#!/bin/bash

set -eu
set -o pipefail

backup_id=""
filesystem_directory=""
save_directory=""
skip_tar_diff="False"
components_to_strip=0

function usage
{
    retval="$1"
    case "$retval" in
        0)
            ;;
        *)
            exec 1>&2
            ;;
    esac

    echo "This script checks for:"
    echo "   1) filenames in the backup but not in the filesystem"
    echo "   2) filenames in the filesystem but not in the backup"
    echo "   3) files in both, but with different content (optional - turn off with --skip-tar-diff)"
    echo
    echo "Usage: $0"
    echo "   --backup-id bi"
    echo "   --filesystem-directory /dir/name"
    echo "   --save-directory /backshift/save-directory"
    echo "   --skip-tar-diff"
    echo "   --strip-components number"
    echo
    echo "This script assumes that you:"
    echo "   1) Have no filenames with newlines in them"
    echo "   2) Haven't gotten fancy with the find -print0 you piped in while creating the backup"
    echo
    echo "tar --diff (which this script uses by default) is pretty slow, so --skip-tar-diff says to skip this part"
    echo "of the verification."
    echo
    echo "--filesystem-directory /dir indicates that /dir should be fully recursed for comparison with the"
    echo "specified backup (by --backup-id).  This recursive traversal is limited only by -xdev, so it"
    echo "will not traverse into other filesystems."
    echo
    echo "Note that --filesystem-directory does Not restrict which part of the backup is examined; the entire"
    echo "backup is always inspected, irrespective of --filesystem-directory"
    exit "$retval"
}

while [ "$#" -ge 1 ]
do
    case "$1" in
        --backup-id)
            backup_id="$2"
            shift
            ;;
        --filesystem-directory)
            filesystem_directory="$2"
            shift
            ;;
        --save-directory)
            save_directory="$2"
            shift
            ;;
        --strip-components)
            components_to_strip="$2"
            shift
            ;;
        --skip-tar-diff)
            skip_tar_diff="True"
            ;;
        --help|-h)
            usage 0
            ;;
        *)
            echo "$0: Unrecognized option: $1" 1>&2
            usage 1
            ;;
    esac
    shift
done

case "$backup_id" in
    "")
        echo "$0: --backup-id is a required option" 1>&2
        usage 1
        ;;
esac

case "$filesystem_directory" in
    "")
        echo "$0: --filesystem-directory is a required option" 1>&2
        usage 1
        ;;
    "/"*)
        # shellcheck disable=SC2001
        filesystem_directory=".$filesystem_directory"
        ;;
    *)
        echo "$0: --filesystem-directory must start with a leading /" 1>&2
        usage 1
        ;;
esac

case "$save_directory" in
    "")
        echo "$0: --save-directory is a required option" 1>&2
        usage 1
        ;;
esac

case "$skip_tar_diff" in
    True) ;;
    False) ;;
    *)
        echo "$0: Internal error: (1) \$skip_tar_diff is not True or False: $skip_tar_diff" 1>&2
        exit 1
        ;;
esac

basename="$(basename "$0")"

tempdir="/tmp/$basename.$$"

# shellcheck disable=SC2064
trap "rm -rf \"$tempdir\"" $(seq 0 15)

mkdir "$tempdir"
backshift \
    --save-directory "$save_directory" \
    --list-backup-simply \
    --backup-id "$backup_id" 2>&1 \
    | count -c \
    | sed -e 's#/$##' -e '/^$/d' \
    | strip-components --count "$components_to_strip" > "$tempdir/in-backup"

in_backup_count=$(wc -l "$tempdir/in-backup")
echo "Got $in_backup_count files in backup"

cd /

find "$filesystem_directory" -xdev -print | sed -e 's#^\./##' -e '#^$#d' > "$tempdir/currently-in-filesystem"

in_filesystem_count=$(wc -l "$tempdir/currently-in-filesystem")
echo "Got $in_filesystem_count files currently in filesystem"

echo "Files in backup but not currently in filesystem:"
set-arithmetic --difference "$tempdir/in-backup" "$tempdir/currently-in-filesystem" | sort | sed 's/^/   /'

echo
echo "Files currently in filesystem but not in backup:"
set-arithmetic --difference "$tempdir/currently-in-filesystem" "$tempdir/in-backup" | sort | sed 's/^/   /'

case "$skip_tar_diff" in
    True) ;;
    False)
        echo
        echo "Verifying file content of all files in backup:"

        if [ "${DISPLAY:-}" = "" ]
        then
            # Just do the comparison; do not try to give progress info.
            # We could use reblock-backshift-extract, but then stderr would get clobbered by progress info.
            backshift \
                --save-directory "$save_directory" \
                --produce-tar \
                --backup-id "$backup_id"
        else
            # This pops up a GUI for how long the backshift --produce-tar | tar --diff is going to take.
            gprog-backshift-extract \
                --save-directory "$save_directory" \
                --tar-format gnu \
                --backup-id "$backup_id"
        fi \
            | (
                # See https://www.gnu.org/software/tar/manual/html_section/tar_51.html for what's going on here.
                cd / && tar --show-transformed-names --strip-components="$components_to_strip" --diff 2>&1 | sed 's/^/   /'
            )
        ;;
    *)
        echo "$0: Internal error: (2) \$skip_tar_diff is not True or False: $skip_tar_diff" 1>&2
        exit 1
        ;;
esac