Files
odysseus/tests/conftest.py
T
Joshua Valderrama 35b4dd2824 fix: session context drifting — messages leaking between chats (#135) (#267)
* docs: add implementation plan for fixing chat context drifting (#135)

* fix: make Session.history immutable + fix {}.history crash

- Session.history now exposes a COPY of the internal _history list
- add_message() replaces history with a fresh copy each time
- get_context_messages() derives from _history directly
- replace_messages() updates both _history and history
- truncate_messages() updates both _history and history
- _persist_message() line 207: fixed {}.history fallback crash
- Added 11 tests for session isolation and edge cases

Addresses #135 root cause #1: shared mutable references

* fix: task scheduler uses SessionManager methods instead of overwriting sessions

- Added ensure_task_session() to SessionManager (checks cache first)
- Task scheduler now uses ensure_task_session() instead of direct dict assignment
- Task scheduler now uses SessionManager.add_message() for message persistence
- Removed direct sess_obj.history.append() that was silently losing data

Addresses #135 root causes #2 and #3

* fix: add age guard to cleanup_empty_sessions — don't delete sessions <1h old

Prevents the cleanup task from deleting sessions that were just created
and haven't received any messages yet (message_count == 0).

Addresses #135 root cause #5

* test: comprehensive session isolation tests (10/10 passing)

* refactor: consolidate _session_manager into singleton pattern

- Added set_session_manager_instance / get_session_manager_instance to core/models
- kept backward-compat aliases (set_session_manager, get_session_manager)
- session_manager.py re-exports the singleton functions
- ai_interaction.set_session_manager now syncs with the core singleton
- context_compactor uses get_session_manager_instance() instead of getattr hack
- app.py initializes the singleton once

Addresses #135 root cause #4: fragile global wiring

* test: add concurrent session isolation integration tests

Verifies:
- Concurrent add_message to different sessions doesn't cross-contaminate
- Rapid parallel writes maintain isolation
- Read-write concurrent access is safe

All 3 async tests pass, proving the immutable history fix works under concurrency

* fix: pre-import core.models in conftest to prevent test pollution

test_agent_loop.py stubs sys.modules['core.models'] = MagicMock() at
module level during collection. Any test collected after it imports
Session as a MagicMock. Pre-importing core.models in conftest.py
before test_agent_loop.py's module-level code runs prevents this.

* fix: make .history authoritative mutable list, address PR review

Per review feedback: keep .history as the authoritative mutable list so
existing code doing .history.pop(), .history = [...], etc. still works.
Fix the cross-contamination bug by ensuring __post_init__() gives each
Session its OWN unique history list (never shared).

Changes:
- core/models.py: .history IS the authoritative list. _history aliases it.
  Each Session gets its own list in __post_init__.
- core/session_manager.py: add_message() delegates to Session.add_message()
  instead of appending directly — no double-append, single source of truth.
- tests/test_session_manager.py: updated test to reflect that .history
  references see new messages (same list, not a snapshot).
- docs/plans/2026-06-01-fix-chat-context-drifting.md: removed (not for
  shipping — useful design context but too much process/doc to ship).

All 272 tests pass (3 pre-existing failures unrelated).

* Fix session manager message persistence

* Fix session history alias regressions

* Fix session history aliasing and task delivery
2026-06-09 14:12:52 +01:00

95 lines
4.2 KiB
Python

"""Shared test configuration - ensure project root is on sys.path and stub heavy deps."""
import sys
import os
import types
import importlib.util
from unittest.mock import MagicMock
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
# Importing core.database below runs init_db() at import time, and its default
# (sqlite:///./data/app.db) can't be opened in a clean worktree because SQLite
# won't create the missing ./data parent dir - pytest then dies during
# collection, before any test module loads. Default to an in-memory DB for the
# test session so collection is deterministic and writes no repo-local
# artifacts. An explicit DATABASE_URL (a real test/CI database) is preserved.
# This only unblocks collection/import-time init; it does not provide a shared
# file-backed DB across processes - tests needing that must set DATABASE_URL.
os.environ.setdefault("DATABASE_URL", "sqlite:///:memory:")
# Pre-import real heavy modules BEFORE any test file's module-level stubs can
# replace them with MagicMock. Some test files (e.g. test_llm_core_sanitize_*)
# stub sqlalchemy/core.database at module scope with `if mod not in sys.modules`,
# which fires during collection. If the real module hasn't been imported yet,
# the stub wins and contaminates every subsequent test that needs the real ORM.
try:
import sqlalchemy # noqa: F401
import sqlalchemy.orm # noqa: F401
import core.database # noqa: F401
except ImportError:
pass # not installed - the stubs below will handle it
def _has_module(mod_name: str) -> bool:
try:
return importlib.util.find_spec(mod_name) is not None
except (ImportError, ValueError):
return False
# Stub optional dependencies only when they are not installed. Do not replace
# real FastAPI/Starlette/Pydantic modules: route tests import their subpackages.
for mod_name in [
"sqlalchemy", "sqlalchemy.orm", "sqlalchemy.types", "sqlalchemy.ext", "sqlalchemy.ext.declarative",
"sqlalchemy.ext.hybrid", "sqlalchemy.sql", "sqlalchemy.sql.expression",
"sqlalchemy.sql.sqltypes", "bcrypt", "pyotp",
"httpx", "fastapi", "fastapi.responses", "fastapi.routing",
"starlette", "starlette.responses", "starlette.middleware", "starlette.middleware.base",
"pydantic",
]:
if mod_name not in sys.modules and not _has_module(mod_name):
sys.modules[mod_name] = MagicMock()
if "src.database" not in sys.modules:
_db = types.ModuleType("src.database")
_db.SessionLocal = MagicMock()
_db.ModelEndpoint = MagicMock()
sys.modules["src.database"] = _db
# Pre-import core.models before test_agent_loop.py's module-level stubs
# run (it replaces sys.modules['core.models'] with a MagicMock during
# collection, which breaks session import in subsequent tests).
import core.models # noqa: E402
def pytest_configure(config):
"""Register the dynamic taxonomy ``sub_*`` markers before collection.
The stable ``area_*`` markers are declared in ``pyproject.toml``. The
per-file ``sub_*`` markers are derived from the test filenames here so that
unknown-mark warnings still surface genuine typos outside the taxonomy. This
only registers marker names; it imports no production module.
"""
import pathlib
from tests._taxonomy import discover_markers
tests_dir = pathlib.Path(__file__).parent
paths = list(tests_dir.rglob("test_*.py")) + list(tests_dir.rglob("*_test.py"))
for marker_name in discover_markers(paths):
if marker_name.startswith("sub_"):
config.addinivalue_line("markers", f"{marker_name}: taxonomy sub-area marker")
def pytest_collection_modifyitems(config, items):
"""Tag each collected test with its taxonomy ``area_*`` and ``sub_*`` markers.
Collection-time only: this adds markers and nothing else. It does not skip,
reorder, or deselect tests, mutate fixtures or the environment, or import any
production module. See ``tests/_taxonomy.py`` for the classification rules.
"""
import pytest
from tests._taxonomy import markers_for_path
for item in items:
path = getattr(item, "path", None) or item.fspath
for marker_name in markers_for_path(path):
item.add_marker(getattr(pytest.mark, marker_name))