Add auto detection of driver for .geojson, .geojsonl and .geojsons files

pyogrio/raw.py

@@ -14,6 +14,9 @@
     ".gpkg": "GPKG",
     ".shp": "ESRI Shapefile",
     ".json": "GeoJSON",
+    ".geojson": "GeoJSON",
+    ".geojsons": "GeoJSONSeq",
+    ".geojsonl": "GeoJSONSeq",
     ".fgb": "FlatGeobuf",
 }
 

pyogrio/tests/conftest.py

@@ -8,7 +8,7 @@
 
 
 _data_dir = Path(__file__).parent.resolve() / "fixtures"
-ALL_EXTS = [".shp", ".gpkg", ".json"]
+ALL_EXTS = [".shp", ".gpkg", ".geojson", ".geojsonl"]
 
 
 def pytest_report_header(config):

pyogrio/tests/test_geopandas_io.py

@@ -448,21 +448,36 @@ def test_write_dataframe(tmpdir, naturalearth_lowres, driver, ext):
 
 @pytest.mark.filterwarnings("ignore:.*Layer .* does not have any features to read")
 @pytest.mark.parametrize(
-    "driver,ext", [("ESRI Shapefile", "shp"), ("GeoJSON", "geojson"), ("GPKG", "gpkg")]
+    "ext", [ext for ext in ALL_EXTS if ext not in [".geojsonl", ".geojsons"]]
 )
-def test_write_empty_dataframe(tmpdir, driver, ext):
+def test_write_empty_dataframe(tmp_path, ext):
     expected = gp.GeoDataFrame(geometry=[], crs=4326)
 
-    filename = os.path.join(str(tmpdir), f"test.{ext}")
-    write_dataframe(expected, filename, driver=driver)
-
-    assert os.path.exists(filename)
+    filename = tmp_path / f"test{ext}"
+    write_dataframe(expected, filename)
 
+    assert filename.exists()
     df = read_dataframe(filename)
-
     assert_geodataframe_equal(df, expected)
 
 
+@pytest.mark.parametrize("ext", [".geojsonl", ".geojsons"])
+def test_write_read_empty_dataframe_unsupported(tmp_path, ext):
+    # Writing empty dataframe to .geojsons or .geojsonl results logically in a 0 byte
+    # file, but gdal isn't able to read those again at the time of writing.
+    # Issue logged here: https://github.com/geopandas/pyogrio/issues/94
+    expected = gp.GeoDataFrame(geometry=[], crs=4326)
+
+    filename = tmp_path / f"test{ext}"
+    write_dataframe(expected, filename)
+
+    assert filename.exists()
+    with pytest.raises(
+        Exception, match=".* not recognized as a supported file format."
+    ):
+        _ = read_dataframe(filename)
+
+
 def test_write_dataframe_gdalparams(tmp_path, naturalearth_lowres):
     original_df = read_dataframe(naturalearth_lowres)
 

pyogrio/tests/test_raw_io.py

@@ -6,8 +6,9 @@
 import pytest
 
 from pyogrio import list_layers, list_drivers
-from pyogrio.raw import read, write
+from pyogrio.raw import DRIVERS, read, write
 from pyogrio.errors import DataSourceError, DataLayerError, FeatureError
+from pyogrio.tests.conftest import prepare_testfile
 
 
 def test_read(naturalearth_lowres):
@@ -33,6 +34,32 @@ def test_read(naturalearth_lowres):
     assert geometry[0][:6] == b"\x01\x06\x00\x00\x00\x03"
 
 
+@pytest.mark.parametrize("ext", DRIVERS)
+def test_read_autodetect_driver(tmp_path, naturalearth_lowres, ext):
+    # Test all supported autodetect drivers
+    testfile = prepare_testfile(naturalearth_lowres, dst_dir=tmp_path, ext=ext)
+
+    assert testfile.suffix == ext
+    assert testfile.exists()
+    meta, _, geometry, fields = read(testfile)
+
+    assert meta["crs"] == "EPSG:4326"
+    assert meta["geometry_type"] in ("Polygon", "Unknown")
+    assert meta["encoding"] == "UTF-8"
+    assert meta["fields"].shape == (5,)
+
+    assert meta["fields"].tolist() == [
+        "pop_est",
+        "continent",
+        "name",
+        "iso_a3",
+        "gdp_md_est",
+    ]
+
+    assert len(fields) == 5
+    assert len(geometry) == len(fields[0])
+
+
 def test_read_invalid_layer(naturalearth_lowres):
     with pytest.raises(DataLayerError, match="Layer 'invalid' could not be opened"):
         read(naturalearth_lowres, layer="invalid")
@@ -407,4 +434,3 @@ def test_read_data_types_numeric_with_null(test_gpkg_nulls):
             assert field.dtype == "float32"
         else:
             assert field.dtype == "float64"
-