Bugfix/unknow coordinates (ecmwf#72)

* allow for unknown coordinates

CHANGELOG.md

@@ -12,6 +12,10 @@ Keep it human-readable, your future self will thank you!
 
 ## [0.5.5](https://github.com/ecmwf/anemoi-datasets/compare/0.5.4...0.5.5) - 2024-10-04
 
+### Changed
+
+- Allow for unknown coordinates when parsing CF input
+
 ## [Add support for (semi-)constant fields](https://github.com/ecmwf/anemoi-datasets/compare/0.5.1...0.5.2) - 2024-10-03
 
 ### Changed

src/anemoi/datasets/create/functions/sources/xarray/__init__.py

@@ -42,7 +42,7 @@ def load_one(emoji, context, dates, dataset, options={}, flavour=None, **kwargs)
     We have seen this bug triggered when we run many clients in parallel, for example, when we create a new dataset using `xarray-zarr`.
     """
 
-    context.trace(emoji, dataset, options)
+    context.trace(emoji, dataset, options, kwargs)
 
     if isinstance(dataset, str) and ".zarr" in dataset:
         data = xr.open_zarr(name_to_zarr_store(dataset), **options)

src/anemoi/datasets/create/functions/sources/xarray/coordinates.py

@@ -244,3 +244,9 @@ class ScalarCoordinate(Coordinate):
     @property
     def mars_names(self):
         return (self.variable.name,)
+
+
+class UnsupportedCoordinate(Coordinate):
+    @property
+    def mars_names(self):
+        return (self.variable.name,)

src/anemoi/datasets/create/functions/sources/xarray/flavour.py

@@ -18,6 +18,7 @@
 from .coordinates import ScalarCoordinate
 from .coordinates import StepCoordinate
 from .coordinates import TimeCoordinate
+from .coordinates import UnsupportedCoordinate
 from .coordinates import XCoordinate
 from .coordinates import YCoordinate
 from .coordinates import is_scalar
@@ -157,11 +158,13 @@ def _guess(self, c, coord):
         if c.shape in ((1,), tuple()):
             return ScalarCoordinate(c)
 
-        raise NotImplementedError(
+        LOG.warning(
             f"Coordinate {coord} not supported\n{axis=}, {name=},"
             f" {long_name=}, {standard_name=}, units\n\n{c}\n\n{type(c.values)} {c.shape}"
         )
 
+        return UnsupportedCoordinate(c)
+
     def grid(self, coordinates, variable):
         lat = [c for c in coordinates if c.is_lat]
         lon = [c for c in coordinates if c.is_lon]