mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-29 18:46:59 +00:00
feat(tools): full Speech-to-Text configurability in hermes tools + GUI
STT previously had no configuration surface outside hand-editing config.yaml — no category in the hermes tools picker, no provider matrix in the GUI capabilities tab, no status line in hermes setup. - TOOL_CATEGORIES['stt']: 7 provider rows (Local Whisper, Nous Subscription managed, OpenAI, Groq, xAI, ElevenLabs Scribe, DeepInfra) with key prompts, badges, and post-setup hooks - stt_provider marker wired through _write_provider_config, _configure_provider, _reconfigure_provider, and _is_provider_active — GUI and CLI share one write path (apply_provider_selection) - STT model picker (_configure_stt_model + STT_MODEL_CATALOG) runs after provider pick: local sizes, Groq whisper family, OpenAI whisper-1/gpt-4o-*/gpt-transcribe, ElevenLabs scribe (model_id key) - faster_whisper post-setup hook auto-installs the local backend; registered in _POST_SETUP_READY - stt is CONFIG-ONLY (_CONFIG_ONLY_TOOLSETS): it ships no tool schemas, so it is excluded from the per-platform enable checklist; the GUI toolset toggle writes stt.enabled instead of platform_toolsets - hermes setup shows a Speech-to-Text status line per provider - Mistral row omitted (mistralai PyPI quarantine), mirroring the dashboard stt.provider options Tests: tests/hermes_cli/test_stt_picker.py (20 cases) incl. invariant checks against agent.transcription_registry builtins and the runtime OPENAI_MODELS/GROQ_MODELS sets.
This commit is contained in:
Teknium
parent
c80acad5b7
commit
96bf65a6f7
4 changed files with 421 additions and 5 deletions
|
|
@ -547,6 +547,34 @@ def _print_setup_summary(config: dict, hermes_home):
|
|||
else:
|
||||
tool_status.append(("Text-to-Speech (Edge TTS)", True, None))
|
||||
|
||||
# STT — show configured provider
|
||||
stt_provider = cfg_get(config, "stt", "provider", default="local") or "local"
|
||||
if getattr(subscription_features, "stt", None) and subscription_features.stt.managed_by_nous:
|
||||
tool_status.append(("Speech-to-Text (OpenAI via Nous subscription)", True, None))
|
||||
elif stt_provider == "openai" and (
|
||||
get_env_value("VOICE_TOOLS_OPENAI_KEY") or get_env_value("OPENAI_API_KEY")
|
||||
):
|
||||
tool_status.append(("Speech-to-Text (OpenAI)", True, None))
|
||||
elif stt_provider == "groq" and get_env_value("GROQ_API_KEY"):
|
||||
tool_status.append(("Speech-to-Text (Groq Whisper)", True, None))
|
||||
elif stt_provider == "elevenlabs" and get_env_value("ELEVENLABS_API_KEY"):
|
||||
tool_status.append(("Speech-to-Text (ElevenLabs Scribe)", True, None))
|
||||
elif stt_provider == "xai":
|
||||
tool_status.append(("Speech-to-Text (xAI)", True, None))
|
||||
elif stt_provider == "deepinfra" and get_env_value("DEEPINFRA_API_KEY"):
|
||||
tool_status.append(("Speech-to-Text (DeepInfra)", True, None))
|
||||
else:
|
||||
try:
|
||||
fw_ok = importlib.util.find_spec("faster_whisper") is not None
|
||||
except Exception:
|
||||
fw_ok = False
|
||||
if fw_ok:
|
||||
tool_status.append(("Speech-to-Text (Local Whisper)", True, None))
|
||||
else:
|
||||
tool_status.append(
|
||||
("Speech-to-Text (Local Whisper — not installed)", False, "run 'hermes tools' → Speech-to-Text")
|
||||
)
|
||||
|
||||
if subscription_features.modal.managed_by_nous:
|
||||
tool_status.append(("Modal Execution (Nous subscription)", True, None))
|
||||
elif cfg_get(config, "terminal", "backend") == "modal":
|
||||
|
|
|
|||
|
|
@ -104,6 +104,7 @@ CONFIGURABLE_TOOLSETS = [
|
|||
("video_gen", "🎬 Video Generation", "video_generate (text/image/reference)"),
|
||||
("x_search", "🐦 X (Twitter) Search", "x_search (requires xAI OAuth or XAI_API_KEY)"),
|
||||
("tts", "🔊 Text-to-Speech", "text_to_speech"),
|
||||
("stt", "🎙️ Speech-to-Text", "voice transcription (gateway voice messages + voice mode)"),
|
||||
("skills", "📚 Skills", "list, view, manage"),
|
||||
("todo", "📋 Task Planning", "todo"),
|
||||
("memory", "💾 Memory", "persistent memory across sessions"),
|
||||
|
|
@ -153,6 +154,15 @@ def gui_toolset_label(label: str) -> str:
|
|||
_DEFAULT_OFF_TOOLSETS = {"homeassistant", "spotify", "discord", "discord_admin", "video", "video_gen", "x_search"}
|
||||
|
||||
|
||||
# Config-only capabilities: they appear in `hermes tools` for provider/API-key
|
||||
# configuration (TOOL_CATEGORIES) but are NOT model toolsets — they ship zero
|
||||
# tool schemas and their on/off switch lives in their own config section
|
||||
# (e.g. ``stt.enabled``), not ``platform_toolsets``. Excluded from the
|
||||
# per-platform enable/disable checklist; configured via the "Reconfigure an
|
||||
# existing tool" flow and the GUI provider matrix instead.
|
||||
_CONFIG_ONLY_TOOLSETS = {"stt"}
|
||||
|
||||
|
||||
def _xai_credentials_present() -> bool:
|
||||
"""Cheap, side-effect-free check for usable xAI credentials.
|
||||
|
||||
|
|
@ -273,6 +283,7 @@ def _checklist_toolset_keys(platform: str) -> Set[str]:
|
|||
ts_key
|
||||
for ts_key, _, _ in _get_effective_configurable_toolsets()
|
||||
if _toolset_allowed_for_platform(ts_key, platform)
|
||||
and ts_key not in _CONFIG_ONLY_TOOLSETS
|
||||
}
|
||||
|
||||
# Platform display config — derived from the canonical registry so every
|
||||
|
|
@ -384,6 +395,76 @@ TOOL_CATEGORIES = {
|
|||
},
|
||||
],
|
||||
},
|
||||
"stt": {
|
||||
"name": "Speech-to-Text",
|
||||
"icon": "🎙️",
|
||||
"providers": [
|
||||
{
|
||||
"name": "Local Whisper",
|
||||
"badge": "★ recommended · free",
|
||||
"tag": "faster-whisper on-device, no API key",
|
||||
"env_vars": [],
|
||||
"stt_provider": "local",
|
||||
"post_setup": "faster_whisper",
|
||||
},
|
||||
{
|
||||
"name": "Nous Subscription",
|
||||
"badge": "subscription",
|
||||
"tag": "Managed OpenAI transcription billed to your subscription",
|
||||
"env_vars": [],
|
||||
"stt_provider": "openai",
|
||||
"requires_nous_auth": True,
|
||||
"managed_nous_feature": "stt",
|
||||
"override_env_vars": ["VOICE_TOOLS_OPENAI_KEY", "OPENAI_API_KEY"],
|
||||
},
|
||||
{
|
||||
"name": "OpenAI",
|
||||
"badge": "paid",
|
||||
"tag": "whisper-1, gpt-4o-transcribe, gpt-transcribe",
|
||||
"env_vars": [
|
||||
{"key": "VOICE_TOOLS_OPENAI_KEY", "prompt": "OpenAI API key", "url": "https://platform.openai.com/api-keys"},
|
||||
],
|
||||
"stt_provider": "openai",
|
||||
},
|
||||
{
|
||||
"name": "Groq",
|
||||
"badge": "free tier",
|
||||
"tag": "Whisper large-v3 family — very fast",
|
||||
"env_vars": [
|
||||
{"key": "GROQ_API_KEY", "prompt": "Groq API key", "url": "https://console.groq.com/keys"},
|
||||
],
|
||||
"stt_provider": "groq",
|
||||
},
|
||||
{
|
||||
"name": "xAI",
|
||||
"tag": "grok-stt — uses xAI Grok OAuth or XAI_API_KEY",
|
||||
"env_vars": [],
|
||||
"stt_provider": "xai",
|
||||
"post_setup": "xai_grok",
|
||||
},
|
||||
{
|
||||
"name": "ElevenLabs Scribe",
|
||||
"badge": "paid",
|
||||
"tag": "scribe_v2 — diarization + audio-event tagging",
|
||||
"env_vars": [
|
||||
{"key": "ELEVENLABS_API_KEY", "prompt": "ElevenLabs API key", "url": "https://elevenlabs.io/app/settings/api-keys"},
|
||||
],
|
||||
"stt_provider": "elevenlabs",
|
||||
},
|
||||
# Mistral Voxtral STT intentionally omitted — mistralai PyPI
|
||||
# package quarantined (malicious 2.4.6 release, 2026-05-12).
|
||||
# Restore alongside the dashboard stt.provider option.
|
||||
{
|
||||
"name": "DeepInfra",
|
||||
"badge": "paid",
|
||||
"tag": "Live STT catalog from api.deepinfra.com",
|
||||
"env_vars": [
|
||||
{"key": "DEEPINFRA_API_KEY", "prompt": "DeepInfra API key", "url": "https://deepinfra.com/dash/api_keys"},
|
||||
],
|
||||
"stt_provider": "deepinfra",
|
||||
},
|
||||
],
|
||||
},
|
||||
"web": {
|
||||
"name": "Web Search & Extract",
|
||||
"setup_title": "Select Search Provider",
|
||||
|
|
@ -1539,6 +1620,29 @@ def _run_post_setup(post_setup_key: str):
|
|||
elif post_setup_key == "cua_driver":
|
||||
install_cua_driver(upgrade=False)
|
||||
|
||||
elif post_setup_key == "faster_whisper":
|
||||
import subprocess
|
||||
try:
|
||||
__import__("faster_whisper")
|
||||
_print_success(" faster-whisper is already installed")
|
||||
return
|
||||
except ImportError:
|
||||
pass
|
||||
_print_info(" Installing faster-whisper (model ~150MB downloads on first use)...")
|
||||
try:
|
||||
result = _pip_install(["-U", "faster-whisper", "--quiet"], timeout=300)
|
||||
if result.returncode == 0:
|
||||
_print_success(" faster-whisper installed")
|
||||
_print_info(" Model sizes: tiny, base (default), small, medium, large-v3")
|
||||
_print_info(" Change via stt.local.model in ~/.hermes/config.yaml")
|
||||
else:
|
||||
_print_warning(" faster-whisper install failed:")
|
||||
_print_info(f" {(result.stderr or '').strip()[:300]}")
|
||||
_print_info(" Run manually: uv pip install -U faster-whisper")
|
||||
except subprocess.TimeoutExpired:
|
||||
_print_warning(" faster-whisper install timed out (>5min)")
|
||||
_print_info(" Run manually: uv pip install -U faster-whisper")
|
||||
|
||||
elif post_setup_key == "kittentts":
|
||||
try:
|
||||
__import__("kittentts")
|
||||
|
|
@ -2290,7 +2394,7 @@ def _toolset_has_keys(
|
|||
except Exception:
|
||||
return False
|
||||
|
||||
if ts_key in {"web", "image_gen", "video_gen", "tts", "browser"}:
|
||||
if ts_key in {"web", "image_gen", "video_gen", "tts", "stt", "browser"}:
|
||||
features = get_nous_subscription_features(config, force_fresh=force_fresh)
|
||||
feature = features.features.get(ts_key)
|
||||
if feature and (feature.available or feature.managed_by_nous):
|
||||
|
|
@ -2385,10 +2489,12 @@ def _prompt_toolset_checklist(
|
|||
tool_tokens = _estimate_tool_tokens()
|
||||
|
||||
effective_all = _get_effective_configurable_toolsets()
|
||||
# Drop platform-scoped toolsets that don't apply to this platform.
|
||||
# Drop platform-scoped toolsets that don't apply to this platform, and
|
||||
# config-only capabilities (stt) that have no per-platform toggle.
|
||||
effective = [
|
||||
(k, l, d) for (k, l, d) in effective_all
|
||||
if _toolset_allowed_for_platform(k, platform)
|
||||
and k not in _CONFIG_ONLY_TOOLSETS
|
||||
]
|
||||
|
||||
labels = []
|
||||
|
|
@ -2908,6 +3014,7 @@ def _camofox_installed() -> bool:
|
|||
_POST_SETUP_READY: dict = {
|
||||
"kittentts": lambda: _module_installed("kittentts"),
|
||||
"piper": lambda: _module_installed("piper"),
|
||||
"faster_whisper": lambda: _module_installed("faster_whisper"),
|
||||
"ddgs": lambda: _module_installed("ddgs"),
|
||||
"langfuse": lambda: _module_installed("langfuse"),
|
||||
"agent_browser": lambda: _agent_browser_installed(),
|
||||
|
|
@ -3236,6 +3343,11 @@ def _is_provider_active(
|
|||
feature.managed_by_nous
|
||||
and cfg_get(config, "tts", "provider") == provider["tts_provider"]
|
||||
)
|
||||
if provider.get("stt_provider"):
|
||||
return (
|
||||
feature.managed_by_nous
|
||||
and cfg_get(config, "stt", "provider") == provider["stt_provider"]
|
||||
)
|
||||
if "browser_provider" in provider:
|
||||
current = cfg_get(config, "browser", "cloud_provider")
|
||||
return feature.managed_by_nous and provider["browser_provider"] == current
|
||||
|
|
@ -3246,6 +3358,10 @@ def _is_provider_active(
|
|||
|
||||
if provider.get("tts_provider"):
|
||||
return cfg_get(config, "tts", "provider") == provider["tts_provider"]
|
||||
if provider.get("stt_provider"):
|
||||
# Default stt.provider is "local" — an unset key means Local Whisper.
|
||||
current = cfg_get(config, "stt", "provider") or "local"
|
||||
return current == provider["stt_provider"]
|
||||
if "browser_provider" in provider:
|
||||
current = cfg_get(config, "browser", "cloud_provider")
|
||||
return provider["browser_provider"] == current
|
||||
|
|
@ -3609,6 +3725,49 @@ def _configure_videogen_model_for_plugin(plugin_name: str, config: dict) -> None
|
|||
_print_success(f" Model set to: {chosen}")
|
||||
|
||||
|
||||
# Per-provider STT model catalogs for the interactive picker. Keys are
|
||||
# ``stt.<provider>`` config sections; the first entry is the default.
|
||||
# Kept in sync with the dashboard selects (hermes_cli/web_server.py
|
||||
# _CONFIG_FIELD_META) and the desktop settings enums
|
||||
# (apps/desktop/src/app/settings/constants.ts).
|
||||
STT_MODEL_CATALOG = {
|
||||
"local": ["base", "tiny", "small", "medium", "large-v3"],
|
||||
"groq": ["whisper-large-v3-turbo", "whisper-large-v3", "distil-whisper-large-v3-en"],
|
||||
"openai": ["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-transcribe", "gpt-transcribe"],
|
||||
"elevenlabs": ["scribe_v2", "scribe_v1"],
|
||||
}
|
||||
|
||||
# ElevenLabs historically uses ``model_id`` instead of ``model``.
|
||||
_STT_MODEL_CONFIG_KEY = {"elevenlabs": "model_id"}
|
||||
|
||||
|
||||
def _configure_stt_model(stt_provider: str, config: dict) -> None:
|
||||
"""Prompt for the STT model after a provider pick (when a catalog exists).
|
||||
|
||||
Providers without a static catalog (xai, deepinfra) skip the prompt —
|
||||
xAI has a single model and DeepInfra resolves from its live catalog.
|
||||
"""
|
||||
catalog = STT_MODEL_CATALOG.get(stt_provider)
|
||||
if not catalog:
|
||||
return
|
||||
stt_cfg = config.setdefault("stt", {})
|
||||
if not isinstance(stt_cfg, dict):
|
||||
stt_cfg = {}
|
||||
config["stt"] = stt_cfg
|
||||
prov_cfg = stt_cfg.setdefault(stt_provider, {})
|
||||
if not isinstance(prov_cfg, dict):
|
||||
prov_cfg = {}
|
||||
stt_cfg[stt_provider] = prov_cfg
|
||||
model_key = _STT_MODEL_CONFIG_KEY.get(stt_provider, "model")
|
||||
current = str(prov_cfg.get(model_key) or "").strip()
|
||||
ordered = list(catalog)
|
||||
default_idx = ordered.index(current) if current in ordered else 0
|
||||
idx = _prompt_choice(" Select STT model:", ordered, default_idx)
|
||||
chosen = ordered[idx]
|
||||
prov_cfg[model_key] = chosen
|
||||
_print_success(f" STT model set to: {chosen}")
|
||||
|
||||
|
||||
def _select_plugin_video_gen_provider(plugin_name: str, config: dict, *, use_gateway: bool = False) -> None:
|
||||
"""Persist a plugin-backed video generation provider selection."""
|
||||
vid_cfg = config.setdefault("video_gen", {})
|
||||
|
|
@ -3639,6 +3798,12 @@ def _write_provider_config(provider: dict, config: dict, *, managed_feature) ->
|
|||
tts_cfg["provider"] = provider["tts_provider"]
|
||||
tts_cfg["use_gateway"] = bool(managed_feature)
|
||||
|
||||
# Set STT provider in config if applicable
|
||||
if provider.get("stt_provider"):
|
||||
stt_cfg = config.setdefault("stt", {})
|
||||
stt_cfg["provider"] = provider["stt_provider"]
|
||||
stt_cfg["use_gateway"] = bool(managed_feature)
|
||||
|
||||
# Set browser cloud provider in config if applicable
|
||||
if "browser_provider" in provider:
|
||||
bp = provider["browser_provider"]
|
||||
|
|
@ -3655,7 +3820,7 @@ def _write_provider_config(provider: dict, config: dict, *, managed_feature) ->
|
|||
|
||||
# For tools without a specific config key (e.g. image_gen), still
|
||||
# track use_gateway so the runtime knows the user's intent.
|
||||
if managed_feature and managed_feature not in {"web", "tts", "browser"}:
|
||||
if managed_feature and managed_feature not in {"web", "tts", "stt", "browser"}:
|
||||
config.setdefault(managed_feature, {})["use_gateway"] = True
|
||||
elif not managed_feature:
|
||||
# User picked a non-gateway provider — find which category this
|
||||
|
|
@ -3778,6 +3943,10 @@ def _configure_provider(
|
|||
tts_cfg["provider"] = provider["tts_provider"]
|
||||
tts_cfg["use_gateway"] = bool(managed_feature)
|
||||
|
||||
# Set STT provider in config if applicable
|
||||
if provider.get("stt_provider"):
|
||||
_print_success(f" STT provider set to: {provider['stt_provider']}")
|
||||
|
||||
# Set browser cloud provider in config if applicable
|
||||
if "browser_provider" in provider:
|
||||
bp = provider["browser_provider"]
|
||||
|
|
@ -3823,6 +3992,10 @@ def _configure_provider(
|
|||
img_cfg = config.setdefault("image_gen", {})
|
||||
if isinstance(img_cfg, dict) and img_cfg.get("provider") not in {None, "", "fal"}:
|
||||
img_cfg["provider"] = "fal"
|
||||
# STT providers prompt for model selection after provider pick
|
||||
# (skipped for managed rows — the gateway pins the model).
|
||||
if provider.get("stt_provider") and not managed_feature:
|
||||
_configure_stt_model(provider["stt_provider"], config)
|
||||
return
|
||||
|
||||
# Prompt for each required env var
|
||||
|
|
@ -3899,6 +4072,9 @@ def _configure_provider(
|
|||
img_cfg = config.setdefault("image_gen", {})
|
||||
if isinstance(img_cfg, dict) and img_cfg.get("provider") not in {None, "", "fal"}:
|
||||
img_cfg["provider"] = "fal"
|
||||
# STT providers prompt for model selection after env vars are in.
|
||||
if provider.get("stt_provider") and not managed_feature:
|
||||
_configure_stt_model(provider["stt_provider"], config)
|
||||
|
||||
|
||||
def _configure_vision_backend() -> None:
|
||||
|
|
@ -4276,6 +4452,12 @@ def _reconfigure_provider(
|
|||
tts_cfg["use_gateway"] = bool(managed_feature)
|
||||
_print_success(f" TTS provider set to: {provider['tts_provider']}")
|
||||
|
||||
if provider.get("stt_provider"):
|
||||
stt_cfg = config.setdefault("stt", {})
|
||||
stt_cfg["provider"] = provider["stt_provider"]
|
||||
stt_cfg["use_gateway"] = bool(managed_feature)
|
||||
_print_success(f" STT provider set to: {provider['stt_provider']}")
|
||||
|
||||
if "browser_provider" in provider:
|
||||
bp = provider["browser_provider"]
|
||||
browser_cfg = config.setdefault("browser", {})
|
||||
|
|
@ -4294,7 +4476,7 @@ def _reconfigure_provider(
|
|||
web_cfg["use_gateway"] = bool(managed_feature)
|
||||
_print_success(f" Web backend set to: {provider['web_backend']}")
|
||||
|
||||
if managed_feature and managed_feature not in {"web", "tts", "browser"}:
|
||||
if managed_feature and managed_feature not in {"web", "tts", "stt", "browser"}:
|
||||
section = config.setdefault(managed_feature, {})
|
||||
if not isinstance(section, dict):
|
||||
section = {}
|
||||
|
|
@ -4332,6 +4514,9 @@ def _reconfigure_provider(
|
|||
if isinstance(img_cfg, dict):
|
||||
img_cfg["provider"] = "fal"
|
||||
img_cfg["use_gateway"] = False
|
||||
# STT providers prompt for model selection on reconfig too.
|
||||
if provider.get("stt_provider") and not managed_feature:
|
||||
_configure_stt_model(provider["stt_provider"], config)
|
||||
return
|
||||
|
||||
for var in env_vars:
|
||||
|
|
@ -4373,6 +4558,10 @@ def _reconfigure_provider(
|
|||
img_cfg["provider"] = "fal"
|
||||
img_cfg["use_gateway"] = False
|
||||
|
||||
# STT providers prompt for model selection on reconfig too.
|
||||
if provider.get("stt_provider") and not managed_feature:
|
||||
_configure_stt_model(provider["stt_provider"], config)
|
||||
|
||||
|
||||
def _reconfigure_simple_requirements(ts_key: str):
|
||||
"""Reconfigure simple env var requirements."""
|
||||
|
|
|
|||
|
|
@ -15962,6 +15962,7 @@ async def update_skill_content(body: SkillContentUpdate):
|
|||
@app.get("/api/tools/toolsets")
|
||||
async def get_toolsets(profile: Optional[str] = None):
|
||||
from hermes_cli.tools_config import (
|
||||
_CONFIG_ONLY_TOOLSETS,
|
||||
_get_effective_configurable_toolsets,
|
||||
_get_platform_tools,
|
||||
_toolset_configuration_platform,
|
||||
|
|
@ -15992,7 +15993,16 @@ async def get_toolsets(profile: Optional[str] = None):
|
|||
except Exception:
|
||||
tools = []
|
||||
target_platform = _toolset_configuration_platform(name)
|
||||
is_enabled = name in enabled_by_platform[target_platform]
|
||||
if name in _CONFIG_ONLY_TOOLSETS:
|
||||
# Config-only capabilities (stt) have no per-platform toolset —
|
||||
# their switch is their own config section (e.g. stt.enabled).
|
||||
from utils import is_truthy_value
|
||||
|
||||
section = config.get(name)
|
||||
section = section if isinstance(section, dict) else {}
|
||||
is_enabled = is_truthy_value(section.get("enabled", True), default=True)
|
||||
else:
|
||||
is_enabled = name in enabled_by_platform[target_platform]
|
||||
result.append({
|
||||
"name": name,
|
||||
"label": gui_toolset_label(label),
|
||||
|
|
@ -16025,6 +16035,7 @@ async def toggle_toolset(name: str, body: ToolsetToggle, profile: Optional[str]
|
|||
to ``body.profile`` when provided. Returns 400 for unknown toolset keys.
|
||||
"""
|
||||
from hermes_cli.tools_config import (
|
||||
_CONFIG_ONLY_TOOLSETS,
|
||||
_get_effective_configurable_toolsets,
|
||||
_get_platform_tools,
|
||||
_save_platform_tools,
|
||||
|
|
@ -16036,6 +16047,23 @@ async def toggle_toolset(name: str, body: ToolsetToggle, profile: Optional[str]
|
|||
raise HTTPException(status_code=400, detail=f"Unknown toolset: {name}")
|
||||
|
||||
target_platform = _toolset_configuration_platform(name)
|
||||
if name in _CONFIG_ONLY_TOOLSETS:
|
||||
# Config-only capabilities (stt) toggle their own config section's
|
||||
# ``enabled`` flag — there is no platform_toolsets entry to write.
|
||||
with _profile_scope(body.profile or profile):
|
||||
config = load_config()
|
||||
section = config.setdefault(name, {})
|
||||
if not isinstance(section, dict):
|
||||
section = {}
|
||||
config[name] = section
|
||||
section["enabled"] = bool(body.enabled)
|
||||
save_config(config)
|
||||
return {
|
||||
"ok": True,
|
||||
"name": name,
|
||||
"platform": target_platform,
|
||||
"enabled": body.enabled,
|
||||
}
|
||||
with _profile_scope(body.profile or profile):
|
||||
config = load_config()
|
||||
enabled = set(
|
||||
|
|
|
|||
171
tests/hermes_cli/test_stt_picker.py
Normal file
171
tests/hermes_cli/test_stt_picker.py
Normal file
|
|
@ -0,0 +1,171 @@
|
|||
"""Tests for the Speech-to-Text category in `hermes tools` (tools_config).
|
||||
|
||||
Covers the STT provider picker rows, config writes (stt.provider /
|
||||
use_gateway), the model picker catalog, config-only checklist exclusion,
|
||||
and the faster_whisper post-setup readiness hook.
|
||||
"""
|
||||
|
||||
import sys
|
||||
import types
|
||||
from pathlib import Path
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
import pytest
|
||||
|
||||
sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
|
||||
|
||||
from hermes_cli.tools_config import ( # noqa: E402
|
||||
_CONFIG_ONLY_TOOLSETS,
|
||||
CONFIGURABLE_TOOLSETS,
|
||||
STT_MODEL_CATALOG,
|
||||
TOOL_CATEGORIES,
|
||||
_checklist_toolset_keys,
|
||||
_configure_stt_model,
|
||||
_is_provider_active,
|
||||
_write_provider_config,
|
||||
apply_provider_selection,
|
||||
)
|
||||
|
||||
|
||||
def _stt_cat():
|
||||
return TOOL_CATEGORIES["stt"]
|
||||
|
||||
|
||||
def _stt_provider_named(name):
|
||||
return next(p for p in _stt_cat()["providers"] if p["name"] == name)
|
||||
|
||||
|
||||
class TestSttCategory:
|
||||
def test_stt_category_exists(self):
|
||||
cat = _stt_cat()
|
||||
assert cat["name"] == "Speech-to-Text"
|
||||
assert len(cat["providers"]) >= 5
|
||||
|
||||
def test_stt_in_configurable_toolsets(self):
|
||||
keys = {k for k, _, _ in CONFIGURABLE_TOOLSETS}
|
||||
assert "stt" in keys
|
||||
|
||||
def test_every_row_writes_stt_provider(self):
|
||||
for prov in _stt_cat()["providers"]:
|
||||
assert prov.get("stt_provider"), f"row {prov['name']} missing stt_provider"
|
||||
|
||||
def test_provider_keys_cover_runtime_dispatch(self):
|
||||
"""Every picker row's stt_provider must be a runtime built-in."""
|
||||
from agent.transcription_registry import _BUILTIN_NAMES
|
||||
|
||||
for prov in _stt_cat()["providers"]:
|
||||
assert prov["stt_provider"] in _BUILTIN_NAMES
|
||||
|
||||
def test_managed_row_shares_tts_coverage_category(self):
|
||||
from hermes_cli.nous_subscription import MANAGED_FEATURE_COVERAGE_CATEGORY
|
||||
|
||||
managed = [p for p in _stt_cat()["providers"] if p.get("managed_nous_feature")]
|
||||
assert managed, "expected a Nous Subscription row"
|
||||
for p in managed:
|
||||
assert p["managed_nous_feature"] == "stt"
|
||||
assert MANAGED_FEATURE_COVERAGE_CATEGORY["stt"] == "openai-audio"
|
||||
|
||||
|
||||
class TestConfigWrites:
|
||||
def test_write_provider_config_sets_stt_provider(self):
|
||||
config = {}
|
||||
prov = _stt_provider_named("Groq")
|
||||
_write_provider_config(prov, config, managed_feature=None)
|
||||
assert config["stt"]["provider"] == "groq"
|
||||
assert config["stt"]["use_gateway"] is False
|
||||
|
||||
def test_write_provider_config_managed_sets_gateway(self):
|
||||
config = {}
|
||||
prov = _stt_provider_named("Nous Subscription")
|
||||
_write_provider_config(prov, config, managed_feature="stt")
|
||||
assert config["stt"]["provider"] == "openai"
|
||||
assert config["stt"]["use_gateway"] is True
|
||||
# Managed stt must not create a bogus top-level section via the
|
||||
# generic use_gateway fallback (it's in the exclusion set).
|
||||
assert set(config.keys()) == {"stt"}
|
||||
|
||||
def test_apply_provider_selection_stt(self):
|
||||
config = {}
|
||||
with patch(
|
||||
"hermes_cli.tools_config.get_nous_subscription_features"
|
||||
) as feats:
|
||||
feats.return_value = MagicMock(
|
||||
nous_auth_present=False, account_info=None
|
||||
)
|
||||
apply_provider_selection("stt", "OpenAI", config)
|
||||
assert config["stt"]["provider"] == "openai"
|
||||
|
||||
def test_byok_pick_clears_stale_gateway_flag(self):
|
||||
config = {"stt": {"provider": "openai", "use_gateway": True}}
|
||||
prov = _stt_provider_named("Groq")
|
||||
_write_provider_config(prov, config, managed_feature=None)
|
||||
assert config["stt"]["use_gateway"] is False
|
||||
|
||||
|
||||
class TestActiveDetection:
|
||||
def test_active_matches_config(self):
|
||||
config = {"stt": {"provider": "groq"}}
|
||||
assert _is_provider_active(_stt_provider_named("Groq"), config)
|
||||
assert not _is_provider_active(_stt_provider_named("OpenAI"), config)
|
||||
|
||||
def test_unset_provider_defaults_to_local(self):
|
||||
assert _is_provider_active(_stt_provider_named("Local Whisper"), {})
|
||||
|
||||
|
||||
class TestModelPicker:
|
||||
def test_catalog_includes_gpt_transcribe(self):
|
||||
assert "gpt-transcribe" in STT_MODEL_CATALOG["openai"]
|
||||
|
||||
def test_catalog_matches_runtime_model_sets(self):
|
||||
from tools.transcription_tools import GROQ_MODELS, OPENAI_MODELS
|
||||
|
||||
assert set(STT_MODEL_CATALOG["openai"]) == OPENAI_MODELS
|
||||
assert set(STT_MODEL_CATALOG["groq"]) == GROQ_MODELS
|
||||
|
||||
def test_configure_stt_model_writes_model(self):
|
||||
config = {}
|
||||
with patch("hermes_cli.tools_config._prompt_choice", return_value=3):
|
||||
_configure_stt_model("openai", config)
|
||||
assert config["stt"]["openai"]["model"] == STT_MODEL_CATALOG["openai"][3]
|
||||
|
||||
def test_configure_stt_model_elevenlabs_uses_model_id(self):
|
||||
config = {}
|
||||
with patch("hermes_cli.tools_config._prompt_choice", return_value=0):
|
||||
_configure_stt_model("elevenlabs", config)
|
||||
assert config["stt"]["elevenlabs"]["model_id"] == "scribe_v2"
|
||||
assert "model" not in config["stt"]["elevenlabs"]
|
||||
|
||||
def test_configure_stt_model_skips_uncataloged_provider(self):
|
||||
config = {}
|
||||
with patch("hermes_cli.tools_config._prompt_choice") as pc:
|
||||
_configure_stt_model("xai", config)
|
||||
_configure_stt_model("deepinfra", config)
|
||||
pc.assert_not_called()
|
||||
assert config == {}
|
||||
|
||||
def test_configure_stt_model_defaults_to_current(self):
|
||||
config = {"stt": {"openai": {"model": "gpt-transcribe"}}}
|
||||
with patch(
|
||||
"hermes_cli.tools_config._prompt_choice", return_value=0
|
||||
) as pc:
|
||||
_configure_stt_model("openai", config)
|
||||
# default index should point at the currently configured model
|
||||
args = pc.call_args[0]
|
||||
assert args[2] == STT_MODEL_CATALOG["openai"].index("gpt-transcribe")
|
||||
|
||||
|
||||
class TestConfigOnlyExclusion:
|
||||
def test_stt_is_config_only(self):
|
||||
assert "stt" in _CONFIG_ONLY_TOOLSETS
|
||||
|
||||
def test_stt_excluded_from_checklist_universe(self):
|
||||
assert "stt" not in _checklist_toolset_keys("cli")
|
||||
# sanity: tts (a real toolset) stays in
|
||||
assert "tts" in _checklist_toolset_keys("cli")
|
||||
|
||||
|
||||
class TestPostSetup:
|
||||
def test_faster_whisper_in_post_setup_ready(self):
|
||||
from hermes_cli.tools_config import _POST_SETUP_READY
|
||||
|
||||
assert "faster_whisper" in _POST_SETUP_READY
|
||||
Loading…
Add table
Add a link
Reference in a new issue