mempool/unfurler

Mempool Link Unfurler Service

This is a standalone nodejs service which implements the Open Graph protocol for Mempool instances. It performs two main tasks:

1. Serving Open Graph html meta tags to social media link crawler bots.
2. Rendering link preview images for social media sharing.

Some additional server configuration is required to properly route requests (see section 4 below).

Setup

0. Install deps

For Linux, in addition to NodeJS/npm you'll need at least:

• nginx
• cups
• chromium-bsu
• libatk1.0
• libatk-bridge2.0
• libxkbcommon-dev
• libxcomposite-dev
• libxdamage-dev
• libxrandr-dev
• libgbm-dev
• libpango1.0-dev
• libasound-dev

1. Clone Mempool Repository

Get the latest Mempool code:

git clone https://github.com/mempool/mempool
cd mempool

Check out the latest release:

latestrelease=$(curl -s https://api.github.com/repos/mempool/mempool/releases/latest|grep tag_name|head -1|cut -d '"' -f4)
git checkout $latestrelease

2. Prepare the Mempool Unfurler

Install

Install dependencies with npm and build the backend:

cd unfurler
npm install

The npm install may fail if your system does not support automatic installation of Chromium for Puppeteer. In that case, manually install Puppeteer without Chromium first:

PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=true npm install puppeteer
npm install

Configure

In the unfurler folder, make a copy of the sample config file:

cp config.sample.json config.json

Edit config.json as needed:

variable	usage
SERVER.HOST	the host where this service will be served
SERVER.HTTP_PORT	the port on which this service should run
MEMPOOL.HTTP_HOST	the host where the Mempool frontend is being served
MEMPOOL.HTTP_PORT	the port on which the Mempool frontend is running (or null)
PUPPETEER.CLUSTER_SIZE	the maximum number of Chromium browser instances to run in parallel, for rendering link previews
PUPPETEER.EXEC_PATH	(optional) an absolute path to the Chromium browser executable, e.g. /usr/local/bin/chrome. Only required when using a manual installation of Chromium

Build

npm run build

3. Run the Mempool Unfurler

npm run start

4. Server configuration

To enable social media link previews, the system serving the Mempool frontend should detect requests from social media crawler bots and proxy those requests to this service instead.

Precise implementation is left as an exercise to the reader, but the following snippet may be of some help for Nginx users:

map $http_user_agent $crawler {
    default 0;
    ~*facebookexternalhit 1;
    ~*twitterbot 1;
    ~*slackbot 1;
    ~*redditbot 1;
    ~*linkedinbot 1;
    ~*pinterestbot 1;
}