Unwanted filling of inner polygons when clipping a shapefile with another shapefile in Python

Question

I'm trying to clip a shapefile with another .shp using pure python.

Original shp I need to clip:

Area of Interest for clipping (shown overlapped to the original .shp for convenience):

If I invoke clipvectorbypolygon from QGIS toolbox via processing.run, I get the desired result:

params = {'INPUT': original_shape,
          'MASK': area_of_interest,
          'OPTIONS':'',
          'OUTPUT': result_shape
         }

processing.run("gdal:clipvectorbypolygon", parameters = params)

But, when trying to replicate it using pure python, I always obtain a clipped polygon with the inner areas filled:

I've tried by using geopandas.clip

input = gpd.read_file(original_shape)

clipped = gpd.clip(input, area_of_interest)

clipped.to_file(result_shape)

and geopandas.GeoSeries.intersection:

input = gpd.read_file(original_shape)

clipped = gpd.GeoSeries.intersection(input, area_of_interest)

clipped.to_file(result_shape)

Both produce the same result:

exploding the original .shp gives a slightly different output, but same problem:

input = gpd.read_file(original_shape).explode()

for index, row in input['geometry'].iteritems():
    clipped = gpd.clip(area_of_interest, row)
    clipped.to_file(os.path.join(output_path, f'{index}.shp'))

The only way of not getting filled polygons is passing as mask just the extent of the Area of Interest with total_bounds, but (predictably) the output is a shape with more geometry as needed:

input = gpd.read_file(original_shape)

clipped = gpd.clip(input, area_of_interest.total_bounds)

clipped.to_file(result_shape)

EDIT: Additional info

The problem seems to be when the result is written to a file. Plotting clipped apparently gives the expected outcome:

clipped.plot().axis('off')
clipped.boundary.plot().axis('off')

Answer 1

In case of invalid geometry, you might get a result like this.

Try make_valid function of shapely

from shapely.validation import make_valid

Before clipping:

input_df["geometry"] = input_df.geometry.apply(make_valid)

or after clipping:

clipped["geometry"] = clipped.geometry.apply(make_valid)

clipped.to_file(result_shape)