warm-cache.sh

#!/usr/bin/env bash
set -euo pipefail

# Parameters
# $1: The scheme and host of the site for which the cache should be warmed

warmCache () {
  # E.g. https://www.example.com
  local schemeAndHost="${1:?'The schemeAndHost parameter is missing'}"

  printf '\e[38;5;33m=> Warming cache of %s by reading the /sitemap.xml file and sending a request to every listed url\e[0m\n' "${schemeAndHost}"

  # MacOS doesn't use GNU grep.
  # It can be installed using Homebrew and used as `ggrep`.
  local grepCommand
  if [[ "${OSTYPE}" == "darwin"* ]]; then
    brew install grep
    grepCommand='ggrep'
  else
    grepCommand='grep'
  fi

  # Matches <loc>URL</loc> and excludes the loc tags from the match, so we only have the actual url as match
  local regex="(?<=<loc>)${schemeAndHost}.+?(?=<\/loc>$)"

  local startTime=$(date +%s)
  curl "${schemeAndHost}"/sitemap.xml -H 'Accept-Encoding: gzip, deflate, sdch' -H 'User-Agent: Cache Warmer Script' -H 'Connection: keep-alive' --compressed -s -o - | \
  eval "${grepCommand} -o -P \"${regex}\""  | \
  while read -r line; do
    printf '=> Sending request to \"%s\"\n' "${line}"
    curl "${line}" -H 'Accept-Encoding: gzip, deflate, sdch' -H 'User-Agent: Cache Warmer Script' -H 'Connection: keep-alive' --compressed -s -o /dev/null -w  '==> Received response after %{time_starttransfer} seconds\n'
  done
  local endTime=$(date +%s)

  printf "\e[32m=> Warming cache complete\e[0m\n"
  printf "=> Took %s\n" "$(secs_to_human $(($endTime - $startTime)))"
}

# Prints given seconds as a human readable string
secs_to_human() {
  local seconds=${1}

  echo "$(( (${seconds} / 60) % 60 ))m $(( ${seconds} % 60 ))s"
}

warmCache "${1}"