Store test datasets in repo

conftest.py

@@ -12,6 +12,57 @@ def pytest_addoption(parser):
     )
 
 
+def generate_test_filepath_dict() -> dict:
+    return {
+        "full_time_netcdf4": "virtualizarr/tests/data/test_ds_netcdf4_full_time.nc",
+        "time_1_netcdf4": "virtualizarr/tests/data/test_ds_netcdf4_split_time1.nc",
+        "time_2_netcdf4": "virtualizarr/tests/data/test_ds_netcdf4_split_time2.nc",
+        "netcdf3": "virtualizarr/tests/data/test_ds_netcdf3.nc",
+        "netcdf4_group": "virtualizarr/tests/data/test_ds_netcdf4_group.nc",
+        "netcdf4_non_standard_time": "virtualizarr/tests/data/test_ds_non_datetime_time.nc",
+    }
+
+
+def generate_small_xr_datasets():
+    """This function can be used to re-generate the locally stored dataset for testing
+    It's 43kB instead of the full 31MB air_temperature tutorial dataset"""
+    import numpy as np
+
+    # building our test dataset from the air_temp dataset, but saving a subset
+    ds = xr.tutorial.open_dataset("air_temperature").isel(time=slice(0, 4))
+
+    lats = np.arange(-90, 90, 1)
+    lons = np.arange(-180, 180, 1)
+
+    data = np.random.randint(0, 2, size=(4, 180, 360), dtype=np.int16)
+
+    # create a dataset with non-standard time
+    non_standard_date_ds = xr.Dataset(
+        data_vars=dict(air=(["time", "lat", "lon"], data)),
+        coords=dict(time=[0, 1, 2, 3], lat=lats, lon=lons),
+    )
+
+    # Add attributes to the time coordinate
+    non_standard_date_ds.time.attrs["units"] = "days since '2000-01-01'"
+    non_standard_date_ds.time.attrs["calendar"] = "standard"
+
+    # write datasets
+    ds.to_netcdf("virtualizarr/tests/data/test_ds_netcdf4_full_time.nc")
+    ds.isel(time=slice(0, 2)).to_netcdf(
+        "virtualizarr/tests/data/test_ds_netcdf4_split_time1.nc"
+    )
+    ds.isel(time=slice(2, 4)).to_netcdf(
+        "virtualizarr/tests/data/test_ds_netcdf4_split_time2.nc"
+    )
+
+    ds.to_netcdf("virtualizarr/tests/data/test_ds_netcdf3.nc", engine="scipy")
+    ds.to_netcdf("virtualizarr/tests/data/test_ds_netcdf4_group.nc", group="test/group")
+
+    non_standard_date_ds.to_netcdf(
+        "virtualizarr/tests/data/test_ds_non_datetime_time.nc"
+    )
+
+
 def pytest_runtest_setup(item):
     # based on https://stackoverflow.com/questions/47559524
     if "network" in item.keywords and not item.config.getoption("--run-network-tests"):
@@ -21,49 +72,19 @@ def pytest_runtest_setup(item):
 
 
 @pytest.fixture
-def netcdf4_file(tmpdir):
-    # Set up example xarray dataset
-    ds = xr.tutorial.open_dataset("air_temperature")
-
-    # Save it to disk as netCDF (in temporary directory)
-    filepath = f"{tmpdir}/air.nc"
-    ds.to_netcdf(filepath, format="NETCDF4")
-    ds.close()
-
-    return filepath
+def netcdf4_file() -> str:
+    return generate_test_filepath_dict()["full_time_netcdf4"]
 
 
 @pytest.fixture
-def hdf5_groups_file(tmpdir):
-    # Set up example xarray dataset
-    ds = xr.tutorial.open_dataset("air_temperature")
-
-    # Save it to disk as netCDF (in temporary directory)
-    filepath = f"{tmpdir}/air.nc"
-    ds.to_netcdf(filepath, format="NETCDF4", group="test/group")
-    ds.close()
-
-    return filepath
+def hdf5_groups_file() -> str:
+    return generate_test_filepath_dict()["netcdf4_group"]
 
 
 @pytest.fixture
-def netcdf4_files(tmpdir):
-    # Set up example xarray dataset
-    ds = xr.tutorial.open_dataset("air_temperature")
-
-    # split inrto equal chunks so we can concatenate them back together later
-    ds1 = ds.isel(time=slice(None, 1460))
-    ds2 = ds.isel(time=slice(1460, None))
-
-    # Save it to disk as netCDF (in temporary directory)
-    filepath1 = f"{tmpdir}/air1.nc"
-    filepath2 = f"{tmpdir}/air2.nc"
-    ds1.to_netcdf(filepath1)
-    ds2.to_netcdf(filepath2)
-    ds1.close()
-    ds2.close()
-
-    return filepath1, filepath2
+def netcdf4_files():
+    test_filepath_dict = generate_test_filepath_dict()
+    return test_filepath_dict["time_1_netcdf4"], test_filepath_dict["time_2_netcdf4"]
 
 
 @pytest.fixture

virtualizarr/tests/test_integration.py

@@ -12,8 +12,8 @@
 @pytest.mark.parametrize(
     "inline_threshold, vars_to_inline",
     [
-        (5e2, ["lat", "lon"]),
-        (5e4, ["lat", "lon", "time"]),
+        (500, ["time", "lat", "lon"]),
+        (5e3, ["lat", "lon", "time"]),
         pytest.param(
             5e7,
             ["lat", "lon", "time", "air"],
@@ -36,7 +36,6 @@ def test_numpy_arrays_to_inlined_kerchunk_refs(
         netcdf4_file, loadable_variables=vars_to_inline, indexes={}
     )
     refs = vds.virtualize.to_kerchunk(format="dict")
-
     # TODO I would just compare the entire dicts but kerchunk returns inconsistent results - see https://github.com/TomNicholas/VirtualiZarr/pull/73#issuecomment-2040931202
     # assert refs == expected
     assert refs["refs"]["air/0.0.0"] == expected["refs"]["air/0.0.0"]
@@ -47,15 +46,11 @@ def test_numpy_arrays_to_inlined_kerchunk_refs(
 
 @pytest.mark.parametrize("format", ["dict", "json", "parquet"])
 class TestKerchunkRoundtrip:
-    def test_kerchunk_roundtrip_no_concat(self, tmpdir, format):
-        # set up example xarray dataset
-        ds = xr.tutorial.open_dataset("air_temperature", decode_times=False)
-
-        # save it to disk as netCDF (in temporary directory)
-        ds.to_netcdf(f"{tmpdir}/air.nc")
+    def test_kerchunk_roundtrip_no_concat(self, netcdf4_file, tmpdir, format):
+        ds = xr.open_dataset(netcdf4_file, decode_times=False)
 
         # use open_dataset_via_kerchunk to read it as references
-        vds = open_virtual_dataset(f"{tmpdir}/air.nc", indexes={})
+        vds = open_virtual_dataset(netcdf4_file, indexes={})
 
         if format == "dict":
             # write those references to an in-memory kerchunk-formatted references dictionary
@@ -76,12 +71,17 @@ def test_kerchunk_roundtrip_no_concat(self, tmpdir, format):
         xrt.assert_identical(roundtrip, ds)
 
     @pytest.mark.parametrize("decode_times,time_vars", [(False, []), (True, ["time"])])
-    def test_kerchunk_roundtrip_concat(self, tmpdir, format, decode_times, time_vars):
+    def test_kerchunk_roundtrip_concat(
+        self, netcdf4_file, netcdf4_files, tmpdir, format, decode_times, time_vars
+    ):
+        netcdf1, netcdf2 = netcdf4_files
+
         # set up example xarray dataset
-        ds = xr.tutorial.open_dataset("air_temperature", decode_times=decode_times)
+        ds = xr.open_dataset(netcdf4_file, decode_times=decode_times)
 
         # split into two datasets
-        ds1, ds2 = ds.isel(time=slice(None, 1460)), ds.isel(time=slice(1460, None))
+        ds1 = xr.open_dataset(netcdf1, decode_times=decode_times)
+        ds2 = xr.open_dataset(netcdf2, decode_times=decode_times)
 
         # save it to disk as netCDF (in temporary directory)
         ds1.to_netcdf(f"{tmpdir}/air1.nc")

virtualizarr/tests/test_xarray.py

@@ -230,10 +230,7 @@ def test_combine_by_coords(self, netcdf4_files):
             vds1 = open_virtual_dataset(filepath1)
         with pytest.warns(UserWarning, match="will create in-memory pandas indexes"):
             vds2 = open_virtual_dataset(filepath2)
-
-        combined_vds = xr.combine_by_coords(
-            [vds2, vds1],
-        )
+        combined_vds = xr.combine_by_coords([vds2, vds1])
 
         assert combined_vds.xindexes["time"].to_pandas_index().is_monotonic_increasing
 
@@ -278,7 +275,7 @@ def local_to_s3_url(old_local_path: str) -> str:
         renamed_vds = vds.virtualize.rename_paths(local_to_s3_url)
         assert (
             renamed_vds["air"].data.manifest.dict()["0.0.0"]["path"]
-            == "s3://bucket/air.nc"
+            == "s3://bucket/test_ds_netcdf4_full_time.nc"
         )
 
     def test_invalid_type(self, netcdf4_file):