geozarr-toolkit 0.1.2

GeoZarr convention models, validation, and creation utilities

geozarr-toolkit

Python library for creating and validating GeoZarr-compliant metadata. Provides Pydantic models, helper functions, and a CLI for the modular Zarr conventions:

• spatial: -- Coordinate transforms between array indices and spatial coordinates
• proj: -- Coordinate Reference System (CRS) via EPSG codes, WKT2, or PROJJSON
• multiscales -- Pyramid structures and resolution levels

Installation

pip install "geozarr-toolkit"

Quick start

Create complete GeoZarr-compliant attributes in one call:

from geozarr_toolkit import create_geozarr_attrs

attrs = create_geozarr_attrs(
    dimensions=["Y", "X"],
    crs="EPSG:32633",
    transform=[10.0, 0.0, 500000.0, 0.0, -10.0, 5000000.0],
    bbox=[500000.0, 4900000.0, 600000.0, 5000000.0],
    shape=[10000, 10000],
)

Or build conventions individually:

from geozarr_toolkit import (
    create_spatial_attrs,
    create_proj_attrs,
    create_zarr_conventions,
    SpatialConventionMetadata,
    ProjConventionMetadata,
)

# Spatial convention
spatial_attrs = create_spatial_attrs(
    dimensions=["Y", "X"],
    transform=[10.0, 0.0, 500000.0, 0.0, -10.0, 5000000.0],
)

# Proj convention
proj_attrs = create_proj_attrs(code="EPSG:32633")

# Combine with zarr_conventions array
attrs = {**spatial_attrs, **proj_attrs}
attrs["zarr_conventions"] = create_zarr_conventions(
    SpatialConventionMetadata(),
    ProjConventionMetadata(),
)

Validation

Validate attributes or Zarr stores against conventions:

from geozarr_toolkit import validate_spatial, validate_proj, detect_conventions

# Validate individual conventions
is_valid, errors = validate_spatial({"spatial:dimensions": ["Y", "X"]})

# Auto-detect and validate all conventions in a dict
from geozarr_toolkit import validate_attrs
results = validate_attrs(attrs)

Working with existing data

Extract convention metadata from rioxarray or GDAL GeoTransforms:

from geozarr_toolkit import from_rioxarray, from_geotransform

# From rioxarray DataArray
attrs = from_rioxarray(da)

# From GDAL GeoTransform
attrs = from_geotransform(
    geotransform=(500000.0, 10.0, 0.0, 5000000.0, 0.0, -10.0),
    crs_wkt='PROJCS["UTM zone 33N"]',
    shape=(10000, 10000),
)

Pydantic models

Use the models directly for type-safe construction and validation:

from geozarr_toolkit import Spatial, Proj

spatial = Spatial(**{"spatial:dimensions": ["Y", "X"]})
proj = Proj(**{"proj:code": "EPSG:4326"})

# Serialize with convention-prefixed keys
attrs = spatial.model_dump(by_alias=True, exclude_none=True)

CLI

# Validate a Zarr store
geozarr validate data.zarr
geozarr validate data.zarr --conventions spatial proj

# Inspect a Zarr store
geozarr info data.zarr
geozarr info data.zarr --json

metazarr

metazarr/ is a client-side JavaScript library and web app for exploring Zarr store hierarchies and validating GeoZarr convention compliance. It runs entirely in the browser with no server required. Deployed at inspect.geozarr.org.

Features:

• Opens remote Zarr v2 and v3 stores (consolidated metadata, directory crawling, or manual path entry)
• Displays full array metadata: shape, dtype, chunks, fill value, codecs/compressor, dimension names, chunk key encoding, memory order, and computed statistics (uncompressed size, chunk count, chunk size)
• Detects sharding (v3 sharding_indexed codec)
• Auto-detects GeoZarr conventions (spatial:, proj:, multiscales, CF) via zarr_conventions or attribute prefixes
• Validates nodes against convention JSON Schemas (Draft-07 and Draft 2020-12)

cd metazarr
npm install
npm run dev    # Start dev server at http://localhost:5173
npm test       # Run test suite

Also usable as an ES module:

import { openStore, buildTree, detectConventions, validateNode, buildNodeDocument } from "metazarr";

const result = await openStore("https://example.com/data.zarr");
const tree = buildTreeFromV3(result.v3Entries);
const conventions = detectConventions(tree.attrs);

Development

git clone https://github.com/zarr-developers/geozarr-toolkit.git
cd geozarr-toolkit
uv sync
uv run pytest

License

MIT