Merge pull request #2220 from MemPalace/fix/windows-hybrid-search-empty-results

fix(tests): harden hybrid search against empty Windows Chroma reads
This commit is contained in:
Igor Lins e Silva 2026-08-11 11:47:26 -03:00 committed by GitHub
commit 9a3afc9d64
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 34 additions and 7 deletions

View File

@ -8,6 +8,7 @@ search would have found.
"""
from mempalace.palace import (
get_backend_for_palace,
get_closets_collection,
get_collection,
upsert_closet_lines,
@ -15,6 +16,29 @@ from mempalace.palace import (
from mempalace.searcher import _hybrid_rank, search_memories
def _close_palace(palace_path: str) -> None:
"""Release chromadb client handles so the next open rebuilds from disk.
Windows CI intermittently returns zero hybrid hits right after a fast
seed write (same class of flake as "Nothing found on disk" on tiny
closet collections). Closing the cached client forces the next
``search_memories`` open to re-read segments that have been flushed.
"""
try:
get_backend_for_palace(palace_path).close_palace(palace_path)
except Exception:
pass
def _search(query: str, palace: str, **kwargs):
"""Search, retrying once after a client reopen if results are empty."""
result = search_memories(query, palace, **kwargs)
if result.get("results"):
return result
_close_palace(palace)
return search_memories(query, palace, **kwargs)
def _seed_drawers(palace_path):
"""Insert 4 short drawers with deterministic content."""
col = get_collection(palace_path, create=True)
@ -33,6 +57,7 @@ def _seed_drawers(palace_path):
{"wing": "backend", "room": "queue", "source_file": "fixture_D4.md"},
],
)
_close_palace(palace_path)
def _seed_strong_closet_for(palace_path, drawer_id, source_file, topics):
@ -65,6 +90,7 @@ def _seed_strong_closet_for(palace_path, drawer_id, source_file, topics):
}
],
)
_close_palace(palace_path)
# ── core invariant: closets can only HELP, never HIDE ─────────────────────
@ -75,7 +101,7 @@ class TestHybridInvariant:
palace = str(tmp_path / "palace")
_seed_drawers(palace)
# No closets created.
result = search_memories("Kafka rebalance timeout", palace, n_results=3)
result = _search("Kafka rebalance timeout", palace, n_results=3)
ids = [h["source_file"] for h in result["results"]]
assert ids, "should return results"
assert "fixture_D4.md" in ids, "direct drawer search alone should surface the Kafka drawer"
@ -92,7 +118,7 @@ class TestHybridInvariant:
source_file="fixture_D3.md",
topics=["Kafka queue tuning", "consumer rebalance config"],
)
result = search_memories("Kafka consumer rebalance timeout", palace, n_results=5)
result = _search("Kafka consumer rebalance timeout", palace, n_results=5)
ids = [h["source_file"] for h in result["results"]]
assert "fixture_D4.md" in ids, (
"D4 must appear — direct drawer search alone would rank it first. "
@ -110,8 +136,9 @@ class TestHybridInvariant:
source_file="fixture_D1.md",
topics=["JWT auth tokens", "session expiry", "authentication service"],
)
result = search_memories("JWT auth tokens expiry", palace, n_results=3)
result = _search("JWT auth tokens expiry", palace, n_results=3)
ids = [h["source_file"] for h in result["results"]]
assert ids, f"expected hybrid hits after seeding drawers+closets; got {result!r}"
assert ids[0] == "fixture_D1.md"
top = result["results"][0]
assert top["matched_via"] == "drawer+closet"
@ -131,7 +158,7 @@ class TestClosetMetadata:
source_file="fixture_D1.md",
topics=["JWT auth tokens", "session expiry", "authentication service"],
)
result = search_memories("JWT auth tokens expiry", palace, n_results=2)
result = _search("JWT auth tokens expiry", palace, n_results=2)
top = result["results"][0]
assert top["source_file"] == "fixture_D1.md"
assert top["matched_via"] == "drawer+closet"
@ -142,7 +169,7 @@ class TestClosetMetadata:
palace = str(tmp_path / "palace")
_seed_drawers(palace)
# No closets
result = search_memories("TanStack Query", palace, n_results=2)
result = _search("TanStack Query", palace, n_results=2)
assert result["results"]
for h in result["results"]:
assert h["matched_via"] == "drawer"
@ -157,7 +184,7 @@ class TestSourceFileFilter:
def test_source_file_filter_excludes_other_sources(self, tmp_path):
palace = str(tmp_path / "palace")
_seed_drawers(palace)
result = search_memories(
result = _search(
"Kafka consumer rebalance timeout",
palace,
n_results=5,
@ -179,7 +206,7 @@ class TestSourceFileFilter:
source_file="fixture_D1.md",
topics=["Kafka queue tuning", "consumer rebalance config"],
)
result = search_memories(
result = _search(
"Kafka consumer rebalance",
palace,
n_results=5,