Files
FabledScribe/tests/test_services_retrieval_telemetry.py
T
bvandeusen 807f478cac
CI & Build / Python lint (push) Successful in 2s
CI & Build / integration (push) Successful in 16s
CI & Build / TypeScript typecheck (push) Successful in 32s
CI & Build / Python tests (push) Successful in 44s
CI & Build / Build & push image (push) Successful in 58s
feat(search): retrieval telemetry — log every semantic retrieval
Add retrieval_logs (migration 0068) + services/retrieval_telemetry with a
fire-and-forget record_retrieval(), wired into the MCP search tool
(source=mcp_search) and the REST search route (source=rest_search). Captures
query, effective params, and the per-result score distribution so KB-injection
thresholds can be tuned from data rather than guessed.

Scribe: project 2, milestone 93, task 1032.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Xz4j1H7pjYSjKsEpgcNH5E
2026-06-22 20:10:15 -04:00

113 lines
4.0 KiB
Python

"""Tests for services.retrieval_telemetry.
_build_payload is pure (no DB, no loop) and gets unit coverage. The persistence
path (_insert_retrieval_log + the RetrievalLog model / JSONB roundtrip) is an
integration test against real Postgres.
"""
from types import SimpleNamespace
import pytest
import pytest_asyncio
from scribe.services.retrieval_telemetry import (
_build_payload,
record_retrieval,
)
def _note(nid):
"""Minimal stand-in — _build_payload only reads .id."""
return SimpleNamespace(id=nid)
# ─── _build_payload (pure) ───────────────────────────────────────────────────
def test_build_payload_ranks_and_score_bounds():
results = [(0.91, _note(11)), (0.72, _note(22)), (0.55, _note(33))]
p = _build_payload(
user_id=7, source="mcp_search", query="hello", threshold=0.45,
limit=10, project_id=3, is_task=None, results=results, duration_ms=12.345,
)
assert p["result_count"] == 3
assert p["top_score"] == 0.91
assert p["min_score"] == 0.55
assert [it["rank"] for it in p["result_ids"]] == [0, 1, 2]
assert [it["id"] for it in p["result_ids"]] == [11, 22, 33]
assert p["duration_ms"] == 12.35 # rounded to 2dp
assert p["user_id"] == 7 and p["project_id"] == 3 and p["threshold"] == 0.45
def test_build_payload_empty_results():
p = _build_payload(
user_id=1, source="rest_search", query="x", threshold=0.3,
limit=5, project_id=None, is_task=False, results=[], duration_ms=None,
)
assert p["result_count"] == 0
assert p["top_score"] is None and p["min_score"] is None
assert p["result_ids"] == []
assert p["duration_ms"] is None
def test_build_payload_rounds_scores_to_5dp():
p = _build_payload(
user_id=1, source="mcp_search", query="q", threshold=0.45,
limit=1, project_id=None, is_task=None,
results=[(0.123456789, _note(1))], duration_ms=0.0,
)
assert p["result_ids"][0]["score"] == 0.12346
def test_record_retrieval_without_event_loop_is_safe():
"""Called from a sync context (no running loop) it must swallow and return,
never raise — telemetry can't be allowed to break a caller."""
# No event loop running in this plain sync test.
assert record_retrieval(
user_id=1, source="mcp_search", query="q", threshold=0.45,
limit=10, project_id=None, is_task=None,
results=[(0.9, _note(1))],
) is None
# ─── persistence (integration) ───────────────────────────────────────────────
@pytest_asyncio.fixture
async def _dispose_engine():
from scribe.models import engine
yield
await engine.dispose()
@pytest.mark.integration
@pytest.mark.asyncio
async def test_insert_retrieval_log_roundtrip(_dispose_engine):
from sqlalchemy import delete, select
from scribe.models import async_session
from scribe.models.retrieval_log import RetrievalLog
from scribe.services.retrieval_telemetry import _insert_retrieval_log
payload = _build_payload(
user_id=990001, source="mcp_search", query="pgvector tuning",
threshold=0.45, limit=10, project_id=None, is_task=None,
results=[(0.88, _note(501)), (0.61, _note(502))], duration_ms=9.9,
)
await _insert_retrieval_log(payload)
async with async_session() as s:
row = (
await s.execute(
select(RetrievalLog).where(RetrievalLog.user_id == 990001)
)
).scalars().first()
assert row is not None
assert row.source == "mcp_search"
assert row.result_count == 2
assert row.top_score == 0.88
# JSONB roundtrips as a list of dicts with the expected shape.
assert row.result_ids[0] == {"id": 501, "score": 0.88, "rank": 0}
assert row.created_at is not None # server_default now()
await s.execute(delete(RetrievalLog).where(RetrievalLog.user_id == 990001))
await s.commit()