RegionCoverer with Polygon in Python

[Kuan Butts]

Hello

Thanks in advance for any guidance that can be provided. I have a the coordinates of a Polygon and would like to identify the set of S2 cells that best represents it (a la what this tool provides: http://s2.sidewalklabs.com/regioncoverer/).

Right now, from the Py docs, there is guidance on how to get the S2 cells that cover a rectangle based on two LatLngs:

r = s2sphere.RegionCoverer()
p1 = s2sphere.LatLng.from_degrees(33, -122)
p2 = s2sphere.LatLng.from_degrees(33.1, -122.1)
cell_ids = r.get_covering(s2sphere.LatLngRect.from_point_pair(p1, p2))

I'd like to do something similar, but provide an array of coordinates instead, representing a Polygon and get the covering for that, from Py. Is that possible with current bindings?

Thanks again
Kuan

[Jesse Rosenstock]

On Monday, April 20, 2020 at 8:30:08 PM UTC+2, Kuan Butts wrote:

Thanks in advance for any guidance that can be provided. I have a the coordinates of a Polygon and would like to identify the set of S2 cells that best represents it (a la what this tool provides: http://s2.sidewalklabs.com/regioncoverer/).

[...] 

I'd like to do something similar, but provide an array of coordinates instead, representing a Polygon and get the covering for that, from Py. Is that possible with current bindings?

Yes. See the python tests that use S2RegionCoverer and S2Polygon. You will initialize the S2Polygon using an S2Loop.

https://github.com/google/s2geometry/blob/master/src/python/pywraps2_test.py#L69

https://github.com/google/s2geometry/blob/master/src/python/pywraps2_test.py#L92

https://github.com/google/s2geometry/blob/master/src/python/pywraps2_test.py#L151

 

[Kuan Butts]

Thanks Jesse

This is embarrassing but I only just now realized that this repo (GH google/s2geometry) is different than S2Sphere from SWL (https://github.com/sidewalklabs/s2sphere).

[Kuan Butts]

* Different in that it only has part of the C library exposed in the SWL Py version vs using the Py extensions that the main S2Geom repo exposes w/ SWIG. Easier install though (via pip) so that's why I gravitated towards it at first.