backup-to-tar

#!/bin/bash

ADDITIONAL_TAR_ARGS_DIRECTIVE="TAR_ARGS="

function help() {
    echo "Backup the directories and files listed in ~/.backup.conf.
input file, i.e. ~/.backup.conf should be a list of absolute paths of directories and files
to backup.


Caches and version control directories are excluded from the backup. see -g flag to include '.git' directories.
    
    -i    Optional input file instead of ~/.backup.conf.
    -o    Optional place to output backup. Default is in /tmp
    -g    Optional include '.git' directories. Default behaviour is to exclude '.git' to save space under the assumption the contents is on a remote somewhere.
          The parent directory to '.git' is backed up regardless.

.backup.conf format:
- Lines beginning with '#' in ~/.backup.conf are treated as comments.
- Additional tar args can be added with i.e. '$ADDITIONAL_TAR_ARGS_DIRECTIVE --exclude='*.foo' --exclude'*.bar'
"
}

while getopts "hi:o:g" arg; do
    case $arg in
        h) # display Help
            help
            exit;;
        i) # optionl config file
            BACKUP_CONF=$OPTARG
            ;;
        o) # optional specify output
            BACKUP_PATH_ROOT=$OPTARG
            ;;
        g) # optional include .git
            INCLUDE_GIT_DIR=1
            ;;
        \j?) # Invalid arg
            echo "Error: Invalid arg"
            help
            exit;;
    esac
done

if [ -z $BACKUP_CONF ]; then
    BACKUP_CONF="$HOME/.backup.conf"
fi


if [ ! -z $BACKUP_PATH_ROOT ]; then
    BACKUP_PATH=$BACKUP_PATH_ROOT/backup-$(date -I)
    BACKUP_SINGLE_FILES_PATH=$BACKUP_PATH_ROOT/backup-$(date -I)-single-files
else
    BACKUP_PATH=/tmp/backup-$(date -I)
    BACKUP_SINGLE_FILES_PATH=/tmp/backup-$(date -I)-single-files
fi

if [[ -d $BACKUP_SINGLE_FILES_PATH ]]; then rm -r $BACKUP_SINGLE_FILES_PATH; fi  # remove from previous runs


ADDITIONAL_TAR_ARGS=$(grep $ADDITIONAL_TAR_ARGS_DIRECTIVE $BACKUP_CONF | sed "s#$ADDITIONAL_TAR_ARGS_DIRECTIVE##g" | grep -v '#')

if [[ $ADDITIONAL_TAR_ARGS != "" ]]; then
    echo "Using additional tar arguments: $ADDITIONAL_TAR_ARGS"
fi

TAR_ARGS="--exclude-backups --exclude-caches-all --exclude-vcs --exclude='.venv' --exclude='__pycache__'"
if [ -z $INCLUDE_GIT_DIR ]; then
    TAR_ARGS="$TAR_ARGS --exclude='.git'"
fi
TAR_ARGS="$TAR_ARGS $ADDITIONAL_TAR_ARGS"

echo "Using tar with arguments \"$TAR_ARGS\""


# contents of BACKUP_CONF expected to be list of paths, where each line is a path and nothing more

mkdir $BACKUP_PATH
while read -r path; do
    if [[ $path == "" ]]; then continue; fi  # ignore empty lines
    # Skip comments
    if [[ $(echo $path | awk '{printf $1}') == "#" ]]; then continue; fi
    # Skip additional args directive
    if [[ $(echo $path | awk '{printf $1}') == "$ADDITIONAL_TAR_ARGS_DIRECTIVE" ]]; then continue; fi

    # tar directories
    if [ -d $path ]; then
        echo "tarring : $path"
        echo
        # echo "tar -czvf $BACKUP_PATH/$(basename $path).tar.gz $TAR_ARGS $path"
        # tar is sketchy, and doesn't seem to work with TAR_ARGS macro expansion, so eval fixes this.
        eval "tar -czvf $BACKUP_PATH/$(basename $path).tar.gz $TAR_ARGS $path"
    # save individual files
    elif [ -f $path ]; then
        echo "copying : $path"
        echo
        if [[ ! -d $BACKUP_SINGLE_FILES_PATH ]]; then
            mkdir $BACKUP_SINGLE_FILES_PATH
        fi
        mkdir -p $BACKUP_SINGLE_FILES_PATH/$(dirname $path)
        cp $path $BACKUP_SINGLE_FILES_PATH/$path
    fi

done < $BACKUP_CONF

#  Tar individual files together
if [[ -d $BACKUP_SINGLE_FILES_PATH ]]; then
    tar -czvf $BACKUP_PATH/other-files.tar.gz $BACKUP_SINGLE_FILES_PATH
fi

echo "tarring the tars"
echo
tar -cvf $BACKUP_PATH.tar.gz $BACKUP_PATH  # Note: Not passing through gzip
echo ""
echo "Created backup at $BACKUP_PATH / $BACKUP_PATH.tar.gz"