#!/bin/bash

mirror_dir=$(readlink -e '/srv/http/arch-mirror')

find "${mirror_dir}" -type f -printf '%h\n' \
| uniq \
| sort -u \
| while read -r repo_dir; do
  paccache -v -r -k 2 -c "${repo_dir}"
  if [ -z "${repo_dir##*/os/*}" ]; then
    repo="${repo_dir%%/os/*}"
    repo="${repo##*/}"
  else
    repo="${repo_dir##*/}"
  fi
  curl \
    --resolve 'arch-mirror.home.eckner.net:80:127.0.0.1' \
    -Ss 'http://arch-mirror.home.eckner.net/'"${repo_dir#${mirror_dir}/}/${repo}"'.db' \
  | tar -Oxz \
  | sed -n '
    /^%\(FILENAME\|CSIZE\|SHA256SUM\)%$/,/^$/p
  ' \
  | grep -v '^%.*%$' \
  | grep -vxF '' \
  | sed '
    N;N
    s@\n@ @g
  ' \
  | while read -r file size sum; do
    [ -f "${repo_dir}/${file}" ] || continue
    if stat -c'%s' "${repo_dir}/${file}" \
    | grep -qxF "${size}" \
    && printf '%s  %s/%s\n' "${sum}" "${repo_dir}" "${file}" \
    | sha256sum -c --quiet; then
      continue
    fi
    rm "${repo_dir}/${file}"
  done
done