mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-04-25 00:51:20 +00:00
Group A (3 tests): 'No LLM provider configured' RuntimeError - test_user_message_surrogates_sanitized, test_counters_initialized_in_init, test_openai_prompt_tokens_unchanged - Root cause: AIAgent.__init__ now requires base_url alongside api_key to skip resolve_provider_client() (which returns None when API keys are blanked in CI). Added base_url='http://localhost:1234/v1' to test agent construction. Group B (5 tests): Discord slash command auto-registration - test_auto_registers_missing_gateway_commands, test_auto_registered_command_*, test_register_skill_group_* - Root cause: xdist workers that loaded a discord mock WITHOUT app_commands.Command/Group caused _register_slash_commands() to fail silently. Added comprehensive shared discord mock in tests/gateway/conftest.py (same pattern as existing telegram mock). Group C (5 errors): Discord reply mode 'NoneType has no DMChannel' - All TestReplyToText tests - Root cause: FakeDMChannel was not a subclass of real discord.DMChannel, so isinstance() checks in _handle_message failed when running in full suite (real discord installed). Made FakeDMChannel inherit from discord.DMChannel when available. Removed fragile monkeypatch approach. Group D (2 tests): detect_provider_for_model wrong provider - test_openrouter_slug_match (got 'ai-gateway'), test_bare_name_gets_ openrouter_slug (got 'copilot') - Root cause: ai-gateway, copilot, and kilocode are multi-vendor aggregators that list other providers' models (OpenRouter-style slugs). They were being matched in Step 1 before OpenRouter. Added all three to _AGGREGATORS set so they're skipped like nous/openrouter. Group E (1 test): model_flow_custom StopIteration - test_model_flow_custom_saves_verified_v1_base_url - Root cause: 'Display name' prompt was added after the test was written. The input iterator had 5 answers but the flow now asks 6 questions. Added 6th empty string answer. Group F (1 test): Telegram proxy env assertion - test_uses_proxy_env_for_primary_and_fallback_transports - Root cause: _resolve_proxy_url() now checks TELEGRAM_PROXY first (via resolve_proxy_url('TELEGRAM_PROXY')). Test didn't clear this env var, allowing potential leakage from other tests in xdist workers. Added TELEGRAM_PROXY to the cleanup list.
128 lines
5.1 KiB
Python
128 lines
5.1 KiB
Python
"""Tests for context token tracking in run_agent.py's usage extraction.
|
|
|
|
The context counter (status bar) must show the TOTAL prompt tokens including
|
|
Anthropic's cached portions. This is an integration test for the token
|
|
extraction in run_conversation(), not the ContextCompressor itself (which
|
|
is tested in tests/agent/test_context_compressor.py).
|
|
"""
|
|
|
|
import sys
|
|
import types
|
|
from types import SimpleNamespace
|
|
|
|
sys.modules.setdefault("fire", types.SimpleNamespace(Fire=lambda *a, **k: None))
|
|
sys.modules.setdefault("firecrawl", types.SimpleNamespace(Firecrawl=object))
|
|
sys.modules.setdefault("fal_client", types.SimpleNamespace())
|
|
|
|
import run_agent
|
|
|
|
|
|
def _patch_bootstrap(monkeypatch):
|
|
monkeypatch.setattr(run_agent, "get_tool_definitions", lambda **kwargs: [{
|
|
"type": "function",
|
|
"function": {"name": "t", "description": "t", "parameters": {"type": "object", "properties": {}}},
|
|
}])
|
|
monkeypatch.setattr(run_agent, "check_toolset_requirements", lambda: {})
|
|
|
|
|
|
class _FakeAnthropicClient:
|
|
def close(self):
|
|
pass
|
|
|
|
|
|
class _FakeOpenAIClient:
|
|
"""Fake OpenAI client returned by mocked resolve_provider_client."""
|
|
api_key = "fake-codex-key"
|
|
base_url = "https://api.openai.com/v1"
|
|
_default_headers = None
|
|
|
|
|
|
def _make_agent(monkeypatch, api_mode, provider, response_fn):
|
|
_patch_bootstrap(monkeypatch)
|
|
if api_mode == "anthropic_messages":
|
|
monkeypatch.setattr("agent.anthropic_adapter.build_anthropic_client", lambda k, b=None: _FakeAnthropicClient())
|
|
if provider == "openai-codex":
|
|
monkeypatch.setattr(
|
|
"agent.auxiliary_client.resolve_provider_client",
|
|
lambda *a, **kw: (_FakeOpenAIClient(), "test-model"),
|
|
)
|
|
|
|
class _A(run_agent.AIAgent):
|
|
def __init__(self, *a, **kw):
|
|
kw.update(skip_context_files=True, skip_memory=True, max_iterations=4)
|
|
super().__init__(*a, **kw)
|
|
self._cleanup_task_resources = self._persist_session = lambda *a, **k: None
|
|
self._save_trajectory = self._save_session_log = lambda *a, **k: None
|
|
|
|
def run_conversation(self, msg, conversation_history=None, task_id=None):
|
|
self._interruptible_api_call = lambda kw: response_fn()
|
|
self._disable_streaming = True
|
|
return super().run_conversation(msg, conversation_history=conversation_history, task_id=task_id)
|
|
|
|
return _A(model="test-model", api_key="test-key", base_url="http://localhost:1234/v1", provider=provider, api_mode=api_mode)
|
|
|
|
|
|
def _anthropic_resp(input_tok, output_tok, cache_read=0, cache_creation=0):
|
|
usage_fields = {"input_tokens": input_tok, "output_tokens": output_tok}
|
|
if cache_read:
|
|
usage_fields["cache_read_input_tokens"] = cache_read
|
|
if cache_creation:
|
|
usage_fields["cache_creation_input_tokens"] = cache_creation
|
|
return SimpleNamespace(
|
|
content=[SimpleNamespace(type="text", text="ok")],
|
|
stop_reason="end_turn",
|
|
usage=SimpleNamespace(**usage_fields),
|
|
model="claude-sonnet-4-6",
|
|
)
|
|
|
|
|
|
# -- Anthropic: cached tokens must be included --
|
|
|
|
def test_anthropic_cache_read_and_creation_added(monkeypatch):
|
|
agent = _make_agent(monkeypatch, "anthropic_messages", "anthropic",
|
|
lambda: _anthropic_resp(3, 10, cache_read=15000, cache_creation=2000))
|
|
agent.run_conversation("hi")
|
|
assert agent.context_compressor.last_prompt_tokens == 17003 # 3+15000+2000
|
|
assert agent.session_prompt_tokens == 17003
|
|
|
|
|
|
def test_anthropic_no_cache_fields(monkeypatch):
|
|
agent = _make_agent(monkeypatch, "anthropic_messages", "anthropic",
|
|
lambda: _anthropic_resp(500, 20))
|
|
agent.run_conversation("hi")
|
|
assert agent.context_compressor.last_prompt_tokens == 500
|
|
|
|
|
|
def test_anthropic_cache_read_only(monkeypatch):
|
|
agent = _make_agent(monkeypatch, "anthropic_messages", "anthropic",
|
|
lambda: _anthropic_resp(5, 15, cache_read=17666, cache_creation=15))
|
|
agent.run_conversation("hi")
|
|
assert agent.context_compressor.last_prompt_tokens == 17686 # 5+17666+15
|
|
|
|
|
|
# -- OpenAI: prompt_tokens already total --
|
|
|
|
def test_openai_prompt_tokens_unchanged(monkeypatch):
|
|
resp = lambda: SimpleNamespace(
|
|
choices=[SimpleNamespace(index=0, message=SimpleNamespace(
|
|
role="assistant", content="ok", tool_calls=None, reasoning_content=None,
|
|
), finish_reason="stop")],
|
|
usage=SimpleNamespace(prompt_tokens=5000, completion_tokens=100, total_tokens=5100),
|
|
model="gpt-4o",
|
|
)
|
|
agent = _make_agent(monkeypatch, "chat_completions", "openrouter", resp)
|
|
agent.run_conversation("hi")
|
|
assert agent.context_compressor.last_prompt_tokens == 5000
|
|
|
|
|
|
# -- Codex: no cache fields, getattr returns 0 --
|
|
|
|
def test_codex_no_cache_fields(monkeypatch):
|
|
resp = lambda: SimpleNamespace(
|
|
output=[SimpleNamespace(type="message", content=[SimpleNamespace(type="output_text", text="ok")])],
|
|
usage=SimpleNamespace(input_tokens=3000, output_tokens=50, total_tokens=3050),
|
|
status="completed", model="gpt-5-codex",
|
|
)
|
|
agent = _make_agent(monkeypatch, "codex_responses", "openai-codex", resp)
|
|
agent.run_conversation("hi")
|
|
assert agent.context_compressor.last_prompt_tokens == 3000
|