logdirsizes

#!/usr/bin/env bash
set -eu

SCRIPT=$(readlink -m $(type -p $0))
SCRIPTDIR=$(dirname $SCRIPT)

usage() {
    cat <<EOF
    Usage:

        $(basename $0) <dir> [depth]

EOF
}

logfilename() {
    local dir=$1
    local depth=$2
    local logdir=$SCRIPTDIR/_data/logdirsizes
    local datestamp=$(date +"%Y%m%d")
    local prefix=$(echo $dir | sed 's,\/,_,g' | sed 's,_$,,' | sed 's,^_,,')
    local logfile="${prefix}-dirsizes-$depth-${datestamp}.csv"
    mkdir -p $logdir
    echo "$logdir/$logfile"
}

[[ $# -lt 1 || $1 == "-h" || $1 == "--help" ]] && { usage; exit 1; }

depth=3
[ -n "${2-}" ] && depth=$2

remote=false
IFS=":" read server remotepath <<< "$1"
if [ -n "${remotepath-}" ]; then
    remote=true
    dir=$(ssh $server "readlink -f $remotepath")
else
    dir=$(readlink -f "$1")
fi

read -d '' cmd << EOF || true
du --time -b --max-depth $depth $dir |
    awk  '{ print strftime("%Y-%m-%d %H:%M:%S"),\$1,\$1/1024/1024/1024,\$4,\$2" "\$3 }' OFS='|' |
    head -n -1
EOF
echo $cmd

if $remote; then
    results=$(ssh $server "$cmd")
else
    results=$(eval "$cmd")
fi
echo "results: "
echo $results

log=$(logfilename $dir $depth)
if [ -n "${results-}" ]; then
    echo "Datetime, Size, SizeG, Directory, Last Modified" > $log
    echo -e "$results" | python $SCRIPTDIR/dsv2csv.py >> $log
    echo "Made $log"
else
    >&2 echo "No output found for $dir"
    touch $log
fi