grapes 0.10.4

Helper for dataflow based programming

grapes

A simple library for dataflow programming in python. It allows you to organize your computation as a directed acyclic graph.

Quickstart

For a more detailed explanation, refer to the documentation. Here's what you need to know to get started quickly.

Install grapes from PyPI with

pip install grapes

Define a graph by adding steps and binding functions to them:

import grapes as gr
g = gr.Graph()
gr.add_step(g, "b", "compute_b", "a")
gr.add_step(g, "c", "compute_c", "b")
gr.update_internal_context(g,
    {
        "compute_b": lambda a: 2*a,
        "compute_c": lambda b: b+1
    }
)
gr.finalize_definition(g)

Execute the graph to find your target, starting from your input context.

context = {"a": 3}
target = "c"
result = gr.execute_graph_from_context(g, context, target)
print(result["c"])  # 7

Additional remarks

The bulk of grapes development was done by Giulio Foletto in his spare time. See LICENSE.txt and NOTICE.txt for details on how grapes is distributed.

grapes is inspired by pythonflow but with substantial modifications.

It relies internally on networkx for graph management and on pygraphviz for graph visualization.

Most of the development of grapes was done before AI coding tools were available. However, recent edits (September 2025) were assisted by GitHub Copilot, especially for writing docstrings.