||||
clean_text = kt_config.get("clean_text", True)
|
||||
|
||||
global _kittentts_model_cache
|
||||
if model_name not in _kittentts_model_cache:
|
||||
logger.info("[KittenTTS] Loading model: %s", model_name)
|
||||
_kittentts_model_cache[model_name] = KittenTTS(model_name)
|
||||
|
||||
model = _kittentts_model_cache[model_name]
|
||||
audio = model.generate(text, voice=voice, speed=speed, clean_text=clean_text)
|
||||
|
||||
import soundfile as sf
|
||||
|
||||
wav_path = output_path
|
||||
if not output_path.endswith(".wav"):
|
||||
wav_path = output_path.rsplit(".", 1)[0] + ".wav"
|
||||
|
||||
sf.write(wav_path, audio, 24000)
|
||||
|
||||
if wav_path != output_path:
|
||||
ffmpeg = shutil.which("ffmpeg")
|
||||
if ffmpeg:
|
||||
conv_cmd = [ffmpeg, "-i", wav_path, "-y", "-loglevel", "error", output_path]
|
||||
subprocess.run(conv_cmd, check=True, timeout=30)
|
||||
os.remove(wav_path)
|
||||
else:
|
||||
os.rename(wav_path, output_path)
|
||||
|
||||
return output_path
|
||||
|
||||
|
||||
# ===========================================================================
|
||||
# Main tool function
|
||||
# ===========================================================================
|
||||
@@ -685,19 +622,6 @@ def text_to_speech_tool(
|
||||
logger.info("Generating speech with NeuTTS (local)...")
|
||||
_generate_neutts(text, file_str, tts_config)
|
||||
|
||||
elif provider == "kittentts":
|
||||
try:
|
||||
_import_kittentts()
|
||||
except ImportError:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "KittenTTS provider selected but 'kittentts' package not installed. "
|
||||
"Run 'hermes setup tts' and choose KittenTTS, or install manually: "
|
||||
"pip install https://github.com/KittenML/KittenTTS/releases/download/0.8.1/kittentts-0.8.1-py3-none-any.whl"
|
||||
}, ensure_ascii=False)
|
||||
logger.info("Generating speech with KittenTTS (local, lightweight)...")
|
||||
_generate_kittentts(text, file_str, tts_config)
|
||||
|
||||
else:
|
||||
# Default: Edge TTS (free), with NeuTTS as local fallback
|
||||
edge_available = True
|
||||
@@ -734,10 +658,10 @@ def text_to_speech_tool(
|
||||
"error": f"TTS generation produced no output (provider: {provider})"
|
||||
}, ensure_ascii=False)
|
||||
|
||||
# Try Opus conversion for Telegram compatibility only.
|
||||
# Outside Telegram, preserve the caller's explicit output format.
|
||||
# Try Opus conversion for Telegram compatibility
|
||||
# Edge TTS outputs MP3, NeuTTS outputs WAV — both need ffmpeg conversion
|
||||
voice_compatible = False
|
||||
if want_opus and provider in ("edge", "neutts", "minimax", "kittentts") and not file_str.endswith(".ogg"):
|
||||
if provider in ("edge", "neutts", "minimax") and not file_str.endswith(".ogg"):
|
||||
opus_path = _convert_to_opus(file_str)
|
||||
if opus_path:
|
||||
file_str = opus_path
|
||||
@@ -818,8 +742,6 @@ def check_tts_requirements() -> bool:
|
||||
pass
|
||||
if _check_neutts_available():
|
||||
return True
|
||||
if _check_kittentts_available():
|
||||
return True
|
||||
return False
|
||||
|
||||
|
||||
|
||||
@@ -10,7 +10,7 @@ Hermes Agent supports both text-to-speech output and voice message transcription
|
||||
|
||||
## Text-to-Speech
|
||||
|
||||
Convert text to speech with seven providers:
|
||||
Convert text to speech with six providers:
|
||||
|
||||
| Provider | Quality | Cost | API Key |
|
||||
|----------|---------|------|---------|
|
||||
@@ -20,7 +20,6 @@ Convert text to speech with seven providers:
|
||||
| **MiniMax TTS** | Excellent | Paid | `MINIMAX_API_KEY` |
|
||||
| **Mistral (Voxtral TTS)** | Excellent | Paid | `MISTRAL_API_KEY` |
|
||||
| **NeuTTS** | Good | Free | None needed |
|
||||
| **KittenTTS** | Good | Free (local) | None needed |
|
||||
|
||||
### Platform Delivery
|
||||
|
||||
@@ -36,7 +35,7 @@ Convert text to speech with seven providers:
|
||||
```yaml
|
||||
# In ~/.hermes/config.yaml
|
||||
tts:
|
||||
provider: "edge" # "edge" | "elevenlabs" | "openai" | "minimax" | "mistral" | "neutts" | "kittentts"
|
||||
provider: "edge" # "edge" | "elevenlabs" | "openai" | "minimax" | "mistral" | "neutts"
|
||||
speed: 1.0 # Global speed multiplier (provider-specific settings override this)
|
||||
edge:
|
||||
voice: "en-US-AriaNeural" # 322 voices, 74 languages
|
||||
@@ -63,11 +62,6 @@ tts:
|
||||
ref_text: ''
|
||||
model: neuphonic/neutts-air-q4-gguf
|
||||
device: cpu
|
||||
kittentts:
|
||||
model: KittenML/kitten-tts-nano-0.8-int8 # 25MB int8 default; also micro and mini variants
|
||||
voice: Jasper # Jasper, Bella, Luna, Bruno, Rosie, Hugo, Kiki, Leo
|
||||
speed: 1.0
|
||||
clean_text: true
|
||||
```
|
||||
|
||||
**Speed control**: The global `tts.speed` value applies to all providers by default. Each provider can override it with its own `speed` setting (e.g., `tts.openai.speed: 1.5`). Provider-specific speed takes precedence over the global value. Default is `1.0` (normal speed).
|
||||
@@ -80,7 +74,6 @@ Telegram voice bubbles require Opus/OGG audio format:
|
||||
- **Edge TTS** (default) outputs MP3 and needs **ffmpeg** to convert:
|
||||
- **MiniMax TTS** outputs MP3 and needs **ffmpeg** to convert for Telegram voice bubbles
|
||||
- **NeuTTS** outputs WAV and also needs **ffmpeg** to convert for Telegram voice bubbles
|
||||
- **KittenTTS** outputs WAV and also needs **ffmpeg** to convert for Telegram voice bubbles
|
||||
|
||||
```bash
|
||||
# Ubuntu/Debian
|
||||
@@ -93,7 +86,7 @@ brew install ffmpeg
|
||||
sudo dnf install ffmpeg
|
||||
```
|
||||
|
||||
Without ffmpeg, Edge TTS, MiniMax TTS, NeuTTS, and KittenTTS audio are sent as regular audio files (playable, but shown as a rectangular player instead of a voice bubble).
|
||||
Without ffmpeg, Edge TTS, MiniMax TTS, and NeuTTS audio are sent as regular audio files (playable, but shown as a rectangular player instead of a voice bubble).
|
||||
|
||||
:::tip
|
||||
If you want voice bubbles without installing ffmpeg, switch to the OpenAI, ElevenLabs, or Mistral provider.
|
||||
|
||||
Reference in New Issue
Block a user