Hi David,
No idea if this is best practice or if hobu and Sean will be shaking their
heads in disappointment in a second, but RTree allows one to store an object
(anything pickleable I think) in the index. You can store a dictionary with
attributes and the geometry from a fiona feature like this:
geom = geometry.shape(feature['geometry'])
idx.insert(int(feature['id']), geom.bounds, obj={'properties':
feature['properties'], 'geom':geom})
Now I can iterate over the results of the RTree intersection query with shapely:
for riding in idx.intersection((point.y, point.x), objects="raw"):
if riding['geom'].contains(point):
return riding
There may be a better way to do this, but in my experience the above works
quite well.
Do I get bonus points for also using Fiona in my solution?
Thanks,
Michael
On 2013-03-26, at 2:17 PM, David Fawcett <[email protected]> wrote:
> I am thinking through the logic for efficiently overlaying ~10,000 points on
> ~100 polys. What I want in the end is a list of dicts holding the string ID
> for the point and string ID for the intersecting Poly. I have an example
> with prepared geometries, but now I am working up an RTree example to
> benchmark.
>
> Here is the logic:
>
> - I use RTree to create an index on my polys
> - at the same time, I create a dict to lookup the string ID (stationID) for
> the integer index value
>
> - I loop through the points in my collection and test them against the index.
> - if only one intersection is returned, I grab the id and look up my countyID.
>
> - if my intersection operation on the index returns multiple poly IDs, I need
> to test against the actual geometries.
> This is where I am struggling. I don't see a way to access a specific
> feature from a shapely collection based on a property. My first idea is to
> store the polygon features in a dictionary after reading them from the
> shapefile. This would give me access by ID. At the same time, I am trying
> to work within the Shapely data structures.
>
> Does anyone have any suggestions on best practices for integrating RTree and
> Shapely? Specifically, how you relate geometry collections to RTree indexes.
>
>
> I will post a code example after I modify my current example to not use
> external data sources.
>
> Thanks,
>
> David.
>
>
>
> _______________________________________________
> Community mailing list
> [email protected]
> http://lists.gispython.org/mailman/listinfo/community
_______________________________________________
Community mailing list
[email protected]
http://lists.gispython.org/mailman/listinfo/community
