ValueError when running "corners" and "squareness"

[iboates]

Describe the problem

When I run momepy.corners, I get this error:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/me/code/quackosm-test/venv/lib/python3.12/site-packages/momepy/functional/_shape.py", line 516, in corners
    cc.index = geometry.index
    ^^^^^^^^
  File "/home/me/code/quackosm-test/venv/lib/python3.12/site-packages/pandas/core/generic.py", line 6220, in __setattr__
    return object.__setattr__(self, name, value)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "pandas/_libs/properties.pyx", line 69, in pandas._libs.properties.AxisProperty.__set__
  File "/home/me/code/quackosm-test/venv/lib/python3.12/site-packages/pandas/core/generic.py", line 766, in _set_axis
    self._mgr.set_axis(axis, labels)
  File "/home/me/code/quackosm-test/venv/lib/python3.12/site-packages/pandas/core/internals/managers.py", line 273, in set_axis
    self._validate_set_axis(axis, new_labels)
  File "/home/me/code/quackosm-test/venv/lib/python3.12/site-packages/pandas/core/internals/managers.py", line 288, in _validate_set_axis
    raise ValueError(
ValueError: Length mismatch: Expected axis has 925 elements, new values have 926 elements

Steps to reproduce

1. Download my data: corners.zip
2. Execute:

import geopandas as gpd
import momepy

gdf = gpd.read_file("/path/to/corners.gpkg")
momepy.corners(gdf)

Versions of your packages

momepy: 0.11.0
geopandas: 1.1.2

Your operating system

Distributor ID: Ubuntu Description: Ubuntu 24.04.2 LTS Release: 24.04 Codename: noble

Additional context

No response

[iboates]

Seems to also happen with "squareness"

[martinfleis]

Thanks for the report! These functions do not support MultiPolygons at the moment when include_interiors=False. All your geometries are MultiPolygons with a single part each, resulting in this weird issue.

We could potentially extend support to MultiPolygons or raise...

[martinfleis]

This is how corners would look like

    if include_interiors:
        coords = geometry.reset_index(drop=True).get_coordinates(index_parts=False)
        cc = coords.groupby(level=0).apply(_count_corners, eps=eps)
    else:
        if (geometry.geom_type == "MultiPolygon").any():
            cc = (
                geometry.explode(index_parts=True)
                .exterior.get_coordinates(index_parts=False)
                .groupby(level=[0, 1])  # to avoid counting across parts
                .apply(_count_corners, eps=10)
                .groupby(level=0)  # sum corners of parts
                .sum()
            )
        else:
            coords = geometry.reset_index(drop=True).exterior.get_coordinates(
                index_parts=False
            )
            cc = coords.groupby(level=0).apply(_count_corners, eps=eps)

    cc.index = geometry.index
    return cc

For squareness, it is slightly more involved as we can't just do mean over mean when aggregating parts.

So, on one hand we could make the effort. On the other, a lot of other functions are simply not well defined for MultiPolygons either (even theoretically) so I am not sure if it is worth it over raising with an informative message. @jGaboardi thoughts?

[iboates]

Thanks for the report! These functions do not support MultiPolygons at the moment when include_interiors=False. All your geometries are MultiPolygons with a single part each, resulting in this weird issue.

We could potentially extend support to MultiPolygons or raise...

Thanks for the quick reply!

I will have another look soon to see if I can convert all to polygons or use the parameter as you said.

I actually just came back here to report that centroid_corner_distance is also affected

[martinfleis]

Yeah, anything that uses the detection of corners will showcase the same behaviour. In your specific case, you can explode it on-the-fly as it is all single part multipolygons (momepy.squareness(gdf.explode())) to get a valid Series back but if there are actual MultiPart objects with multiple independent polygons, I would generally recommend resolving that prior using momepy.

[iboates]

@martinfleis It seems that exploding the geodataframe has no effect for momepy.centroid_corner_distance, see the following file, it fails despite confirming that all geometries are polygons

broken_centroid_corner_distance.zip

[martinfleis]

@iboates I am afraid I cannot reproduce that one.

In [1]: import geopandas as gpd
   ...: import momepy

In [2]: gdf = gpd.read_file('/Users/martin/Downloads/broken_centroid_corner_distance.zip')

In [3]: momepy.centroid_corner_distance(gdf)
Out[3]: 
          mean       std
0    10.195623  0.002888
1     9.333861  0.002552
2    11.768333  3.449638
3     3.951355  0.002126
4     3.654988  0.000912
..         ...       ...
922  13.376573  7.427144
923   6.029147  2.375218
924  10.647982  1.462695
925  10.621362  0.983441
926  17.616961  1.160778

Can you provide a full Traceback?