Split a folder into multiple subfolders in terminal/bash script

This solution worked for me on MacOS:

i=0; for f in *; do d=dir_$(printf %03d $((i/100+1))); mkdir -p $d; mv "$f" $d; let i++; done

It creates subfolders of 100 elements each.

Try something like this:

for i in `seq 1 20`; do mkdir -p "folder$i"; find . -type f -maxdepth 1 | head -n 2000 | xargs -i mv "{}" "folder$i"; done

Full script version:

#!/bin/bash

dir_size=2000
dir_name="folder"
n=$((`find . -maxdepth 1 -type f | wc -l`/$dir_size+1))
for i in `seq 1 $n`;
do
    mkdir -p "$dir_name$i";
    find . -maxdepth 1 -type f | head -n $dir_size | xargs -i mv "{}" "$dir_name$i"
done

---

For dummies:

1. create a new file: vim split_files.sh
2. update the dir_size and dir_name values to match your desires
   • note that the dir_name will have a number appended
3. navigate into the desired folder: cd my_folder
4. run the script: sh ../split_files.sh

The code below assumes that the filenames do not contain linefeeds, spaces, tabs, single quotes, double quotes, or backslashes, and that filenames do not start with a dash. It also assumes that IFS has not been changed, because it uses while read instead of while IFS= read, and because variables are not quoted. Add setopt shwordsplit in Zsh.

i=1;while read l;do mkdir $i;mv $l $((i++));done< <(ls|xargs -n2000)

The code below assumes that filenames do not contain linefeeds and that they do not start with a dash. -n2000 takes 2000 arguments at a time and {#} is the sequence number of the job. Replace {#} with '{=$_=sprintf("%04d",$job->seq())=}' to pad numbers to four digits.

ls|parallel -n2000 mkdir {#}\;mv {} {#}

The command below assumes that filenames do not contain linefeeds. It uses the implementation of rename by Aristotle Pagaltzis which is the rename formula in Homebrew, where -p is needed to create directories, where --stdin is needed to get paths from STDIN, and where $N is the number of the file. In other implementations you can use $. or ++$::i instead of $N.

ls|rename --stdin -p 's,^,1+int(($N-1)/2000)."/",e'

This solution can handle names with whitespace and wildcards and can be easily extended to support less straightforward tree structures. It will look for files in all direct subdirectories of the working directory and sort them into new subdirectories of those. New directories will be named 0, 1, etc.:

#!/bin/bash

maxfilesperdir=20

# loop through all top level directories:
while IFS= read -r -d $'\0' topleveldir
do
        # enter top level subdirectory:
        cd "$topleveldir"

        declare -i filecount=0 # number of moved files per dir
        declare -i dircount=0  # number of subdirs created per top level dir

        # loop through all files in that directory and below
        while IFS= read -r -d $'\0' filename
        do
                # whenever file counter is 0, make a new dir:
                if [ "$filecount" -eq 0 ]
                then
                        mkdir "$dircount"
                fi

                # move the file into the current dir:
                mv "$filename" "${dircount}/"
                filecount+=1

                # whenever our file counter reaches its maximum, reset it, and
                # increase dir counter:
                if [ "$filecount" -ge "$maxfilesperdir" ]
                then
                        dircount+=1
                        filecount=0
                fi
        done < <(find -type f -print0)

        # go back to top level:
        cd ..
done < <(find -mindepth 1 -maxdepth 1 -type d -print0)

The find -print0/read combination with process substitution has been stolen from another question.

It should be noted that simple globbing can handle all kinds of strange directory and file names as well. It is however not easily extensible for multiple levels of directories.