jamesandariese / profanity

HTTP API to check text for profanity

Profanity API

Check text for profanity.

Usage

A sample test:

docker build -t profanity .
docker run --name profanity-api -p 57824:5000 -d --rm profanity
sleep 20 # it takes a bit to load the model!
curl -XPOST localhost:57824 -d $'this\ntext\nis\nclean'
docker rm -f profanity-api

To use the API, simply pass a body to it via POST. You will receive back a JSON blob containing the following:

key	type	description
overall	number	score of all text combined
lines	[number]	score of individual lines
average	number	average score of individual lines
max	number	max score of individual lines
max5	[string]	contents of the highest scoring 5 lines

Scores are integers from 0 to 100 representing the confidence that there is profanity present. Higher numbers mean more confidence that there is profanity.

Docker/OCI

This uses containers exclusively because the core of this API is a very fast but not terribly up-to-date scikit-learn model. In order to keep your own system secure, it's highly recommended that you do not install the prerequisites to get this running anywhere except in its own isolated world.