Analyse a given Syncthing versions folder and display the disk space used by files grouped by dates.

syncthing__versions_du.sh

#!/bin/bash
echo Analyse a given Syncthing versions folder and display the disk space used by files grouped by dates.
targetDir=$( realpath "$1" )
echo Target directory: "'"$targetDir"'"
echo Start time: $(date)

if ! [[ -d "$targetDir" ]]; then
  echo "Unable to access target folder: '$targetDir'"
  exit 1
fi

human_readable_size() {
  local size="$1"
  local units=("B" "KB" "MB" "GB" "TB" "PB" "EB" "ZB" "YB")
  local i=0

  if [[ "$size" -lt 1024 ]]; then
    echo "$size ${units[0]}"
    return
  fi

  while [[ "$size" -ge 1024 ]]; do
    size=$((size / 1024))
    ((i++))
  done

  echo "$size ${units[$i]}"
}

# Favor using '/tmp' first, or what the environment exposes instead.
# This is because EntWare uses '/opt/tmp' but I'd rather use '/tmp' first...
if [[ -d "/tmp" ]]; then
  TMPDIR=/tmp
else
  TMPDIR=${TMPDIR:-${TEMP:-${TMP:-/tmp}}}
fi

tmpDir=$TMPDIR/syncthing_versions_du
mkdir "$tmpDir" 2> /dev/null

if ! [[ -d "$tmpDir" ]]; then
  echo Unable to create cache folder: "$tmpDir"
  exit 2
fi

# Pattern: file name with "~dddddd-dddddd" regardless of the extension
# We could also check that the name ends with an extension `\(\.[^.+]+\)?` but I'm not sure how to handle files with multiple \. (or how Syncthing will name them)
versionFiles=$( find "$targetDir" -type f -regex '.*~[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9]-[0-9][0-9][0-9][0-9][0-9][0-9].*' )

versionDates=$( echo "$versionFiles" | sed 's/.*~\([0-9]\{8\}\).*/\1/g' | sort | uniq )

# DEBUG: only show 3 first versions
# versionDates=$( echo "$versionDates" | head -3 )

versionDateCount=$(echo "$versionDates" | wc -l)
echo "Disk usage by version date ($versionDateCount items):"

results=()
cachedResultNb=0

while read versionDate; do
  # versionDiskUsage=$( find . -type f -name '*~'"$versionDate"'-*' | xargs -d '\n' du -bc | tail -1 | sed 's/ *total//' )

  versionFilesByDate=$( echo "$versionFiles" | grep '.*~'"$versionDate"'-.*' )
  versionFilesByDateHash=$(echo "$versionFilesByDate" | md5sum | sed 's/\([^ ]\+\) .*/\1/g' )

  versionDiskUsageCache="$tmpDir/diskUsageByDate-${versionDate}__${versionFilesByDateHash}.txt"

  if [[ -f "$versionDiskUsageCache" ]]; then
    #DEBUG
    # echo "Cached result used: $versionDiskUsageCache"
    usage=$( cat "$versionDiskUsageCache" )
    echo "$usage"
    results+=( "$usage" )
    cachedResultNb=$(($cachedResultNb + 1))
  else
    versionDiskUsage=$( 
      echo "$versionFilesByDate" \
      | xargs -d '\n' du -bc \
      | tail -1 \
      | sed 's/\([0-9]\+\).*/\1/g'
    )

    usage="$versionDate: $versionDiskUsage: $(human_readable_size $versionDiskUsage)"
    echo "$usage" | tee "$versionDiskUsageCache"
    results+=( "$usage" )
  fi
done < <( echo "$versionDates" )

echo
echo "Found $cachedResultNb cached results."

# Echo the array with newline separators
IFS=$'\n' # Set the Internal Field Separator to a newline
echo
echo "Disk usage per version date, in descending order ($versionDateCount items):"
# Warning: ensure that the " , " separator has a space so that `sort -rn` can parse the byte size as a number, instead of doing an alphabetical sort
echo "${results[*]}" | awk -F': ' '{print $2 " , " $1 " , " $3}' \
| sort -rn \
| awk -F' , ' '{print $2 ": " $1 " ("$3")"}'

echo
echo Tips to view files for a given date:
echo "  " find "'"$targetDir"'" "-type f -name '*~YYYYMMDD-*'"
echo "  " find "'"$targetDir"'" "-type f -name '*~YYYYMMDD-*' | sort | xargs -d '\n' du -bch"
echo
echo End time: $(date)