ngUML API Mapping Framework

Description

This project is made as a backbone for my master thesis for ICT in Business and the Public Sector and is currently still very much work in progress. The project aims to connect applications to each other through their APIs in a generic way. The main (research) purpose was to connect the ngUML project [1] to third-party low-code vendors in a generic way. But this tool can be used to make a connection between any pair applications!

Features

• Based on OpenAPI documents
• Full OpenAPI generation by simply calling your APIs
• API visualisation and connection in the easy to use the High Level Flow editor
• API connection recommendations powered by Machine Learning and Natural Language processing
• Full custom JSON visualisation and simple to make connections between JSON data element the Low Level Mapping Interface
• Immediately synchronise the configured API connections with the build in Sync Server
• No SDK needed, this is generated based on your OpenAPI document or the generated OpenAPI document
• Possibility to use custom python code with a provided boilerplate in the interface for complex connections

Screenshots

OpenAPI document generation by simply calling the API

OpenAPI document results

High Level Flow editor with all available APIs within each application

Low Level Flow editor showing the JSON data schema, the shape of the JSON that that API is going to respond or request

Synchronization server to automatically transport the data between the applications

Usage

The setup relies on docker and docker-compose If you do not already have that you can find a tutorial on how to install it here.

Then build the docker images by running: (depending on your way of installing docker and docker-compose you may need to add sudo in front of the commands)

docker compose build

This can take up to 10 minutes to finish. After building you start the whole socker stack by running:

docker compose up

When this is running without errors you can reach the frontend on:

http://localhost:3000

To Do

• Finish documentation and type hinting
• Try neural network as the provider for the ML model
• Add publication of thesis, when available, for added context
• Writing tests

References

[1] Ramackers, G. J., Griffioen, P. P., Schouten, M. B., & Chaudron, M. R. (2021, October). From prose to prototype: Synthesising executable uml models from natural language. In 2021 ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C) (pp. 380-389). IEEE.