1. Create a PostgreSQL database;
2. Either download a database dump from the current website to fill the initial values or use the files under postgres/ to generate the initial schema.

1. Serve the data directly from the database using PostgREST and the following configuration:

db-uri = "postgres://yourpostgresdatabaseuri"
db-schema = "public"
db-anon-role = "web_anon"
db-pool = 7
server-host = "127.0.0.1"
server-port = 11936
max-rows = 10000

2. Run npm install && make to build the client-side JavaScript. All files will be in the static/ directory.
3. Now you need an HTTP server proxy to serve the PostgREST API and the static site under the same domain. This at the naked path and that under the /api/ path. Nginx will do it, but I use Caddy with the following Caddyfile:

ln.bigsun.xyz {
  handle /api/* {
    uri strip_prefix /api
    reverse_proxy localhost:11936 {
      header_up Accept application/json
      header_up Content-Type application/json
    }
  }
  handle {
    root * /home/fiatjaf/lnchannels/static
    try_files {path} /index.html
    file_server
  }
}

1. Run sparko and generate a key for it with at least the permissions to call listchannels and listnodes.
2. Run bitcoind with txindex=1.
3. Somehow set the following environment variables:

POSTGRES_URL=postgres://yourpostgresdatabaseuri
SPARK_URL=http://yoursparkoaddress:port/rpc
SPARK_TOKEN=accesskeygeneratedonstep1
BITCOIN_RPC_ADDRESS=http://localhost:8332
BITCOIN_RPC_USER=bitcoinrpcuser
BITCOIN_RPC_PASSWORD=bitcoinrpcpass

4. You can place all of the above in a file called .env and later user a program like godotenv to run things while setting them.
5. Install Python (must be python3.8 or greater I believe) dependencies from requirements.txt using any method you like (I do virtualenv venv && venv/bin/pip install -r requirements.txt).
6. Run python -m getdata (or godotenv python -m getdata if you're using an .env file or godotenv venv/bin/python -m getdata if you're using a virtualenv) once every day or hour or week, depending on how often you want to fetch new data -- the greater the interval between runs the more you'll miss shortlived channels, the smaller the interval more you'll clog your database with useless fee changes, also the process takes a long time to finish so I only run it once a day.

Public domain, except you can't use for shitcoins.