feat: complete catalog-driven preprocessing

harness/tests/test_qdrant_vector_store.py

@@ -188,6 +188,194 @@ def _store(
     )
 
 
+def test_explicit_collections_route_reference_and_memory_independently():
+    calls = []
+
+    def request(method, url, *, json=None, timeout=None):
+        calls.append((method, url, json))
+        if method == "GET" and url.endswith("/collections/demo"):
+            return FakeResponse(404, {"status": "error"})
+        if method == "GET":
+            return FakeResponse(200, {
+                "result": {
+                    "config": {"params": {"vectors": {"size": 1024, "distance": "Cosine"}}},
+                    "payload_schema": {
+                        field: {"data_type": "keyword"}
+                        for field in (
+                            "content_hash", "document_id", "kind", "record_key", "record_kind",
+                            "vector_generation", "workspace_id", "workspace_revision",
+                        )
+                    },
+                }
+            })
+        if method == "POST" and url.endswith("/points/query"):
+            return FakeResponse(200, {"result": {"points": []}})
+        if method == "PUT" and "/points" in url:
+            return FakeResponse(200, {"result": {"status": "acknowledged"}})
+        raise AssertionError((method, url, json))
+
+    store = QdrantVectorStore(
+        base_url="http://qdrant:6333",
+        collections={"reference": "demo-reference", "memory": "demo-memory"},
+        workspace_id="demo",
+        workspace_revision="a" * 40,
+        expected_dimension=1024,
+        request=request,
+    )
+    store.search(["schema_records"], [0.2] * 1024, limit=1)
+    store.search(["memory"], [0.2] * 1024, limit=1)
+    store.upsert("memory", [_write_record("memory:1", "memory")])
+
+    urls = [url for _method, url, _payload in calls]
+    assert any("/collections/demo-reference/points/query" in url for url in urls)
+    assert any("/collections/demo-memory/points/query" in url for url in urls)
+    assert any("/collections/demo-memory/points" in url for url in urls)
+    assert not any("/collections/demo-reference/points?" in url for url in urls)
+
+
+def test_clear_drops_reference_without_deleting_memory_collection():
+    calls = []
+
+    def request(method, url, *, json=None, timeout=None):
+        calls.append((method, url))
+        if method == "GET" and url.endswith("/collections/demo"):
+            return FakeResponse(404, {"status": "error"})
+        if method == "GET" and url.endswith("/collections/demo-reference"):
+            return FakeResponse(200, {"result": {}})
+        if method == "DELETE" and url.endswith("/collections/demo-reference"):
+            return FakeResponse(200, {"status": "ok"})
+        raise AssertionError((method, url, json))
+
+    store = QdrantVectorStore(
+        base_url="http://qdrant:6333",
+        collections={"reference": "demo-reference", "memory": "demo-memory"},
+        workspace_id="demo",
+        expected_dimension=1024,
+        request=request,
+    )
+
+    assert store.clear_reference() is True
+    assert ("DELETE", "http://qdrant:6333/collections/demo-reference") in calls
+    assert not any(method == "DELETE" and url.endswith("/collections/demo-memory") for method, url in calls)
+
+
+def test_clear_can_create_memory_collection_before_payload_indexes_become_visible():
+    calls = []
+    memory_created = False
+
+    def request(method, url, *, json=None, timeout=None):
+        nonlocal memory_created
+        calls.append((method, url, json))
+        if method == "GET" and url.endswith("/collections/demo"):
+            return FakeResponse(200, {"result": {}})
+        if method == "GET" and url.endswith("/collections/demo-memory"):
+            if not memory_created:
+                return FakeResponse(404, {"status": "error"})
+            return FakeResponse(200, {
+                "result": {
+                    "config": {"params": {"vectors": {"size": 1024, "distance": "Cosine"}}},
+                    # Qdrant exposes newly-created payload indexes asynchronously.
+                    "payload_schema": {},
+                }
+            })
+        if method == "PUT" and url.endswith("/collections/demo-memory"):
+            memory_created = True
+            return FakeResponse(200, {"status": "ok"})
+        if method == "PUT" and url.endswith("/collections/demo-memory/index"):
+            return FakeResponse(200, {"status": "ok"})
+        if method == "POST" and url.endswith("/collections/demo/points/scroll"):
+            return FakeResponse(200, {
+                "result": {"points": [], "next_page_offset": None}
+            })
+        if method == "DELETE" and url.endswith("/collections/demo"):
+            return FakeResponse(200, {"status": "ok"})
+        if method == "GET" and url.endswith("/collections/demo-reference"):
+            return FakeResponse(404, {"status": "error"})
+        raise AssertionError((method, url, json))
+
+    store = QdrantVectorStore(
+        base_url="http://qdrant:6333",
+        collections={"reference": "demo-reference", "memory": "demo-memory"},
+        workspace_id="demo",
+        expected_dimension=1024,
+        collection_lifecycle="require_existing",
+        request=request,
+    )
+
+    assert store.clear_reference() is True
+    assert memory_created is True
+    assert any(method == "DELETE" and url.endswith("/collections/demo") for method, url, _ in calls)
+
+
+def test_clear_migrates_legacy_memory_before_retiring_shared_collection():
+    calls = []
+    legacy_point = {
+        "id": "legacy-memory-id",
+        "vector": [0.25] * 1024,
+        "payload": {
+            "workspace_id": "demo",
+            "kind": "memory",
+            "record_kind": "memory",
+            "record_key": "memory:legacy",
+            "content_hash": "sha256:" + "a" * 64,
+        },
+    }
+    ready_memory = {
+        "result": {
+            "config": {"params": {"vectors": {"size": 1024, "distance": "Cosine"}}},
+            "payload_schema": {
+                field: {"data_type": "keyword"}
+                for field in (
+                    "content_hash", "document_id", "kind", "record_key", "record_kind",
+                    "vector_generation", "workspace_id", "workspace_revision",
+                )
+            },
+        }
+    }
+
+    def request(method, url, *, json=None, timeout=None):
+        calls.append((method, url, json))
+        if method == "GET" and url.endswith("/collections/demo"):
+            return FakeResponse(200, {"result": {}})
+        if method == "GET" and url.endswith("/collections/demo-memory"):
+            return FakeResponse(200, ready_memory)
+        if method == "POST" and url.endswith("/collections/demo/points/scroll"):
+            assert json["with_vector"] is True
+            return FakeResponse(200, {
+                "result": {"points": [legacy_point], "next_page_offset": None}
+            })
+        if method == "PUT" and "/collections/demo-memory/points?wait=true" in url:
+            return FakeResponse(200, {"status": "ok"})
+        if method == "DELETE" and url.endswith("/collections/demo"):
+            return FakeResponse(200, {"status": "ok"})
+        if method == "GET" and url.endswith("/collections/demo-reference"):
+            return FakeResponse(200, {"result": {}})
+        if method == "DELETE" and url.endswith("/collections/demo-reference"):
+            return FakeResponse(200, {"status": "ok"})
+        raise AssertionError((method, url, json))
+
+    store = QdrantVectorStore(
+        base_url="http://qdrant:6333",
+        collections={"reference": "demo-reference", "memory": "demo-memory"},
+        workspace_id="demo",
+        expected_dimension=1024,
+        request=request,
+    )
+
+    assert store.clear_reference() is True
+    migrated = next(
+        payload for method, url, payload in calls
+        if method == "PUT" and "/collections/demo-memory/points?wait=true" in url
+    )
+    assert migrated == {"points": [legacy_point]}
+    deleted = [url for method, url, _payload in calls if method == "DELETE"]
+    assert deleted == [
+        "http://qdrant:6333/collections/demo",
+        "http://qdrant:6333/collections/demo-reference",
+    ]
+    assert not any(url.endswith("/collections/demo-memory") for url in deleted)
+
+
 def _ready_collection_with_bm25(fake: FakeQdrantHttp) -> None:
     fake.collection = {"vectors": {"size": 1024, "distance": "Cosine"}}
     fake.payload_indexes = {