Cleanup files of well known type

Run this bash script from a directory full of files of well known type (e.g. doc, docx, pdf, jpg etc) and it will move those that linux file command doesn't recognize to a subdir PROBABLY-JUNK

This is usefull after recovery of deleted files where a lot of them are incorectly guessed as "good".

#!/bin/bash

function testfile() {
  echo Scanning dir `pwd`
  ls | sed -e 's/^.*[.]//' | sort | uniq | while read EXT; do
  echo "  found extension $EXT"
  COUNT_TOTAL=`ls | grep "$EXT$" | wc -l`
  echo "    $COUNT_TOTAL total .$EXT files. SCANNING..."
  file *.$EXT | grep "$EXT: data$" |sed -e "s/: .*//" > /tmp/to-delete.txt
  COUNT_DEL=`cat /tmp/to-delete.txt | wc -l`
  echo "    $COUNT_DEL files to delete"
  if grep '[.]' /tmp/to-delete.txt; then
  echo "      moving files to PROBABLY-JUNK"
     mkdir -p PROBABLY-JUNK && cat /tmp/to-delete.txt | xargs mv -t PROBABLY-JUNK
  fi
  done
}

testfile