Limiting Voronoi polygons from shapely.voronoi_polygons to a shape with extend_to

Question

I'm trying to get some voronoi polygons from a set of coordinates distributed within a country, I don't need said polygons to extend out of the country, I tried using extend_to from voronoi_polygons documentation and downloaded an shp file from gadm and reading it with fiona as a geometry, but when passing it as argument it won't take it in count and the voronoi polygons will extend out of the country.

with fiona.open("./gadm41_BOL_0.shp") as file:
    country_shape = shape(file[0]["geometry"])

data = pd.read_csv('data.csv')
points = MultiPoint(data[['long', 'lat']].values)

vor_polygons = voronoi_polygons(points, extend_to=country_shape)

It's worth noting that when displaying country_shape on a jupyter notebook it'll show a small contour / silhouette of the country properly, might as well add that the type of country_shape is shapely.geometry.multipolygon.MultiPolygon, I'm also sure with a high degree of confidence that none of the coordinates in data.csv is outside the country, hence extend_to shouldn't fail.

Answer 1

If you have a close look at the documentation of shapely.voronoi_polygons, you can see for extend_to that the "diagram will be extended to cover the envelope of this geometry". The "envelope" is synonym to the "bounds" or "minimum bounding rectange", so it is normal that the voronoi polygons don't follow the exact boundaries of the country. Relevant extract from the documentation:

extend_to:  Geometry or array_like, optional: 
    If provided, the diagram will be extended to cover the 
    envelope of this geometry (unless this envelope is smaller than the 
    input geometry)

If you want the voronoi polygons to be clipped to the country, you can calculate the intersection between them:

vor_polygons_in_country = shapely.intersection(vor_polygons, country_shape)