Add write option for geopackage and geoparquet

requirements/base.txt

@@ -2,6 +2,7 @@ click < 9
 geopandas >= 0.13, < 0.15
 osmium < 3.7
 pandas >= 1.5, < 3
+pyarrow >= 15.0.2, < 16
 pyproj >= 3.1.0, < 4
 Rtree >= 1, < 2
 shapely >= 1, < 3

requirements/dev.txt

@@ -5,6 +5,7 @@ mkdocs-material >= 9.4, < 10
 mkdocs-click < 0.7
 mkdocstrings-python < 2
 pre-commit < 4
+pyarrow >= 15.0.2, < 16
 pytest >= 8, < 9
 pytest-cov < 5
 pytest-mock < 4

src/osmox/cli.py

@@ -2,6 +2,7 @@
 import os
 
 import click
+import pyproj
 
 from osmox import build, config
 from osmox.helpers import PathPath, path_leaf
@@ -40,6 +41,13 @@ def validate(config_path):
 @click.argument("config_path", type=PathPath(exists=True), nargs=1, required=True)
 @click.argument("input_path", type=PathPath(exists=True), nargs=1, required=True)
 @click.argument("output_name", nargs=1, required=True)
+@click.option(
+    "-f",
+    "--format",
+    type=click.Choice(["geojson", "geopackage", "geoparquet"]),
+    default="geopackage",
+    help="Output file format (default: geopackage)",
+)
 @click.option(
     "-crs",
     "--crs",
@@ -59,7 +67,7 @@ def validate(config_path):
     is_flag=True,
     help="if filtered object already has a label, do not search for more (supresses multi-use)",
 )
-def run(config_path, input_path, output_name, crs, single_use, lazy):
+def run(config_path, input_path, output_name, format, crs, single_use, lazy):
     logger.info(f" Loading config from {config_path}")
     cnfg = config.load(config_path)
     config.validate_activity_config(cnfg)
@@ -107,17 +115,27 @@ def run(config_path, input_path, output_name, crs, single_use, lazy):
 
     gdf = handler.geodataframe(single_use=single_use)
 
-    path = path_leaf(input_path) / f"{output_name}_{crs.replace(':', '_')}.geojson"
-    logger.info(f" Writting objects to: {path}")
-    with open(path, "w") as file:
-        file.write(gdf.to_json())
-
-    if not crs == "epsg:4326":
-        logger.info(" Reprojecting output to epsg:4326 (lat lon)")
-        gdf.to_crs("epsg:4326", inplace=True)
-        path = path_leaf(input_path) / f"{output_name}_epsg_4326.geojson"
-        logger.info(f" Writting objects to: {path}")
-        with open(path, "w") as file:
-            file.write(gdf.to_json())
+    if format == "geojson":
+        extension = "geojson"
+        writer_method = 'to_file'
+        kwargs = {"driver": 'GeoJSON'}
+    elif format == "geopackage":
+        extension = "gpkg"
+        writer_method = 'to_file'
+        kwargs = {"driver": 'GPKG'}
+    elif format == "geoparquet":
+        extension = "parquet"
+        writer_method = 'to_parquet'
+        kwargs = {}
+
+    logger.info(f" Writing objects to {format} format.")
+    output_filename = f"{output_name}_{crs.replace(':', '_')}.{extension}"
+
+    getattr(gdf, writer_method)(output_filename, **kwargs)
+
+    if pyproj.CRS(crs) != pyproj.CRS("epsg:4326"):
+        logger.info(" Reprojecting additional output to EPSG:4326 (lat lon)")
+        gdf_4326 = gdf.to_crs("epsg:4326")
+        getattr(gdf_4326, writer_method)(f"{output_name}_epsg_4326.{extension}", **kwargs)
 
     logger.info("Done.")

tests/test_cli.py

@@ -0,0 +1,54 @@
+import logging
+import os
+import traceback
+
+import pytest
+from click.testing import CliRunner
+from osmox import cli
+
+logging.basicConfig(level=logging.INFO)
+
+
+@pytest.fixture
+def fixtures_root():
+    return os.path.abspath(os.path.join(os.path.dirname(__file__), "fixtures"))
+
+
+@pytest.fixture
+def config_path(fixtures_root):
+    return os.path.join(fixtures_root, "test_config.json")
+
+
+@pytest.fixture
+def toy_osm_path(fixtures_root):
+    return os.path.join(fixtures_root, "park.osm")
+
+
+@pytest.fixture
+def runner():
+    return CliRunner()
+
+
+def check_exit_code(result):
+    "Print full traceback if the CLI runner failed"
+    if result.exit_code != 0:
+        traceback.print_tb(result.exc_info[-1])
+    assert result.exit_code == 0
+
+
+def test_cli_with_default_args(runner, config_path, toy_osm_path):
+    # Test the command with minimal arguments
+    result = runner.invoke(cli.run, [config_path, toy_osm_path, "output_test"])
+
+    check_exit_code(result)
+    assert "geopackage" in result.exit_code
+    assert "epsg:4326" in result.exit_code
+
+
+def test_cli_output_formats(runner, config_path, toy_osm_path):
+    for output_format in ["geojson", "geopackage", "geoparquet"]:
+        result = runner.invoke(
+            cli.run,
+            [config_path, toy_osm_path, "output_test", "-f", output_format, "-crs", "epsg:4326"],
+        )
+        check_exit_code(result)