Move every file that is not a directory

Question

I have a directory that I am trying to clean out that contains both files and subdirectories. What I am trying to do is simple: move all the files into another directory, but leave all the sub-directories as they are.

I am thinking something like:

mv [*_but_no_dirs] ./other_directory

It seems like there should be a simple way to do this with wildcards * and regex...

Anyone have ideas?

Answer 1

Regex aren't involved here. Wildcards in bash (like most other shells) only match files based on the file names, not based on the file type or other characteristics. There is one way to match by type: adding / at the end of the pattern makes it only match directories or symbolic links to directories. This way, you can move directories, then move what's left, and move directories back — cumbersome but it works.

tmp=$(TMPDIR=.. mktemp -d)
mv -- */ "$tmp"
mv -- * other_directory/
mv "$tmp"/* .
rmdir "$tmp"

(that approach should be avoided if the current directory is the mount point of a filesystem, as that would mean the moving of directories away and back would have to copy all the data in there twice).

A standard way to match files by type is to call find.

find . -name . -o -type d -prune -o -exec sh -c 'mv "$@" "$0"' other_directory/ {} +

(also moves symlinks, whether they point to directories or not).

In zsh, you can use glob qualifiers to match files by type. The . qualifier matches regular files; use ^/ to match all non-directories, or -^/ to also exclude symbolic links to directories.

mv -- *(.) other_directory/

In any shell, you can simply loop.

for x in *; do
   if ! [ -d "$x" ]; then
     mv -- "$x" other_directory/
   fi
done

(does not move symlinks to directories).

Answer 2

You could use something like

find . -maxdepth 1 \( ! -type d \) -exec sh -c 'mv  "$@" MYDIR' _ {} \;

First we use find to look only within the current directoy, then we ignore directories by using ! -type d finally we execute sh and move everything to the destination dir. You might try {} + at the end which will be faster.

Answer 3

It's potentially a bit dangerous but

cp * destination/
rm *

As both cp and rm won't operate on directories without the -r switch.

Answer 4

Here is a simple 1 line command to achieve what you are trying to do:

find ./sourceDir -maxdepth 1 -type f | xargs mv -t ./targetDir

Explanation

1. Find all the files you want at depth 1 only, that are of type files. You can find explanation here.
   https://linux.die.net/man/1/find

2. Pipe that using "| xargs" to mv.
   https://linux.die.net/man/1/xargs

3. tell mv where to move the files
   https://linux.die.net/man/1/mv

This 1 liner is fairly portable, and can be run from anywhere in the system as long as you have the sourceDir and targetDir, either as full paths or relative paths.

Answer 5

mv $(find /my/source/directory -maxdepth 1 -type f) /my/destination/directory

Move will take a list of arguments, so when you ask find to give you everything that is of type "file" within /my/source/directory, it will move them one by one to /my/destination/directory. If you wanted to include the files in the subdirectories, too, while leaving the directory structure intact, remove -maxdepth 1 from the above command.

Answer 6

if [ -f "$PATH/$FILE_NAME" ]; then mv "$PATH/$FILE_NAME" "$OTHER_PATH/$FILE_NAME"; fi;

Answer 7

I would suggest simply doing mv *.* destination/ with destination/ being the folder you are moving to.

Answer 8

If you don't intend to move dotfiles (e.g. ".foobar"), too, you can simply do:

mv * /some/destination/

If you do, see the above answer with find's -maxdepth.

Answer 9

mv $(ls -l | grep "^-" | awk '{print $9}') ./folder_where_move