mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-04-25 00:51:20 +00:00
2558d28a9b
Production fixes: - Add clear_session_context() to hermes_logging.py (fixes 48 teardown errors) - Add clear_session() to tools/approval.py (fixes 9 setup errors) - Add SyncError M_UNKNOWN_TOKEN check to Matrix _sync_loop (bug fix) - Fall back to inline api_key in named custom providers when key_env is absent (runtime_provider.py) Test fixes: - test_memory_user_id: use builtin+external provider pair, fix honcho peer_name override test to match production behavior - test_display_config: remove TestHelpers for non-existent functions - test_auxiliary_client: fix OAuth tokens to match _is_oauth_token patterns, replace get_vision_auxiliary_client with resolve_vision_provider_client - test_cli_interrupt_subagent: add missing _execution_thread_id attr - test_compress_focus: add model/provider/api_key/base_url/api_mode to mock compressor - test_auth_provider_gate: add autouse fixture to clean Anthropic env vars that leak from CI secrets - test_opencode_go_in_model_list: accept both 'built-in' and 'hermes' source (models.dev API unavailable in CI) - test_email: verify email Platform enum membership instead of source inspection (build_channel_directory now uses dynamic enum loop) - test_feishu: add bot_added/bot_deleted handler mocks to _Builder - test_ws_auth_retry: add AsyncMock for sync_store.get_next_batch, add _pending_megolm and _joined_rooms to Matrix adapter mocks - test_restart_drain: monkeypatch-delete INVOCATION_ID (systemd sets this in CI, changing the restart call signature) - test_session_hygiene: add user_id to SessionSource - test_session_env: use relative baseline for contextvar clear check (pytest-xdist workers share context)
144 lines
4.9 KiB
Python
144 lines
4.9 KiB
Python
"""Tests for focus_topic flowing through the compressor.
|
|
|
|
Verifies that _generate_summary and compress accept and use the focus_topic
|
|
parameter correctly. Inspired by Claude Code's /compact <focus>.
|
|
"""
|
|
|
|
from unittest.mock import MagicMock, patch
|
|
|
|
from agent.context_compressor import ContextCompressor
|
|
|
|
|
|
def _make_compressor():
|
|
"""Create a ContextCompressor with minimal state for testing."""
|
|
compressor = ContextCompressor.__new__(ContextCompressor)
|
|
compressor.protect_first_n = 2
|
|
compressor.protect_last_n = 5
|
|
compressor.tail_token_budget = 20000
|
|
compressor.context_length = 200000
|
|
compressor.threshold_percent = 0.80
|
|
compressor.threshold_tokens = 160000
|
|
compressor.max_summary_tokens = 10000
|
|
compressor.quiet_mode = True
|
|
compressor.compression_count = 0
|
|
compressor.last_prompt_tokens = 0
|
|
compressor._previous_summary = None
|
|
compressor._summary_failure_cooldown_until = 0.0
|
|
compressor.summary_model = None
|
|
compressor.model = "test-model"
|
|
compressor.provider = "test"
|
|
compressor.base_url = "http://localhost"
|
|
compressor.api_key = "test-key"
|
|
compressor.api_mode = "chat_completions"
|
|
return compressor
|
|
|
|
|
|
def test_focus_topic_injected_into_summary_prompt():
|
|
"""When focus_topic is provided, the LLM prompt includes focus guidance."""
|
|
compressor = _make_compressor()
|
|
turns = [
|
|
{"role": "user", "content": "Tell me about the database schema"},
|
|
{"role": "assistant", "content": "The schema has tables: users, orders, products."},
|
|
]
|
|
|
|
captured_prompt = {}
|
|
|
|
def mock_call_llm(**kwargs):
|
|
captured_prompt["messages"] = kwargs["messages"]
|
|
resp = MagicMock()
|
|
resp.choices = [MagicMock()]
|
|
resp.choices[0].message.content = "## Goal\nUnderstand DB schema."
|
|
return resp
|
|
|
|
with patch("agent.context_compressor.call_llm", mock_call_llm):
|
|
result = compressor._generate_summary(turns, focus_topic="database schema")
|
|
|
|
assert result is not None
|
|
prompt_text = captured_prompt["messages"][0]["content"]
|
|
assert 'FOCUS TOPIC: "database schema"' in prompt_text
|
|
assert "PRIORITISE" in prompt_text
|
|
assert "60-70%" in prompt_text
|
|
|
|
|
|
def test_no_focus_topic_no_injection():
|
|
"""Without focus_topic, the prompt doesn't contain focus guidance."""
|
|
compressor = _make_compressor()
|
|
turns = [
|
|
{"role": "user", "content": "Hello"},
|
|
{"role": "assistant", "content": "Hi"},
|
|
]
|
|
|
|
captured_prompt = {}
|
|
|
|
def mock_call_llm(**kwargs):
|
|
captured_prompt["messages"] = kwargs["messages"]
|
|
resp = MagicMock()
|
|
resp.choices = [MagicMock()]
|
|
resp.choices[0].message.content = "## Goal\nGreeting."
|
|
return resp
|
|
|
|
with patch("agent.context_compressor.call_llm", mock_call_llm):
|
|
result = compressor._generate_summary(turns)
|
|
|
|
prompt_text = captured_prompt["messages"][0]["content"]
|
|
assert "FOCUS TOPIC" not in prompt_text
|
|
|
|
|
|
def test_compress_passes_focus_to_generate_summary():
|
|
"""compress() passes focus_topic through to _generate_summary."""
|
|
compressor = _make_compressor()
|
|
|
|
# Track what _generate_summary receives
|
|
received_kwargs = {}
|
|
original_generate = compressor._generate_summary
|
|
|
|
def tracking_generate(turns, **kwargs):
|
|
received_kwargs.update(kwargs)
|
|
return "## Goal\nTest."
|
|
|
|
compressor._generate_summary = tracking_generate
|
|
|
|
messages = [
|
|
{"role": "system", "content": "System prompt"},
|
|
{"role": "user", "content": "first"},
|
|
{"role": "assistant", "content": "reply1"},
|
|
{"role": "user", "content": "second"},
|
|
{"role": "assistant", "content": "reply2"},
|
|
{"role": "user", "content": "third"},
|
|
{"role": "assistant", "content": "reply3"},
|
|
{"role": "user", "content": "fourth"},
|
|
{"role": "assistant", "content": "reply4"},
|
|
]
|
|
|
|
compressor.compress(messages, current_tokens=100000, focus_topic="authentication flow")
|
|
|
|
assert received_kwargs.get("focus_topic") == "authentication flow"
|
|
|
|
|
|
def test_compress_none_focus_by_default():
|
|
"""compress() passes None focus_topic by default."""
|
|
compressor = _make_compressor()
|
|
|
|
received_kwargs = {}
|
|
|
|
def tracking_generate(turns, **kwargs):
|
|
received_kwargs.update(kwargs)
|
|
return "## Goal\nTest."
|
|
|
|
compressor._generate_summary = tracking_generate
|
|
|
|
messages = [
|
|
{"role": "system", "content": "System prompt"},
|
|
{"role": "user", "content": "first"},
|
|
{"role": "assistant", "content": "reply1"},
|
|
{"role": "user", "content": "second"},
|
|
{"role": "assistant", "content": "reply2"},
|
|
{"role": "user", "content": "third"},
|
|
{"role": "assistant", "content": "reply3"},
|
|
{"role": "user", "content": "fourth"},
|
|
{"role": "assistant", "content": "reply4"},
|
|
]
|
|
|
|
compressor.compress(messages, current_tokens=100000)
|
|
|
|
assert received_kwargs.get("focus_topic") is None
|