Esri
Empty SHAPE
I'd like the spatial dataframes geometry to have empty geometry
when reading data from database(sqls) we can often see empty geometry columns.
having an empty geometry "Imagine that a system creates a trigger to insert a new row, than an webhook does an etl to get data and geometrie from an external api"
The geoaccessor can read empty geometries, but creates the row with None. And the column becomes object.
geom = sdf.SHAPE.iloc[0]
geom.isEmpty
# would return True
Gracefully handle empty geom would be a good thing, if possible.
Describe the solution you'd like Proposed Changes I suggest that to implement the following changes:
Modify the spatial dataframe structure to accommodate empty geometries.
Ensure that operations on the dataframes (like transformations, visualizations, and exports, validation, svg) gracefully handle empty geometries. Provide clear documentation on how empty geometries are represented and managed within the dataframes. I believe that incorporating these changes will enhance the robustness and flexibility of our spatial data handling capabilities. I look forward to discussing this further and exploring the best ways to integrate this feature.
Describe alternatives you've considered Implement ORM with sqlalchemy. Or geoaccessor and split the spatial dataframe. Additional context
@achapkowski
Right now i'm testing in my local env. But in the client it is something like this:
@achapkowski
To my understanding, using arcpy would necessitate this initial fix:
Upon reviewing the Geometry Factory, I noticed that it already has the is_empty method implemented correctly. Perhaps this should be tagged as a bug.
Another enhancement or bug is the extension of the numpy array. In my understanding, sdf.isna() and sdf.isnull() should return True for empty geometries.
The implementation might look something like this, if I'm not mistaken.
P.S.: I tested this with arcpy as the backend. https://github.com/geopandas/geopandas/blob/main/geopandas/array.py