tests: check client server schema compatibility

carbonserver/tests/api/test_schema_compatibility.py

@@ -0,0 +1,120 @@
+import dataclasses
+import importlib.util
+from pathlib import Path
+
+import pytest
+
+from carbonserver.api import schemas as server_schemas
+
+
+def load_client_schemas():
+    repo_root = Path(__file__).resolve().parents[3]
+    client_schema_path = repo_root / "codecarbon" / "core" / "schemas.py"
+    spec = importlib.util.spec_from_file_location(
+        "codecarbon_client_schemas", client_schema_path
+    )
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+client_schemas = load_client_schemas()
+
+
+CREATE_SCHEMA_PAIRS = [
+    (client_schemas.EmissionCreate, server_schemas.EmissionCreate),
+    (client_schemas.RunCreate, server_schemas.RunCreate),
+    (client_schemas.ExperimentCreate, server_schemas.ExperimentCreate),
+    (client_schemas.ProjectCreate, server_schemas.ProjectCreate),
+    (client_schemas.OrganizationCreate, server_schemas.OrganizationCreate),
+]
+
+
+@pytest.mark.parametrize(("client_schema", "server_schema"), CREATE_SCHEMA_PAIRS)
+def test_client_create_schemas_match_server_fields(client_schema, server_schema):
+    client_fields = {field.name for field in dataclasses.fields(client_schema)}
+    server_fields = set(server_schema.model_fields)
+
+    assert client_fields == server_fields
+
+
+@pytest.mark.parametrize(
+    ("client_payload", "server_schema"),
+    [
+        (
+            client_schemas.EmissionCreate(
+                timestamp="2021-04-04T08:43:00+02:00",
+                run_id="40088f1a-d28e-4980-8d80-bf5600056a14",
+                duration=98745,
+                emissions_sum=1544.54,
+                emissions_rate=1.548444,
+                cpu_power=0.3,
+                gpu_power=0.0,
+                ram_power=0.15,
+                cpu_energy=55.21874,
+                gpu_energy=0.0,
+                ram_energy=2.0,
+                energy_consumed=57.21874,
+                cpu_utilization_percent=12.5,
+                gpu_utilization_percent=34.5,
+                ram_utilization_percent=56.5,
+                wue=0.8,
+            ),
+            server_schemas.EmissionCreate,
+        ),
+        (
+            client_schemas.RunCreate(
+                timestamp="2021-04-04T08:43:00+02:00",
+                experiment_id="8edb03e1-9a28-452a-9c93-a3b6560136d7",
+                os="macOS-10.15.7-x86_64-i386-64bit",
+                python_version="3.8.0",
+                codecarbon_version="2.1.3",
+                cpu_count=12,
+                cpu_model="Intel(R) Core(TM) i7-8850H CPU @ 2.60GHz",
+                gpu_count=4,
+                gpu_model="NVIDIA",
+                longitude=-7.6174,
+                latitude=33.5822,
+                region="EUROPE",
+                provider="AWS",
+                ram_total_size=83948.22,
+                tracking_mode="Machine",
+            ),
+            server_schemas.RunCreate,
+        ),
+        (
+            client_schemas.ExperimentCreate(
+                timestamp="2021-04-04T08:43:00+02:00",
+                name="Run on AWS",
+                description="AWS API for Code Carbon",
+                country_name="France",
+                country_iso_code="FRA",
+                region="france",
+                on_cloud=True,
+                cloud_provider="aws",
+                cloud_region="eu-west-1a",
+                project_id="8edb03e1-9a28-452a-9c93-a3b6560136d7",
+            ),
+            server_schemas.ExperimentCreate,
+        ),
+        (
+            client_schemas.ProjectCreate(
+                name="API Code Carbon",
+                description="API for Code Carbon",
+                organization_id="8edb03e1-9a28-452a-9c93-a3b6560136d7",
+            ),
+            server_schemas.ProjectCreate,
+        ),
+        (
+            client_schemas.OrganizationCreate(
+                name="Code Carbon",
+                description="Save the world, one run at a time.",
+            ),
+            server_schemas.OrganizationCreate,
+        ),
+    ],
+)
+def test_client_create_payloads_validate_against_server_schemas(
+    client_payload, server_schema
+):
+    server_schema.model_validate(dataclasses.asdict(client_payload))

codecarbon/core/api_client.py

@@ -227,6 +227,10 @@ def add_emission(self, carbon_emission: dict):
             gpu_energy=carbon_emission["gpu_energy"],
             ram_energy=carbon_emission["ram_energy"],
             energy_consumed=carbon_emission["energy_consumed"],
+            cpu_utilization_percent=carbon_emission.get("cpu_utilization_percent"),
+            gpu_utilization_percent=carbon_emission.get("gpu_utilization_percent"),
+            ram_utilization_percent=carbon_emission.get("ram_utilization_percent"),
+            wue=carbon_emission.get("wue", 0),
         )
         try:
             payload = dataclasses.asdict(emission)

codecarbon/core/schemas.py

@@ -22,6 +22,10 @@ class EmissionBase:
     gpu_energy: float
     ram_energy: float
     energy_consumed: float
+    cpu_utilization_percent: Optional[float] = None
+    gpu_utilization_percent: Optional[float] = None
+    ram_utilization_percent: Optional[float] = None
+    wue: Optional[float] = 0
 
 
 class EmissionCreate(EmissionBase):

tests/test_api_call.py

@@ -125,8 +125,17 @@ def test_call_api(self):
                 latitude=33.5822,
                 ram_total_size=83948.22,
                 tracking_mode="Machine",
+                cpu_utilization_percent=12.5,
+                gpu_utilization_percent=34.5,
+                ram_utilization_percent=56.5,
+                wue=0.8,
             )
             assert api.add_emission(dataclasses.asdict(carbon_emission))
+            payload = m.last_request.json()
+            assert payload["cpu_utilization_percent"] == 12.5
+            assert payload["gpu_utilization_percent"] == 34.5
+            assert payload["ram_utilization_percent"] == 56.5
+            assert payload["wue"] == 0.8
 
     def test_check_auth_returns_none_on_error(self):
         with requests_mock.Mocker() as m: