Files

308 lines
10 KiB
Python
Raw Permalink Normal View History

"""Tests for setup_model_provider — verifies the delegation to
select_provider_and_model() and config dict sync."""
import json
import sys
import types
from hermes_cli.auth import get_active_provider
from hermes_cli.config import load_config, save_config
from hermes_cli.setup import setup_model_provider
def _maybe_keep_current_tts(question, choices):
if question != "Select TTS provider:":
return None
assert choices[-1].startswith("Keep current (")
return len(choices) - 1
def _clear_provider_env(monkeypatch):
for key in (
"NOUS_API_KEY",
"OPENROUTER_API_KEY",
"OPENAI_BASE_URL",
"OPENAI_API_KEY",
"LLM_MODEL",
):
monkeypatch.delenv(key, raising=False)
def _stub_tts(monkeypatch):
"""Stub out TTS prompts so setup_model_provider doesn't block."""
monkeypatch.setattr("hermes_cli.setup.prompt_choice", lambda q, c, d=0: (
_maybe_keep_current_tts(q, c) if _maybe_keep_current_tts(q, c) is not None
else d
))
monkeypatch.setattr("hermes_cli.setup.prompt_yes_no", lambda *a, **kw: False)
def _write_model_config(tmp_path, provider, base_url="", model_name="test-model"):
"""Simulate what a _model_flow_* function writes to disk."""
cfg = load_config()
m = cfg.get("model")
if not isinstance(m, dict):
m = {"default": m} if m else {}
cfg["model"] = m
m["provider"] = provider
if base_url:
m["base_url"] = base_url
if model_name:
m["default"] = model_name
save_config(cfg)
def test_setup_delegates_to_select_provider_and_model(tmp_path, monkeypatch):
"""setup_model_provider calls select_provider_and_model and syncs config."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
def fake_select():
_write_model_config(tmp_path, "custom", "http://localhost:11434/v1", "qwen3.5:32b")
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
save_config(config)
reloaded = load_config()
assert isinstance(reloaded["model"], dict)
assert reloaded["model"]["provider"] == "custom"
assert reloaded["model"]["base_url"] == "http://localhost:11434/v1"
assert reloaded["model"]["default"] == "qwen3.5:32b"
def test_setup_syncs_openrouter_from_disk(tmp_path, monkeypatch):
"""When select_provider_and_model saves OpenRouter config to disk,
the wizard's config dict picks it up."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
assert isinstance(config.get("model"), str) # fresh install
def fake_select():
_write_model_config(tmp_path, "openrouter", model_name="anthropic/claude-opus-4.6")
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
save_config(config)
reloaded = load_config()
assert isinstance(reloaded["model"], dict)
assert reloaded["model"]["provider"] == "openrouter"
def test_setup_syncs_nous_from_disk(tmp_path, monkeypatch):
"""Nous OAuth writes config to disk; wizard config dict must pick it up."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
def fake_select():
_write_model_config(tmp_path, "nous", "https://inference.example.com/v1", "gemini-3-flash")
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
save_config(config)
feat: overhaul context length detection with models.dev and provider-aware resolution (#2158) Replace the fragile hardcoded context length system with a multi-source resolution chain that correctly identifies context windows per provider. Key changes: - New agent/models_dev.py: Fetches and caches the models.dev registry (3800+ models across 100+ providers with per-provider context windows). In-memory cache (1hr TTL) + disk cache for cold starts. - Rewritten get_model_context_length() resolution chain: 0. Config override (model.context_length) 1. Custom providers per-model context_length 2. Persistent disk cache 3. Endpoint /models (local servers) 4. Anthropic /v1/models API (max_input_tokens, API-key only) 5. OpenRouter live API (existing, unchanged) 6. Nous suffix-match via OpenRouter (dot/dash normalization) 7. models.dev registry lookup (provider-aware) 8. Thin hardcoded defaults (broad family patterns) 9. 128K fallback (was 2M) - Provider-aware context: same model now correctly resolves to different context windows per provider (e.g. claude-opus-4.6: 1M on Anthropic, 128K on GitHub Copilot). Provider name flows through ContextCompressor. - DEFAULT_CONTEXT_LENGTHS shrunk from 80+ entries to ~16 broad patterns. models.dev replaces the per-model hardcoding. - CONTEXT_PROBE_TIERS changed from [2M, 1M, 512K, 200K, 128K, 64K, 32K] to [128K, 64K, 32K, 16K, 8K]. Unknown models no longer start at 2M. - hermes model: prompts for context_length when configuring custom endpoints. Supports shorthand (32k, 128K). Saved to custom_providers per-model config. - custom_providers schema extended with optional models dict for per-model context_length (backward compatible). - Nous Portal: suffix-matches bare IDs (claude-opus-4-6) against OpenRouter's prefixed IDs (anthropic/claude-opus-4.6) with dot/dash normalization. Handles all 15 current Nous models. - Anthropic direct: queries /v1/models for max_input_tokens. Only works with regular API keys (sk-ant-api*), not OAuth tokens. Falls through to models.dev for OAuth users. Tests: 5574 passed (18 new tests for models_dev + updated probe tiers) Docs: Updated configuration.md context length section, AGENTS.md Co-authored-by: Test <test@test.com>
2026-03-20 06:04:33 -07:00
reloaded = load_config()
assert isinstance(reloaded["model"], dict)
assert reloaded["model"]["provider"] == "nous"
assert reloaded["model"]["base_url"] == "https://inference.example.com/v1"
def test_setup_custom_providers_synced(tmp_path, monkeypatch):
"""custom_providers written by select_provider_and_model must survive."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
def fake_select():
_write_model_config(tmp_path, "custom", "http://localhost:8080/v1", "llama3")
cfg = load_config()
cfg["custom_providers"] = [{"name": "Local", "base_url": "http://localhost:8080/v1"}]
save_config(cfg)
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
save_config(config)
feat: overhaul context length detection with models.dev and provider-aware resolution (#2158) Replace the fragile hardcoded context length system with a multi-source resolution chain that correctly identifies context windows per provider. Key changes: - New agent/models_dev.py: Fetches and caches the models.dev registry (3800+ models across 100+ providers with per-provider context windows). In-memory cache (1hr TTL) + disk cache for cold starts. - Rewritten get_model_context_length() resolution chain: 0. Config override (model.context_length) 1. Custom providers per-model context_length 2. Persistent disk cache 3. Endpoint /models (local servers) 4. Anthropic /v1/models API (max_input_tokens, API-key only) 5. OpenRouter live API (existing, unchanged) 6. Nous suffix-match via OpenRouter (dot/dash normalization) 7. models.dev registry lookup (provider-aware) 8. Thin hardcoded defaults (broad family patterns) 9. 128K fallback (was 2M) - Provider-aware context: same model now correctly resolves to different context windows per provider (e.g. claude-opus-4.6: 1M on Anthropic, 128K on GitHub Copilot). Provider name flows through ContextCompressor. - DEFAULT_CONTEXT_LENGTHS shrunk from 80+ entries to ~16 broad patterns. models.dev replaces the per-model hardcoding. - CONTEXT_PROBE_TIERS changed from [2M, 1M, 512K, 200K, 128K, 64K, 32K] to [128K, 64K, 32K, 16K, 8K]. Unknown models no longer start at 2M. - hermes model: prompts for context_length when configuring custom endpoints. Supports shorthand (32k, 128K). Saved to custom_providers per-model config. - custom_providers schema extended with optional models dict for per-model context_length (backward compatible). - Nous Portal: suffix-matches bare IDs (claude-opus-4-6) against OpenRouter's prefixed IDs (anthropic/claude-opus-4.6) with dot/dash normalization. Handles all 15 current Nous models. - Anthropic direct: queries /v1/models for max_input_tokens. Only works with regular API keys (sk-ant-api*), not OAuth tokens. Falls through to models.dev for OAuth users. Tests: 5574 passed (18 new tests for models_dev + updated probe tiers) Docs: Updated configuration.md context length section, AGENTS.md Co-authored-by: Test <test@test.com>
2026-03-20 06:04:33 -07:00
reloaded = load_config()
assert reloaded.get("custom_providers") == [{"name": "Local", "base_url": "http://localhost:8080/v1"}]
2026-03-13 21:06:06 -07:00
def test_setup_cancel_preserves_existing_config(tmp_path, monkeypatch):
"""When the user cancels provider selection, existing config is preserved."""
2026-03-13 21:06:06 -07:00
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
# Pre-set a provider
_write_model_config(tmp_path, "openrouter", model_name="gpt-4o")
2026-03-13 21:06:06 -07:00
config = load_config()
assert config["model"]["provider"] == "openrouter"
2026-03-13 21:06:06 -07:00
def fake_select():
pass # user cancelled — nothing written to disk
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
2026-03-13 21:06:06 -07:00
setup_model_provider(config)
save_config(config)
reloaded = load_config()
assert isinstance(reloaded["model"], dict)
assert reloaded["model"]["provider"] == "openrouter"
assert reloaded["model"]["default"] == "gpt-4o"
def test_setup_exception_in_select_gracefully_handled(tmp_path, monkeypatch):
"""If select_provider_and_model raises, setup continues with existing config."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
def fake_select():
raise RuntimeError("something broke")
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
# Should not raise
setup_model_provider(config)
def test_setup_keyboard_interrupt_gracefully_handled(tmp_path, monkeypatch):
"""KeyboardInterrupt during provider selection is handled."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
_clear_provider_env(monkeypatch)
_stub_tts(monkeypatch)
config = load_config()
def fake_select():
raise KeyboardInterrupt()
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
2026-03-13 21:06:06 -07:00
def test_codex_setup_uses_runtime_access_token_for_live_model_list(tmp_path, monkeypatch):
"""Codex model list fetching uses the runtime access token."""
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
2026-03-13 21:06:06 -07:00
monkeypatch.setenv("OPENROUTER_API_KEY", "or-test-key")
_clear_provider_env(monkeypatch)
2026-03-13 21:06:06 -07:00
monkeypatch.setenv("OPENROUTER_API_KEY", "or-test-key")
config = load_config()
_stub_tts(monkeypatch)
def fake_select():
_write_model_config(tmp_path, "openai-codex", "https://api.openai.com/v1", "gpt-4o")
monkeypatch.setattr("hermes_cli.main.select_provider_and_model", fake_select)
setup_model_provider(config)
2026-03-13 21:06:06 -07:00
save_config(config)
2026-03-13 21:06:06 -07:00
reloaded = load_config()
assert isinstance(reloaded["model"], dict)
assert reloaded["model"]["provider"] == "openai-codex"
def test_modal_setup_can_use_nous_subscription_without_modal_creds(tmp_path, monkeypatch, capsys):
monkeypatch.setenv("HERMES_ENABLE_NOUS_MANAGED_TOOLS", "1")
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
config = load_config()
def fake_prompt_choice(question, choices, default=0):
if question == "Select terminal backend:":
return 2
if question == "Select how Modal execution should be billed:":
return 0
raise AssertionError(f"Unexpected prompt_choice call: {question}")
def fake_prompt(message, *args, **kwargs):
assert "Modal Token" not in message
raise AssertionError(f"Unexpected prompt call: {message}")
monkeypatch.setattr("hermes_cli.setup.prompt_choice", fake_prompt_choice)
monkeypatch.setattr("hermes_cli.setup.prompt", fake_prompt)
monkeypatch.setattr("hermes_cli.setup._prompt_container_resources", lambda config: None)
monkeypatch.setattr(
"hermes_cli.setup.get_nous_subscription_features",
lambda config: type("Features", (), {"nous_auth_present": True})(),
)
monkeypatch.setitem(
sys.modules,
"tools.managed_tool_gateway",
types.SimpleNamespace(
is_managed_tool_gateway_ready=lambda vendor: vendor == "modal",
resolve_managed_tool_gateway=lambda vendor: None,
),
)
from hermes_cli.setup import setup_terminal_backend
setup_terminal_backend(config)
out = capsys.readouterr().out
assert config["terminal"]["backend"] == "modal"
assert config["terminal"]["modal_mode"] == "managed"
assert "bill to your subscription" in out
def test_modal_setup_persists_direct_mode_when_user_chooses_their_own_account(tmp_path, monkeypatch):
monkeypatch.setenv("HERMES_ENABLE_NOUS_MANAGED_TOOLS", "1")
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
monkeypatch.delenv("MODAL_TOKEN_ID", raising=False)
monkeypatch.delenv("MODAL_TOKEN_SECRET", raising=False)
config = load_config()
def fake_prompt_choice(question, choices, default=0):
if question == "Select terminal backend:":
return 2
if question == "Select how Modal execution should be billed:":
return 1
raise AssertionError(f"Unexpected prompt_choice call: {question}")
prompt_values = iter(["token-id", "token-secret", ""])
monkeypatch.setattr("hermes_cli.setup.prompt_choice", fake_prompt_choice)
monkeypatch.setattr("hermes_cli.setup.prompt", lambda *args, **kwargs: next(prompt_values))
monkeypatch.setattr("hermes_cli.setup._prompt_container_resources", lambda config: None)
monkeypatch.setattr(
"hermes_cli.setup.get_nous_subscription_features",
lambda config: type("Features", (), {"nous_auth_present": True})(),
)
monkeypatch.setitem(
sys.modules,
"tools.managed_tool_gateway",
types.SimpleNamespace(
is_managed_tool_gateway_ready=lambda vendor: vendor == "modal",
resolve_managed_tool_gateway=lambda vendor: None,
),
)
monkeypatch.setitem(sys.modules, "swe_rex", object())
from hermes_cli.setup import setup_terminal_backend
setup_terminal_backend(config)
assert config["terminal"]["backend"] == "modal"
assert config["terminal"]["modal_mode"] == "direct"