This is a C/Python library that wraps geometry functions in GEOS in numpy ufuncs. This project is still in a mock-up phase: the API will most likely change.
GEOS geometry objects are stored in a Python extension type pygeos.GEOSGeometry, that keeps the python interpreter out of the numpy ufunc inner loop. This object calls the GEOS destroy function to deallocate memory just before the wrapping GEOSGeometry object is deallocated.
Ufuncs only act on these pygeos.GEOSGeometry objects. Construct these as follows. This operation copies the underlying C object so that we can safely deallocate it once point is garbage collected:
>>> from pygeos import GEOSGeometry
>>> from shapely.geometry import Point
>>> pointer_to_geometry = Point(i, j)._geom
>>> geometry = GEOSGeometry(pointer_to_geometry)Or simply:
>>> from pygeos import points
>>> point = points(5.2, 52.1)Compare an grid of points with a polygon:
>>> geoms = points(*np.indices((4, 4)))
>>> polygon = box(0, 0, 2, 2)
>>> contains(polygon, geoms)
array([[False, False, False, False],
[False, True, False, False],
[False, False, False, False],
[False, False, False, False]])Compute the area of all possible intersections of two lists of polygons:
>>> from pygeos import box, area, intersection
>>> polygons_x = box(range(5), 0, range(10, 15), 10)
>>> polygons_y = box(0, range(5), 10, range(10, 15))
>>> area(intersection(polygons_x[:, np.newaxis], polygons_y[np.newaxis, :]))
array([[100., 90., 80., 70., 60.],
[ 90., 81., 72., 63., 54.],
[ 80., 72., 64., 56., 48.],
[ 70., 63., 56., 49., 42.],
[ 60., 54., 48., 42., 36.]])Pygeos uses shapely's installing scripts. If you have libgeos at a standard location, the following should work:
$ pip install pygeos
Clone the package:
$ git clone https://github.com/caspervdw/pygeos.git
Install it using pip:
$ pip install -e .
Run the unittests:
$ python -m pytest pygeos/test.py
- GEOS: http://trac.osgeo.org/geos
- Shapely: https://shapely.readthedocs.io/en/latest/
- Numpy ufuncs: https://docs.scipy.org/doc/numpy/reference/ufuncs.html
- Joris van den Bossche's blogpost: https://jorisvandenbossche.github.io/blog/2017/09/19/geopandas-cython/
- Matthew Rocklin's blogpost: http://matthewrocklin.com/blog/work/2017/09/21/accelerating-geopandas-1
Copyright (c) 2019, Casper van der Wel. BSD 3-Clause license.