tanyaleepr / anime_crime-watchers

Hanzai Watchers is a crime alert + game MVP that provides safety to our society. We used MERN stack, Tailwind CSS, FBI and Random User API, React-Bootstrap, Mongoose, and Filmora X

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The Hanzai Project

By Tanya, Ashley, Don-EL, and Paul

The Mission

The Hanzai Project, is a Full Stack MERN application that we decided on for our final project to display what we've learned throughout the bootcamp:

The purpose of this application is to report and alert others to crime in your area in hopes of providing awareness in your community. We would like our society to be safer and with the more crime tips reported the more points you earn. With enough points you'll become your community's #1 Samurai!

Home Page

Home Page ScreenShot

Weapons Used

  • FrontEnd: React.JS, Redux Library, Bootstrap, HTML/CSS
  • Backend: Node.JS, Express.JS
  • Database: MongoDB, Mongoose

Installation

You will need node and npm to install and run this project.

  1. git clone https://github.com/DonL44/anime_crime-watchers.git cde && cd cde
  2. npm install

Running the app locally

The application expects a few environment variables to interact with the API:

  • CDE_API - this should be the URL for the API. To use the public API, set this to https://api.usa.gov/crime/fbi/sapi
  • API_KEY - this should match the key used by the API. If you are using the public API, sign up for an API key at https://api.data.gov/signup/

You can copy the env.sample file (cp env.sample .env), fill in your own values, and then make sure to source .env before running the build process.

Use npm run watch to start the continuous webpack processes and a webserver.

Running tests

You can lint the code with npm run lint and run tests with npm run test. RunTest

About

Hanzai Watchers is a crime alert + game MVP that provides safety to our society. We used MERN stack, Tailwind CSS, FBI and Random User API, React-Bootstrap, Mongoose, and Filmora X


Languages

Language:JavaScript 72.3%Language:CSS 25.1%Language:HTML 2.5%