add tests for converter

packages/utils/tests/data/test_converter.py

@@ -13,6 +13,9 @@
 # limitations under the License.
 
 
+import numpy as np
+import pandas as pd
+import pytest
 import xarray as xr
 
 from pyearthtools.utils.data import converter
@@ -21,6 +24,24 @@
 SIMPLE_DATA_SET = xr.Dataset({"Entry": SIMPLE_DATA_ARRAY})
 
 
+def test_NumpyConverterTuple():
+    """
+    Checks NumpyConverter using Tuples
+    """
+    dataset_tuple = (SIMPLE_DATA_SET.copy(), SIMPLE_DATA_SET.copy())
+    nc = converter.NumpyConverter()
+    np_array = nc.convert_from_xarray(dataset_tuple)
+    xr_ds = nc.convert_to_xarray(np_array)
+
+    assert isinstance(xr_ds, tuple)
+    assert isinstance(xr_ds[0], xr.Dataset)
+    assert isinstance(xr_ds[-1], xr.Dataset)
+
+    xr_ds1 = xr_ds[0]
+    assert "Entry" in xr_ds1.data_vars
+    xr.testing.assert_identical(xr_ds1["Entry"], SIMPLE_DATA_ARRAY)
+
+
 def test_NumpyConverter():
     """
     Checks conversion from xarray to numpy and back
@@ -36,6 +57,48 @@ def test_NumpyConverter():
     assert "Entry" in xr_ds
     xr.testing.assert_identical(xr_ds["Entry"], SIMPLE_DATA_ARRAY)
 
+    # Data that hasn't been converted yet throws runtime error
+    with pytest.raises(RuntimeError):
+        nc = converter.NumpyConverter()
+        _xr_ds = nc.convert_to_xarray(np.array([0, 1, 2]))
+
+    # Data with an empty dataset throws runtime error
+    with pytest.raises(RuntimeError):
+        nc = converter.NumpyConverter()
+        _np_array = nc.convert_from_xarray(SIMPLE_DATA_SET)
+        _xr_ds = nc.convert_to_xarray(np.empty(0))
+
+    # String data type throws error
+    with pytest.raises(TypeError):
+        nc = converter.NumpyConverter()
+        _xr_ds = nc.convert_from_xarray(["wrong type"])
+
+    nc = converter.NumpyConverter()
+
+    # Create a DataArray with a 'time' dimension but no 'x' coordinate variable
+    data_array = xr.DataArray(np.random.rand(5, 3), dims=["time", "x"], coords={"time": np.random.rand(5)})
+
+    # Create a Dataset from this DataArray
+    ds = xr.Dataset({"my_variable": data_array})
+    np_array = nc.convert_from_xarray(ds)
+
+    # Value error converting back to numpy array due to missing coord
+    with pytest.raises(ValueError):
+        _xr_ds = nc.convert_to_xarray(np_array)
+
+    ds = xr.Dataset(
+        data_vars={"data_var": ("x", np.array([1, 2, 3]))},
+        coords={
+            "x": np.arange(3),
+            "empty_coord_dim": np.arange(2),  # Coordinate with new dim but no data var
+        },
+    )
+    nc = converter.NumpyConverter()
+
+    # Throw runtime error that cannot record coordinate
+    with pytest.raises(RuntimeError):
+        _xr_ds = nc.convert_from_xarray(ds)
+
 
 def test_DaskConverter():
     """
@@ -50,3 +113,88 @@ def test_DaskConverter():
     assert isinstance(xr_ds, xr.Dataset)
     assert "Entry" in xr_ds
     xr.testing.assert_identical(xr_ds["Entry"], SIMPLE_DATA_ARRAY)
+
+    dataset_tuple = (SIMPLE_DATA_SET.copy(), SIMPLE_DATA_SET.copy())
+    da_array = dc.convert_from_xarray(dataset_tuple)
+
+    xr_ds = dc.convert_to_xarray(da_array)
+    assert isinstance(xr_ds, tuple)
+    assert "Entry" in xr_ds
+    xr.testing.assert_identical(xr_ds[0]["Entry"], SIMPLE_DATA_ARRAY)
+
+
+def test_save_and_load_records(tmpdir):
+    """
+    Test save and load records functionality.
+    """
+    tmp_path = tmpdir.mkdir("sub").join("nc.records")
+
+    time_index = pd.date_range("2025-01-01", periods=3, freq="D")
+
+    data = np.random.rand(3, 3, 3)
+
+    coords = {"time": time_index, "x": [np.nan, np.nan, np.nan], "y": np.random.randn(3)}
+
+    test_data_array = xr.DataArray(
+        data,
+        coords=coords,
+        dims=["time", "x", "y"],
+    )
+
+    test_dataset = xr.Dataset({"entry": test_data_array})
+
+    nc = converter.NumpyConverter()
+    nc.convert_from_xarray(test_data_array)
+    nc.save_records(tmp_path)
+    saved_records = nc.records.copy()
+
+    # Add extra record to the numpy converter
+    nc.convert_from_xarray(test_data_array)
+
+    assert len(saved_records) == 1
+    assert len(nc.records) == 2
+    assert nc.records != saved_records  # Assert saved records are not the same as converted records
+
+    assert nc.load_records(tmp_path)  # Assert it loads correctly
+    assert len(nc.records) == 1
+
+    loaded_records = nc.records.copy()
+    loaded_vars = loaded_records[0]["coords"]["x"]
+    saved_vars = saved_records[0]["coords"]["x"]
+
+    # Check if variables are equal
+    np.testing.assert_equal(saved_vars, loaded_vars)
+
+    # Broken path returns False
+    assert not nc.load_records("/broken/path")
+
+    # Trigger datetime instance in save_records.parse
+    nc = converter.NumpyConverter()
+    nc.convert_from_xarray(test_data_array)
+    nc.save_records(tmp_path)
+
+    # Trigger np.isnan(v).all() in save_records.parse
+    nc = converter.NumpyConverter()
+    da_array = nc.convert_from_xarray(test_dataset)
+    nc.convert_to_xarray(da_array)
+    nc.save_records(tmp_path)
+
+
+def test_non_shared_coordinates_throws_value_error():
+    """
+    Test if coordinates are not shared will throw an appropriate value error
+    """
+    x_coords = np.array([0, 1, 2])
+    y_coords = np.array([0, 1, 2])
+
+    data_1 = np.random.rand(len(x_coords), len(y_coords))
+    data_2 = np.random.rand(len(x_coords))
+    missing_coord_ds = xr.Dataset(
+        {"data_1": (("x", "y"), data_1), "data_2": (("x"), data_2)},
+        coords={"x": x_coords, "y": y_coords},
+    )
+    nc = converter.NumpyConverter()
+
+    # Cannot stack variables so will raise a value error
+    with pytest.raises(ValueError):
+        nc.convert_from_xarray(missing_coord_ds)