dotfiles/scripts/.local/bin/archive

#!/usr/bin/env bash
#
# Archives files and directories to a central depository via hardlinks
# Takes either a file or directory as argument and links them to the archive.
#
# Archiving does *not* duplicate the data, it just provides another permanent
# pointer to it, so the file will not be lost even if you delete it from the
# rest of your folders.

show_help() {
    printf """
archive.sh: Hard linking your stuff.

archive.sh is not backup software, rather it will create hardlinks for your files.
What this accomplishes is multiple links into the same data on your hard drive, 
so wherever you put the original file another pointer to it will reside in your archive
directory. You can think of it almost like the opposite of backups: You still only 
have one copy of the file, but multiple places where you link to it. 
This allows you to make the file in question the single source of truth in your system,
good for stopping duplicate downloading or contributing continued upload to 
decentralized download protocols while still organizing your files.

Usage:  archive.sh [OPTION] <file|directory>

Point it to a file or directory that you want linked. 

Options:

	-h		Display this help.

	-d		Directory to archive to.

	-f		Force archive to link. By default, will skip file entries already
			existing in archive and not link the current file at all.
			This forces the archive to link the files. It will not overwrite
			the already existing entry, instead creating a new unique archive
			entry by prepending the current unix timestamp.

	-a		Absolute linking. By default, will rebuild the same directory
			structure as in the original in the archive. Using this flag tells
			archive.sh to instead link everything directly in the root of the 
			archive directory, creating a flat directory without any nesting.
			Can create a more coherent archive, but runs the danger of 
			accidentally running into file conflicts.

	-n		Dryrun. Do not link anything yet but print the commands that would
			be executed.
"""
}

while getopts "nahfd:" opt; do
    case "$opt" in
    # v)  verbose=1
    # ;;
    f)
        FORCE="true"
        ;;
    a)
        ABSOLUTE="true"
        ;;
    d)
        ARCHIVEDIR="$OPTARG"
        ;;
    n)
        DRYRUN="true"
        ;;
    h | \? | *)
        show_help
        exit 0
        ;;
    esac
done
shift $((OPTIND - 1))

main() {
    archivedir="${ARCHIVEDIR:-/mnt/dietpi_userdata/minio-data/videos/archive}"
    source="${1:-"."}"
    _ensuredir "${archivedir}"

    find "$source" -type f | while read -r file; do
        if [ "$ABSOLUTE" = true ]; then
            fname="$(basename "$file")"
        else
            fname="$file"
            _ensuredir "${archivedir}/$(dirname "$fname")"
        fi

        ERROR=$(_link "$file" "${archivedir}/${fname}")
        stat="$?"

        #force mode: also add duplicates, but inform user
        if [ "$FORCE" = "true" ] && [ "$stat" -eq 1 ] && [ -z "${ERROR##*File exists}" ]; then

            duplicatef=$(_prepend_date "${fname}")

            _link "${file}" "${archivedir}/${duplicated}/${duplicatef}"
            printf "ERROR: File %s exists. Linked as %s. Check for duplicates.\n" "$fname" "$duplicatef"
        elif [ "$stat" -gt 0 ]; then
            printf "ERROR $stat: File %s exists. Nothing done. Use -f to force relinking new file.\n" "$fname"
            printf %s "$ERROR"
        fi
    done
}

_prepend_date() {
    fname="$1"
    duplicated=$(dirname "$fname")
    duplicatef=$(basename "${fname}")
    printf "%s/%s-%s" "$duplicated" "$(date +%s)" "$duplicatef"
}

_link() {
    #try to create a hard link to file
    if [ "$DRYRUN" = "true" ]; then
        echo ln "${1}" "${2}"
        return
    fi
    { ln "${1}" "${2}" >/dev/null; } 2>&1
}

_ensuredir() {
    if [ "$DRYRUN" = "true" ]; then
        echo mkdir -p "${1}"
        return
    fi
    mkdir -p "${1}"
}

main "$@"