In ExpressJS, you always have to start with scratch for things like CSRF / JWT Token, Exception Handling, Clusters, Queues, Cache, etc. So, we thought to make is simple for everyone by building a boilerplate using Typescript to start off with.
The repository is available at GeekyAnt's GitHub under the name Express-Typescript.
Why we need this Boilerplate?
1. Exception Handling
Rather than restarting the entire node app repeatedly for minor bugs, we found a way to handle exceptions & errors using Nodes' process events.
2. Fork a New Cluster on Exit of Old Cluster
Rather than sending the entire load onto a single Core / CPU, we try to take advantage of a multi-core system to handle load. Here, if the cluster dies for some reason, we create / fork a new one immediately.
3. Auth (using Email OR Social Accounts)
Using PassportJS as an interface, we keep adding configurations & registering them in the app for social authentication.
A Single Log class with methods like info, warn, error & custom. These methods send the log string to a file under <YYYY-MM-DD>.log directory of your project root and keep creating these log files on a daily basis rather than having a single log file.
Use of .env, to load environment variables/constants for the entire app. So now, we have a single centralized constants location.
We prefer using CSRF token for our Web routes and JWT token for API routes. So, we create middlewares for both tokens and register them against Web and API routes respectively.
There was a bit of a discussion on the issue of structuring of folders/files in ExpressJS. It was hard to establish where to organize and which files to organize, or if a perfect structure even existed and was possible to achieve.
At last, I came up with the following folder structure.
Note: This structure contains just parent folders.
There are various things in a backend logic that does not really need to happen upon a request like sending out E-Mails or SMS texts, etc. In short, things which require a connection to any third party service are overhead in some conditions
To fix this, I have used KueJS, a priority based job queue that runs on a Redis driver.
For caching of web routes, I have used a simple in-memory cache using memory-cache.