Generate "realistic" nested folders of (random-content) files on macOS

README.md

Usage: mkdummy [-f files] [-n bytes] [-x bytes] [-s bytes] [-d depth] [-r [0..99)] [path]
  -d: [D]epth of recursive folder nesting
  -f: maximum number of [F]iles
  -n: mi[N]imum file size (in bytes)
  -r: probabilty of [R]ecursive folder nesting [0..100), default 10
  -s: approximate amount of disk[S]pace to consume (in bytes)
  -x: ma[X]imum file size (in bytes)

Requires installation of no new dependencies in macOS. Generates files of random content with quasi-realistic file names and extensions, within a nested folder hiearchy.

mkdummy

#!/bin/bash

DIR=$(pwd)
DNAMES=(stuff things homework)
FNAMES=(file test data "my paper")
FEXTS=(docx xlsx pptx pdf png jpg gif mp4)
FSIZE_MIN=$(bc <<<"1048*2") #2K
FSIZE_MAX=$(bc <<<"1024*500") #500K
FILES_MAX=0
DISKSPACE_MAX=0
RECURSE_PROB=10
RECURSE_MAX=3

DIRS=0
FILES=0
DISKSPACE=0

# https://gist.github.com/gnif/cadc611f04998706559def45318a0129
function humanReadableSize()
{
  local SIZE=$1
  local UNITS="B K M G T P"
  for F in $UNITS; do
    local UNIT=$F
    test ${SIZE%.*} -lt 1024 && break;
    SIZE=$(echo "$SIZE / 1024" | bc -l)
  done

  if [ "$UNIT" == "B" ]; then
    printf "%1.0f%s\n" $SIZE $UNIT
  else
    printf "%1.02f%s\n" $SIZE $UNIT
  fi
}

# Finder-style copy renaming if file or directory exists
finderCopy() {
	local BASE="$1"
	local EXT="$2"
	if [ -n "$EXT" ]; then
		EXT=".$EXT"
	fi
	local FNAME="$BASE$EXT"
	local i=0
	while [[ -f "$FNAME" || -d "$FNAME" ]]; do
		if [ $i -eq 0 ]; then
			FNAME="$BASE copy$EXT"
		else
			FNAME="$BASE copy $((i+1))$EXT"
		fi
		let "i++"
	done
	echo $FNAME
}

generate() {
	local FPATH=$1
	local FPATH_DEPTH=${2:-1}
	local FPATH_FILES_MAX=${3:-0}	
	local FPATH_FILES=0
	while [[ ($DISKSPACE_MAX -eq 0 || $DISKSPACE -lt $DISKSPACE_MAX) && \
			($FPATH_FILES_MAX -eq 0 || $FPATH_FILES -lt $FPATH_FILES_MAX) && \
			( $FILES_MAX -eq 0  || $FILES -lt $FILES_MAX ) ]] ; do
		if [[ $FPATH_DEPTH -ge $RECURSE_MAX || $(jot -r 1 0 100) -gt $RECURSE_PROB ]]; then
			local FSIZE=$(jot -r 1 $FSIZE_MIN $FSIZE_MAX)
			local FNAME=$(finderCopy "${FNAMES[$RANDOM % ${#FNAMES[@]}]}" "${FEXTS[$RANDOM % ${#FEXTS[@]}]}")
			dd if=/dev/urandom bs=1 count=$FSIZE of="$FPATH/$FNAME" >/dev/null 2>&1
			echo "$FPATH/$FNAME ($(humanReadableSize $FSIZE))"
			DISKSPACE=$(bc <<<"$DISKSPACE + $FSIZE")
			let "FILES++"
			let "FPATH_FILES++"
		else
			local SUB=$(finderCopy "$FPATH/${DNAMES[$RANDOM % ${#DNAMES[@]}]}")
			mkdir "$SUB"
			let "DIRS++"
			generate "$SUB" $((FPATH_DEPTH+1)) $(jot -r 1 1 100)
		fi
	done
}

print_usage() {
	echo "Usage: $(basename $0) [-f files] [-n bytes] [-x bytes] [-s bytes] [-d depth] [-r {0..99}] [path]"
	echo "  -d: [D]epth of recursive folder nesting"
	echo "  -f: maximum number of [F]iles"
	echo "  -n: mi[N]imum file size (in bytes)"
	echo "  -r: probabilty of [R]ecursive folder nesting [0..100), default 10"
	echo "  -s: approximate amount of disk[S]pace to consume (in bytes)"
	echo "  -x: ma[X]imum file size (in bytes) "
}

while getopts 'd:f:n:x:s:r:' flag; do
	case "${flag}" in
		f) FILES_MAX="${OPTARG}" ;;
		x) FSIZE_MAX="${OPTARG}" ;;
		n) FSIZE_MIN="${OPTARG}" ;;
		s) DISKSPACE_MAX="${OPTARG}" ;;
		d) RECURSE_MAX="${OPTARG}" ;;
		r) RECURSE_PROB=$(bc <<<"${OPTARG} % 100") ;;
		*) print_usage
		   exit 1 ;;
	esac
done
shift $(($OPTIND - 1)) # assign positional args to $1, $2, ...$n
DIR=${1:-$(pwd)}

if [[ $DISKSPACE_MAX -eq 0 && $FILES_MAX -eq 0 ]]; then
	DISKSPACE_MAX=$(bc <<<"1024*1024")
fi

mkdir -p $DIR >/dev/null 2>&1
generate $DIR
echo "$FILES files ($(humanReadableSize $DISKSPACE))"