A project to see whether a graph database such as Neo4J can be used to power a forum-like feature with branching replies.
This project was inspired by some product conversations I had regarding a 'forum' feature. There are two popular product models in vogue:
- the 'traditional' forum, with threads containing a set of linear, time-ordered replies
- the 'branching' forum, with threads containing replies that can have replies, up to an (in principle) unlimited level of nesting
Model (1) is perfectly good for most products (including the one my colleagues ended up building), but plenty of social media sites (e.g. Reddit, Twitter) have apparently-arbitrary nesting of replies/forking of conversations - model (2). I thought it would be interesting to follow this route.
As a 'data' person, my first thought was 'what datastore should I use for this?' To me, the branching conversations make me think immediately of graph databases. One very well-known graph database is Neo4J, so that's what I decided to use.
Along the way I got acquainted with neomodel, an Object Graph Mapper.
I've added an MIT license to the project, so feel free to make use of it under those terms. But please note that this was a learning project; I threw in some boilerplate infrastructure stuff, but haven't created a test suite ... and (since I don't plan to deploy this anywhere) I haven't added authentication to any of the routes. If it were deployed as-is, anyone would be able to call any of the routes!
Let's start by preparing the python environment and the graph database:
pipenv run setupdocker compose up graphdb
This second command just brings up the Neo4J container. Data are persisted in volumes across restarts (unless you do docker compose down --volumes).
The first time you start the database, the user neo4j will be set up with a password defined via an environment variable $NEO4J_AUTH=neo4j/<your password>. However, a new password should be subsequently set and stored in the .env file as the environment variable $DBPASS.
This is best done by logging into the Neo4J browser at http://localhost:7474/browser/ and running ALTER CURRENT USER SET PASSWORD FROM '<your password>' TO '<your new password>'.
Next (i.e. before you add any data) you want to apply any database constraints defined in the src/models.py file. For convenience, a script has been defined in the Pipfile. pipenv run constraints will apply the constraints defined in the file.
To bring up the database and API, use docker compose up. At the moment the API is configured to be available at http://localhost:8765/.