Fix ICON Reader (#1430)

* o Fix ICON Reader added to _icon_to_ugrid_dims

* Fix cross section coords handling and further fix ICON reader dim parser

* test: add icon cross_section test case

---------

Co-authored-by: Orhan Eroglu <[email protected]>
Co-authored-by: erogluorhan <[email protected]>

Author: rajeeja

test/io/test_icon.py

@@ -8,3 +8,17 @@ def test_read_icon_grid(gridpath):
 def test_read_icon_dataset(gridpath):
     grid_path = gridpath("icon", "R02B04", "icon_grid_0010_R02B04_G.nc")
     uxds = ux.open_dataset(grid_path, grid_path)
+
+def test_icon_cross_section(gridpath):
+    grid_path = gridpath("icon", "R02B04", "icon_grid_0010_R02B04_G.nc")
+    uxds = ux.open_dataset(grid_path, grid_path)
+
+    # Test cross_section with cell_area variable
+    result = uxds.cell_area.cross_section(start=(0, -90), end=(0, 90))
+    assert result is not None
+    assert len(result) == 100  # 100 steps by default
+    assert result.dims == ('steps',)
+    assert all(result.coords['lon'] == 0.0)  # Constant longitude
+    assert result.coords['lat'].min() == -90.0
+    assert result.coords['lat'].max() == 90.0
+    assert result.attrs['units'] == 'steradian'

uxarray/cross_sections/dataarray_accessor.py

@@ -132,7 +132,11 @@ def __call__(
         data = np.moveaxis(filled, -1, dim_axis)
 
         # Build coords dict: keep everything except 'n_face'
-        coords = {d: self.uxda.coords[d] for d in self.uxda.coords if d != "n_face"}
+        coords = {
+            name: self.uxda.coords[name]
+            for name in self.uxda.coords
+            if name != "n_face" and "n_face" not in self.uxda.coords[name].dims
+        }
         # index along the arc
         coords[new_dim] = np.arange(steps)
 

uxarray/io/_icon.py

@@ -5,8 +5,66 @@
 
 
 def _icon_to_ugrid_dims(in_ds):
-    source_dims_dict = {"vertex": "n_node", "edge": "n_edge", "cell": "n_face"}
-    return source_dims_dict
+    """Parse ICON dimension names and map them to UGRID conventions."""
+    source_dims_dict = {}
+
+    # Coordinate-driven mappings
+    if "vlat" in in_ds:
+        source_dims_dict[in_ds["vlat"].dims[0]] = ugrid.NODE_DIM
+    if "vlon" in in_ds:
+        source_dims_dict[in_ds["vlon"].dims[0]] = ugrid.NODE_DIM
+
+    if "elat" in in_ds:
+        source_dims_dict[in_ds["elat"].dims[0]] = ugrid.EDGE_DIM
+    if "elon" in in_ds:
+        source_dims_dict[in_ds["elon"].dims[0]] = ugrid.EDGE_DIM
+
+    if "clat" in in_ds:
+        source_dims_dict[in_ds["clat"].dims[0]] = ugrid.FACE_DIM
+    if "clon" in in_ds:
+        source_dims_dict[in_ds["clon"].dims[0]] = ugrid.FACE_DIM
+
+    # Connectivity-driven mappings
+    if "vertex_of_cell" in in_ds:
+        n_max_face_nodes_dim, face_dim = in_ds["vertex_of_cell"].dims
+        source_dims_dict.setdefault(face_dim, ugrid.FACE_DIM)
+        source_dims_dict.setdefault(n_max_face_nodes_dim, ugrid.N_MAX_FACE_NODES_DIM)
+
+    if "edge_of_cell" in in_ds:
+        n_max_face_edges_dim, face_dim = in_ds["edge_of_cell"].dims
+        source_dims_dict.setdefault(face_dim, ugrid.FACE_DIM)
+        source_dims_dict.setdefault(
+            n_max_face_edges_dim, ugrid.FACE_EDGE_CONNECTIVITY_DIMS[1]
+        )
+
+    if "neighbor_cell_index" in in_ds:
+        n_max_face_faces_dim, face_dim = in_ds["neighbor_cell_index"].dims
+        source_dims_dict.setdefault(face_dim, ugrid.FACE_DIM)
+        source_dims_dict.setdefault(
+            n_max_face_faces_dim, ugrid.FACE_FACE_CONNECTIVITY_DIMS[1]
+        )
+
+    if "adjacent_cell_of_edge" in in_ds:
+        two_dim, edge_dim = in_ds["adjacent_cell_of_edge"].dims
+        source_dims_dict.setdefault(edge_dim, ugrid.EDGE_DIM)
+        source_dims_dict.setdefault(two_dim, ugrid.EDGE_FACE_CONNECTIVITY_DIMS[1])
+
+    if "edge_vertices" in in_ds:
+        two_dim, edge_dim = in_ds["edge_vertices"].dims
+        source_dims_dict.setdefault(edge_dim, ugrid.EDGE_DIM)
+        source_dims_dict.setdefault(two_dim, ugrid.EDGE_NODE_CONNECTIVITY_DIMS[1])
+
+    # Fall back to common ICON dimension names if they were not detected above
+    for dim, ugrid_dim in {
+        "vertex": ugrid.NODE_DIM,
+        "edge": ugrid.EDGE_DIM,
+        "cell": ugrid.FACE_DIM,
+    }.items():
+        if dim in in_ds.dims:
+            source_dims_dict.setdefault(dim, ugrid_dim)
+
+    # Keep only dims that actually exist on the dataset
+    return {dim: name for dim, name in source_dims_dict.items() if dim in in_ds.dims}
 
 
 def _primal_to_ugrid(in_ds, out_ds):