Allow any cell measure (#115)

* allow extra cell measures

* update docs

* no fancy note

* manually lint notebook

* break tests

* comment valid keys check

* do not pass through _get_measure

* restore original _get_measure docstring

Author: malmans2

cf_xarray/accessor.py

@@ -21,6 +21,7 @@
 from xarray import DataArray, Dataset
 
 from .helpers import bounds_to_vertices
+from .utils import parse_cell_methods_attr
 
 #: Classes wrapped by cf_xarray.
 _WRAPPED_CLASSES = (
@@ -369,11 +370,8 @@ def _get_measure(da: Union[DataArray, Dataset], key: str) -> List[str]:
         )
 
     attr = da.attrs["cell_measures"]
-    strings = [s for scolons in attr.split(":") for s in scolons.split()]
-    if len(strings) % 2 != 0:
-        raise ValueError(f"attrs['cell_measures'] = {attr!r} is malformed.")
-    measures = dict(zip(strings[slice(0, None, 2)], strings[slice(1, None, 2)]))
-    results = measures.get(key, [])
+    measures = parse_cell_methods_attr(attr)
+    results: Union[str, List] = measures.get(key, [])
     if isinstance(results, str):
         return [results]
     return results
@@ -937,6 +935,7 @@ def cell_measures(self) -> Dict[str, List[str]]:
         Returns
         -------
         Dictionary of valid cell measure names that can be used with __getitem__ or .cf[key].
+        Will be ("area", "volume") or a subset thereof.
         """
         assert isinstance(self._obj, DataArray), "this only works with DataArrays"
 
@@ -1012,13 +1011,7 @@ def get_associated_variable_names(self, name: Hashable) -> Dict[str, List[str]]:
 
         if "cell_measures" in attrs_or_encoding:
             coords["cell_measures"] = list(
-                itertools.chain(
-                    *[
-                        _get_measure(self._obj[name], measure)
-                        for measure in _CELL_MEASURES
-                        if measure in attrs_or_encoding["cell_measures"]
-                    ]
-                )
+                parse_cell_methods_attr(attrs_or_encoding["cell_measures"]).values()
             )
 
         if (

cf_xarray/tests/test_accessor.py

@@ -53,10 +53,13 @@ def test_coordinates():
 
 
 def test_cell_measures():
-    da = airds["air"].copy()
-    da.attrs["cell_measures"] += " dummy: dummy"
+    ds = airds.copy(deep=True)
+    ds["foo"] = xr.DataArray(ds["cell_area"], attrs=dict(standard_name="foo_std_name"))
+    ds["air"].attrs["cell_measures"] += " foo_measure: foo"
+    assert "foo_std_name" in ds.cf["air_temperature"].cf
+
     expected = dict(area=["cell_area"])
-    actual = da.cf.cell_measures
+    actual = ds["air"].cf.cell_measures
     assert actual == expected
 
     with pytest.raises(AssertionError, match=r"this only works with DataArrays"):

cf_xarray/utils.py

@@ -1,4 +1,4 @@
-from typing import Any, Hashable, Mapping, Optional, TypeVar, cast
+from typing import Any, Dict, Hashable, Mapping, Optional, TypeVar, cast
 
 K = TypeVar("K")
 V = TypeVar("V")
@@ -47,3 +47,23 @@ def __get__(self, obj, cls):
             return self._accessor
 
         return self._accessor(obj)
+
+
+def parse_cell_methods_attr(attr: str) -> Dict[str, str]:
+    """
+    Parse cell_methods attributes (format is 'measure: name').
+
+    Parameters
+    ----------
+    attr: str
+        String to parse
+
+    Returns
+    -------
+    Dictionary mapping measure to name
+    """
+    strings = [s for scolons in attr.split(":") for s in scolons.split()]
+    if len(strings) % 2 != 0:
+        raise ValueError(f"attrs['cell_measures'] = {attr!r} is malformed.")
+
+    return dict(zip(strings[slice(0, None, 2)], strings[slice(1, None, 2)]))

doc/examples/introduction.ipynb

@@ -555,6 +555,16 @@
     "pop.cf.coordinates"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "**Note:** Although it is possible to assign additional coordinates and cell\n",
+    "measures, `.cf.coordinates` and `.cf.cell_measures` only return a subset of\n",
+    "`(\"longitude\", \"latitude\", \"vertical\", \"time\")` and `(\"area\", \"volume\")`,\n",
+    "respectively.\n"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -1021,7 +1031,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.6"
+   "version": "3.7.3"
   },
   "toc": {
    "base_numbering": 1,