cleanup_SynoFiles

#!/bin/bash

# check parameters
if [[ $1 == -l ]] ; then lst=y; shift ; fi
if (( $# == 0 )) ; then err=y ; fi
# make sure $@ only contains directories or find will create havoc if you pass -exec in $@ by accident
for d in "$@" ; do if [[ ! -d $d ]] ; then echo "$d: not a directory" >&2 ; err=y ; fi ; done
xat=xattrs.lst
if [[ ! -f $xat ]] ; then echo "$xat not found - don't know which xattrs are not bogus" >&2 ; exit 1 ; fi
if [[ -n $err ]] ; then
	echo "usage: $0 [-l] dir [dir ...]" >&2
	echo "       delete or list (-l) in specified directories:" >&2
	echo "       - the 'stray' (extraneous) @SynoEAStream and @SynoResource files" >&2
	echo "       - any subdirectories inside the @eaDir directories (which are used by Synology to store picture previews" >&2
	echo "         (Photo Station), SYNO_DTIME delete timestamps (when using the Trash), and other paraphernalia)" >&2
	echo "       - the 'bogus' @SynoEAStream and @SynoResource files ('bogus' means the files don't have any of the meaningful xattrs listed in $xat)" >&2
	echo "       - when deleting, it also cleans up any empty @eaDir directories that are left behind" >&2
	echo "       prints found files and directories on stdout; prints messages on stderr" >&2
	exit 1
fi

# find stray @SynoEAStream files, i.e. @SynoEAStream files that don't have the parent file it is supposed to be associated with
echo "finding stray @SynoEAStream files" >&2
find "$@" -type f -name '*@SynoEAStream' -print | while read f ; do
	p=${f/@eaDir\/} ; p=${p%@SynoEAStream}
	if [[ ! -e "$p" ]] ; then
		if [[ -n $lst ]] ; then echo "$f" ; else rm -v "$f" ; fi
	fi
done

# find stray @SynoResource files, i.e. @SynoResource files that don't have the parent file it is supposed to be associated with
echo "finding stray @SynoResource files" >&2
find "$@" -type f -name '*@SynoResource' -print | while read f ; do
	p=${f/@eaDir\/} ; p=${p%@SynoResource}
	if [[ ! -e "$p" ]] ; then
		if [[ -n $lst ]] ; then echo "$f" ; else rm -v "$f" ; fi
	fi
done

# find @eaDir subdirectories, but skip the #recycle bin because that contains useful SYNO_DTIME timestamps
echo "finding @eaDir subdirectories (*/@eaDir/*/)" >&2
find "$@" -depth -type d -path '*/@eaDir/*' ! -path '*/#recycle/*' -print | while read d ; do
	if [[ -n $lst ]] ; then echo "$d/" ; else rm -rfv "$d" ; fi
done
# now scan the #recycle bin and find any _stray_ @eaDir subdirectories - these are no good
find "$@" -depth -type d -path '*/#recycle/@eaDir/*' -print | while read d ; do
	p=${d/@eaDir\/}
	if [[ ! -e "$p" ]] ; then
		if [[ -n $lst ]] ; then echo "$d/" ; else rm -rfv "$d" ; fi
	fi
done

# find @SynoEAStream files that don't have any of the relevant (non-bogus) xattrs inside
echo "finding bogus @SynoEAStream files (and their @SynoResource siblings)" >&2
# use find followed by ! grep -qF -f because grep -vqFr -f doesn't produce the same result
find "$@" -type f -name '*@SynoEAStream' -print | while read f ; do
	if ! grep -qF -f $xat "$f" ; then
		if true ; then
			# final paranoia check
			for i in $(cat $xat) ; do
				if [[ -n $(get_attr -x -q "$i" "$f") ]] ; then
					echo "error: $f has a $i xattr - there's something wrong - exit" >&2 ; exit 1
				fi
			done
		fi
		# if this @SynoEAStream is bogus, and happens to have a @SynoResource at its side, it's not necessary either (can be considered bogus as well)
		s="${f%@SynoEAStream}@SynoResource"
		if [[ -n $lst ]] ; then
			echo "$f"
			if [[ -f "$s" ]] ; then echo "$s" ; fi
		else
			rm -v "$f"
			if [[ -f "$s" ]] ; then rm -v "$s" ; fi
		fi
	fi
done

# find remaining @SynoEAStream files that do have any of the relevant (non-bogus) xattrs inside, but they all have an empty value
echo "finding more bogus @SynoEAStream files (and their @SynoResource siblings)" >&2
grep -rlF -f $xat "$@" --include='*@SynoEAStream' | while read f ; do
	# this filters all files with one or more of the (non-bogus) xattrs, but they could be empty (all having an empty value)
	found=n
	# visit all non-bogus($xat) attributes and check on their values, if one non-empty attr is found, the metadata (and as a result, the file as a whole) is not bogus
	for i in $(cat $xat) ; do
		val=$(get_attr -x -q "$i" "$f")
		# on an empty xattr, the value could still be either empty or an empty bplist in case of com.apple.metadata:_kMDItemUserTags
		if [[ -n $val && $val != "62706c6973743030a0080000000000000101000000000000000100000000000000000000000000000009" ]] ; then
			found=y # found (at least one) non-empty non-bogus xattr
		fi
	done
	# check that none of the xat attributes had meaningful data in the value
	if [[ $found == "n" ]] ; then
		if true ; then
			# final paranoia check
			if [[ -n $(tag "$f") ]] ; then
				echo "error: $f has a custom tag - there's something wrong - exit" >&2 ; exit 1
			fi
			for i in $(cat $xat | grep -v com.apple) ; do
				if [[ -n $(get_attr -x -q "$i" "$f") ]] ; then
					echo "error: $f has a $i xattr - there's something wrong - exit" >&2 ; exit 1
				fi
			done
		fi
		# this @SynoEAStream is bogus, and if it happens to have a @SynoResource at its side, it's not necessary either (can be considered bogus as well)
		s="${f%@SynoEAStream}@SynoResource"
		if [[ -n $lst ]] ; then
			echo "$f"
			if [[ -f "$s" ]] ; then echo "$s" ; fi
		else
			rm -v "$f"
			if [[ -f "$s" ]] ; then rm -v "$s" ; fi
		fi
	fi
done

# find any remaining @SynoResource files that don't have a @SynoEAStream file at their side
echo "finding bogus @SynoResource files" >&2
find "$@" -type f -name '*@SynoResource' -print | while read f ; do
	# if this @SynoResource is on its own, it's not necessary either (can be considered bogus as well)
	s="${f%@SynoResource}@SynoEAStream"
	if [[ ! -f "$s" ]] ; then
		if [[ -n $lst ]] ; then echo "$f" ; else rm -v "$f" ; fi
	fi
done

# find leftover @eaDir directories
if [[ ! -n $lst ]] ; then
	echo "finding leftover (empty) @eaDir directories" >&2
	find "$@" -depth -type d -name '@eaDir' -empty -exec rmdir -v {} \;
fi

#EOF