#!/usr/bin/env bash
# Backup the home directory to a remote SSH location using rsync.
#
# Excludes large, regenerable, or already-replicated data (caches, trash,
# build artifacts, git checkouts, syncthing, media libraries, VM images,
# etc.). See the EXCLUDES array below for the full list.
#
# Usage:
#   home-backup                 # Run the backup (real run)
#   home-backup -n              # Dry run (show what would transfer)
#   home-backup -d user@host:/path/backup   # Override destination
#
# The destination can also be set via the BACKUP_DEST environment variable.

set -euo pipefail

declare -r PROGRAM=${0##*/}

# Remote SSH destination. Defaults to the encrypted ZFS dataset on f2
# (zdata/enc/earth-backup, inherits aes-256-gcm encryption from zdata/enc).
# Override with the BACKUP_DEST env var or the -d flag.
declare DEST=${BACKUP_DEST:-"paul@f2.lan:/data/enc/earth-backup"}

# Source directory (trailing slash matters for rsync: copy contents of $HOME).
declare -r SRC="$HOME/"

declare DRY_RUN=""

# Paths/patterns to exclude from the backup. Patterns are relative to $HOME.
declare -ra EXCLUDES=(
  # --- Explicitly excluded by request ---
  "/git/"
  "/syncthing/"
  "/Pictures/"
  "/Music/"
  "/.gradle/"
  "/.thunderbird/"

  # --- Caches / trash / regenerable data ---
  "/.cache/"
  "/.local/share/Trash/"
  "/.local/share/baloo/"
  "/.local/share/containers/"
  "/.local/share/gnome-boxes/"
  "/.npm/"
  "/go/"
  "/.cargo/"
  "/.bun/"
  "/.android/"

  # --- Common regenerable junk anywhere in the tree ---
  "node_modules/"
  ".cache/"
  "*.tmp"
  "/.var/app/*/cache/"

  # --- Runtime / volatile ---
  "/.gvfs/"
  "/.dbus/"
)

log() {
  printf '%s\n' "$*"
}

error() {
  log "$*" >&2
}

usage() {
  cat <<EOF
Usage: $PROGRAM [-n] [-d DEST]

  -n           Dry run (show what would be transferred, transfer nothing)
  -d DEST      Remote SSH destination (user@host:/path), overrides BACKUP_DEST
  -h           Show this help

Destination may also be set via the BACKUP_DEST environment variable
or by editing the DEST variable at the top of this script.
EOF
}

main() {
  while getopts ':nd:h' opt; do
    case "$opt" in
      n) DRY_RUN="--dry-run" ;;
      d) DEST="$OPTARG" ;;
      h) usage; exit 0 ;;
      :) error "Option -$OPTARG requires an argument."; usage; exit 2 ;;
      ?) error "Unknown option: -$OPTARG"; usage; exit 2 ;;
    esac
  done

  if [[ "$DEST" == CHANGEME_* || -z "$DEST" ]]; then
    error "No backup destination set."
    error "Set BACKUP_DEST, pass -d user@host:/path, or edit DEST in $PROGRAM."
    exit 1
  fi

  # Build the rsync exclude arguments.
  local -a exclude_args=()
  local pattern
  for pattern in "${EXCLUDES[@]}"; do
    exclude_args+=("--exclude=$pattern")
  done

  log "Source:      $SRC"
  log "Destination: $DEST"
  [[ -n "$DRY_RUN" ]] && log "Mode:        DRY RUN (no data transferred)"
  log ""

  # rsync flags:
  #   -a  archive (recursive, perms, times, symlinks, etc.)
  #   -H  preserve hard links
  #   -A  preserve ACLs
  #   -X  preserve extended attributes
  #   --numeric-ids        don't map uid/gid by name on the remote
  #   --partial --progress resumable transfers with progress output
  #   --delete             mirror: remove remote files that no longer exist locally
  #   --delete-excluded    also remove previously-backed-up excluded paths
  #   --human-readable     human-readable sizes
  #   -e ssh               transport over SSH
  rsync \
    -aHAX \
    --numeric-ids \
    --partial --progress \
    --delete --delete-excluded \
    --human-readable \
    ${DRY_RUN:+$DRY_RUN} \
    -e ssh \
    "${exclude_args[@]}" \
    "$SRC" \
    "$DEST"

  log ""
  log "Backup complete."
}

main "$@"