Installation

APbase's mapping pipeline is distributed with compiled numerical extensions. Install the integration for the language and runtime you use.

For production deployments, install APbase in the runtime used by your API service, worker, MCP server, agent backend, or batch container. See Production and cloud deployment for cloud and worker sizing guidance.

Requirements

  • Python 3.12+
  • numpy, pyproj, shapely (installed automatically as dependencies)

Install

pip install apbase

Note

If apbase is available from your package index, the command above is enough. Until it is published to a public index, install from a local distribution or the project repository.

pip install /path/to/apbase-python

pip install git+https://github.com/ap-base/apbase-python.git

Optional extras

Some Python helpers accept GeoDataFrame or GeoSeries inputs when geopandas is installed. You can skip this extra when passing plain NumPy arrays:

pip install geopandas

Verify the install

import apbase

apbase.config["n_threads"] = 4
print(apbase.Map, apbase.create_map)

Using it in production

The package already includes its compiled native extensions. Install it in the production runtime with the package manager; the application image does not need a Fortran compiler or a separate build stage.

Configure CPU parallelism once when the process starts:

export APBASE_N_THREADS=2

or in Python before handling work:

import apbase

apbase.config["n_threads"] = 2

Keep workers * APBASE_N_THREADS at or below the vCPU allocated to the service to avoid CPU oversubscription.