dsize

#!/usr/bin/env bash
# dsize – terminal tool that shows human-readable table of directory sizes, with mount point info.
# Author: Mitch Wilson - https://csimw.com
# GitHub repo: https://github.com/emo333/dsize
# Usage: source dsize              # runs in current directory
#        bash dsize /path          # optional path argument
#
# Requires: bash 4+, GNU coreutils (du, stat, find).

set -euo pipefail

# ── Configuration ────────────────────────────────────────────────────────────
readonly PADDING=2

# ANSI color thresholds (first match wins)
readonly COL_GREY=$'\e[38;5;244m'   # < 1 MB
readonly COL_GREEN=$'\e[38;5;40m'   # ≥ 1 MB
readonly COL_YELLOW=$'\e[38;5;220m' # ≥ 1 GB
readonly COL_ORANGE=$'\e[38;5;208m' # ≥ 5 GB
readonly COL_RED=$'\e[38;5;160m'    # ≥ 10 GB
readonly RESET=$'\e[0m'

# ── Helpers ──────────────────────────────────────────────────────────────────

die() {
  printf 'error: %s\n' "$*" >&2
  exit 1
}

# Human-readable bytes — pure shell arithmetic + one awk call for division.
human_bytes() {
  local b=$1
  ((b < 0)) && die "negative bytes: $b"
  if ((b >= 1073741824)); then
    awk "BEGIN{printf \"%.1fG\", $b/1073741824}"
  elif ((b >= 1048576)); then
    awk "BEGIN{printf \"%.1fM\", $b/1048576}"
  elif ((b >= 1024)); then
    awk "BEGIN{printf \"%.1fK\", $b/1024}"
  else
    printf '%d' "$b"
  fi
}

# Strip ANSI codes from a string (GNU sed).
strip_ansi() {
  printf '%s' "$1" | sed $'s/\x1b\[[0-9;]*[a-zA-Z]//g'
}

# Colorize a size value based on its magnitude.
size_colorize() {
  local val="$1" n u b=0
  # Extract numeric part and unit suffix.
  n="${val%%[^0-9.]*}"
  u="${val##*[0-9.]}"
  u="${u,,}" # lowercase (bash 4+)

  case "$u" in
  g) b=$(awk "BEGIN{printf \"%d\", $n * 1073741824}") ;;
  m) b=$(awk "BEGIN{printf \"%d\", $n * 1048576}") ;;
  k) b=$(awk "BEGIN{printf \"%d\", $n * 1024}") ;;
  *) b=${n:-0} ;;
  esac

  if ((b >= 10737418240)); then
    color="$COL_RED"
  elif ((b >= 5368709120)); then
    color="$COL_ORANGE"
  elif ((b >= 1073741824)); then
    color="$COL_YELLOW"
  elif ((b >= 1048576)); then
    color="$COL_GREEN"
  else
    color="$COL_GREY"
  fi

  printf '%s%s%s' "$color" "$val" "$RESET"
}

# ── Cleanup trap ─────────────────────────────────────────────────────────────
_TMPFILE=""
_cleanup() { [[ -n $_TMPFILE && -f $_TMPFILE ]] && rm -f "$_TMPFILE"; }
trap _cleanup EXIT INT TERM

# ── Main logic ───────────────────────────────────────────────────────────────

TARGET_DIR="${1:-.}"

[[ -d $TARGET_DIR ]] || die "'$TARGET_DIR' is not a directory"

cd -- "$TARGET_DIR"

_TMPFILE="$(mktemp)" || die "failed to create temp file"

# ── 1. Gather data (single du pass, dedup by inode) ────────────────────────
declare -A _seen_inode=()
declare -a szs=() fds=()
max_sz_width=0 max_fd_width=0
total_raw_bytes=0

# Count direct file sizes in . for the root entry (excludes subdirectories).
_root_file_bytes="$(find . -maxdepth 1 -type f \
  -exec stat --format='%s' {} + 2>/dev/null | awk '{s+=$1} END{print s+0}')"

# Process each du output line.
while IFS=$'\t' read -r size path; do
  # Normalise relative paths; skip the root "." entry (we compute it separately).
  case "$path" in
  .) continue ;;
  ./*) path="${path#./}" ;; # strip leading "./"
  esac

  # Deduplicate by real device:inode (handles bind mounts, symlinks).
  _real="$(readlink -f "$path" 2>/dev/null)" || continue
  _devino="$(stat --format='%d:%i' "$_real" 2>/dev/null)" || continue
  [[ -n "${_seen_inode[$_devino]+_}" ]] && continue
  _seen_inode["$_devino"]=1

  # Track visible widths (from raw human strings, before any coloring).
  ((${#size} > max_sz_width)) && max_sz_width=${#size}
  ((${#path} > max_fd_width)) && max_fd_width=${#path}

  szs+=("$size")
  fds+=("$path")

  # Accumulate raw bytes for total.
  _n="${size%%[^0-9.]*}"
  _u="${size##*[0-9.]}"
  _u="${_u,,}"
  case "$_u" in
  g) total_raw_bytes=$((total_raw_bytes + $(awk "BEGIN{printf \"%d\", $_n*1073741824}"))) ;;
  m) total_raw_bytes=$((total_raw_bytes + $(awk "BEGIN{printf \"%d\", $_n*1048576}"))) ;;
  k) total_raw_bytes=$((total_raw_bytes + $(awk "BEGIN{printf \"%d\", $_n*1024}"))) ;;
  *) total_raw_bytes=$((total_raw_bytes + _n)) ;;
  esac

done < <(du -h --max-depth=1 . 2>/dev/null | sort -rh)

# Add root entry (files only, no children).
_root_size="$(human_bytes "$_root_file_bytes")"
szs=("$_root_size" "${szs[@]}")
fds=("." "${fds[@]}")
((${#_root_size} > max_sz_width)) && max_sz_width=${#_root_size}

# Include root file bytes in the total.
total_raw_bytes=$((total_raw_bytes + _root_file_bytes))

total_human="$(human_bytes "$total_raw_bytes")"

[[ ${#szs[@]} -eq 0 ]] && die "no directories found in '$TARGET_DIR'"

# ── 2. Mount point info ─────────────────────────────────────────────────────
# df columns: Filesystem Size Used Avail Use% Mounted-on (with -B1)
read -r _mount_dev df_total_kb df_used_kb df_avail_kb _use_pct _mount_pt <<<"$(df -B1 . | tail -1)"
df_total_kb=${df_total_kb:-0}
[[ $df_total_kb =~ ^[0-9]+$ ]] || df_total_kb=0
df_used_kb=${df_used_kb:-0}
[[ $df_used_kb =~ ^[0-9]+$ ]] || df_used_kb=0
df_avail_kb=${df_avail_kb:-0}
[[ $df_avail_kb =~ ^[0-9]+$ ]] || df_avail_kb=0

if ((df_total_kb > 0)); then
  usage_pct=$((df_used_kb * 100 / df_total_kb))
else
  usage_pct=0
fi

mount_name="$(df . | tail -1 | awk '{print $1}')"
[[ -z $mount_name ]] && mount_name="unknown"

# Pre-format mount sizes so we don't call human_bytes inside the output loop.
mount_total_human="$(human_bytes "$df_total_kb")"
mount_used_human="$(human_bytes "$df_used_kb")"
mount_avail_human="$(human_bytes "$df_avail_kb")"

# ── 3. Compute table dimensions ─────────────────────────────────────────────
# Table 1 (directory sizes)
col_width=$((max_sz_width + PADDING))
fd_width=$((max_fd_width + PADDING))
line_width=$((col_width + fd_width + 7)) # "| " ... " |" + internal sep = 4 + 3
separator="$(printf '%*s' "$line_width" '' | tr ' ' '-')"

# Table 2 (mount point info)
m_sz_w=${#mount_total_human}
[[ ${#mount_used_human} -gt m_sz_w ]] && m_sz_w=${#mount_used_human}
[[ ${#mount_avail_human} -gt m_sz_w ]] && m_sz_w=${#mount_avail_human}
m_sz_w=$((m_sz_w + PADDING))
m_fd_w=12
m_line=$((m_sz_w + m_fd_w + 7))
mount_sep="$(printf '%*s' "$m_line" '' | tr ' ' '-')"

# Mount colour thresholds.
m_used_color="" m_avail_color=""
if ((usage_pct >= 90)); then
  m_used_color="$COL_RED"
  m_avail_color="$COL_GREY"
elif ((usage_pct >= 75)); then
  m_used_color=""
  m_avail_color="$COL_ORANGE"
else
  m_used_color=""
  m_avail_color=""
fi
if ((usage_pct >= 50 && usage_pct < 75)); then
  m_avail_color="$COL_YELLOW"
fi

# ── 4. Write output ────────────────────────────────────────────────────────
fmt() { printf "| %-*s | %-*s |\n" "$col_width" "$1" "$fd_width" "$2"; }
m_fmt() { printf "| %-*s | %-*s |\n" "$m_sz_w" "$1" "$m_fd_w" "$2"; }

{
  # ── Table 1: directory sizes ──
  echo "$separator"
  fmt Size Folder
  echo "$separator"

  for ((i = 0; i < ${#szs[@]}; i++)); do
    colored="$(size_colorize "${szs[$i]}")"
    visible="$(strip_ansi "$colored")"
    pad=$((col_width - ${#visible}))
    [[ $pad -lt 0 ]] && pad=0
    printf "| %s%*s | %-*s |\n" "$colored" "$pad" "" "$fd_width" "${fds[$i]}"
  done

  echo "$separator"
  tc="$(size_colorize "$total_human")"
  tv="$(strip_ansi "$tc")"
  pad=$((col_width - ${#tv}))
  [[ $pad -lt 0 ]] && pad=0
  printf "| %s%*s | %-*s |\n" "$tc" "$pad" "" "$fd_width" "Total:"
  echo "$separator"

  # ── Table 2: mount point info ──
  echo ""

  echo "$mount_sep"
  # Header spans full width of data lines (m_line chars total)
  printf "| %-$((m_line - 4))s |\n" "Mount: $mount_name"
  echo "$mount_sep"

  m_fmt "$mount_total_human" "Total"

  if [[ -n "$m_used_color" ]]; then
    printf "| %s%*s\e[0m | %-*s |\n" "$m_used_color" "$m_sz_w" "$mount_used_human" "$m_fd_w" "Used"
  else
    m_fmt "$mount_used_human" "Used"
  fi

  if [[ -n "$m_avail_color" ]]; then
    printf "| %s%*s\e[0m | %-*s |\n" "$m_avail_color" "$m_sz_w" "$mount_avail_human" "$m_fd_w" "Avail"
  else
    m_fmt "$mount_avail_human" "Avail"
  fi

  echo "$mount_sep"
} >"$_TMPFILE"

# ── 5. Display ───────────────────────────────────────────────────────────────
cat "$_TMPFILE"