mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-31 19:16:29 +00:00
6b81590c55
Systematic prune per AGENTS.md test policy, one pass over every major test tree (gateway, hermes_cli, tools, agent, run_agent, plugins, cli, cron, tui_gateway, honcho/openviking, root-level): - DELETE: source-reading tests (read_text/getsource on prod files), change-detector tests (exact catalog counts, model-name snapshots, config version literals), mock-echo tests (assert a mock returns what it was told), assertion-free/trivial tests, near-duplicate parametrizations (boundaries + one representative kept), async/sync twin duplicates, cosmetic within-file variations. - KEEP (mandatory): security/redaction/approval guards, message-role alternation invariants, prompt-caching/deterministic-call-id invariants, issue-number regression tests (deduped), E2E tests. - 6 test files deleted outright (script-style/no-assert or fully redundant); conftest.py, fakes/, fixtures/ untouched. - tests/acp/conftest.py added: autouse fixture stubs the live models.dev/GitHub/Copilot/Anthropic inventory fetches that ACP server tests performed on every session create — test_server.py 147s → 3.4s, and the tests are now genuinely hermetic. - Sleep-based slowness shrunk where safe (codex_ttfb_watchdog, compression_concurrent_fork, etc.); no wall-clock assertion tightened. Verification: full hermetic suite via scripts/run_tests.sh — 2439 files, 31,130 tests passed, 0 failed, 0 flaky retries, 315s wall (baseline: 583s wall, 13,564s subprocess CPU).
122 lines
3.7 KiB
Python
122 lines
3.7 KiB
Python
"""Tests for the non-stream stale-call detector context estimator.
|
|
|
|
Covers:
|
|
- ``estimate_request_context_tokens`` for Chat Completions, Responses API,
|
|
bare lists, and mixed-shape dicts.
|
|
- ``AIAgent._compute_non_stream_stale_timeout`` with both legacy ``messages``
|
|
list and full ``api_kwargs`` dicts.
|
|
- The May 2026 default-base change (300s -> 90s) and the lowered
|
|
context-tier ceilings (450/600 -> 150/240).
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from pathlib import Path
|
|
|
|
|
|
|
|
def _write_config(tmp_path: Path, body: str) -> None:
|
|
hermes_home = tmp_path
|
|
(hermes_home / "config.yaml").write_text(body or "{}\n", encoding="utf-8")
|
|
|
|
|
|
def _make_agent(tmp_path: Path, **overrides):
|
|
from run_agent import AIAgent
|
|
kwargs = dict(
|
|
model="gpt-5.5",
|
|
provider="openai-codex",
|
|
api_key="sk-dummy",
|
|
base_url="https://chatgpt.com/backend-api/codex",
|
|
quiet_mode=True,
|
|
skip_context_files=True,
|
|
skip_memory=True,
|
|
platform="cli",
|
|
)
|
|
kwargs.update(overrides)
|
|
return AIAgent(**kwargs)
|
|
|
|
|
|
# ── estimator ──────────────────────────────────────────────────────────────
|
|
|
|
|
|
|
|
|
|
def test_estimator_responses_api_input():
|
|
from agent.chat_completion_helpers import estimate_request_context_tokens
|
|
payload = {
|
|
"model": "gpt-5.5",
|
|
"instructions": "i" * 1000,
|
|
"input": "x" * 4000,
|
|
"tools": [{"name": "t", "description": "d" * 200}],
|
|
}
|
|
# input(4000) + instructions(1000) + tools (~stringified) -> well over 1000 tokens
|
|
tokens = estimate_request_context_tokens(payload)
|
|
assert tokens >= 1200, f"Responses API estimator returned {tokens}"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_estimator_empty_inputs():
|
|
from agent.chat_completion_helpers import estimate_request_context_tokens
|
|
assert estimate_request_context_tokens({}) == 0
|
|
assert estimate_request_context_tokens([]) == 0
|
|
assert estimate_request_context_tokens(None) == 0
|
|
|
|
|
|
|
|
|
|
# ── default base + tier scaling ────────────────────────────────────────────
|
|
|
|
|
|
def test_default_base_is_90s(monkeypatch, tmp_path):
|
|
"""Default base stale timeout dropped from 300s to 90s (May 2026)."""
|
|
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
|
|
(tmp_path / ".env").write_text("", encoding="utf-8")
|
|
monkeypatch.delenv("HERMES_API_CALL_STALE_TIMEOUT", raising=False)
|
|
_write_config(tmp_path, "")
|
|
|
|
agent = _make_agent(tmp_path)
|
|
base, implicit = agent._resolved_api_call_stale_timeout_base()
|
|
assert base == 90.0
|
|
assert implicit is True
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_explicit_user_config_overrides_default(monkeypatch, tmp_path):
|
|
"""If the user explicitly sets a stale_timeout, the new defaults don't apply."""
|
|
monkeypatch.setenv("HERMES_HOME", str(tmp_path))
|
|
(tmp_path / ".env").write_text("", encoding="utf-8")
|
|
_write_config(tmp_path, """\
|
|
providers:
|
|
openai-codex:
|
|
stale_timeout_seconds: 1800
|
|
""")
|
|
monkeypatch.delenv("HERMES_API_CALL_STALE_TIMEOUT", raising=False)
|
|
|
|
import importlib
|
|
from hermes_cli import timeouts as to_mod
|
|
importlib.reload(to_mod)
|
|
|
|
agent = _make_agent(tmp_path)
|
|
assert agent._compute_non_stream_stale_timeout({"input": "hi"}) == 1800.0
|
|
|
|
|
|
# ── openai-codex gateway-scale stale floor ────────────────────────────────
|
|
|
|
|
|
|
|
|
|
def test_openai_codex_stale_floor_tiers():
|
|
from agent.chat_completion_helpers import openai_codex_stale_timeout_floor
|
|
|
|
assert openai_codex_stale_timeout_floor(55_000) == 900.0
|
|
assert openai_codex_stale_timeout_floor(120_000) == 1200.0
|