cellmap_flow.io
Reading datasets: what cellmap-flow adds over worldarray.
The reading is worldarray’s: wa.open and wa.open_pyramid open arrays
and pyramids by path (zarr v2 and v3, N5, precomputed, on disk or at a URL),
.meta and wa.metadata say what an array is and where it sits,
list_chunk_keys lists the chunks that exist, read_index and read
read voxel and world boxes. These modules keep what cellmap-flow needs on top:
metadata opens a path the way every reader does (open_level) and
gives nanometers per unit; multiscale names the raw level a model reads
(closest_raw_scale); paths undoes a shell-escaped space; raw opens
the level a model reads and reads it through the input chain (open_raw,
RawInput). Writing is worldarray’s too (wa.create, write,
write_index, write_chunk), at the writers themselves.
The submodules import nothing from the rest of cellmap_flow – raw
excepted, which reads the process’s chain (cellmap_flow.process_chain)
– and in particular not cellmap_flow.globals, so they can be used
without configuring logging, loading a Flow or importing
flask/neuroglancer/torch; importing worldarray brings numpy and tensorstore.
Import the submodule you need; this package imports none of them itself.
Submodules
Attributes
Package Contents
- cellmap_flow.io.CHANNEL_AXIS_NAMES = ('c', 'c^', 'channel')