geozl

Python bindings to libgeozl, the geospatial codec extensions for OpenZL.

import numpy as np
import geozl

tile = np.fromfile("b04.raw", dtype=np.uint16).reshape(1024, 1024)

g = geozl.graph(tile, "planar>zigzag>transpose>entropy")
frame = geozl.compress(tile, graph=g)
back = geozl.decompress(frame).view(np.uint16).reshape(1024, 1024)

Pass a positive number as error for a fixed absolute bound, or a percentage string for a relative bound. None and zero are lossless. Full LINEAR, LOG and SQRT recipes remain available.

absolute = geozl.graph(tile, "planar>zigzag>entropy", error=2)
relative = geozl.graph(tile, "planar>zigzag>entropy", error="1%")

profile returns a list-like result. Print it to see the input geometry, benchmark settings and ranked graphs together.

rows = geozl.profile(tile, prior=None)
print(rows)
best = rows[0]["graph"]

Fourteen documented codecs, lossless and lossy, reachable two ways. geozl.graph takes a recipe string and builds the graph for you, and geozl.compress runs one tile through it. geozl.lossless and geozl.lossy hand you the individual nodes to place in an OpenZL graph yourself, and geozl.register_decoders teaches a DCtx to read what they wrote.

Wheels carry a prebuilt native library. A source install has to build it first, see the build instructions in the repository.

The full documentation, the codec catalogue and the wire format live in the repository.