simplify

.github/workflows/scripts.yml

@@ -33,7 +33,6 @@ jobs:
         run: |
           python -m pip install pytest
           cd tests
-          python generate_test_data.py
           pytest test_json_schema.py -v
 
   test-json-metadata:

.gitignore

@@ -1,4 +1,3 @@
 # Ignore GeoPackage file used in conversion to GeoParquet
 *.gpkg*
-tests/valid/*
-tests/invalid/*
+tests/data/*

tests/generate_test_data.py

@@ -103,6 +103,10 @@ def write_metadata_json(metadata, name, case="valid"):
 metadata["columns"]["geometry"]["crs"] = None
 write_metadata_json(metadata, "crs_null")
 
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["crs"] = "EPSG:4326"
+write_metadata_json(metadata, "crs_string", case="invalid")
+
 
 # Orientation
 

tests/test_json_schema.py

@@ -1,3 +1,5 @@
+
+import copy
 import json
 import pathlib
 
@@ -11,14 +13,139 @@
 SCHEMA = json.loads(SCHEMA_SRC.read_text())
 
 
-@pytest.mark.parametrize(
-    "path", list((HERE / "valid").iterdir()), ids=lambda path: path.name
-)
-def test_valid_schema(path):
+# # Define test cases
+
+valid_cases = {}
+invalid_cases = {}
+
+
+metadata_template = {
+    "version": "0.5.0-dev",
+    "primary_column": "geometry",
+    "columns": {
+        "geometry": {
+            "encoding": "WKB",
+            "geometry_types": [],
+        },
+    },
+}
+
+
+# Minimum required metadata
+
+metadata = copy.deepcopy(metadata_template)
+valid_cases["minimal"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata.pop("version")
+invalid_cases["missing_version"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata.pop("primary_column")
+invalid_cases["missing_primary_column"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata.pop("columns")
+invalid_cases["missing_columns"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"] = {}
+invalid_cases["missing_columns_entry"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"].pop("encoding")
+invalid_cases["missing_geometry_encoding"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"].pop("geometry_types")
+invalid_cases["missing_geometry_type"] = metadata
+
+
+# Geometry column name
+
+metadata = copy.deepcopy(metadata_template)
+metadata["primary_column"] = "geom"
+metadata["columns"]["geom"] = metadata["columns"].pop("geometry")
+valid_cases["geometry_column_name"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["primary_column"] = ""
+invalid_cases["geometry_column_name_empty"] = metadata
+
+
+# Encoding
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["encoding"] = "WKT"
+invalid_cases["encoding"] = metadata
+
+
+# Geometry type - non-empty list
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["geometry_types"] = ["Point"]
+valid_cases["geometry_type_list"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["geometry_types"] = "Point"
+invalid_cases["geometry_type_string"] = metadata
 
-    with open(path, "r") as f:
-        metadata = json.load(f)["geo"]
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["geometry_types"] = ["Curve"]
+invalid_cases["geometry_type_nonexistent"] = metadata
 
+# CRS - explicit null
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["crs"] = None
+valid_cases["crs_null"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["crs"] = "EPSG:4326"
+invalid_cases["crs_string"] = metadata
+
+
+# Orientation
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["orientation"] = "counterclockwise"
+valid_cases["orientation"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["orientation"] = "clockwise"
+invalid_cases["orientation"] = metadata
+
+# Edges
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["edges"] = "planar"
+valid_cases["edges_planar"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["edges"] = "spherical"
+valid_cases["edges_spherical"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["edges"] = "ellipsoid"
+invalid_cases["edges"] = metadata
+
+# Epoch
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["epoch"] = 2015.1
+valid_cases["epoch"] = metadata
+
+metadata = copy.deepcopy(metadata_template)
+metadata["columns"]["geometry"]["epoch"] = "2015.1"
+invalid_cases["epoch_string"] = metadata
+
+
+# # Tests
+
+@pytest.mark.parametrize(
+    "metadata", valid_cases.values(), ids=valid_cases.keys()
+)
+def test_valid_schema(request, metadata):
     errors = Draft7Validator(SCHEMA).iter_errors(metadata)
 
     msgs = []
@@ -31,25 +158,41 @@ def test_valid_schema(path):
         msgs.append(msg)
 
     if not valid:
-        raise AssertionError(f"Error while validation {path.name}:\n" + "\n".join(msgs))
+        raise AssertionError(
+            f"Error while validating '{request.node.callspec.id}':\n"
+            + json.dumps({"geo": metadata}, indent=2, sort_keys=True)
+            + "\n\nErrors:\n" + "\n".join(msgs)
+        )
 
 
 @pytest.mark.parametrize(
-    "path", list((HERE / "invalid").iterdir()), ids=lambda path: path.name
+    "metadata", invalid_cases.values(), ids=invalid_cases.keys()
 )
-def test_invalid_schema(request, path):
-    if "missing_columns_entry" in path.name:
+def test_invalid_schema(request, metadata):
+    if "missing_columns_entry" in request.node.callspec.id:
         request.node.add_marker(
                 pytest.mark.xfail(reason="Not yet working", strict=True)
             )
 
-    with open(path, "r") as f:
-        metadata = json.load(f)["geo"]
-
     errors = Draft7Validator(SCHEMA).iter_errors(metadata)
 
     if not len(list(errors)):
         raise AssertionError(
             "This is an invalid GeoParquet file, but no validation error "
-            f"occurred for {path.name}."
+            f"occurred for '{request.node.callspec.id}':\n"
+            + json.dumps({"geo": metadata}, indent=2, sort_keys=True)
         )
+
+
+if __name__ == "__main__":
+    (HERE / "data").mkdir(exist_ok=True)
+
+    def write_metadata_json(metadata, name):
+        with open(HERE / "data" / ("metadata_" + name + ".json"), "w") as f:
+            json.dump({"geo": metadata}, f, indent=2, sort_keys=True)
+
+    for case, metadata in valid_cases.items():
+        write_metadata_json(metadata, "valid_" + case)
+
+    for case, metadata in invalid_cases.items():
+        write_metadata_json(metadata, "invalid_" + case)