rmdup_snapprefs

#!/bin/bash

# Do not specify a $DUP_NAME with a space.
DUP_NAME=Duplicate
DUP_DIR="$(pwd)/$DUP_NAME"

function enter_subfolder {
	DIR=$(find ./ -maxdepth 1 -type d)
	COUNT=$(echo "$DIR" | wc -l)

	# We check if we can go deeper.
	if [[ "$COUNT" == "1" ]]; then
		# Execute the filter since we are at the last folder of the tree.
		execute_filter
	else
		# We loop through each folder.
		while read LINE; do
			# We ignore the current folder.
			if [[ "$LINE" == "./" ]]; then
				continue
			fi

			# We ignore the $DUP_DIR folder.
			if [[ "$LINE" == "./$DUP_NAME" ]]; then
				cd "$LINE"
				CHECK=$(pwd)
				cd ..

				if [[ "$CHECK" == "$DUP_DIR" ]]; then
					continue
				fi
			fi
			
			# We recursively go into folder until we are at the last one.
			cd "$LINE"
			enter_subfolder
			cd ..			
		done < <(echo "$DIR")
	fi
}

function execute_filter {
	# Snapchat uses *.jpg and *.mp4 files.
	FILES=$(find ./ -maxdepth 1 -type f -regex ".*/.*\.\(mp4\|jpg\)")
	HASHES=()
	COUNT=$(echo "$FILES" | wc -l)

	# We make sure that there is at least one file.
	if [[ "$COUNT" != 0 ]]; then
		# We loop through each file.
		while read LINE; do
			# We ignore the file if, for some reason, the string is empty.
			if [[ -z "$LINE" ]]; then
				continue
			fi
			
			# We hash the file.
			HASH=$(sha512sum "$LINE" | awk '{print $1;}')
			printf "$HASH\t"
			
			# We store each hash in order to lookup if the hash already exists.
			if [[ "${HASHES[@]}" =~ "$HASH" ]]; then
				# We move the file if it is a duplicated file.
				printf "DUPLICATE DETECTED\n"
				mv "$LINE" "$DUP_DIR"
			else
				# We store the hash if it is the first time we encountered it.
				printf "\n"
				HASHES+=("$HASH")
			fi
		done < <(echo "$FILES")
	fi
}

# The execution starts here.
mkdir -p "$DUP_DIR"
enter_subfolder
exit 0