5c0r/profanity-check-api

Just another assignment on a windy day

Introduction

This is just an assignment. But trying out new stuffs is cool ;)

Requirements

• .NET Core 3.1 SDK
• Docker to run Dockerfile locally
• Heroku CLI if needed to deploy from local machine
• Postman for testing request

Developing and running locally

• Go to src folder, run dotnet restore , dotnet build, dotnet run --project .\ProfanityCheck.WebAPI\ProfanityCheck.WebAPI.csproj to start the web project
• Using ngRok/anything to tunnel requests to your localhost ( if you are testing againts Salesforce client)
• Use the postman_collection.json , remember to re-attach text file in POST request

Deploying backend API ( to Heroku )

• Currently using GitHub Actions for deployment , use .github/workflows/main.yml as reference
• HEROKU_API_KEY is found from https://dashboard.heroku.com/account , and should be set as secret
• Any changes from master branch should invoke the Action

Deploying ( to Salesforce )

• All triggers and classes should be deployed , starting with classes then triggers

Configuration ( Salesforce )

• GlobalSettings.apcx needs to set to the Backend API URL
• Remote Site Settings to authorize Backend API
• Email attachment settings might need some tweak

What has been done

• Apex Triggers and Apex Queueable Job
• Backend API with list of banned words from Github and local file (Banned_Words.txt)
• Tests for Backend API
• Deploy-able Backend API , with Github Actions and Dockerfile
• Tests for Salesforce Client App ( ~70% )
• propose or even implement the logic to notify the user when s/he has attached files with disallowed content, => This can be done using either email (implemented) or Notification Bell
• propose or even implement the logic to process files with the size bigger than 12MB (you know Salesforce limits, don’t you?). => Sounds like Apex Governor Limits/ Heap size limits , I am thinking should it be about splitting up the file into multiple chunks (being sent with contentDocumentId), then chaining the queuable jobs ? File should be merged on the other end and to be processed

How to test things out

• Create a new Case
• Attach a text file that has/no banned words ( current supports csv,txt and log files)
• Upload a new version with banned words || Upload new file with banned words
• File should be pernamently deleted, even from Recycle Bin
• Owner should receive an email with violated file

What were challenging

• Figuring out what is the right Trigger to use ( ContentDocument,ContentVersion, ContentDocumentLink)
• Trigger.new does not show everything. ContentVersion.VersionData needs to be fetched again in Trigger. https://salesforce.stackexchange.com/questions/268436/trigger-on-content-document-to-retrieve-the-body
• Uploading files using Apex was a pain, maybe there was a better approach
• Testing things asynchronously, still figuring out how to test Trigger with Queueable..

What can be improved further

• Notification Bell to user instead of email
• Bulk handling of files , not one file per notification email
• Instead of removing files, maybe replacing banned words with ****