241 lines
8.3 KiB
Python
241 lines
8.3 KiB
Python
from __future__ import annotations
|
|
|
|
import os
|
|
from collections.abc import AsyncIterator, Iterable
|
|
from dataclasses import replace
|
|
from uuid import uuid4
|
|
|
|
import pytest
|
|
from qdrant_client import AsyncQdrantClient
|
|
from sqlalchemy.ext.asyncio import AsyncEngine
|
|
|
|
from ragcore.application.indexes import IndexBuildService, IndexBuildState
|
|
from ragcore.config import Settings
|
|
from ragcore.infrastructure.database import create_database_engine, create_session_factory
|
|
from ragcore.infrastructure.database.indexes import PostgresIndexOperationStore
|
|
from ragcore.infrastructure.qdrant import (
|
|
CollectionKind,
|
|
CollectionProfile,
|
|
IndexPayload,
|
|
IndexPoint,
|
|
QdrantIndexAdapter,
|
|
)
|
|
from ragcore.infrastructure.qdrant.rebuild import QdrantRebuildStore
|
|
|
|
|
|
class FixtureSource:
|
|
def __init__(self, points: Iterable[IndexPoint]) -> None:
|
|
self.points = tuple(points)
|
|
|
|
async def iter_points(self, profile: CollectionProfile) -> AsyncIterator[IndexPoint]:
|
|
for point in self.points:
|
|
yield point
|
|
|
|
|
|
class QdrantGoldenSmoke:
|
|
def __init__(self, client: AsyncQdrantClient) -> None:
|
|
self.client = client
|
|
|
|
async def run(self, profile: CollectionProfile) -> None:
|
|
response = await self.client.query_points(
|
|
collection_name=profile.collection_name,
|
|
query=[0.1, 0.2, 0.3, 0.4],
|
|
using="dense",
|
|
limit=1,
|
|
with_payload=True,
|
|
timeout=30,
|
|
)
|
|
if not response.points:
|
|
raise AssertionError("golden smoke query returned no points")
|
|
|
|
|
|
def make_profile() -> CollectionProfile:
|
|
return CollectionProfile(
|
|
kind=CollectionKind.DENSE,
|
|
profile="rebuild-test",
|
|
version=uuid4().hex[:12],
|
|
dimensions=4,
|
|
model_digest="a" * 64,
|
|
)
|
|
|
|
|
|
def make_point(profile: CollectionProfile, index: int) -> IndexPoint:
|
|
point_id = uuid4()
|
|
return IndexPoint(
|
|
point_id=point_id,
|
|
dense_vector=(0.1 + index, 0.2, 0.3, 0.4),
|
|
sparse_indices=(index + 1, index + 9),
|
|
sparse_values=(0.8, 0.3),
|
|
payload=IndexPayload(
|
|
workspace_id=uuid4(),
|
|
space_id=uuid4(),
|
|
document_id=uuid4(),
|
|
document_version_id=uuid4(),
|
|
chunk_id=point_id,
|
|
parent_chunk_id=None,
|
|
source_id=uuid4(),
|
|
status="published",
|
|
content_type="text/markdown",
|
|
language="en",
|
|
trust_level="source",
|
|
location={"page": index + 1},
|
|
parser_version="text-v1",
|
|
chunk_profile_version=profile.version,
|
|
embedding_model_digest=profile.model_digest,
|
|
text=f"A safe indexed passage {index}.",
|
|
excerpt=f"A safe indexed passage {index}.",
|
|
),
|
|
)
|
|
|
|
|
|
@pytest.fixture
|
|
async def qdrant() -> AsyncIterator[AsyncQdrantClient]:
|
|
client = AsyncQdrantClient(
|
|
url=os.getenv("RAGCORE_QDRANT_URL", "http://qdrant:6333"),
|
|
api_key=os.getenv("RAGCORE_QDRANT_API_KEY", "replace-me"),
|
|
timeout=60,
|
|
)
|
|
try:
|
|
yield client
|
|
finally:
|
|
await client.close()
|
|
|
|
|
|
@pytest.fixture
|
|
async def db_engine() -> AsyncIterator[AsyncEngine]:
|
|
engine = create_database_engine(Settings(env="test"))
|
|
try:
|
|
yield engine
|
|
finally:
|
|
await engine.dispose()
|
|
|
|
|
|
@pytest.fixture
|
|
def operations(db_engine: AsyncEngine) -> PostgresIndexOperationStore:
|
|
return PostgresIndexOperationStore(create_session_factory(db_engine))
|
|
|
|
|
|
@pytest.mark.integration
|
|
async def test_rebuild_from_empty_validates_and_activates(
|
|
qdrant: AsyncQdrantClient,
|
|
operations: PostgresIndexOperationStore,
|
|
) -> None:
|
|
profile = make_profile()
|
|
points = (make_point(profile, 0), make_point(profile, 1))
|
|
store = QdrantRebuildStore(qdrant)
|
|
service = IndexBuildService(store, operations)
|
|
try:
|
|
report = await service.build(profile, FixtureSource(points), QdrantGoldenSmoke(qdrant))
|
|
|
|
assert report.state is IndexBuildState.ACTIVE
|
|
assert report.previous_collection is None
|
|
assert report.validation.expected_count == 2
|
|
assert report.validation.actual_count == 2
|
|
assert report.validation.passed
|
|
assert await store.current_alias_target(profile.alias_name) == profile.collection_name
|
|
finally:
|
|
await store.rollback(profile.alias_name, None)
|
|
await qdrant.delete_collection(profile.collection_name)
|
|
|
|
|
|
@pytest.mark.integration
|
|
async def test_alias_rollback_restores_previous_active_collection(
|
|
qdrant: AsyncQdrantClient,
|
|
operations: PostgresIndexOperationStore,
|
|
) -> None:
|
|
first = make_profile()
|
|
second = CollectionProfile(
|
|
kind=first.kind,
|
|
profile=first.profile,
|
|
version=uuid4().hex[:12],
|
|
dimensions=first.dimensions,
|
|
model_digest=first.model_digest,
|
|
)
|
|
first_points = (make_point(first, 0),)
|
|
second_points = (make_point(second, 1),)
|
|
store = QdrantRebuildStore(qdrant)
|
|
service = IndexBuildService(store, operations)
|
|
try:
|
|
await service.build(first, FixtureSource(first_points), QdrantGoldenSmoke(qdrant))
|
|
second_report = await service.build(
|
|
second, FixtureSource(second_points), QdrantGoldenSmoke(qdrant)
|
|
)
|
|
assert second_report.previous_collection == first.collection_name
|
|
|
|
await service.rollback(second_report)
|
|
|
|
assert await store.current_alias_target(first.alias_name) == first.collection_name
|
|
finally:
|
|
await store.rollback(first.alias_name, None)
|
|
await qdrant.delete_collection(first.collection_name)
|
|
await qdrant.delete_collection(second.collection_name)
|
|
|
|
|
|
@pytest.mark.integration
|
|
async def test_rebuild_rejects_direct_payload_tampering(
|
|
qdrant: AsyncQdrantClient,
|
|
operations: PostgresIndexOperationStore,
|
|
) -> None:
|
|
profile = make_profile()
|
|
point = make_point(profile, 0)
|
|
store = QdrantRebuildStore(qdrant)
|
|
service = IndexBuildService(store, operations)
|
|
try:
|
|
await service.build(profile, FixtureSource((point,)), QdrantGoldenSmoke(qdrant))
|
|
await qdrant.set_payload(
|
|
collection_name=profile.collection_name,
|
|
payload={"text": "tampered indexed content"},
|
|
points=[str(point.point_id)],
|
|
wait=True,
|
|
timeout=30,
|
|
)
|
|
valid, malformed = await store.list_points(profile)
|
|
assert valid == ()
|
|
assert len(malformed) == 1
|
|
assert malformed[0].point_id == point.point_id
|
|
assert "wire integrity" in malformed[0].reason
|
|
finally:
|
|
await store.rollback(profile.alias_name, None)
|
|
await qdrant.delete_collection(profile.collection_name)
|
|
|
|
|
|
@pytest.mark.integration
|
|
async def test_reconciliation_repairs_missing_changed_and_orphan_points(
|
|
qdrant: AsyncQdrantClient,
|
|
operations: PostgresIndexOperationStore,
|
|
) -> None:
|
|
profile = make_profile()
|
|
points = (make_point(profile, 0), make_point(profile, 1))
|
|
rogue = make_point(profile, 99)
|
|
store = QdrantRebuildStore(qdrant)
|
|
adapter = QdrantIndexAdapter(qdrant)
|
|
service = IndexBuildService(store, operations)
|
|
try:
|
|
await service.build(profile, FixtureSource(points), QdrantGoldenSmoke(qdrant))
|
|
await adapter.delete_points(profile, (points[0].point_id,))
|
|
await adapter.upsert_points(profile, (rogue,))
|
|
changed = replace(
|
|
points[1],
|
|
payload=replace(
|
|
points[1].payload,
|
|
text="A changed indexed passage.",
|
|
excerpt="A changed indexed passage.",
|
|
),
|
|
)
|
|
await adapter.upsert_points(profile, (changed,))
|
|
|
|
report = await service.reconcile(profile, FixtureSource(points))
|
|
|
|
assert report.missing_point_ids == (points[0].point_id,)
|
|
assert report.orphan_point_ids == (rogue.point_id,)
|
|
assert report.changed_point_ids == (points[1].point_id,)
|
|
# repaired_point_ids is deterministically sorted by str(uuid), not by
|
|
# fixture insertion order, since both ids are independently random.
|
|
assert set(report.repaired_point_ids) == {points[0].point_id, points[1].point_id}
|
|
assert report.repaired_point_ids == tuple(
|
|
sorted((points[0].point_id, points[1].point_id), key=str)
|
|
)
|
|
assert await store.count_points(profile) == 2
|
|
finally:
|
|
await store.rollback(profile.alias_name, None)
|
|
await qdrant.delete_collection(profile.collection_name) |