pys2index

Python / NumPy compatible geographical index based on s2geometry.

This project doesn't provide Python wrappers for the whole s2geometry library. Instead, it aims to provide some index wrappers with an API similar to scipy.spatial.cKDTree.

Build Dependencies

• C++17 compiler
• CMake
• s2geometry
• scikit-build-core
• xtensor-python
• pybind11
• Python
• NumPy

Installation

Using conda

pys2index can be installed using conda (or mamba):

$ conda install pys2index -c conda-forge

From source

First, clone this repository:

$ git clone https://github.com/benbovy/pys2index
$ cd pys2index

You can install all the dependencies using conda (or mamba):

$ conda install python cxx-compiler numpy s2geometry pybind11 xtensor-python cmake scikit-build-core -c conda-forge

Build and install this library

$ python -m pip install . --no-build-isolation

Usage

In [1]: import numpy as np

In [2]: from pys2index import S2PointIndex

In [3]: latlon_points = np.array([[40.0, 15.0],
   ...:                           [-20.0, 53.0],
   ...:                           [81.0, 153.0]])
   ...:

In [4]: index = S2PointIndex(latlon_points)

In [5]: query_points = np.array([[-10.0, 60.0],
   ...:                          [45.0, -20.0],
   ...:                          [75.0, 122.0]])
   ...:

In [6]: distances, positions = index.query(query_points)

In [7]: distances
Out[7]: array([12.06534671, 26.07312392,  8.60671311])

In [8]: positions
Out[8]: array([1, 0, 2])

In [9]: index.get_cell_ids()
Out[9]: array([1386017682036854979, 2415595305706115691, 6525033740530229539],
              dtype=uint64)

Running the tests

Running the tests requires pytest (it is also available on conda-forge).

$ pytest .