HartOMat/scripts/repo_hygiene.sh

#!/usr/bin/env bash
set -euo pipefail

APPLY=0
DEEP=0

usage() {
  cat <<'EOF'
Usage:
  ./scripts/repo_hygiene.sh [options]

Options:
  --apply       Delete the reported hygiene artifacts.
  --deep        Also remove heavy local environments (`frontend/node_modules`, `backend/.venv`).
  -h, --help    Show this help.

Behavior:
  Without --apply, the script only reports what it would clean.
  The default cleanup is conservative and removes debug/output/cache artifacts only.

Examples:
  ./scripts/repo_hygiene.sh
  ./scripts/repo_hygiene.sh --apply
  ./scripts/repo_hygiene.sh --apply --deep
EOF
}

log() {
  printf '\n[%s] %s\n' "$(date '+%H:%M:%S')" "$*"
}

warn() {
  printf '[warn] %s\n' "$*" >&2
}

ownership_candidate_find_expr() {
  cat <<'EOF'
find . \
  \( -path './.git' -o -path './backend/.venv' -o -path './frontend/node_modules' \) -prune -o \
  \( \
    -path './tmp' -o \
    -path './frontend/dist' -o \
    -path './backend/.pytest_cache' -o \
    -path './backend/celerybeat-schedule' -o \
    -type d -name '__pycache__' -o \
    -type f \( -name '*.pyc' -o -name '*.pyo' -o -name 'core' \) \
  \) \
  -print
EOF
}

repair_backend_ownership_with_docker() {
  if ! command -v docker >/dev/null 2>&1; then
    return 1
  fi
  if ! docker info >/dev/null 2>&1; then
    return 1
  fi

  log "Repairing backend artifact ownership through Docker"
  docker run --rm \
    -v "$REPO_ROOT/backend:/target" \
    alpine:3.20 \
    sh -lc "chown -R $(id -u):$(id -g) /target"
}

repair_repo_ownership_with_docker() {
  if ! command -v docker >/dev/null 2>&1; then
    return 1
  fi
  if ! docker info >/dev/null 2>&1; then
    return 1
  fi

  log "Repairing generated artifact ownership across the repository through Docker"
  docker run --rm \
    -v "$REPO_ROOT:/target" \
    alpine:3.20 \
    sh -lc "
      find /target \\
        \\( -path /target/.git -o -path /target/frontend/node_modules -o -path /target/backend/.venv \\) -prune -o \\
        \\( -type d -name __pycache__ -o -type f \\( -name '*.pyc' -o -name '*.pyo' -o -name 'celerybeat-schedule' -o -name 'core' \\) \\) \\
        -exec chown $(id -u):$(id -g) {} +
    "
}

ensure_repo_root() {
  if ! REPO_ROOT="$(git rev-parse --show-toplevel 2>/dev/null)"; then
    echo "This script must be run inside the git repository." >&2
    exit 1
  fi
  cd "$REPO_ROOT"
}

parse_args() {
  while [ "$#" -gt 0 ]; do
    case "$1" in
      --apply)
        APPLY=1
        ;;
      --deep)
        DEEP=1
        ;;
      -h|--help)
        usage
        exit 0
        ;;
      *)
        echo "Unknown option: $1" >&2
        usage >&2
        exit 1
        ;;
    esac
    shift
  done
}

print_candidates() {
  local label="$1"
  shift
  local cmd=("$@")
  log "$label"
  "${cmd[@]}"
}

print_non_writable_artifacts() {
  log "Generated artifacts not owned by the current user"
  bash -lc "$(ownership_candidate_find_expr) | while IFS= read -r path; do [ -e \"\$path\" ] || continue; if [ ! -O \"\$path\" ]; then printf '%s\n' \"\$path\"; fi; done | sort"
}

repo_find_expr() {
  cat <<'EOF'
find . \
  \( -path './.git' -o -path './backend/.venv' -o -path './frontend/node_modules' \) -prune -o
EOF
}

cleanup_safe_targets() {
  local permission_failures=0
  local target
  PERMISSION_FAILURE_PATHS=()
  local targets=(
    tmp
    frontend/dist
    backend/.pytest_cache
    backend/celerybeat-schedule
  )

  for target in "${targets[@]}"; do
    [ -e "$target" ] || continue
    if ! rm -rf "$target" 2>/dev/null; then
      warn "Could not remove $target. Ownership or permissions need to be fixed first."
      PERMISSION_FAILURE_PATHS+=("$target")
      permission_failures=1
    fi
  done

  while IFS= read -r target; do
    if ! rm -rf "$target" 2>/dev/null; then
      warn "Could not remove $target. Ownership or permissions need to be fixed first."
      PERMISSION_FAILURE_PATHS+=("$target")
      permission_failures=1
    fi
  done < <(find . -type d -name __pycache__ -print)

  while IFS= read -r target; do
    if ! rm -f "$target" 2>/dev/null; then
      warn "Could not remove $target. Ownership or permissions need to be fixed first."
      PERMISSION_FAILURE_PATHS+=("$target")
      permission_failures=1
    fi
  done < <(find . -type f \( -name '*.pyc' -o -name '*.pyo' \) -print)

  while IFS= read -r target; do
    if ! rm -f "$target" 2>/dev/null; then
      warn "Could not remove $target. Ownership or permissions need to be fixed first."
      PERMISSION_FAILURE_PATHS+=("$target")
      permission_failures=1
    fi
  done < <(find . -type f -name 'core' -print)

  return "$permission_failures"
}

cleanup_deep_targets() {
  rm -rf frontend/node_modules
  rm -rf backend/.venv
}

main() {
  parse_args "$@"
  ensure_repo_root

  print_candidates \
    "Core debug and cache artifacts" \
    bash -lc "find tmp frontend/dist backend/.pytest_cache -mindepth 0 -maxdepth 0 2>/dev/null | sort"

  print_candidates \
    "Python cache directories" \
    bash -lc "$(repo_find_expr) -type d -name '__pycache__' -print | sort"

  print_candidates \
    "Python bytecode files" \
    bash -lc "$(repo_find_expr) -type f \\( -name '*.pyc' -o -name '*.pyo' \\) -print | sort | sed -n '1,200p'"

  print_non_writable_artifacts

  if [ "$DEEP" -eq 1 ]; then
    print_candidates \
      "Heavy local environments" \
      bash -lc "find frontend/node_modules backend/.venv -mindepth 0 -maxdepth 0 2>/dev/null | sort"
  fi

  if [ "$APPLY" -eq 0 ]; then
    log "Dry run only. Re-run with --apply to delete the reported artifacts."
    if [ "$DEEP" -eq 1 ]; then
      echo "Deep mode is enabled: node_modules and backend/.venv would also be removed."
    fi
    exit 0
  fi

  log "Deleting conservative hygiene artifacts"
  CLEANUP_PERMISSION_FAILURES=0
  if ! cleanup_safe_targets; then
    CLEANUP_PERMISSION_FAILURES=1
  fi

  if [ "$DEEP" -eq 1 ]; then
    log "Deleting deep local environments"
    cleanup_deep_targets
  fi

  if [ "$CLEANUP_PERMISSION_FAILURES" -eq 1 ]; then
    if repair_repo_ownership_with_docker || repair_backend_ownership_with_docker; then
      log "Retrying conservative hygiene cleanup after ownership repair"
      CLEANUP_PERMISSION_FAILURES=0
      cleanup_safe_targets || CLEANUP_PERMISSION_FAILURES=1
    fi
  fi

  if [ "$CLEANUP_PERMISSION_FAILURES" -eq 1 ]; then
    log "Some artifacts could not be removed because they are not writable by the current user."
    printf 'Blocked paths:\n'
    printf '  %s\n' "${PERMISSION_FAILURE_PATHS[@]}" | sort -u
    echo
    echo "Run this once, then re-run the hygiene script:"
    echo "  $(ownership_candidate_find_expr | sed 's/-print$/-print0/' | tr '\n' ' ' | sed 's/  */ /g') | sudo xargs -0 chown -R \"\$USER:\$USER\""
  fi

  log "Remaining git status"
  git status --short
}

main "$@"