Click Here to Visit the Project Planning Google Document
Click Here to Visit the Website tldr.fly.dev
- From this project we gained a better understanding of the flask log library. The library is much more powerful than we both used in project 1 part 2 as we discovered there are multiple ways to approach logging in. It was interesting because the
/route logs the user out and redirects to the main page. This was an interesting fix we had to come up with as for some reason even if the server was restarted the previous user was still logged in and it would create certain bugs. This logout on/fixed these issues. - Something else we learned is HuggingFace. HuggingFace is actually the first of its kind and it is a company dedicated to OpenSource-AI. This allows for some cool projects like the one we have presented. In our search for AI APIs HuggingFace was the only legitmate solution we found.
- From this project Ali and I (Saad) both were not looking forward to the fly.io deployment. This is just because of the issues it has caused us in the past. However, even though we were not looking forward to it deployment was easier than anticpated which was nice.
- From this project Ali and I both would have liked to see Javascript as a requirment. Of course we had to freedom to implement it but due to classes and other responsibilites we made due without it. Having it as a requriment would have forced us to learn it which is a good thing. I (Saad) believe adding any sort of extra techical requriment would have been best as this project was essentially the same as the last.
- App Runs on Flask Server Written in Python
- Postgres Database
- REST API
- User Login
- Beautification
- VSCode
- GitHub
- WSL
- Flask
- Fly
- json
- os
- requests
- flask
- random
- dotenv
- Hugging Face Bart Large CNN API
Beging by cloning the repository in WSL the installing the required dependencies by running:
python -m pip install -r requirements.txtCreate a .env file in the same directory as TLDR.py and create three variables secret_key, ENDPOINT, DATABASE_URI
Set secret_key to any value; this will be used for Flask to be able to perform Flashes
Next create an account for (Hugging Face)[https://huggingface.co/] then navigate to settings -> (Access Tokens)[https://huggingface.co/settings/tokens] to create an API key. Finall set ENDPOINT to the given API key.
Next create a postgres cluster following the directions on this github repo. In the process of creation you will get a Database URI link. Set the DATABASE_URI variable to the given link.
Naviate to fly.io and create an account. Then in WSL run the following commands:
curl -L https://fly.io/install.sh | sh
flyctl auth loginNote: You may need to export to Fly to path after running the curl command
Once logged into fly on WSL run flyctl launch. When prompted for a name that will be name of your website. After the questions Procfile and fly.toml will be created. Edit Procfile to say the following:
# Modify this Procfile to fit your needs
web: gunicorn web_server:app
Then we are ready to deploy by running the following command.
flyctl deploy
We can then check status by doing flyctl status which should look something like this:
App
Name = tldr
Owner = personal
Version = 1
Status = running
Hostname = tldr.fly.dev
Platform = nomad
Deployment Status
ID = a283e4c6-f656-a445-f103-8f88fcd6187b
Version = v1
Status = successful
Description = Deployment completed successfully
Instances = 1 desired, 1 placed, 1 healthy, 0 unhealthy
...
We are concerned with the Hostname which is the link to the website that is now deployed, tldr.fly.dev. We are also considered with Status and Description which should say successful.