Installation

bagkit is a single executable with SQLite compiled in, so it needs no database server or system libraries.

Install script

On macOS and Linux:

curl -fsSL https://dstotijn.github.io/bagkit/install.sh | sh

The script downloads the latest release for your platform, verifies its SHA-256 checksum and installs it to ~/.local/bin. Set BAGKIT_VERSION to pin a release, such as v0.1.0, and BAGKIT_INSTALL_DIR to install elsewhere:

curl -fsSL https://dstotijn.github.io/bagkit/install.sh | BAGKIT_VERSION=v0.1.0 BAGKIT_INSTALL_DIR=/opt/bin sh

Release binaries cover macOS and Linux on arm64 and x86_64, and Windows on x86_64; the Linux binaries are statically linked. On Windows, download bagkit-x86_64-pc-windows-msvc.zip from the releases page and put bagkit.exe on your PATH.

Cargo

With Rust 1.88 or later:

cargo install --locked bagkit

From source

The repository pins Rust, Python and Node with mise:

git clone https://github.com/dstotijn/bagkit
cd bagkit
mise trust && mise install
mise exec -- cargo build --release
./target/release/bagkit --help

mise run check runs formatting, lints and tests.

Library

Add bagkit to a Rust project. Without default features it reads extracts and writes CSV, with no SQLite, Parquet or HTTP dependencies:

Cargo.toml

[dependencies]
bagkit = { version = "0.1", default-features = false }

See Using the Rust library.

Crate features

Feature	Adds	Default
sqlite	SQLite output, lookups and the spatial index	yes
parquet	Parquet and GeoParquet output	yes
server	The HTTP lookup server	yes
cli	The bagkit command	yes

Requirements

• Memory: a full national conversion peaks at about 1 GiB.
• Disk: room for the 3.6 GB extract, the output (up to 6 GiB for SQLite), and a second copy of the output while it replaces an existing one. SQLite sorts spill to temporary files, and so do ZIPs nested more than one level deep that are larger than 64 MiB; the national extract has none.
• CPU: every core is used by default; --threads limits that.