Spelling Corrector

This repo contains two main tools:

• The command line interface, located in the cli/ directory. See cli/README.MD for more details.
• The web app:
  • For the back-end we have used fastapi and uvicorn to set up an API server using Python. This is located in the server/ directory.
  • For the front-end we have used Next.js with a TypeScript template. This is located in the web directory (or web_new in the develop branch). See web/README.md or web_new/README.md (in develop branch) for more details.

Command Line Interface

The command line tool allows you to use the Spelling Corrector on the terminal. See cli/README.MD for details.

Web App

There are currently two active branches in this repository: main and develop.

Using your own local pre-trained model

The main branch currently uses two API endpoints hosted on angocair.garg.ed.ac.uk to run the two pre-trained transformer models. You currently cannot run your own pre-trained transformer model on the web app in the main branch.

The develop branch has three models available through the web_new/ directory: the ones hosted in angocair.garg.ed.ac.uk plus the local pre-trained model. To make the local model accessible, you first need to deploy the API server locally using uvicorn. All relevant back-end infrastructure is located in the server/ directory:

• The local pre-trained model should be placed in server/models/.
• The script server/api/api.py loads the model and hosts the API.

First, make sure you have installed the necessary Python packages in your environment:

pip install -r requirements.txt

Now you can run the server in the background using:

nohup uvicorn server.main:app --reload --host 0.0.0.0 &

The above command deploys the local model and makes it accessible to the web app through the address http://localhost:8000. You can open http://localhost:8000/docs in your browser to view the API user interface.

Note that you will need to do the following in order to kill the process running in port 8000, once you are done with it. In your terminal, run:

lsof -i:8000

Copy the PID of the process and then run:

kill -9 PID

Run the Web App locally

You can access the operational web app on http://angocair.garg.ed.ac.uk/. However, if you want to run the web app locally (e.g. for further development), you can do so using the yarn package manager. Prerequisites: install Node.js and Yarn - see https://classic.yarnpkg.com/lang/en/docs/install/.

First navigate to the web/ or web_new/ (in the develop branch) directories and install all dependencies listed within package.json:

cd web
yarn install

To run the development server:

yarn dev

Open http://localhost:3000 with your browser to see the result.

To create a production-ready build of the application and export it as a set of static HTML files:

yarn build && yarn export

The output files will be written to out/.

Check web_new/README.MD (in the develop branch) for more details.

Setup Environment using caddy

TODO: this section needs reviewing.

To allow people to access publically, we need to setup caddy or other web hosting server.

# download and install caddy
curl -O caddy https://caddyserver.com/api/download?os=linux&arch=arm64&p=github.com%2Fcaddy-dns%2Fcloudflare&idempotency=91119776165967
sudo apt install -y debian-keyring debian-archive-keyring apt-transport-https
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/gpg.key' | sudo gpg --dearmor -o /usr/share/keyrings/caddy-stable-archive-keyring.gpg
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/debian.deb.txt' | sudo tee /etc/apt/sources.list.d/caddy-stable.list
apt update
apt install caddy
systemctl status caddy

# change Caddyfile
nvim /etc/caddy/Caddyfile
systemctl reload caddy
systemctl status caddy

# copy the html
cp -r ./web/out/. /usr/share/caddy/

TODO

• API
• UI
• Feature1: suggestion
• Feature2: sentence
• Make UI prettier
• Diffchecker