geopandas / pyogrio

Vectorized vector I/O using OGR
https://pyogrio.readthedocs.io
MIT License
259 stars 22 forks source link

TST: Replace test datasets with pyogrio-generated files where possible #441

Open brendan-ward opened 3 days ago

brendan-ward commented 3 days ago

Per #433, we were including test data files from other parties and may not have been in great compliance with their licenses, even though we were usually using a tiny extract of the source files.

See the updates to tests/fixtures/README.md first for updated guidance on managing test files, and steps used to create those still present as files.

Where we were using GeoJSON files, I migrated those directly into conftest.py so they are generated using code, which should make the licensing of those files more clear (i.e., part of Pyogrio source code so our MIT license is clear). I don't have a strong opinion about whether these should be files or in code, but putting them in code seemed reasonable; let me know if you disagree.

This also made it easier to create an invalid GeoJSON file for testing polygons with insufficient coordinates (replaces data file with problematic license).

Where we were using 3rd party FGDB datasets, I created new datasets to mimic how we were using those. I used QGIS to hand-digitize (super simple) LineString ZM, Polygon ZM, Curve, CurvePolygon, and MultiSurface datasets, since we were using one of the FGDB datasets to verify that they were correctly downgraded to supported geometry types.