test: preserve batched context ties and duplicate hits

This commit is contained in:
Teknium
2026-09-07 02:32:55 -07:00
parent 20af238599
commit 7e73ef4676
2 changed files with 18 additions and 1 deletions
@@ -25,3 +25,20 @@ def test_search_context_batches_hits_without_cross_session_neighbors(tmp_path):
projected = db.search_messages("needleprobe", fields=["id"], limit=20)
assert len(projected) == len(rows)
assert not any("WITH target AS" in q for q in sql)
def test_context_batches_keep_tied_timestamp_order_and_duplicate_hits(tmp_path):
with SessionDB(db_path=tmp_path / "ties.db") as db:
db.create_session("ties", "cli")
db.append_messages_batch("ties", [
{"role": "user", "content": f"needleprobe {i}", "timestamp": 10.0}
for i in range(502)
])
rows = db.search_messages("needleprobe", limit=502)
assert len(rows) == 502
by_id = sorted(rows, key=lambda row: row["id"])
for i, row in enumerate(by_id):
expected = [f"needleprobe {j}" for j in range(max(0, i - 1), min(502, i + 2))]
assert [r["content"] for r in row["context"]] == expected
duplicate = db._finalize_search_matches([dict(by_id[250]), dict(by_id[250])])
assert duplicate[0]["context"] == duplicate[1]["context"] == by_id[250]["context"]