Replace climetlab with earthkit-data (#8)

* Closes #7
- Replaces climetlab with earthkit

* Fix: Remove testing exception

* Fix: Address review comments
- ekd instead of earthkit.data
- Use 'valid_time' instead of 'base_time'

* Fix: Address valid_time retrieval comment

CHANGELOG.md

@@ -11,10 +11,13 @@ Keep it human-readable, your future self will thank you!
 ## [Unreleased]
 
 ### Added
+- earthkit-data replaces climetlab
 
 ### Changed
 
 ### Removed
+- climetlab
+
 
 ## [0.1.10] Fix missing constants
 

pyproject.toml

@@ -52,6 +52,7 @@ dynamic = [
 dependencies = [
   "anemoi-utils>=0.3",
   "anytree",
+  "earthkit-data>=0.9",
   "numpy",
   "pyyaml",
   "semantic-version",

src/anemoi/inference/runner.py

@@ -36,10 +36,10 @@
 
 
 def forcing_and_constants(source, date, param):
-    import climetlab as cml
+    import earthkit.data as ekd
 
-    ds = cml.load_source(
-        "constants",
+    ds = ekd.from_source(
+        "forcings",
         source,
         date=date,
         param=param,
@@ -119,7 +119,7 @@ def run(
 
         LOGGER.info("Loading input: %d fields (lagged=%d)", len(input_fields), len(self.lagged))
 
-        input_fields_numpy = input_fields.to_numpy(dtype=np.float32, reshape=False)
+        input_fields_numpy = input_fields.to_numpy(dtype=np.float32)
         print(input_fields_numpy.shape)
 
         input_fields_numpy = input_fields_numpy.reshape(
@@ -224,7 +224,8 @@ def run(
         ]
 
         if start_datetime is None:
-            start_datetime = input_fields.order_by(valid_datetime="ascending")[-1].datetime()
+            start_datetime_str = input_fields.order_by(valid_datetime="ascending")[-1].metadata("valid_datetime")
+            start_datetime = datetime.datetime.fromisoformat(start_datetime_str)
 
         constants = forcing_and_constants(
             source=input_fields[:1],
@@ -290,14 +291,14 @@ def run(
 
         # Write dynamic fields
         def get_most_recent_datetime(input_fields):
-            datetimes = [f.valid_datetime() for f in input_fields]
+            datetimes = [f.datetime()["valid_time"] for f in input_fields]
             latest = datetimes[-1]
             for d in datetimes:
                 assert d <= latest, (datetimes, d, latest)
             return latest
 
         most_recent_datetime = get_most_recent_datetime(input_fields)
-        reference_fields = [f for f in input_fields if f.valid_datetime() == most_recent_datetime]
+        reference_fields = [f for f in input_fields if f.datetime()["valid_time"] == most_recent_datetime]
         precip_template = reference_fields[self.checkpoint.variable_to_index["2t"]]
 
         accumulated_output = np.zeros(
@@ -335,8 +336,8 @@ def get_most_recent_datetime(input_fields):
 
             for n, (m, param) in enumerate(zip(prognostic_data_from_retrieved_fields_mask, prognostic_params)):
                 template = reference_fields[m]
-                assert template.valid_datetime() == most_recent_datetime, (
-                    template.valid_datetime(),
+                assert template.datetime()["valid_time"] == most_recent_datetime, (
+                    template.datetime()["valid_time"],
                     most_recent_datetime,
                 )
                 output_callback(
@@ -350,8 +351,8 @@ def get_most_recent_datetime(input_fields):
             if len(diagnostic_output_mask):
                 for n, param in enumerate(self.checkpoint.diagnostic_params):
                     accumulated_output[n] += np.maximum(0, diagnostic_fields_numpy[:, n])
-                    assert precip_template.valid_datetime() == most_recent_datetime, (
-                        precip_template.valid_datetime(),
+                    assert precip_template.datetime()["valid_time"] == most_recent_datetime, (
+                        precip_template.datetime()["valid_time"],
                         most_recent_datetime,
                     )
                     output_callback(