Skip to content

Installation

GraFlo is a Python package. This page installs it, with or without the extras for development, documentation and diagrams, and tells you where to get a database to load into.

What you need

  • Python 3.11 or newer.
  • A graph database to load into, unless you use the file backend, which writes the graph to a directory. See Get a database below.

Install the package

With pip:

pip install graflo

With uv, into a project of yours:

uv add graflo

From a clone of the repository, which also gives you the examples and the database containers:

git clone https://github.com/growgraph/graflo.git
cd graflo
uv sync

The default install includes the clients for every supported database, RDF and SPARQL support, and the Kafka client.

Optional extras

The extras add tooling only; they do not switch ingestion features on or off.

Extra What it adds
dev Tests and checks: pytest and its plugins, hypothesis, ty, pre-commit
docs Building this site: ProperDocs and its plugins
plot Drawing the diagrams of graflo plot-manifest and the --plot figures of graflo merge and graflo merge3, as SVG, PDF or PNG

With pip, name the extras you want:

pip install "graflo[plot]"
pip install "graflo[dev,docs,plot]"

From a clone, name every extra you want in one command, because uv sync removes the extras you leave out:

uv sync --extra dev --extra docs --extra plot

The plot extra needs no system packages: Graphviz runs as WebAssembly inside Python, and PDF and PNG text uses fonts shipped with graflo, so it works the same on a desktop and in a slim container image. Those fonts cover Latin, Greek and Cyrillic; other scripts show in SVG, which leaves text to the viewer's fonts. On Alpine, install libgcc (apk add libgcc) for the WebAssembly runtime. Drawing a manifest explains the figures.

Check the installation

graflo --version

graflo, version 1.15.1

The version you see is the one you installed. In a clone, run it as uv run graflo --version.

Get a database

GraFlo writes to ArangoDB, Neo4j, TigerGraph, FalkorDB, Memgraph, NebulaGraph and PostgreSQL. A clone of the repository has a Docker Compose setup for each of them, plus Apache Fuseki and Kafka as sources to read from and SeaweedFS as object storage. docker/README.md explains how to start one or all of them; the examples read their connection settings from those containers.

If you do not want to run a database yet, use the file backend: it writes the graph to a directory, and you can load it into a database later. The file backend example (14) shows how.