chore: Add benchmark test for long running subscription (strawberry-g…

…raphql#3593)

tests/benchmarks/api.py

@@ -73,6 +73,11 @@ class Subscription:
     async def something(self) -> AsyncIterator[str]:
         yield "Hello World!"
 
+    @strawberry.subscription
+    async def long_running(self, count: int) -> AsyncIterator[int]:
+        for i in range(count):
+            yield i
+
 
 @strawberry.directive(locations=[DirectiveLocation.FIELD])
 def uppercase(value: str) -> str:

tests/benchmarks/test_subscriptions.py

@@ -1,6 +1,8 @@
 import asyncio
+from typing import AsyncIterator
 
 import pytest
+from graphql import ExecutionResult
 from pytest_codspeed.plugin import BenchmarkFixture
 
 from .api import schema
@@ -24,3 +26,25 @@ async def _run():
             assert value.data["something"] == "Hello World!"
 
     benchmark(lambda: asyncio.run(_run()))
+
+
+@pytest.mark.benchmark
+@pytest.mark.parametrize("count", [1000, 20000])
+def test_subscription_long_run(benchmark: BenchmarkFixture, count: int) -> None:
+    s = """#graphql
+    subscription LongRunning($count: Int!) {
+        longRunning(count: $count)
+    }
+    """
+
+    async def _run():
+        i = 0
+        aiterator: AsyncIterator[ExecutionResult] = await schema.subscribe(
+            s, variable_values={"count": count}
+        )  # type: ignore[assignment]
+        async for res in aiterator:
+            assert res.data is not None
+            assert res.data["longRunning"] == i
+            i += 1
+
+    benchmark(lambda: asyncio.run(_run()))