Datastore

• Client: google-cloud-datastore (datastore.Client)
• Transport: gRPC. The client connects insecurely when DATASTORE_EMULATOR_HOST is set, so drongo runs an in-process gRPC emulator and points the client at it via that env var.
• Backend: per-project.

Use the normal client with its default transport. Nothing changes.

Firestore in Datastore mode

This mocks the classic Datastore API (entities, kinds, keys, ancestor paths). Typed property values round-trip exactly, and mutations plus queries run against an in-memory entity store. Transactions are accepted and completed synchronously (no isolation is enforced).

Entities: put, get, delete

from drongo import mock_gcp


@mock_gcp
def test_entities():
    from google.cloud import datastore

    client = datastore.Client(project="my-project")

    task = datastore.Entity(key=client.key("Task", "t1"))
    task.update({"title": "Buy milk", "done": False, "priority": 3})
    client.put(task)

    assert client.get(client.key("Task", "t1"))["title"] == "Buy milk"

    # Incomplete keys get an allocated id on put.
    auto = datastore.Entity(key=client.key("Task"))
    auto.update({"title": "Auto"})
    client.put(auto)
    assert auto.key.id

    client.delete(client.key("Task", "t1"))
    assert client.get(client.key("Task", "t1")) is None

Queries

kind, property filters (=, <, <=, >, >=, IN), ordering, and limits work. Filters wrap in composite AND/OR:

@mock_gcp
def test_queries():
    from google.cloud import datastore
    from google.cloud.datastore.query import PropertyFilter

    client = datastore.Client(project="p")
    for name, pri in [("a", 1), ("b", 2), ("c", 3)]:
        e = datastore.Entity(key=client.key("Task", name))
        e["priority"] = pri
        client.put(e)

    q = client.query(kind="Task")
    q.add_filter(filter=PropertyFilter("priority", ">=", 2))
    q.order = ["-priority"]
    assert [e.key.name for e in q.fetch()] == ["c", "b"]

Coverage

Operation	Status
put / get / delete (Commit, Lookup)	Supported
Named + auto-id keys (AllocateIds)	Supported
Typed values (str, int, float, bool, null, bytes, array, timestamp, key)	Supported
Queries: kind, filters, order, limit, offset	Supported
Composite AND / OR, IN / NOT_IN	Supported
Transactions (accepted, completed synchronously)	Supported
Ancestor queries, projections, cursors	Planned
Aggregation queries (count)	Planned