PolyStore

Framework-agnostic storage primitives for scientific applications.

PolyStore owns backend interfaces, execution-local backend registries, FileManager routing, storage formats, ROI values, virtual-workspace source references, and viewer-streaming payload mechanics. Applications retain ownership of domain artifact names and materialization policy.

Quick start

Pass an explicit mapping of backend names to instances. FileManager has no global fallback and does not infer a backend from a path.

from pathlib import Path
import numpy as np

from polystore import DiskBackend, FileManager, MemoryBackend

registry = {
    "disk": DiskBackend(),
    "memory": MemoryBackend(),
}
files = FileManager(registry)

data = np.arange(6).reshape(2, 3)
files.save(data, Path("output.npy"), backend="disk")
loaded = files.load(Path("output.npy"), backend="disk")

For application startup where every discoverable context-free backend is wanted, use the package-owned lazy registry:

from polystore import FileManager, ensure_storage_registry, storage_registry

ensure_storage_registry()
files = FileManager(dict(storage_registry))

Backends requiring context-specific construction, including virtual workspaces and OMERO, must be instantiated by the application and added to its registry.

Nominal backend extension

Concrete backends inherit DataSource, DataSink, StorageBackend, or ReadOnlyBackend and declare their backend key. The authoritative class catalog is BackendBase.__registry__ (also exported as STORAGE_BACKENDS). Do not build a second backend class table.

Installation

python -m pip install polystore

The core install includes NumPy, ArrayBridge, metaclass-registry, file locking, platform-aware cache paths, image I/O, Zarr/OME-Zarr support, and ZMQRuntime. Framework and streaming extras add their corresponding optional runtimes. Install polystore[bioformats] for the package-owned PyImageJ and Bio-Formats bridge. Its first use materializes a checksummed Fiji, JDK, and PyImageJ bridge runtime; later processes reuse that complete local cache without Maven resolution.

Full documentation: polystore.readthedocs.io

Metadata

Release files for polystore 0.6.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for polystore 0.6.2
File	Size	Uploaded
polystore-0.6.2.tar.gz	196.9 kB	Oct 12, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for polystore 0.6.2
File	Interpreter	ABI	Platform	Reset
polystore-0.6.2-py3-none-any.whl 158.3 kB Oct 12, 2026	Python 3	none	any	Details

Total release size: 355.2 kB

Release files / polystore-0.6.2.tar.gz

Download URL	polystore-0.6.2.tar.gz
Size	196.9 kB
Tags	Source
SHA-256 checksum How to use checksums	`1a9d91461f1957fb4cea7c5ef374e8a7de0b84bf8b54b4d7e1ad2c5343a72257`
BLAKE2b-256 checksum How to use checksums	`44c60aa0ec6839e14b5723b76a6d7d229ac41731a5e0419901fa6df4d1cf9fb3`
Upload date	Oct 12, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.12.14`

Release files / polystore-0.6.2-py3-none-any.whl

Download URL	polystore-0.6.2-py3-none-any.whl
Size	158.3 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`acb363c5b77348f8d700183cd401ae25705f1f39ce3bbbdd8b44c970e80993d6`
BLAKE2b-256 checksum How to use checksums	`3dea528b62b9a4967d49ce9bf18c2aaaa8f6f2b4806062406e24260e71d3a321`
Upload date	Oct 12, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.12.14`