Repository navigation
v0.10.7 breaks bounds_to_vertices for cftime objects #581
Description
I just realized my CI broke because, e.g., this does not work since the latest release (worked before):
# Example: xarray Dataset with CF-compliant time bounds using cftime
import numpy as np
import xarray as xr
import cf_xarray as cfxr
import cftime
# Create cftime objects for monthly bounds
periods = 3
start = cftime.DatetimeGregorian(2000, 1, 1)
edges = [cftime.DatetimeGregorian(2000, m, 1) for m in range(1, periods + 2)]
# Bounds as [start, end) for each month
bnds = np.array([[edges[i], edges[i+1]] for i in range(periods)])
mid = np.array([edges[i] + (edges[i+1] - edges[i]) / 2 for i in range(periods)])
# Sample data
values = xr.DataArray(np.arange(periods, dtype=float), dims=("time",), coords={"time": mid})
# Build dataset with CF-style bounds
ds = xr.Dataset(
{"foo": values},
coords={
"time": ("time", mid, {"bounds": "time_bnds"}),
"time_bnds": (("time", "bnds"), bnds),
"bnds": ("bnds", [0, 1])
}
)
# Create vertices from bounds
cfxr.bounds_to_vertices(ds.time_bnds, "bnds")gives...
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
Cell In[3], [line 30](vscode-notebook-cell:?execution_count=3&line=30)
20 ds = xr.Dataset(
21 {"foo": values},
22 coords={
(...) 26 }
27 )
29 # Create vertices from bounds
---> 30 cfxr.bounds_to_vertices(ds.time_bnds, "bnds")
File ~/python/packages/cf-xarray/cf_xarray/helpers.py:181, in bounds_to_vertices(bounds, bounds_dim, core_dims, order)
174 raise ValueError(
175 f"Bounds format not understood. Got {bounds.dims} with shape {bounds.shape}."
176 )
178 core_dim_coords = {
179 dim: bounds.coords[dim].values for dim in core_dims if dim in bounds.coords
180 }
--> 181 core_dim_orders = _get_core_dim_orders(core_dim_coords)
183 return xr.apply_ufunc(
184 _bounds_helper,
185 bounds,
(...) 196 output_dtypes=[bounds.dtype],
197 )
File ~/python/packages/cf-xarray/cf_xarray/helpers.py:240, in _get_core_dim_orders(core_dim_coords)
237 if nonzero_diffs.size == 0:
238 # All values are equal, treat as ascending
239 core_dim_orders[dim] = "ascending"
--> 240 elif np.all(nonzero_diffs > 0):
241 core_dim_orders[dim] = "ascending"
242 elif np.all(nonzero_diffs < 0):
TypeError: '>' not supported between instances of 'datetime.timedelta' and 'int'
It still, however, works with usual datetime objects.
Activity
larsbuntemeyer commentedon Aug 15, 2025
It seems with #579, we need to determine the order of an array of cftime objects or datetime timedelta respectivley in here
cf-xarray/cf_xarray/helpers.py
Lines 200 to 247 in cfe87d0
timedelta does not cast to float which seems to be the problem here...
larsbuntemeyer commentedon Aug 15, 2025
Seems like we could use total_seconds to cast cftime diffs to float.