#!/bin/bash

for pkg in $( \
  curl 'https://crux.eckner.net/REPO' 2> /dev/null | \
    grep '^d:' | \
    cut -d : -f 2
)
do
  inhalt="$( \
    curl "https://crux.nu/portdb/index.php?q=${pkg}&a=search" 2> /dev/null \
  )"
  echo "${inhalt}" | \
    grep -v '>deepthought<' | \
    grep -q '<td><a href="?a=repo&q=' \
  || continue

  minNum=-1
  minPrint=''
  for pkgfile in $( \
    echo "${inhalt}" | \
      tr '\n' ' ' | \
      sed 's|<tr>|\n<tr>|g' | \
      grep -v '>deepthought<' | \
      tr '"' '\n' | \
      grep 'Pkgfile$' \
  )
  do
    remPkgCont="$(
      wget -O - "${pkgfile}" 2> /dev/null
    )"
    numLinesDiff=$( \
      diff \
        <( \
          echo "${remPkgCont}" | \
            sed 's|^\s*||' | \
            sed 's|#.*$||' | \
            grep -v '^$'
        ) \
        <( \
          curl "https://crux.eckner.net/${pkg}/Pkgfile" 2> /dev/null | \
            sed 's|^\s*||' | \
            sed 's|#.*$||' | \
            grep -v '^$'
        ) | \
        grep -c '^[<>]' \
      )
    [ ${minNum} -ge 0 ] && [ ${minNum} -le ${numLinesDiff} ] && continue
    minNum=${numLinesDiff}
    [ ${numLinesDiff} -ge 10 ] && minPrint="${pkg}: ${numLinesDiff}" && continue
    minPrint="$(
      echo "${pkg} gibt es noch mal so ähnlich: ${pkgfile}"
      diff -u3 \
        <( \
          echo "${remPkgCont}" | \
            grep -v '^#' | \
            sed 's|^\s\+| |' \
        ) \
        <( \
          curl "https://crux.eckner.net/${pkg}/Pkgfile" 2> /dev/null | \
            grep -v '^#' | \
            sed 's|^\s\+| |' \
        )
    )"
  done
  [ ${minNum} -ge 0 ] && echo "${minPrint}"

done