Installation¶
zarr-vectors requires Python 3.10 or later and depends on zarr>=3.0,
numpy>=1.24, and numcodecs. All mandatory dependencies are installed
automatically by pip.
Standard install¶
pip install zarr-vectors
This installs the core read/write/validate API for all seven geometry types. It does not include format converters, cloud-store drivers, or Draco mesh compression — those are available as optional extras described below.
Optional extras¶
Ingest and export formats¶
Format converters and the zarr-vectors CLI live in the companion
package zarr-vectors-tools — install that separately if you need
to read/write LAS, PLY, CSV/XYZ, TRK, TCK, TRX, SWC, GraphML, OBJ, or
STL files.
Draco mesh compression¶
pip install "zarr-vectors[draco]"
Enables Google Draco encoding and decoding for the mesh geometry type.
Requires DracoPy. Draco-compressed stores are not readable without this
extra installed.
Cloud object-store backends¶
pip install "zarr-vectors[cloud]"
Enables reading from and writing to Amazon S3, Google Cloud Storage, and
Azure Blob Storage via s3fs and gcsfs. See
Cloud stores for configuration details.
Everything¶
pip install "zarr-vectors[all]"
Installs all optional extras in a single command.
Development install¶
To install from source with all development dependencies:
git clone https://github.com/BRIDGE-Neuroscience/zarr-vectors-py.git
cd zarr-vectors-py
pip install -e ".[all]"
pip install -r docs/requirements-docs.txt # if building the docs locally
Run the test suite to verify the install:
pytest tests/ -v
All tests should pass. The suite does not require network access or external data files.
Verifying the install¶
import zarr_vectors
print(zarr_vectors.__version__)
# Confirm geometry type constants are available
from zarr_vectors.constants import GEOM_POINT_CLOUD, GEOM_STREAMLINE
print(GEOM_POINT_CLOUD, GEOM_STREAMLINE)
Dependency notes¶
zarr-vectors targets Zarr v3 exclusively. Zarr v2 stores are not
supported and cannot be opened with this package. If you have an existing
v2 workflow, migrate the store using zarr’s built-in conversion utilities
before ingesting into the ZVF format.
NumPy 2.x is supported from zarr-vectors 0.2 onward. Earlier releases
require NumPy 1.x.