feat: pristine harness JSON interfaces and require-existing semantic mode (P2)

harness/tests/test_qdrant_vector_store.py

@@ -39,6 +39,7 @@ class FakeQdrantHttp:
         self.malformed_query = False
         self.malformed_scroll = False
         self.scroll_pages: list[dict] | None = None
+        self.drop_collection_on_points = False
 
     def request(self, method, url, *, json=None, timeout=None):
         self.calls.append((method, url, json))
@@ -75,6 +76,9 @@ class FakeQdrantHttp:
             return FakeResponse(200, {"status": "ok"})
 
         if method == "PUT" and path == "/collections/workspace-semantic/points":
+            if self.drop_collection_on_points:
+                self.collection = None
+                return FakeResponse(404, {"status": "error"})
             for point in json["points"]:
                 self.points[point["id"]] = point
             return FakeResponse(200, {"result": {"status": "acknowledged"}})
@@ -161,13 +165,16 @@ def _write_record(record_id: str, kind: str, *, metadata=None):
     )
 
 
-def _store(fake: FakeQdrantHttp) -> QdrantVectorStore:
+def _store(
+    fake: FakeQdrantHttp, *, collection_lifecycle: str = "self_heal"
+) -> QdrantVectorStore:
     return QdrantVectorStore(
         base_url="http://qdrant:6333",
         collection="workspace-semantic",
         workspace_id="demo",
         workspace_revision="a" * 40,
         expected_dimension=1024,
+        collection_lifecycle=collection_lifecycle,
         request=fake.request,
     )
 
@@ -212,6 +219,87 @@ def test_upsert_refuses_collection_dimension_or_distance_mismatch_without_recrea
     assert creates == []
 
 
+def test_upsert_require_existing_refuses_missing_collection_without_creating():
+    fake = FakeQdrantHttp()
+    store = _store(fake, collection_lifecycle="require_existing")
+
+    with pytest.raises(VectorStoreError, match="semantic_index_incompatible"):
+        store.upsert("memory", [_write_record("memory:1", "memory")])
+
+    assert fake.collection is None
+    creates = [
+        call
+        for call in fake.calls
+        if call[0] == "PUT" and call[1].endswith("/collections/workspace-semantic")
+    ]
+    assert creates == []
+
+
+def test_upsert_require_existing_refuses_incompatible_collection_without_mutating():
+    fake = FakeQdrantHttp(dimension=384, distance="Dot")
+    fake.collection = {"vectors": {"size": 384, "distance": "Dot"}}
+    store = _store(fake, collection_lifecycle="require_existing")
+
+    with pytest.raises(VectorStoreError, match="semantic_index_incompatible"):
+        store.upsert("memory", [_write_record("memory:1", "memory")])
+
+    assert fake.payload_indexes == set()
+    mutating = [call for call in fake.calls if call[0] == "PUT"]
+    assert mutating == []
+
+
+def test_upsert_require_existing_writes_to_existing_compatible_collection():
+    fake = FakeQdrantHttp()
+    fake.collection = {"vectors": {"size": 1024, "distance": "Cosine"}}
+    fake.payload_indexes = {
+        "content_hash",
+        "document_id",
+        "kind",
+        "record_key",
+        "record_kind",
+        "vector_generation",
+        "workspace_id",
+        "workspace_revision",
+    }
+    store = _store(fake, collection_lifecycle="require_existing")
+
+    assert store.upsert("memory", [_write_record("memory:1", "memory")]) == 1
+
+    create_or_index = [
+        call
+        for call in fake.calls
+        if call[0] == "PUT" and not call[1].endswith("/points?wait=true")
+    ]
+    assert create_or_index == []
+
+
+def test_upsert_require_existing_fails_if_collection_disappears_after_preflight():
+    fake = FakeQdrantHttp()
+    fake.collection = {"vectors": {"size": 1024, "distance": "Cosine"}}
+    fake.payload_indexes = {
+        "content_hash",
+        "document_id",
+        "kind",
+        "record_key",
+        "record_kind",
+        "vector_generation",
+        "workspace_id",
+        "workspace_revision",
+    }
+    fake.drop_collection_on_points = True
+    store = _store(fake, collection_lifecycle="require_existing")
+
+    with pytest.raises(VectorStoreError, match="HTTP 404"):
+        store.upsert("memory", [_write_record("memory:1", "memory")])
+
+    creates = [
+        call
+        for call in fake.calls
+        if call[0] == "PUT" and call[1].endswith("/collections/workspace-semantic")
+    ]
+    assert creates == []
+
+
 def test_health_fails_when_the_bound_collection_is_missing():
     fake = FakeQdrantHttp()