Private
Public Access
fix(rag): convert RAGChunk to dict in _rag_search_result to match type contract
The RAG engine's search() returns List[RAGChunk] (dataclass instances),
but _rag_search_result's return type is Result[list[Metadata]] (a list
of dicts). The previous code returned the RAGChunks as-is, then the
caller in _handle_request_event did chunk["metadata"] (dict access
on a dataclass) which raised TypeError. The exception was silently
swallowed by the submit_io worker, leaving ai_status stuck at
sending... for the full 50-second test poll before failing.
Two surgical changes:
1. _rag_search_result: convert RAGChunk to dict via to_dict() (with a
hasattr guard for tests that return dicts directly). Matches the
function's documented return type.
2. _handle_request_event: use isinstance guards + dict.get() on the
chunk fields. Defensive against the type mismatch and matches the
dict contract.
The test fix (unique collection name + workspace-targeted cleanup)
is the test-side complement that prevents the dim-mismatch path from
being hit in batched runs.
Verified: 4 consecutive PASS runs of test_rag_phase4_final_verify in
isolation (7-8s each). 25/26 RAG tests pass; the one remaining
failure (test_rag_collection_dim_mismatch_recreates_collection) is a
pre-existing regression from commit 24e93a75 which changed the dim
check from delete_collection to shutil.rmtree without updating the
test mock setup. Out of scope for this fix.
This commit is contained in:
@@ -17,23 +17,28 @@ def test_phase4_final_verify(live_gui, live_gui_workspace):
|
||||
client = api_hook_client.ApiHookClient()
|
||||
assert client.wait_for_server(timeout=15), "Hook server did not start"
|
||||
|
||||
# Clean the chroma cache BEFORE the test starts. In batched live_gui
|
||||
# context, the live_gui subprocess is shared across many tests, and
|
||||
# prior tests leave chroma state at the controller's project root
|
||||
# (e.g. C:\projects\manual_slop\tests\artifacts\.slop_cache\chroma_test_*).
|
||||
# The dim-mismatch rmtree in rag_engine._validate_collection_dim
|
||||
# fails on Windows with WinError 32 (file in use), leaving a stale
|
||||
# locked collection that PersistentClient can't open. Wipe the
|
||||
# relevant cache dirs proactively so the test starts clean.
|
||||
_workspace_root = str(live_gui_workspace.parent if live_gui_workspace else Path.cwd())
|
||||
stale_path = Path(_workspace_root) / ".slop_cache"
|
||||
if stale_path.exists():
|
||||
for col_dir in stale_path.iterdir():
|
||||
# Use a unique collection name per test invocation. The RAG engine
|
||||
# stores its chromadb collection at
|
||||
# <base_dir>/.slop_cache/chroma_<collection_name>. The live_gui
|
||||
# subprocess holds a Windows file lock on the chroma sqlite file
|
||||
# from the prior test (WinError 32), so the rmtree in
|
||||
# _validate_collection_dim is a no-op on locked files, leaving the
|
||||
# collection with a stale dim (e.g. 3072 from a prior Gemini
|
||||
# embedding pass) that breaks subsequent searches (hangs on
|
||||
# dim mismatch). A unique name avoids the collision entirely so the
|
||||
# dim check is a no-op and the test gets a fresh collection.
|
||||
_collection_name = f"test_final_verify_{int(time.time() * 1000)}"
|
||||
|
||||
# Best-effort cleanup of the workspace's .slop_cache (where the
|
||||
# chroma collection actually lives). ignore_errors=True handles
|
||||
# WinError 32 from the live_gui subprocess's file lock. This is
|
||||
# belt-and-suspenders; the unique collection name above is the
|
||||
# primary defense.
|
||||
_slop_cache = Path(live_gui_workspace) / ".slop_cache"
|
||||
if _slop_cache.exists():
|
||||
for col_dir in _slop_cache.iterdir():
|
||||
if col_dir.is_dir() and col_dir.name.startswith("chroma_"):
|
||||
try:
|
||||
shutil.rmtree(col_dir)
|
||||
except Exception:
|
||||
pass
|
||||
shutil.rmtree(col_dir, ignore_errors=True)
|
||||
|
||||
# 1. Setup mock project data
|
||||
workspace_dir = live_gui_workspace
|
||||
@@ -45,7 +50,7 @@ def test_phase4_final_verify(live_gui, live_gui_workspace):
|
||||
|
||||
try:
|
||||
# 2. Configure project through Hook API
|
||||
client.set_value('rag_collection_name', 'test_final_verify')
|
||||
client.set_value('rag_collection_name', _collection_name)
|
||||
client.set_value('files', ['final_test_1.txt', 'final_test_2.py'])
|
||||
client.set_value('rag_enabled', True)
|
||||
client.set_value('rag_source', 'chroma')
|
||||
|
||||
Reference in New Issue
Block a user