First pass at making NetCDFTimeIndex compatible with pydata#1356

Author: spencerkclark

xarray/conventions/netcdftimeindex.py

@@ -101,7 +101,7 @@ def _parsed_string_to_bounds(date_type, resolution, parsed):
 
 def get_date_field(datetimes, field):
     """Adapted from pandas.tslib.get_date_field"""
-    return [getattr(date, field) for date in datetimes]
+    return np.array([getattr(date, field) for date in datetimes])
 
 
 def _field_accessor(name, docstring=None):

xarray/core/accessors.py

@@ -2,7 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
-from .common import is_datetime_like
+from .common import is_np_datetime_like, _contains_datetime_like_objects
 from .pycompat import dask_array_type
 
 from functools import partial
@@ -20,6 +20,20 @@ def _season_from_months(months):
     return seasons[(months // 3) % 4]
 
 
+def _access_through_netcdftimeindex(values, name):
+    """Coerce an array of datetime-like values to a NetCDFTimeIndex 
+    and access requested datetime component
+    """
+    from ..conventions.netcdftimeindex import NetCDFTimeIndex
+    values_as_netcdftimeindex = NetCDFTimeIndex(values)
+    if name == 'season':
+        months = values_as_netcdftimeindex.month
+        field_values = _season_from_months(months)
+    else:
+        field_values = getattr(values_as_netcdftimeindex, name)
+    return field_values.reshape(values.shape)
+
+
 def _access_through_series(values, name):
     """Coerce an array of datetime-like values to a pandas Series and
     access requested datetime component
@@ -52,12 +66,17 @@ def _get_date_field(values, name, dtype):
         Array-like of datetime fields accessed for each element in values
 
     """
+    if is_np_datetime_like(values.dtype):
+        access_method = _access_through_series
+    else:
+        access_method = _access_through_netcdftimeindex
+
     if isinstance(values, dask_array_type):
         from dask.array import map_blocks
-        return map_blocks(_access_through_series,
+        return map_blocks(access_method,
                           values, name, dtype=dtype)
     else:
-        return _access_through_series(values, name)
+        return access_method(values, name)
 
 
 class DatetimeAccessor(object):
@@ -83,9 +102,11 @@ class DatetimeAccessor(object):
 
      """
     def __init__(self, xarray_obj):
-        if not is_datetime_like(xarray_obj.dtype):
+        if not _contains_datetime_like_objects(xarray_obj):
             raise TypeError("'dt' accessor only available for "
-                            "DataArray with datetime64 or timedelta64 dtype")
+                            "DataArray with datetime64 timedelta64 dtype or "
+                            "for arrays containing netcdftime datetime "
+                            "objects.")
         self._obj = xarray_obj
 
     def _tslib_field_accessor(name, docstring=None, dtype=None):
@@ -147,4 +168,4 @@ def f(self, dtype=dtype):
 
     time = _tslib_field_accessor(
         "time", "Timestamps corresponding to datetimes", object
-    )
+    )

xarray/core/common.py

@@ -763,8 +763,25 @@ def ones_like(other, dtype=None):
     return full_like(other, 1, dtype)
 
 
-def is_datetime_like(dtype):
+def is_np_datetime_like(dtype):
     """Check if a dtype is a subclass of the numpy datetime types
     """
     return (np.issubdtype(dtype, np.datetime64) or
             np.issubdtype(dtype, np.timedelta64))
+
+
+def _contains_netcdftime_datetimes(var):
+    """Check if a variable contains netcdftime datetime objects"""
+    from netcdftime._netcdftime import datetime
+    return isinstance(var.data.ravel()[0], datetime)
+
+
+def _contains_datetime_like_objects(var):
+    """Check if a variable contains datetime like objects (either
+    np.datetime64, np.timedelta64, or netcdftime._netcdftime.datetime)"""
+    if is_np_datetime_like(var.dtype):
+        return True
+    try:
+        return _contains_netcdftime_datetimes(var)
+    except ImportError:
+        return False

xarray/core/dataset.py

@@ -21,7 +21,8 @@
 from .alignment import align
 from ..conventions import coding
 from .coordinates import DatasetCoordinates, LevelCoordinatesSource, Indexes
-from .common import ImplementsDatasetReduce, BaseDataObject, is_datetime_like
+from .common import (ImplementsDatasetReduce, BaseDataObject,
+                     _contains_datetime_like_objects)
 from .merge import (dataset_update_method, dataset_merge_method,
                     merge_data_and_coords)
 from .utils import (Frozen, SortedKeysDict, maybe_wrap_array, hashable,
@@ -76,7 +77,7 @@ def _get_virtual_variable(variables, key, level_vars=None, dim_sizes=None):
         virtual_var = ref_var
         var_name = key
     else:
-        if is_datetime_like(ref_var.dtype):
+        if _contains_datetime_like_objects(ref_var):
             ref_var = xr.DataArray(ref_var)
             data = getattr(ref_var.dt, var_name).data
         else: