mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-31 19:16:29 +00:00
39975613b1
Second, deeper pass over tools/gateway/hermes_cli plus first pass over the trees wave 1 missed (acp, acp_adapter, skills, computer_use, docker, dashboard, conformance, monitoring, secret_sources, hermes_state, providers). Same rubric as wave 1 (AGENTS.md test policy); security, alternation/caching invariants, issue-number regressions, and E2E kept. Real test-quality fixes found and rooted out along the way: - tests/tools/test_command_guards.py made real auxiliary-LLM HTTPS calls (DEFAULT_CONFIG smart-approval leaked in) — pinned approval mode=manual via autouse fixture: 17.4s → 0.4s. - test_model_switch_custom_providers.py / test_user_providers_model_switch.py silently probed live provider catalogs (~2s/test) — stubbed cached_provider_model_ids/provider_model_ids/fetch_api_models. - test_telegram_noise_filter.py: 15-platform copy-paste matrix over shared gateway.run logic → 3 representative platforms (55s → 3.9s). - test_gateway_shutdown.py: stop()'s 5s interrupt-deadline loop spun on MagicMock agents — interrupt.side_effect now clears _running_agents (22s → 1.0s). - test_gateway_inactivity_timeout.py poll-harness timings shrunk 3-5x (24s → 1.1s); test_mcp_stability.py backoff/SIGTERM-grace sleeps patched (15.4s → 2.5s); test_async_delegation.py negative-drain wait 5s → 0.5s. - test_telegram_init_deadline.py: loop-block margin restored to 1.0s with rationale comment — the watchdog-dump assertion needs the loop blocked well past deadline+grace under parallel load (flaked once in the 40-worker verification run at a 0.2s margin). Verification: full hermetic suite via scripts/run_tests.sh — 2,438 files, 21,718 tests passed, 0 failed, 293.9s wall. Suite totals vs original baseline: 46,820 → 19,757 test functions (−57.8%), wall 583.5s → 293.9s (−50%), subprocess CPU 13,564s → 11,623s.
55 lines
1.9 KiB
Python
55 lines
1.9 KiB
Python
"""MoA one-shot model override must be restored on both success and failure.
|
|
|
|
These exercise the real ``GatewayRunner._restore_moa_one_shot`` helper that the
|
|
message-handling ``finally`` block calls, so they prove the production logic —
|
|
not a re-implementation of it. The bug being guarded: the restore used to live
|
|
in the ``try`` block, so a turn that raised skipped it and the MoA override
|
|
leaked permanently (every later message silently fanned out through MoA).
|
|
"""
|
|
|
|
from types import SimpleNamespace
|
|
from unittest.mock import MagicMock
|
|
|
|
from gateway.run import GatewayRunner
|
|
|
|
|
|
def _make_runner():
|
|
"""Minimal GatewayRunner with only the fields _restore_moa_one_shot reads."""
|
|
runner = object.__new__(GatewayRunner)
|
|
runner._session_model_overrides = {}
|
|
runner._evict_cached_agent = MagicMock()
|
|
return runner
|
|
|
|
|
|
def _make_event(moa_disable=False, moa_restore=None):
|
|
event = SimpleNamespace()
|
|
if moa_disable:
|
|
event._moa_disable_after_turn = True
|
|
event._moa_restore_override = moa_restore
|
|
return event
|
|
|
|
|
|
def test_restore_runs_from_finally_even_when_turn_raises():
|
|
"""The whole point of the fix: a raising turn still reverts the override.
|
|
|
|
Mirrors the real call site — the restore is invoked from a ``finally`` block,
|
|
so it fires after an exception propagates out of the turn body.
|
|
"""
|
|
runner = _make_runner()
|
|
key = "agent:main:telegram:dm:999"
|
|
runner._session_model_overrides[key] = {"provider": "moa", "model": "default"}
|
|
event = _make_event(
|
|
moa_disable=True,
|
|
moa_restore={"provider": "openrouter", "model": "gpt-4"},
|
|
)
|
|
|
|
with __import__("pytest").raises(RuntimeError):
|
|
try:
|
|
raise RuntimeError("provider error mid-turn")
|
|
finally:
|
|
runner._restore_moa_one_shot(event, key)
|
|
|
|
assert runner._session_model_overrides[key] == {
|
|
"provider": "openrouter",
|
|
"model": "gpt-4",
|
|
}
|