fix(geoarrow-pyarrow): Ensure/prefer ISO WKB where possible

Adds a strict_iso_wkb option to as_wkb()
Exposes this option in io.write_geoparquet()
Ensures geopandas arrays are converted directly to ISO WKB

On my computer this is about 20ms per million points (to verify that all items in an ISO WKB array are actually ISO WKB encoded).

import pyarrow as pa
import geoarrow.pyarrow as ga
import numpy as np

n = int(1e6)
xs = np.random.random(n)
ys = np.random.random(n)
zs = np.random.random(n)
points = ga.point().with_dimensions(ga.Dimensions.XYZ).from_geobuffers(None, xs, ys, zs)
gp = ga.to_geopandas(points)
ewkb = pa.array(gp.to_wkb())
not_ewkb = ga.as_wkb(ewkb, strict_iso_wkb=True)

# Best case detection
%timeit ga._compute._any_ewkb(ewkb)
#> 35.9 µs ± 215 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
# Worst case detection (can probably optimize better in geoarrow-c)
%timeit ga._compute._any_ewkb(not_ewkb)
#> 20.1 ms ± 91.7 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

# Comes out about even to fix it
%timeit ga.as_wkb(not_ewkb, strict_iso_wkb=True)
#> 20.1 ms ± 41.3 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)
%timeit ga.as_wkb(ewkb, strict_iso_wkb=True)
#> 23.8 ms ± 116 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

geoarrow / geoarrow-python

fix(geoarrow-pyarrow): Ensure/prefer ISO WKB where possible #39

Codecov Report