This application utilizes a machine learning model developed by Group 9 in ECS 171 Fall 2021 to take a book description/summary as input and return the predicted genre(s) of the book.
https://book-genre-predictor.wl.r.appspot.com/
Application utiliizes Angular for UI/Frontend element and Python for Backend and Flask for the webserver. Application is configured so that the Angular Frontend is served via Flask to keep the project running as a single service due to restrictions with how Heroku, the application host service, encapsulates and isolates their Dynos which are what they call their processes. Otherwise a flask server for the backend could talk to an express server for the frontend. This approach was taken before we ran up against memory limits on Heroku which forced us to look elsewhere for deployment eventually landing on Google App Engine. Effort could be made to split the front and back end into separate microservices since it is easily supported on the Project level on GAE.
This project was generated with Angular CLI version 12.2.11.
Instructions for Linux:
In the terminal go to the project directory, I recommend setting up a virtual environment to keep the package dependencies controlled, you will need to have nodejs and npm installed in order to handle the packages. At that point, you should be able to run npm install to download package dependecies that are outlined in packages.json.
Once packages are installed, run ng build --configuration=production in order to build the application. After that, run gunicorn wsgi:app to host the application locally. Depending on how you port bind gunicorn it should be available at localhost:8000 in your web-browser.
The trained NN model had to be stored in AWS S3 bucket due to its sheer size. As a result we have to use awscli when running locally and have console aws credentials set via aws configure in order for the code to be able to download the model when it starts up. There is an AWS IAM User, genre-prediction-app, that has a key setup whose credentials are being used. These credentials also have to be set on Heroku as well so it can access the S3 bucket files when the app is deployed. Docs here https://devcenter.heroku.com/articles/s3
You also want to re-build and restart the webserver, in that order, if you want to see any changes you've made to the application appear in the web-browser.
Not sure if this will work the same for MacOS, in that I am not sure if npm will work the same way to download dependecies, once you have the appropriate packages it should be the same though. If you are on Windows I recommend using WSL/WSL2 to run this, which is what I use for all this anyhow.
For application debug gunicorn wsgi:app --capture-output --log-level debug --timeout 90 to run the app will output gunicorn logs as well as the debug logs in the app itself.
The application is hosted on Google App Engine. The way GAE works for a single service is that it only allows one environment setup which we need geared toward the Flask/Python backend. Since the front end is made in a nodejs environment we can just build locally and then make sure the static files that are generated during build in the dist/ directory are included in the deployment upload if UI changes have occurred.
Using the gcloud command line tool, you can deploy from a terminal in the project directory with gcloud app deploy, as mentioned above, if UI changes are made then you need to rebuild the static files first with ng build --configuration=production before deploying.
The deployment configuration for GAE is laid out in app.yaml. The current instance type we are using is F4_1G since it has a 2048 MB memory limit so we don't encounter memory issues with a large Neural Network model in memory. More details on instance types can be found here: https://cloud.google.com/appengine/docs/standard#second-gen-runtimes
Run ng generate component component-name to generate a new component. You can also use ng generate directive|pipe|service|class|guard|interface|enum|module.
Run ng build to build the project. The build artifacts will be stored in the dist/ directory.
To get more help on the Angular CLI use ng help or go check out the Angular CLI Overview and Command Reference page.