Modern Presidential Tweeting

See this project live here: https://www.preztweeting.com

This is a project attempting to define what "modern day presidential" tweets are, as referred to by the USA's 45th president:

My use of social media is not Presidential - it’s MODERN DAY PRESIDENTIAL. Make America Great Again!

— Donald J. Trump (@realDonaldTrump) July 1, 2017

Prerequisites

The following 3rd party software is required to build and run the project:

• Backend API
  • Go 1.8
  • Glide 0.12.3
  • PostgreSQL 9.6
• Web UI
  • NodeJS 8.1.3
  • NPM 5.0.3

The use of Go/NodeJS means the development platform is agnostic of operating system.

Building the Backend

First, you must clone this repository in $GOPATH/src/github.com/fsufitch/prez-tweet.

Next, run the following command to install and vendor all required software:

glide install

To build the statically linked Go executable, run:

go build github.com/fsufitch/prez-tweet

This creates a prez-tweet (or prez-tweet.exe, depending on platform) file that contains everything needed to run the API.

Building the Web UI

The web UI uses NPM scripts and configuration. To install everything needed for building it, run:

npm install

Next, to build the distributable artifacts, run:

npm run webpack

The output JS and other files will be located under dist/dev/.

This command is affected by the ENV environment variable. Current recognized values are dev and prod. This affects several things, including code minification, source map generation, Angular debug mode, and destination of build artifacts. If ENV=prod, then the build artifacts will be placed in dist/prod/.

Configuration

The backend is configured via environment variables; the web UI gets any relevant configuration from the backend. The backend's environment variables are:

Variable	Required	Description
PORT	yes	the port on which the backend should run
UI_RES_URL	yes	HTTP address at which the backend should mirror the web UI from
API_HOST	yes	host and port the web UI should use to contact the backend
DATABASE_URL	yes	postgres:// connection string for the application DB
PROXY_CACHE_TTL	yes	the duration (in seconds) that the backend should cache web UI files; should be 0 when doing web UI development to disable caching
TWITTER_AUTH	yes	Twitter keys/tokens for OAuth authorization, formatted as: api_key::api_secret::access_token::access_token_secret
WEBPACK_DEV_SERVER_PORT	no	the port on which the Webpack dev server should run (default: 8888)

Development Operation

1. Environment Variables

For development, the repository contains some sample environment configuration under scripts/dev-environment.{bat,sh}. TWITTER_AUTH is not included. Inject them into your environment as appropriate for your platform.

2. Database Setup

The database can be set up/torn down for development using these SQL scripts:

• dev-db-setup.sql - create a user and database, and grant full access
• dev-db-drop.sql - drop both user and entire database

In Heroku production, user/database creation is handled by the Heroku PostgreSQL App.

3. Run the Backend

To start the backend server, simply run:

./prez-tweet serve

Or, to both build and run:

go run main.go serve

Both these modes of execution require a manual re-compile/restart to receive code changes.

4. Run the Web UI

To start the web UI development server, run:

npm run webpack:server

This compiles all the necessary resources into a temporary location and serves them on the port specified by WEBPACK_DEV_SERVER_PORT (or 8888). The backend must be configured to recognize this location they are served, just as the web UI must be configured to know where to contact the backend.

5. Import Historical Tweet Data (optional)

The repository includes a repository of old tweets by @BarackObama, @realDonaldTrump, @POTUS44 and @POTUS, most of which cannot be normally accessed using Twitter's API. It is found in tweet-archive.json.

To import this data into your development server, run:

./prez-tweet import-archive tweet-archive.json

The data is generated by a scraping method contained in the twitter-scrape directory. To run it, install the go package github.com/fsufitch/prez-tweet/twitter-scrape/cmd/twitter-scrape, then use the twitter-scrape command (which is fairly self explanatory).

This scraper is inspired by GOT.

6. Visit the Backend's Root URL

Navigate your browser to http://localhost:8080, or whatever port your backend is running on. The backend will serve you the proxied web UI files from there.

Backend API

This is a quick summary of the functionality provided by the backend's endpoints:

• GET /status - Provide server status in JSON format
• GET /latest - Retrieve the tweet IDs of the latest presidential tweets
• GET /tweet/{id} - Retrieve details about a specific tweet and the ones bracketing it in its timeline
• Synchronized tweet browsing - These endpoints enable browsing tweets by keeping their dates aligned (with an optional offset)
  • GET /syncApplyOffset?tweet1=XXX&tweet2=YYY&tweet1_offset_years=ZZZ - Apply the given offset to the pair of tweets in question, and return the new tweet IDs; i.e. "replace the first tweet with a tweet by the same author but set ZZZ years before the second tweet"
  • (Deprecated) GET /syncOlder?tweet1=XXX&tweet2=YYY&tweet1_offset_years=ZZZ - Navigate backward in time, replacing the newer of the two tweets with the next oldest tweet by the same author, and taking into account the offset
  • (Deprecated) GET /syncNewer?tweet1=XXX&tweet2=YYY&tweet1_offset_years=ZZZ - Navigate forward in time, replacing the older of the two tweets with the next newest tweet by the same author, and taking into account the offset
• Tweet pairs - These endpoints provide unique IDs for pairs of tweets, for permalinking
  • POST /pair - Take a pair of tweet IDs and return the pair's ID and shortened ID; create the pair if necessary; expected body: {tweet1_id_str: "string", tweet2_id_str: "string"}
  • GET /pair/{shortID} - Given a shortened ID, return the corresponding pair and associated data.