fix(models): delegate generic provider catalogs to Hermes CLI

This commit is contained in:
Michael Lam
2026-05-05 16:48:56 -07:00
committed by test
parent 857f536f82
commit 63239d5b3c
4 changed files with 253 additions and 17 deletions

View File

@@ -1963,6 +1963,65 @@ def _get_label_for_model(model_id: str, existing_groups: list) -> str:
)
def _read_live_provider_model_ids(provider_id: str) -> list[str]:
"""Return live model IDs from Hermes CLI for a provider, or [] on failure.
WebUI's static ``_PROVIDER_MODELS`` table is only a fallback. The agent CLI
owns the provider registry and catalog-discovery logic, so ordinary picker
groups should ask ``hermes_cli.models.provider_model_ids()`` first (#1240).
Provider aliases are tried as a secondary lookup because WebUI keeps a few
display-facing IDs (for example ``google`` / ``x-ai``) that Hermes CLI may
normalize internally.
"""
pid = str(provider_id or "").strip()
if not pid:
return []
try:
from hermes_cli.models import provider_model_ids as _provider_model_ids
except Exception:
return []
candidates = [pid]
try:
alias = _resolve_provider_alias(pid)
except Exception:
alias = ""
if alias and alias not in candidates:
candidates.append(alias)
seen: set[str] = set()
for candidate in candidates:
try:
live_ids = _provider_model_ids(candidate) or []
except Exception:
logger.debug("Failed to load %s models from hermes_cli", candidate)
continue
result: list[str] = []
for mid in live_ids:
mid_s = str(mid or "").strip()
if mid_s and mid_s not in seen:
seen.add(mid_s)
result.append(mid_s)
if result:
return result
return []
def _models_from_live_provider_ids(provider_id: str, live_ids: list[str]) -> list[dict]:
"""Convert Hermes CLI model ids into WebUI picker model entries."""
formatter = _format_ollama_label if provider_id in ("ollama", "ollama-cloud") else None
models: list[dict] = []
seen: set[str] = set()
for mid in live_ids:
mid_s = str(mid or "").strip()
if not mid_s or mid_s in seen:
continue
seen.add(mid_s)
label = formatter(mid_s) if formatter else _get_label_for_model(mid_s, [])
models.append({"id": mid_s, "label": label})
return models
def _read_visible_codex_cache_model_ids() -> list[str]:
"""Return visible model slugs from Codex's local models_cache.json.
@@ -2912,18 +2971,33 @@ def get_available_models() -> dict:
group_entry["extra_models"] = extras
groups.append(group_entry)
elif pid in _PROVIDER_MODELS or pid in cfg.get("providers", {}):
raw_models = copy.deepcopy(_PROVIDER_MODELS.get(pid, []))
detected_models = auto_detected_models_by_provider.get(pid, [])
if detected_models and not raw_models:
raw_models = copy.deepcopy(detected_models)
provider_cfg = cfg.get("providers", {}).get(pid, {})
raw_models = []
# User-configured model allowlists are explicit local
# source-of-truth and should still beat auto-discovery.
# Otherwise, ask Hermes CLI first so WebUI tracks the same
# live catalog as the agent/CLI picker; WebUI's static
# _PROVIDER_MODELS table is now a fallback only (#1240).
if isinstance(provider_cfg, dict) and "models" in provider_cfg:
cfg_models = provider_cfg["models"]
if isinstance(cfg_models, dict):
raw_models = [{"id": k, "label": k} for k in cfg_models.keys()]
elif isinstance(cfg_models, list):
raw_models = [{"id": k, "label": k} for k in cfg_models]
if not raw_models:
raw_models = _models_from_live_provider_ids(
pid,
_read_live_provider_model_ids(pid),
)
if not raw_models:
raw_models = copy.deepcopy(_PROVIDER_MODELS.get(pid, []))
detected_models = auto_detected_models_by_provider.get(pid, [])
if detected_models and not raw_models:
raw_models = copy.deepcopy(detected_models)
models = _apply_provider_prefix(raw_models, pid, active_provider)
groups.append(
{

View File

@@ -0,0 +1,162 @@
"""Regression tests for #1240 — WebUI model catalog should delegate to Hermes CLI.
The WebUI picker should not freeze ordinary providers to its static
``_PROVIDER_MODELS`` snapshot when Hermes CLI can return a fresher provider
catalog. Static lists remain a fallback only.
"""
from __future__ import annotations
import sys
import types
import api.config as config
_PROVIDER_ENV_VARS = (
"ANTHROPIC_API_KEY",
"OPENAI_API_KEY",
"OPENROUTER_API_KEY",
"GOOGLE_API_KEY",
"GEMINI_API_KEY",
"GLM_API_KEY",
"KIMI_API_KEY",
"DEEPSEEK_API_KEY",
"OPENCODE_ZEN_API_KEY",
"OPENCODE_GO_API_KEY",
"MINIMAX_API_KEY",
"MINIMAX_CN_API_KEY",
"XAI_API_KEY",
"MISTRAL_API_KEY",
"OLLAMA_CLOUD_API_KEY",
"OLLAMA_API_KEY",
"NOUS_API_KEY",
"NVIDIA_API_KEY",
)
def _scrub_provider_env(monkeypatch):
for name in _PROVIDER_ENV_VARS:
monkeypatch.delenv(name, raising=False)
def _install_fake_hermes_cli(monkeypatch, *, provider_id: str, live_ids, raise_on_lookup: bool = False):
"""Install a hermes_cli stub that reports one authenticated provider."""
fake_pkg = types.ModuleType("hermes_cli")
fake_pkg.__path__ = []
fake_models = types.ModuleType("hermes_cli.models")
fake_models.list_available_providers = lambda: [
{"id": provider_id, "authenticated": True}
]
calls: list[str] = []
def provider_model_ids(pid):
calls.append(pid)
if raise_on_lookup:
raise RuntimeError("simulated provider_model_ids failure")
return list(live_ids) if pid == provider_id else []
fake_models.provider_model_ids = provider_model_ids
fake_auth = types.ModuleType("hermes_cli.auth")
def get_auth_status(pid):
if pid == provider_id:
return {"logged_in": True, "key_source": ""}
return {"logged_in": False, "key_source": ""}
fake_auth.get_auth_status = get_auth_status
monkeypatch.setitem(sys.modules, "hermes_cli", fake_pkg)
monkeypatch.setitem(sys.modules, "hermes_cli.models", fake_models)
monkeypatch.setitem(sys.modules, "hermes_cli.auth", fake_auth)
monkeypatch.delitem(sys.modules, "agent.credential_pool", raising=False)
monkeypatch.delitem(sys.modules, "agent", raising=False)
config.invalidate_models_cache()
return calls
def _configure(monkeypatch, tmp_path, *, provider: str, default: str = ""):
monkeypatch.setattr(config, "_get_config_path", lambda: tmp_path / "missing-config.yaml")
monkeypatch.setattr(config, "_models_cache_path", tmp_path / "models_cache.json")
monkeypatch.setattr(
config,
"cfg",
{
"model": {"provider": provider, "default": default},
"providers": {},
"fallback_providers": [],
},
)
monkeypatch.setattr(config, "_cfg_mtime", 0.0)
config.invalidate_models_cache()
def _provider_group(result: dict, provider_id: str) -> dict:
return next(g for g in result["groups"] if g.get("provider_id") == provider_id)
def _ids(group: dict) -> list[str]:
return [m.get("id") for m in group.get("models", [])]
def test_generic_provider_uses_hermes_cli_catalog_before_static_snapshot(monkeypatch, tmp_path):
"""A normal provider should show fresh CLI-discovered models.
``claude-sonnet-5.0`` is intentionally absent from WebUI's static Anthropic
list. Before this fix the group came entirely from ``_PROVIDER_MODELS`` and
this model was invisible even though Hermes CLI knew about it.
"""
_scrub_provider_env(monkeypatch)
calls = _install_fake_hermes_cli(
monkeypatch,
provider_id="anthropic",
live_ids=["claude-opus-4.7", "claude-sonnet-5.0"],
)
_configure(monkeypatch, tmp_path, provider="anthropic", default="claude-opus-4.7")
result = config.get_available_models()
group = _provider_group(result, "anthropic")
assert calls == ["anthropic"]
assert _ids(group) == ["claude-opus-4.7", "claude-sonnet-5.0"]
assert group["models"][1]["label"] == "Claude Sonnet 5.0"
def test_generic_provider_keeps_static_catalog_as_cli_failure_fallback(monkeypatch, tmp_path):
_scrub_provider_env(monkeypatch)
calls = _install_fake_hermes_cli(
monkeypatch,
provider_id="anthropic",
live_ids=[],
raise_on_lookup=True,
)
_configure(monkeypatch, tmp_path, provider="anthropic", default="claude-opus-4.7")
result = config.get_available_models()
group = _provider_group(result, "anthropic")
assert calls == ["anthropic"]
assert "claude-opus-4.7" in _ids(group)
assert "claude-sonnet-4.6" in _ids(group)
def test_generic_provider_prefixes_live_ids_when_not_active_provider(monkeypatch, tmp_path):
"""Provider-qualified live IDs must route through the selected provider."""
_scrub_provider_env(monkeypatch)
calls = _install_fake_hermes_cli(
monkeypatch,
provider_id="anthropic",
live_ids=["claude-sonnet-5.0"],
)
# Anthropic is authenticated via Hermes CLI, but OpenAI is the active
# default. The Anthropic row still has to be pickable/routable.
_configure(monkeypatch, tmp_path, provider="openai", default="gpt-5.5")
result = config.get_available_models()
group = _provider_group(result, "anthropic")
assert "anthropic" in calls
assert _ids(group) == ["@anthropic:claude-sonnet-5.0"]

View File

@@ -116,8 +116,9 @@ class TestConfigYamlModelsLoading:
)
break
def test_provider_in_provider_models_but_no_cfg_override_unchanged(self):
"""When no models key in cfg.providers, hardcoded _PROVIDER_MODELS still used."""
def test_provider_in_provider_models_but_no_cfg_override_uses_static_fallback(self, monkeypatch):
"""When Hermes CLI has no live catalog, _PROVIDER_MODELS remains fallback."""
monkeypatch.setattr(_cfg, "_read_live_provider_model_ids", lambda _pid: [])
cfg = {
"model": {"provider": "anthropic"},
"providers": {
@@ -132,10 +133,9 @@ class TestConfigYamlModelsLoading:
for g in result["groups"]:
if g["provider"] == "Anthropic":
returned_ids = {m["id"] for m in g["models"]}
# Should still have the hardcoded models
overlap = raw_ids & returned_ids
assert overlap, (
f"No _PROVIDER_MODELS models found in Anthropic group. "
f"No _PROVIDER_MODELS fallback models found in Anthropic group. "
f"Expected subset of {raw_ids}, got {returned_ids}"
)
break

View File

@@ -234,18 +234,16 @@ def test_no_duplicate_when_default_model_is_prefixed():
_cfg.cfg.update(old_cfg)
def test_default_provider_models_not_prefixed():
def test_default_provider_models_not_prefixed(monkeypatch):
"""The active provider's models remain bare (no @prefix added)."""
import api.config as _cfg
raw_anthropic_ids = {m['id'] for m in _cfg._PROVIDER_MODELS.get('anthropic', [])}
monkeypatch.setattr(_cfg, "_read_live_provider_model_ids", lambda pid: ["claude-sonnet-5.0"] if pid == "anthropic" else [])
result = _available_models_with_provider('anthropic')
groups = {g['provider']: g['models'] for g in result['groups']}
if 'Anthropic' in groups:
returned_ids = {m['id'] for m in groups['Anthropic']}
for bare_id in raw_anthropic_ids:
assert bare_id in returned_ids, (
f"_PROVIDER_MODELS entry '{bare_id}' is missing from the Anthropic group"
)
assert "claude-sonnet-5.0" in returned_ids
assert not any(mid.startswith('@anthropic:') for mid in returned_ids), returned_ids
# ── get_available_models(): phantom "Custom" group regression ─────────────
@@ -437,8 +435,10 @@ def test_custom_endpoint_uses_model_config_api_key_for_model_discovery(monkeypat
return False
def _fake_urlopen(req, timeout=10):
captured['auth'] = req.get_header('Authorization')
captured['ua'] = req.get_header('User-agent')
url = getattr(req, 'full_url', '')
if 'example.test' in url:
captured['auth'] = req.get_header('Authorization')
captured['ua'] = req.get_header('User-agent')
return _Resp()
monkeypatch.setattr('urllib.request.urlopen', _fake_urlopen)