Recursively find and extract gz files without extension

Question

I have a folder structure (a bad rip of an old website - you maybe know about the old double compression problem), and some files are called somefilename.txt but actually gz compressed. I am looking for a way to identify these files and then unpacking them into the txt files that they actually are. Environment is Linux.

Answer 1

You could use a bash function to test the file type with file and decompress if it is gzip-compressed. And you could call the (exported) function in execdir action of find:

if_gz_decompress () {
  type=$(file -bi "$1")
  if [[ "$type" =~ ^application/gzip\; ]]; then
    tmp=$(mktemp) &&
    gunzip -dc "$1" > "$tmp" &&
    mv -i "$tmp" "$1" ||
    rm -f "$tmp"
  fi
}
export -f if_gz_decompress
find . -type f -execdir bash -c 'if_gz_decompress "$1"' _ {} \;

Because of the -i option of mv you will have to confirm for each found file. Remove it if you are 100% sure it works as you wish (or if you have a backup).

If you have many such files and some are in the same directory we can maybe optimize a bit with:

if_gz_decompress () {
  for f in "$@"; do
    type=$(file -bi "$f")
    if [[ "$type" =~ ^application/gzip\; ]]; then
      tmp=$(mktemp) &&
      gunzip -dc "$f" > "$tmp" &&
      mv -i "$tmp" "$f" ||
      rm -f "$tmp"
    fi
  done
}
export -f if_gz_decompress
find . -type f -execdir bash -c 'if_gz_decompress "$@"' _ {} +