geosplit 0.4.0

Split GeoJSON by feature count or file size and convert GeoJSON to and from GeoPackage

GeoSplit

A focused command-line tool that:

• splits a GeoJSON FeatureCollection by feature count;
• splits it by a strict maximum file size; and
• converts GeoJSON to GeoPackage, or a GeoPackage layer back to GeoJSON.

The splitter uses a lightweight streaming parser, preserves coordinate precision, and bounds memory use to the current output chunk. GeoPackage support is optional.

View general or command-specific help at any time:

geosplit help
geosplit help split
geosplit help convert

Install

Requires Python 3.10 or newer. Install GeoSplit from PyPI:

python -m pip install geosplit

Include conversion support:

python -m pip install "geosplit[gpkg]"

Confirm the installation:

geosplit help

If the geosplit command is not on your system path, use python -m geosplit help and replace geosplit with python -m geosplit in the examples below.

Split GeoJSON

By feature count:

geosplit split world.geojson output --features 1000

Omit the output directory to create world_split . Specifying Output directory is optional:

geosplit split world.geojson --features 1000

By maximum file size:

geosplit split world.geojson output --size 10MB

Preview every output file, feature count, size, warning, and conflict without writing anything:

geosplit split world.geojson output --features 1000 --dry-run

Sizes accept B, KB, KiB, MB, MiB, GB, and GiB. Each resulting file is a complete, compact GeoJSON document whose on-disk size does not exceed the requested limit. If one feature cannot fit by itself, the command stops with a clear error.

Output files are numbered automatically, such as world_001.geojson, world_002.geojson, and so on. Use --prefix countries to customize their names or --force to replace files recorded in GeoSplit's hidden output manifest. Progress is written to stderr; use --quiet to suppress it. Interrupted transactions are recovered on the next run. A source collection's top-level bbox is omitted because it would no longer describe each split collection; other top-level metadata is retained.

Python API

Stream validated collections without writing files:

from geosplit import iter_batches

for collection in iter_batches("world.geojson", features=1000):
    process(collection)

Plan a split, or inspect the result of a completed split:

from geosplit import plan_split, split_geojson

plan = plan_split("world.geojson", features_per_file=1000)
result = split_geojson("world.geojson", features_per_file=1000)

print(plan.files)
print(result.files)
print(result.feature_count)
print(result.total_bytes)

Show the installed version with geosplit --version.

Convert formats

# GeoJSON to GeoPackage
geosplit convert roads.geojson roads.gpkg

# Choose the new GeoPackage layer name
geosplit convert roads.geojson map.gpkg --output-layer roads

# GeoPackage to GeoJSON
geosplit convert map.gpkg roads.geojson --layer roads

If a GeoPackage contains exactly one layer, --layer is optional. Existing destinations are protected unless --force is supplied.

Update

python -m pip install --upgrade geosplit

License

MIT