Create fixture for test files of other file types (#74)

Author: theroggy

pyogrio/tests/conftest.py

@@ -4,7 +4,7 @@
 import pytest
 
 from pyogrio import __gdal_version_string__, __version__, list_drivers
-
+import pyogrio
 
 _data_dir = Path(__file__).parent.resolve() / "fixtures"
 
@@ -18,14 +18,34 @@ def pytest_report_header(config):
     )
 
 
+def prepare_testfile(testfile_path, dst_dir, ext):
+    if ext == testfile_path.suffix:
+        return testfile_path
+
+    dst_path = dst_dir / f"{testfile_path.stem}{ext}"
+    if dst_path.exists():
+        return dst_path
+    gdf = pyogrio.read_dataframe(testfile_path)
+    pyogrio.write_dataframe(gdf, dst_path)
+    return dst_path
+
+
 @pytest.fixture(scope="session")
 def data_dir():
     return _data_dir
 
 
-@pytest.fixture(scope="session")
-def naturalearth_lowres():
-    return _data_dir / Path("naturalearth_lowres/naturalearth_lowres.shp")
+@pytest.fixture(scope="function")
+def naturalearth_lowres(tmp_path, request):
+    ext = getattr(request, "param", ".shp")
+    testfile_path = _data_dir / Path("naturalearth_lowres/naturalearth_lowres.shp")
+
+    return prepare_testfile(testfile_path, tmp_path, ext)
+
+
+@pytest.fixture(scope="function", params=[".shp", ".gpkg", ".json"])
+def naturalearth_lowres_all_ext(tmp_path, naturalearth_lowres, request):
+    return prepare_testfile(naturalearth_lowres, tmp_path, request.param)
 
 
 @pytest.fixture(scope="function")
@@ -44,4 +64,3 @@ def naturalearth_lowres_vsi(tmp_path, naturalearth_lowres):
 @pytest.fixture(scope="session")
 def test_fgdb_vsi():
     return f"/vsizip/{_data_dir}/test_fgdb.gdb.zip"
-

pyogrio/tests/test_geopandas_io.py

@@ -21,8 +21,8 @@
 pytestmark = pytest.mark.skipif(not has_geopandas, reason="GeoPandas not available")
 
 
-def test_read_dataframe(naturalearth_lowres):
-    df = read_dataframe(naturalearth_lowres)
+def test_read_dataframe(naturalearth_lowres_all_ext):
+    df = read_dataframe(naturalearth_lowres_all_ext)
 
     assert isinstance(df, gp.GeoDataFrame)
 
@@ -45,8 +45,19 @@ def test_read_dataframe_vsi(naturalearth_lowres_vsi):
     assert len(df) == 177
 
 
-def test_read_no_geometry(naturalearth_lowres):
-    df = read_dataframe(naturalearth_lowres, read_geometry=False)
+@pytest.mark.parametrize(
+        "naturalearth_lowres, expected_ext",
+        [(".gpkg", ".gpkg"), (".shp", ".shp")],
+        indirect=["naturalearth_lowres"])
+def test_fixture_naturalearth_lowres(naturalearth_lowres, expected_ext):
+    # Test the fixture with "indirect" parameter
+    assert naturalearth_lowres.suffix == expected_ext
+    df = read_dataframe(naturalearth_lowres)
+    assert len(df) == 177
+
+
+def test_read_no_geometry(naturalearth_lowres_all_ext):
+    df = read_dataframe(naturalearth_lowres_all_ext, read_geometry=False)
     assert isinstance(df, pd.DataFrame)
     assert not isinstance(df, gp.GeoDataFrame)
 
@@ -82,9 +93,9 @@ def test_read_layer(test_fgdb_vsi):
     assert "RIVER_MILE" in df.columns
 
 
-def test_read_layer_invalid(naturalearth_lowres):
+def test_read_layer_invalid(naturalearth_lowres_all_ext):
     with pytest.raises(DataLayerError, match="Layer 'wrong' could not be opened"):
-        read_dataframe(naturalearth_lowres, layer="wrong")
+        read_dataframe(naturalearth_lowres_all_ext, layer="wrong")
 
 
 @pytest.mark.filterwarnings("ignore: Measured")
@@ -98,77 +109,87 @@ def test_read_null_values(test_fgdb_vsi):
 
     # make sure that Null values are preserved
     assert df.SEGMENT_NAME.isnull().max() == True
-    assert df.loc[df.SEGMENT_NAME.isnull()].SEGMENT_NAME.iloc[0] == None
+    assert df.loc[df.SEGMENT_NAME.isnull()].SEGMENT_NAME.iloc[0] is None
 
 
-def test_read_fid_as_index(naturalearth_lowres):
+def test_read_fid_as_index(naturalearth_lowres_all_ext):
     kwargs = {"skip_features": 2, "max_features": 2}
 
     # default is to not set FIDs as index
-    df = read_dataframe(naturalearth_lowres, **kwargs)
+    df = read_dataframe(naturalearth_lowres_all_ext, **kwargs)
     assert_index_equal(df.index, pd.RangeIndex(0, 2))
 
-    df = read_dataframe(naturalearth_lowres, fid_as_index=False, **kwargs)
+    df = read_dataframe(naturalearth_lowres_all_ext, fid_as_index=False, **kwargs)
     assert_index_equal(df.index, pd.RangeIndex(0, 2))
 
-    df = read_dataframe(naturalearth_lowres, fid_as_index=True, **kwargs)
-    assert_index_equal(df.index, pd.Index([2, 3], name="fid"))
+    df = read_dataframe(naturalearth_lowres_all_ext, fid_as_index=True, **kwargs)
+    if naturalearth_lowres_all_ext.suffix in ['.gpkg']:
+        # File format where fid starts at 1
+        assert_index_equal(df.index, pd.Index([3, 4], name="fid"))
+    else:
+        # File format where fid starts at 0
+        assert_index_equal(df.index, pd.Index([2, 3], name="fid"))
 
 
 @pytest.mark.filterwarnings("ignore: Layer")
-def test_read_where(naturalearth_lowres):
+def test_read_where(naturalearth_lowres_all_ext):
     # empty filter should return full set of records
-    df = read_dataframe(naturalearth_lowres, where="")
+    df = read_dataframe(naturalearth_lowres_all_ext, where="")
     assert len(df) == 177
 
     # should return singular item
-    df = read_dataframe(naturalearth_lowres, where="iso_a3 = 'CAN'")
+    df = read_dataframe(naturalearth_lowres_all_ext, where="iso_a3 = 'CAN'")
     assert len(df) == 1
     assert df.iloc[0].iso_a3 == "CAN"
 
-    df = read_dataframe(naturalearth_lowres, where="iso_a3 IN ('CAN', 'USA', 'MEX')")
+    df = read_dataframe(naturalearth_lowres_all_ext, where="iso_a3 IN ('CAN', 'USA', 'MEX')")
     assert len(df) == 3
     assert len(set(df.iso_a3.unique()).difference(["CAN", "USA", "MEX"])) == 0
 
     # should return items within range
     df = read_dataframe(
-        naturalearth_lowres, where="POP_EST >= 10000000 AND POP_EST < 100000000"
+        naturalearth_lowres_all_ext, where="POP_EST >= 10000000 AND POP_EST < 100000000"
     )
     assert len(df) == 75
     assert df.pop_est.min() >= 10000000
     assert df.pop_est.max() < 100000000
 
     # should match no items
-    df = read_dataframe(naturalearth_lowres, where="ISO_A3 = 'INVALID'")
+    df = read_dataframe(naturalearth_lowres_all_ext, where="ISO_A3 = 'INVALID'")
     assert len(df) == 0
 
 
-def test_read_where_invalid(naturalearth_lowres):
-    with pytest.raises(ValueError, match="Invalid SQL"):
-        read_dataframe(naturalearth_lowres, where="invalid")
+def test_read_where_invalid(naturalearth_lowres_all_ext):
+    if naturalearth_lowres_all_ext.suffix in [".gpkg"]:
+        # Geopackage doesn't raise, but returns empty df?
+        gdf = read_dataframe(naturalearth_lowres_all_ext, where="invalid")
+        assert len(gdf) == 0
+    else:
+        with pytest.raises(ValueError, match="Invalid SQL"):
+            read_dataframe(naturalearth_lowres_all_ext, where="invalid")
 
 
 @pytest.mark.parametrize("bbox", [(1,), (1, 2), (1, 2, 3)])
-def test_read_bbox_invalid(naturalearth_lowres, bbox):
+def test_read_bbox_invalid(naturalearth_lowres_all_ext, bbox):
     with pytest.raises(ValueError, match="Invalid bbox"):
-        read_dataframe(naturalearth_lowres, bbox=bbox)
+        read_dataframe(naturalearth_lowres_all_ext, bbox=bbox)
 
 
-def test_read_bbox(naturalearth_lowres):
+def test_read_bbox(naturalearth_lowres_all_ext):
     # should return no features
     with pytest.warns(UserWarning, match="does not have any features to read"):
-        df = read_dataframe(naturalearth_lowres, bbox=(0, 0, 0.00001, 0.00001))
+        df = read_dataframe(naturalearth_lowres_all_ext, bbox=(0, 0, 0.00001, 0.00001))
         assert len(df) == 0
 
-    df = read_dataframe(naturalearth_lowres, bbox=(-140, 20, -100, 40))
+    df = read_dataframe(naturalearth_lowres_all_ext, bbox=(-140, 20, -100, 40))
     assert len(df) == 2
     assert np.array_equal(df.iso_a3, ["USA", "MEX"])
 
 
-def test_read_fids(naturalearth_lowres):
+def test_read_fids(naturalearth_lowres_all_ext):
     # ensure keyword is properly passed through
-    fids = np.array([0, 10, 5], dtype=np.int64)
-    df = read_dataframe(naturalearth_lowres, fids=fids, fid_as_index=True)
+    fids = np.array([1, 10, 5], dtype=np.int64)
+    df = read_dataframe(naturalearth_lowres_all_ext, fids=fids, fid_as_index=True)
     assert len(df) == 3
     assert np.array_equal(fids, df.index.values)
 
@@ -238,27 +259,27 @@ def test_write_empty_dataframe(tmpdir, driver, ext):
     assert_geodataframe_equal(df, expected)
 
 
-def test_write_dataframe_gdalparams(tmpdir, naturalearth_lowres):
+def test_write_dataframe_gdalparams(tmp_path, naturalearth_lowres):
     original_df = read_dataframe(naturalearth_lowres)
-    
-    test_noindex_filename = os.path.join(str(tmpdir), f"test_gdalparams_noindex.shp")
+
+    test_noindex_filename = tmp_path / "test_gdalparams_noindex.shp"
     write_dataframe(original_df, test_noindex_filename, SPATIAL_INDEX="NO")
-    assert os.path.exists(test_noindex_filename) is True
-    test_noindex_index_filename = os.path.join(str(tmpdir), f"test_gdalparams_noindex.qix")
-    assert os.path.exists(test_noindex_index_filename) is False
-    
-    test_withindex_filename = os.path.join(str(tmpdir), f"test_gdalparams_withindex.shp")
+    assert test_noindex_filename.exists() is True
+    test_noindex_index_filename = tmp_path / "test_gdalparams_noindex.qix"
+    assert test_noindex_index_filename.exists() is False
+
+    test_withindex_filename = tmp_path / "test_gdalparams_withindex.shp"
     write_dataframe(original_df, test_withindex_filename, SPATIAL_INDEX="YES")
-    assert os.path.exists(test_withindex_filename) is True
-    test_withindex_index_filename = os.path.join(str(tmpdir), f"test_gdalparams_withindex.qix")
-    assert os.path.exists(test_withindex_index_filename) is True
+    assert test_withindex_filename.exists() is True
+    test_withindex_index_filename = tmp_path / "test_gdalparams_withindex.qix"
+    assert test_withindex_index_filename.exists() is True
 
 
 @pytest.mark.filterwarnings(
     "ignore: You will likely lose important projection information"
 )
-def test_custom_crs_io(tmpdir, naturalearth_lowres):
-    df = read_dataframe(naturalearth_lowres)
+def test_custom_crs_io(tmpdir, naturalearth_lowres_all_ext):
+    df = read_dataframe(naturalearth_lowres_all_ext)
     # project Belgium to a custom Albers Equal Area projection
     expected = df.loc[df.name == "Belgium"].to_crs(
         "+proj=aea +lat_1=49.5 +lat_2=51.5 +lon_0=4.3"
@@ -275,4 +296,3 @@ def test_custom_crs_io(tmpdir, naturalearth_lowres):
     assert crs["lat_2"] == 51.5
     assert crs["lon_0"] == 4.3
     assert df.crs.equals(expected.crs)
-