#!/bin/bash

usage() {
    echo "USAGE: $0 [--clobber=none|first|second] [--verbose] [--] search_dir other_dir"
}

verbose=
debug() {
    if [ -n "$verbose" ]
    then
        echo "$@"
    fi
}

clobber=none
while [ "${1:0:2}" == "--" ]
do
    case $1 in
        --)
            break
            ;;
        --verbose)
            verbose=true
            ;;
        --clobber=none)
            ;;
        --clobber=first)
            clobber=first
            ;;
        --clobber=second)
            clobber=second
            ;;
        *)
            >&2 echo "Unrecognized option: $1"
            usage
            exit 1
    esac
    shift
done

if (( $# != 2 ))
then
  usage
  exit 0
fi

from="$1"
to="$2"

hardlink-file-if-duplicate() {
    if [ ! -f "$1" ]
    then
        debug "Not a regular file (or does not exist): $1"
        return
    elif [ ! -f "$2" ]
    then
        debug "Not a regular file (or does not exist): $2"
        return
    elif [ "$1" -ef "$2" ]
    then
        debug "Already a hardlink: $1 => $2"
        return
    elif ! cmp -s -- "$1" "$2"
    then
        debug "Files aren't identical: $1 != $2"
        return
    fi

    # from https://unix.stackexchange.com/a/120866
    if [ "$(stat --format="%h" "$2")" -gt 1 ]
    then
        if [ "$(stat --format="%h" "$1")" -gt 1 ]
        then
            case $clobber in
                none)
                    echo "Skipping: both files have multiple hardlinks. Specify --clobber=first to delete $1 or --clobber=second to delete $2."
                    return
                    ;;
                first)
                    echo "Both files have multiple hardlinks, defaulting to $2 overwritting $1"
                    one_link_file="$1"
                    other_file="$2"
                    ;;
                second)
                    echo "Both files have multiple hardlinks, defaulting to $1 overwritting $2"
                    one_link_file="$2"
                    other_file="$1"
                    ;;
            esac
        else
            one_link_file="$1"
            other_file="$2"
        fi
    else
        one_link_file="$2"
        other_file="$1"
    fi

    ln -v -f -- "$other_file" "$one_link_file"
}

if [ "$from" -ef "$to" ]
then
    echo "WARNING: $from and $to are the same path. Exiting."
    exit 2
elif [ ! -d "$from" ] || [ ! -d "$to" ]
then
    debug hardlink-file-if-duplicate "$from" "$to"
    hardlink-file-if-duplicate "$from" "$to"
    exit
fi

if [[ "${from:0-1}" != "/" ]]
then
    from="$from/"
fi
if [[ "${to:0-1}" != "/" ]]
then
    to="$to/"
fi

debug "Searching $from ..."
# from https://stackoverflow.com/a/9612232
find "$from" -type f -print0 | 
    while IFS= read -r -d '' line; do 
        debug hardlink-file-if-duplicate "$line" "${line/#$from/$to}"
        hardlink-file-if-duplicate "$line" "${line/#$from/$to}"
    done