Merge pull request #55 from nasa/feature/fix-variable-value-matching

Fix variable value matching

CHANGELOG.md

@@ -10,6 +10,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Deprecated
 ### Removed
 ### Fixed
+- [pull-request/55](https://github.com/nasa/ncompare/pull/55): Fix variable value matching
 ### Security
 
 ## [1.2.0]

ncompare/core.py

@@ -127,7 +127,7 @@ def run_through_comparisons(
     _, _, _ = out.lists_diff(list_a, list_b)
 
     # Show the groups in each NetCDF file and evaluate differences.
-    out.print(Fore.LIGHTBLUE_EX + "\nGroups:", add_to_history=True)
+    out.print(Fore.LIGHTBLUE_EX + "\nRoot-level Groups:", add_to_history=True)
     list_a = _get_groups(nc_a)
     list_b = _get_groups(nc_b)
     _, _, _ = out.lists_diff(list_a, list_b)
@@ -158,7 +158,9 @@ def run_through_comparisons(
                     + "\nChecking multiple random values within specified variable <%s>:"
                     % comparison_var_name
                 )
-                compare_multiple_random_values(out, nc_a, nc_b, groupname=comparison_var_group)
+                compare_multiple_random_values(out, nc_a, nc_b,
+                                               groupname=comparison_var_group,
+                                               varname=comparison_var_name)
 
             except KeyError:
                 out.print(
@@ -186,12 +188,13 @@ def compare_multiple_random_values(
     nc_a: Union[str, Path],
     nc_b: Union[str, Path],
     groupname: str,
+    varname: str,
     num_comparisons: int = 100,
 ):
     """Iterate through N random samples, and evaluate whether the differences exceed a threshold."""
     # Open a variable from each NetCDF
-    nc_var_a = xr.open_dataset(nc_a, backend_kwargs={"group": groupname}).varname
-    nc_var_b = xr.open_dataset(nc_b, backend_kwargs={"group": groupname}).varname
+    nc_var_a = xr.open_dataset(nc_a, backend_kwargs={"group": groupname}).variables[varname]
+    nc_var_b = xr.open_dataset(nc_b, backend_kwargs={"group": groupname}).variables[varname]
 
     num_mismatches = 0
     for _ in range(num_comparisons):
@@ -200,6 +203,7 @@ def compare_multiple_random_values(
             out.print(".", colors=False, end="")
         elif match_result is None:
             out.print("n", colors=False, end="")
+            num_mismatches += 1
         else:
             out.print("x", colors=False, end="")
             num_mismatches += 1
@@ -444,14 +448,14 @@ def _var_properties(group: Union[netCDF4.Dataset, netCDF4.Group], varname: str)
 
 
 def _match_random_value(
-    out: Outputter, nc_var_a: netCDF4.Variable, nc_var_b: netCDF4.Variable, thresh: float = 1e-6
+    out: Outputter, nc_var_a: xr.Variable, nc_var_b: xr.Variable, thresh: float = 1e-6
 ) -> Union[bool, None]:
     """Check whether a randomly selected data point matches between two variables.
 
     Returns
     -------
     None or bool
-        None if data point is null for either variable
+        None if data point is null for one and only one of the variables
         True if values match
         False if the difference exceeds the given threshold
     """
@@ -466,7 +470,9 @@ def _match_random_value(
     value_b = nc_var_b.values[rand_index_tuple]
 
     # Check whether null
-    if np.isnan(value_a) or np.isnan(value_b):
+    if np.isnan(value_a) and np.isnan(value_b):
+        return True
+    elif np.isnan(value_a) or np.isnan(value_b):
         return None
 
     # Evaluate difference between values

tests/conftest.py

@@ -6,6 +6,11 @@
 from xarray import Dataset
 
 from . import data_for_tests_dir
+from ncompare.printing import Outputter
+
+@pytest.fixture(scope="session")
+def outputter_obj():
+    return Outputter()
 
 @pytest.fixture(scope="session")
 def temp_data_dir(tmpdir_factory) -> Path:
@@ -14,11 +19,12 @@ def temp_data_dir(tmpdir_factory) -> Path:
 @pytest.fixture(scope="session")
 def ds_3dims_2vars_4coords(temp_data_dir) -> Path:
     ds = Dataset(
-        dict(
+        data_vars=dict(
+            # "normal" (Gaussian) distribution of mean 0 and variance 1
             z1=(["y", "x"], np.random.randn(2, 8)),
             z2=(["time", "y"], np.random.randn(10, 2)),
         ),
-        dict(
+        coords=dict(
             x=("x", np.linspace(0, 1.0, 8)),
             time=("time", np.linspace(0, 1.0, 10)),
             c=("y", ["a", "b"]),
@@ -33,12 +39,13 @@ def ds_3dims_2vars_4coords(temp_data_dir) -> Path:
 @pytest.fixture(scope="session")
 def ds_4dims_3vars_5coords(temp_data_dir):
     ds = Dataset(
-        dict(
-            z1=(["y", "x"], np.random.randn(2, 8)),
+        data_vars=dict(
+            # "normal" (Gaussian) distribution of mean 10 and standard deviation 2.5
+            z1=(["y", "x"], 10 + 2.5 * np.random.randn(2, 8)),
             z2=(["time", "y"], np.random.randn(10, 2)),
             z3=(["y", "z"], np.random.randn(2, 9)),
         ),
-        dict(
+        coords=dict(
             x=("x", np.linspace(0, 1.0, 8)),
             time=("time", np.linspace(0, 1.0, 10)),
             c=("y", ["a", "b"]),

tests/test_netcdf_compare.py

@@ -1,8 +1,18 @@
-from ncompare.core import compare
+import xarray as xr
+
+from ncompare.core import compare, _match_random_value
 
 
 def test_dataset_compare_does_not_raise_exception(ds_3dims_2vars_4coords, ds_4dims_3vars_5coords):
     compare(ds_3dims_2vars_4coords, ds_4dims_3vars_5coords)
 
 def test_dataset_compare_does_not_raise_exception_2(ds_3dims_2vars_4coords, ds_3dims_3vars_4coords_1group):
     compare(ds_3dims_2vars_4coords, ds_3dims_3vars_4coords_1group)
+
+def test_matching_random_values(ds_3dims_2vars_4coords, ds_4dims_3vars_5coords,
+                                ds_3dims_3vars_4coords_1group, outputter_obj):
+    variable_array_1 = xr.open_dataset(ds_3dims_2vars_4coords).variables['z1']
+    variable_array_2 = xr.open_dataset(ds_4dims_3vars_5coords).variables['z1']
+
+    assert _match_random_value(outputter_obj, variable_array_1, variable_array_1, ) is True
+    assert _match_random_value(outputter_obj, variable_array_1, variable_array_2, ) is False

tests/test_printing.py

@@ -1,11 +1,5 @@
 import pytest
 
-from ncompare.printing import Outputter
-
-
-@pytest.fixture
-def outputter_obj():
-    return Outputter()
 
 def test_list_of_strings_diff(outputter_obj):
     left, right, both = outputter_obj.lists_diff(['hey', 'yo', 'beebop'],