Tracking attributes of `changed` edges after `neatify`

[gsagostini]

As discussed in #6 . Some edges with changed status after a neatify loop will not inherit any attribute. Is this expected? Using neatnet 0.1.2

import osmnx as ox
import neatnet

#Collect a graph via osmnx and convert to a GeoDataFrame with a single attribute:
G = ox.graph_from_bbox((-73.86, 40.73,-73.85, 40.74))
Gedges = ox.convert.graph_to_gdfs(G, nodes=False)[['geometry']].reset_index(drop=True).to_crs('EPSG:3857')
Gedges['attribute'] = Gedges.index.astype(int)

#Neatify:
neatified_edges = neatnet.neatify(Gedges)

#There are edges labeled "changed" which do not receive an attribute:
bug_changed = (neatified_edges._status == 'changed')&(neatified_edges.attribute.isna())
print(bug_changed.sum())

This prints 1 for bug_changed.sum() i.e. there is an edge whose status is _changed which does not inherit the attribute from any parent edge. This is happening, both in this example and in others, around bridges or overpasses.

I cannot yet produce an example where the new edge receives an attribute from basic geometries or from an osmnx graph, but this is happening to me on a much larger graph. Will continue to look for an example and update, but figured this issue could be helpful to look at already.

[lenkahas]

I am experiencing a similar issue. I have noticed this happens to me when using the PBF files through pyrosm but not through osmnx, so I have attributed it to the data sources. I'll have a look next week to see if I can get a working example out.

[jGaboardi]

@lenkahas -- this is great info for helping troubleshoot.

[lenkahas]

I apologise, looking through this, my issue is completely unrelated to neatify

[martinfleis]

The case reported above is fixed in #264. @gsagostini I am not sure how representative this case is so I'll let you decide whether this can be closed or not by that PR.