File management sample codes?

defmodule sort_files_by_extension do
  def sort_files do
    File.ls!("./test")
    |> Enum.map(fn filename -> 
         extname = Path.extname(filename)  #=> ".jpg"
         basename = Path.basename(filename, extname)  #=> "IMG_XXXX"
         {basename, extname, filename}
       end)
    |> Enum.map(&make_extension_dir/1)
    |> Enum.map(&dedup/1)
    |> Enum.each(&move_extension_directory/1)

  defp make_extension_dir({_, "." <> extension, _} = arg) do
    File.mkdir_p!("./#{extension}")
    arg  # just return the argument for other `Enum.map`
  end

  defp dedup(arg, suffix \\ 0)

  defp dedup({basename, "." <> extension, filename} = arg, 0) do
    if File.exists?("#{extension}/#{filename}") do
      dedup(arg, 1)
    else
      arg
    end  
  end

  defp dedup({basename, "." <> extension = extname, filename} = arg, suffix) do
    if File.exists?("#{extension}/#{basename}_#{suffix}.#{extension}") do
      dedup(arg, suffix + 1)
    else
      {"#{basename}_#{suffix}", extname, filename}
    end 
  end

  defp move_to_extension_directory({basename, "." <> extension, filename}) do
    File.rename!("./test/#{filename}", "./#{extension}/#{basename}.#{extension}")
  end
end

A few suggestions:

1. Enum.each is for pure side effects (e.g. pure file system operations). It’s not chainable. Use Enum.map instead.
2. You can wrap all the information in a tuple, and pattern match on it in the function parameters.
3. You can trust the file system to do the right thing (e.g. File.mkdir_p!)
4. Hail recursion!

This might not be the answer you are looking for – but I’d reach for sqlite3. It also can be up to 35% faster than raw file access.

Trouble is, current state of the art of sqlite3 in Elixir does not support Ecto 3 (latest version of the de facto DB persistence library); only version 2 which is now quite old. I am working on bringing Ecto 3 to sqlite3 but it definitely is not going to be ready tomorrow.

Using sqlite3 will rid you of the potential file overwriting problem as well since there the ID is… you know, the database ID, not the filename itself. What’s more, an embedded DB like sqlite3 gives you the ability to sort and filter out of the box.

If that doesn’t sound tempting to you then I’d be happy to help you exactly with the File / IO API. However, you should have in mind that functions like stat can have differing behaviours between different OS-es. Which is all the more reason to opt for a database.

Any particular scenario you would like help with?

If you would like something that basically manages uploads to your app then Waffle might be exactly what you are looking for (it can put all stored files into Amazon’s S3 or in your local filesystem).

I’d probably do this:

  def sort_files(path) do
    files_by_dir =
      File.ls!(path)
      |> Enum.reject(&File.dir?/1)
      |> Enum.group_by(&Path.extname/1)

    Enum.each(files_by_dir, fn {"." <> ext, files} ->
      extpath = Path.join(path, ext)
      File.mkdir_p(extpath)

      Enum.each(files, fn file ->
        filepath = Path.join(extpath, file)
        # Keep only a max of 9, could easily make this unbounded though, but eh useful feature to add
        Enum.each(9..2, &File.rename("#{filepath}_#{&1-1}", "#{filepath}_#{&1}"))
        File.rename(filepath, "#{filepath}_1")
        File.rename!(Path.join(path, file), filepath)
      end)
    end)
  end

Could use some error reporting, but eh.