Add Canadian Meterological Center Model Support

src/nwp_consumer/internal/inputs/cmc/client.py

@@ -1,4 +1,4 @@
-"""Implements a client to fetch ICON data from DWD."""
+"""Implements a client to fetch GDPS/GEPS data from CMC."""
 import bz2
 import datetime as dt
 import pathlib
@@ -33,9 +33,9 @@
 
 
 class Client(internal.FetcherInterface):
-    """Implements a client to fetch ICON data from DWD."""
+    """Implements a client to fetch GDPS/GEPS data from CMC."""
 
-    baseurl: str  # The base URL for the ICON model
+    baseurl: str  # The base URL for the GDPS/GEPS model
     model: str  # The model to fetch data for
     parameters: list[str]  # The parameters to fetch
     conform: bool  # Whether to rename parameters to OCF names and clear unwanted coordinates
@@ -59,7 +59,7 @@ def __init__(self, model: str, hours: int = 48, param_group: str = "default") ->
                 self.baseurl += "/ensemble/geps/grib2/raw/"
             case _:
                 raise ValueError(
-                    f"unknown icon model {model}. Valid models are 'gdps' and 'geps'",
+                    f"unknown GDPS/GEPS model {model}. Valid models are 'gdps' and 'geps'",
                 )
 
         match (param_group, model):
@@ -72,6 +72,12 @@ def __init__(self, model: str, hours: int = 48, param_group: str = "default") ->
             case ("full", "gdps"):
                 self.parameters = GDPS_VARIABLES
                 self.conform = False
+            case ("basic", "geps"):
+                self.parameters = GEPS_VARIABLES[:2]
+                self.conform = False
+            case ("basic", "gdps"):
+                self.parameters = GDPS_VARIABLES[:2]
+                self.conform = False
             case (_, _):
                 raise ValueError(
                     f"unknown parameter group {param_group}."
@@ -85,10 +91,10 @@ def listRawFilesForInitTime(self, *, it: dt.datetime) -> list[internal.FileInfoM
         # GDPS data is only available for today's and yesterdays's date. If data hasn't been uploaded for that init
         # time yet, then yesterday's data will still be present on the server.
         if it.date() != dt.datetime.now(dt.timezone.utc).date():
-            raise ValueError("ICON data is only available on today's date")
+            raise ValueError("GDPS/GEPS data is only available on today's date")
             return []
 
-        # The ICON model only runs on the hours [00, 12]
+        # The GDPS/GEPS model only runs on the hours [00, 12]
         if it.hour not in [0, 12]:
             return []
 
@@ -101,7 +107,7 @@ def listRawFilesForInitTime(self, *, it: dt.datetime) -> list[internal.FileInfoM
             # The list of files for the parameter
             parameterFiles: list[internal.FileInfoModel] = []
 
-            # Fetch DWD webpage detailing the available files for the timestep
+            # Fetch CMC webpage detailing the available files for the timestep
             response = requests.get(f"{self.baseurl}/{it.strftime('%H')}/000/", timeout=3)
 
             if response.status_code != 200:
@@ -184,7 +190,7 @@ def mapTemp(self, *, p: pathlib.Path) -> xr.Dataset:  # noqa: D102
             return xr.Dataset()
         # Rename variable to the value, as some have unknown as the name
         if list(ds.data_vars.keys())[0] == "unknown":
-            ds = ds.rename({"unknown": str(p.name).split("_")[3].lower()})
+            ds = ds.rename({"unknown": str(p.name).split("_")[2].lower()})
 
         # Rename variables that are both pressure level and surface
         if "surface" in list(ds.coords):

src/nwp_consumer/internal/inputs/cmc/test_client.py

@@ -22,13 +22,15 @@ def test_mapTemp(self) -> None:
         # Check latitude and longitude are injected
         self.assertTrue("latitude" in out.coords)
         self.assertTrue("longitude" in out.coords)
+        self.assertEqual(len(out["latitude"].values), 1201)
+        self.assertEqual(len(out["longitude"].values), 2400)
         # Check that the dimensions are correctly ordered and renamed
         self.assertEqual(
             out[next(iter(out.data_vars.keys()))].dims,
             ("variable", "init_time", "step", "latitude", "longitude"),
         )
         # Check that the parameter is renamed
-        self.assertEqual(out["variable"].values[0], "v")
+        self.assertEqual(out["variable"].values[0], "v10")
 
         # Test with europe file
         testFilePath: pathlib.Path = (
@@ -39,6 +41,8 @@ def test_mapTemp(self) -> None:
         # Check latitude and longitude are present
         self.assertTrue("latitude" in out.coords)
         self.assertTrue("longitude" in out.coords)
+        self.assertEqual(len(out["latitude"].values), 1201)
+        self.assertEqual(len(out["longitude"].values), 2400)
         # Check that the dimensions are correctly ordered and renamed
         self.assertEqual(
             out[next(iter(out.data_vars.keys()))].dims,

src/test_integration/test_inputs_integration.py

@@ -12,6 +12,7 @@
 from nwp_consumer.internal.inputs.ecmwf._models import ECMWFMarsFileInfo
 from nwp_consumer.internal.inputs.icon._models import IconFileInfo
 from nwp_consumer.internal.inputs.metoffice._models import MetOfficeFileInfo
+from nwp_consumer.internal.inputs.cmc._models import CMCFileInfo
 
 storageClient = outputs.localfs.Client()
 
@@ -92,6 +93,24 @@ def test_downloadsRawGribFileFromICON(self) -> None:
         _, tmpPath = iconClient.downloadToTemp(fi=fileInfo)
         self.assertGreater(tmpPath.stat().st_size, 40000)
 
+    def test_downloadsRawGribFileFromCMC(self) -> None:
+        cmcInitTime: dt.datetime = dt.datetime.now(tz=dt.UTC).replace(
+            hour=0, minute=0, second=0, microsecond=0,
+        )
+
+        cmcClient = inputs.cmc.Client(
+            model="gdps",
+        )
+        fileInfo = CMCFileInfo(
+            it=cmcInitTime,
+            filename=f"CMC_glb_VGRD_ISBL_200_latlon.15x.15_{cmcInitTime.strftime('%Y%m%d%H')}_P120.grib2",
+            currentURL="https://dd.weather.gc.ca/model_gem_global/15km/grib2/lat_lon/00/120",
+            step=1,
+        )
+        _, tmpPath = cmcClient.downloadToTemp(fi=fileInfo)
+        self.assertTrue(tmpPath.name.endswith(".grib2"))
+        self.assertGreater(tmpPath.stat().st_size, 40000)
+
 
 class TestListRawFilesForInitTime(unittest.TestCase):
     def test_getsFileInfosFromCEDA(self) -> None:
@@ -152,6 +171,26 @@ def test_getsFileInfosFromICON(self) -> None:
         self.assertTrue(len(euFileInfos) > 0)
         self.assertNotEqual(fileInfos, euFileInfos)
 
+    def test_getsFileInfosFromCMC(self) -> None:
+        cmcInitTime: dt.datetime = dt.datetime.now(tz=dt.UTC).replace(
+            hour=0, minute=0, second=0, microsecond=0,
+        )
+        cmcClient = inputs.cmc.Client(
+            model="gdps",
+            hours=4,
+            param_group="basic",
+        )
+        fileInfos = cmcClient.listRawFilesForInitTime(it=cmcInitTime)
+        self.assertTrue(len(fileInfos) > 0)
+
+        cmcClient = inputs.cmc.Client(
+            model="geps",
+            hours=4,
+            param_group="basic",
+        )
+        gepsFileInfos = cmcClient.listRawFilesForInitTime(it=cmcInitTime)
+        self.assertTrue(len(gepsFileInfos) > 0)
+        self.assertNotEqual(fileInfos, gepsFileInfos)
 
 if __name__ == "__main__":
     unittest.main()