Regalamiunsorriso/sync/promote-sync.sh

#!/bin/sh

set -eu
IFS=' 
	'

SCRIPT_NAME=$(basename "$0")

STAGE_ROOT=${SYNC_STAGE_ROOT:-/home/marco/regalamiunsorriso}
LIVE_ROOT=${SYNC_LIVE_ROOT:-/home/sites/regalamiunsorriso}
STAGE_USER=${SYNC_STAGE_USER:-marco}
INCOMING_ROOT=${SYNC_INCOMING_ROOT:-$STAGE_ROOT/incoming}
LOG_FILE=${SYNC_LOG_FILE:-$STAGE_ROOT/logs/promote-sync.log}
PERMS_DB=${SYNC_PERMS_DB:-$STAGE_ROOT/metadata/perms.jsonl}

MODE=
PROMOTE_EXECUTE=0
INIT_EXECUTE=0
ONLY_SCOPE=
RESOLVED_LIVE_ROOT=

UPDATED_COUNT=0
SKIPPED_COUNT=0
FAILED_COUNT=0

usage() {
  cat <<EOF
Usage:
  $SCRIPT_NAME --dry-run [--only <scope>] [--perms-db <path>] [--log <path>]
  $SCRIPT_NAME --apply [--only <scope>] [--perms-db <path>] [--log <path>]
  $SCRIPT_NAME --init-sync [--only <scope>] [--perms-db <path>] [--log <path>]
  $SCRIPT_NAME --init-sync --apply-init [--only <scope>] [--perms-db <path>] [--log <path>]

Modes:
  --dry-run       Preview staged file promotion into the live tree.
  --apply         Promote staged files into the live tree.
  --init-sync     Preview initial copy from live tree into staging.
  --apply-init    Execute the initial copy when used with --init-sync.

Options:
  --only <scope>      Limit processing to one scope: www, rus, WEB-INF/classes.
  --perms-db <path>   Metadata database location. Default: $PERMS_DB
  --log <path>        Log file location. Default: $LOG_FILE
  --stage-root <path> Override staging root. Default: $STAGE_ROOT
  --live-root <path>  Override live root. Default: $LIVE_ROOT
  --stage-user <user> Override staging owner. Default: $STAGE_USER
  --help              Show this help.

Environment overrides:
  SYNC_STAGE_ROOT, SYNC_LIVE_ROOT, SYNC_STAGE_USER, SYNC_INCOMING_ROOT,
  SYNC_LOG_FILE, SYNC_PERMS_DB
EOF
}

timestamp_utc() {
  date -u +"%Y-%m-%dT%H:%M:%SZ"
}

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

ensure_log_dir() {
  mkdir -p "$(dirname -- "$LOG_FILE")"
}

log_line() {
  ensure_log_dir
  printf '%s\n' "$1" | tee -a "$LOG_FILE"
}

stat_owner() {
  if stat -f '%Su' "$1" >/dev/null 2>&1; then
    stat -f '%Su' "$1"
  else
    stat -c '%U' "$1"
  fi
}

stat_group() {
  if stat -f '%Sg' "$1" >/dev/null 2>&1; then
    stat -f '%Sg' "$1"
  else
    stat -c '%G' "$1"
  fi
}

stat_mode() {
  _mode=

  if stat -f '%Lp' "$1" >/dev/null 2>&1; then
    _mode=$(stat -f '%Lp' "$1")
  else
    _mode=$(stat -c '%a' "$1")
  fi

  printf '%04d\n' "$_mode"
}

resolve_existing_path() {
  if command -v realpath >/dev/null 2>&1; then
    realpath "$1"
    return
  fi

  _dir_path=$(dirname -- "$1")
  _base_name=$(basename -- "$1")
  (
    cd "$_dir_path"
    printf '%s/%s\n' "$(pwd -P)" "$_base_name"
  )
}

is_under_root() {
  case "$1" in
    "$2"|"$2"/*)
      return 0
      ;;
    *)
      return 1
      ;;
  esac
}

require_root() {
  if [ "$(id -u)" -ne 0 ]; then
    fail "this mode must run under sudo or as root"
  fi
}

is_allowed_scope() {
  case "$1" in
    www|rus|WEB-INF/classes)
      return 0
      ;;
    *)
      return 1
      ;;
  esac
}

run_for_selected_scopes() {
  _callback=$1

  if [ -n "$ONLY_SCOPE" ]; then
    "$_callback" "$ONLY_SCOPE"
    return
  fi

  "$_callback" "www"
  "$_callback" "rus"
  "$_callback" "WEB-INF/classes"
}

is_excluded_relpath() {
  case "$1" in
    www/_img|www/_img/*|www/_news|www/_news/*|www/_tmp|www/_tmp/*|www/mypics|www/mypics/*|www/mypics-archivio|www/mypics-archivio/*|www/mypics2|www/mypics2/*)
      return 0
      ;;
    *)
      return 1
      ;;
  esac
}

is_safe_relpath() {
  case "$1" in
    ''|/*|../*|*/../*|*/..|*//*)
      return 1
      ;;
  esac

  case "$1" in
    */../*|*/..)
      return 1
      ;;
    *)
      return 0
      ;;
  esac
}

json_escape() {
  printf '%s' "$1" | sed 's/\\/\\\\/g; s/"/\\"/g'
}

sha256_file() {
  if command -v sha256 >/dev/null 2>&1; then
    sha256 -q "$1"
  elif command -v sha256sum >/dev/null 2>&1; then
    sha256sum "$1" | awk '{print $1}'
  else
    printf ''
  fi
}

db_backup_existing() {
  _metadata_dir=$(dirname -- "$PERMS_DB")
  mkdir -p "$_metadata_dir"

  if [ -f "$PERMS_DB" ]; then
    _backup_path="$PERMS_DB.$(date -u +%Y%m%dT%H%M%SZ).bak"
    cp "$PERMS_DB" "$_backup_path"
    log_line "$(timestamp_utc) INFO perms-db-backup path=$_backup_path"
  fi
}

db_append_record() {
  printf '{"relpath":"%s","owner":"%s","group":"%s","mode":"%s","type":"%s","sha256":"%s"}\n' \
    "$(json_escape "$1")" \
    "$(json_escape "$2")" \
    "$(json_escape "$3")" \
    "$(json_escape "$4")" \
    "$(json_escape "$5")" \
    "$(json_escape "${6:-}")" >> "$PERMS_DB"
}

db_lookup_line() {
  awk -v target="\"relpath\":\"$1\"" '
    index($0, target) {
      print
      found = 1
      exit
    }
    END {
      if (!found) {
        exit 1
      }
    }
  ' "$PERMS_DB"
}

db_extract_field() {
  printf '%s\n' "$1" | awk -F'"' -v lookup_key="$2" '
    {
      for (i = 2; i < NF; i += 4) {
        if ($i == lookup_key) {
          print $(i + 2)
          exit
        }
      }
    }
  '
}

user_exists() {
  id "$1" >/dev/null 2>&1
}

group_exists() {
  if command -v getent >/dev/null 2>&1; then
    getent group "$1" >/dev/null 2>&1
    return
  fi

  if command -v pw >/dev/null 2>&1; then
    pw groupshow "$1" >/dev/null 2>&1
    return
  fi

  return 0
}

record_live_metadata() {
  _live_path=$1
  _relpath=$2
  _entry_type=$3
  _owner=$(stat_owner "$_live_path")
  _group=$(stat_group "$_live_path")
  _mode=$(stat_mode "$_live_path")
  _checksum=

  if [ "$_entry_type" = "file" ]; then
    _checksum=$(sha256_file "$_live_path")
  fi

  db_append_record "$_relpath" "$_owner" "$_group" "$_mode" "$_entry_type" "$_checksum"
}

log_skip() {
  SKIPPED_COUNT=$((SKIPPED_COUNT + 1))
  log_line "$(timestamp_utc) SKIP $1 reason=$2"
}

log_failure() {
  FAILED_COUNT=$((FAILED_COUNT + 1))
  log_line "$(timestamp_utc) FAIL $1 reason=$2"
}

log_update() {
  UPDATED_COUNT=$((UPDATED_COUNT + 1))
  log_line "$(timestamp_utc) $1 $2 -> $3 OK"
}

validate_path_policy() {
  if ! is_safe_relpath "$1"; then
    return 1
  fi

  if is_excluded_relpath "$1"; then
    return 1
  fi

  case "$1" in
    RUS|RUS/*)
      return 1
      ;;
    *)
      return 0
      ;;
  esac
}

ensure_parent_directory_exists() {
  [ -d "$(dirname -- "$1")" ]
}

copy_live_entry_to_stage() {
  _live_path=$1
  _relpath=$2
  _stage_path=$3
  _entry_type=$4

  if [ "$_entry_type" = "dir" ]; then
    mkdir -p "$_stage_path"
    chown "$STAGE_USER" "$_stage_path"
    chmod 0755 "$_stage_path"
    log_update "INIT-DIR" "$_relpath" "$_stage_path"
    return
  fi

  mkdir -p "$(dirname -- "$_stage_path")"
  cp "$_live_path" "$_stage_path"
  chown "$STAGE_USER" "$_stage_path"
  chmod 0644 "$_stage_path"
  log_update "INIT" "$_relpath" "$_stage_path"
}

run_init_scope() {
  _scope=$1
  _live_scope_root="$LIVE_ROOT/$_scope"
  _stage_scope_root="$INCOMING_ROOT/$_scope"

  if [ ! -e "$_live_scope_root" ]; then
    log_skip "$_scope" "live-scope-missing"
    return
  fi

  _entry_list=$(mktemp "${TMPDIR:-/tmp}/promote-sync.init.XXXXXX")
  find "$_live_scope_root" \( -type d -o -type f -o -type l \) > "$_entry_list"

  while IFS= read -r _entry || [ -n "$_entry" ]; do
    _relpath=${_entry#"$LIVE_ROOT/"}

    if ! validate_path_policy "$_relpath"; then
      log_skip "$_relpath" "excluded-or-unsafe"
      continue
    fi

    if [ -L "$_entry" ]; then
      log_skip "$_relpath" "symlink-excluded"
      continue
    fi

    if ! is_under_root "$(resolve_existing_path "$_entry")" "$RESOLVED_LIVE_ROOT"; then
      log_failure "$_relpath" "outside-live-root"
      continue
    fi

    _stage_path="$INCOMING_ROOT/$_relpath"

    if [ -d "$_entry" ]; then
      if [ "$INIT_EXECUTE" -eq 1 ]; then
        record_live_metadata "$_entry" "$_relpath" "dir"
        copy_live_entry_to_stage "$_entry" "$_relpath" "$_stage_path" "dir"
      else
        log_line "$(timestamp_utc) INIT-DRY-RUN $_relpath -> $_stage_path"
      fi
      continue
    fi

    if [ "$INIT_EXECUTE" -eq 1 ]; then
      record_live_metadata "$_entry" "$_relpath" "file"
      copy_live_entry_to_stage "$_entry" "$_relpath" "$_stage_path" "file"
    else
      log_line "$(timestamp_utc) INIT-DRY-RUN $_relpath -> $_stage_path"
    fi
  done < "$_entry_list"

  rm -f "$_entry_list"

  if [ "$INIT_EXECUTE" -eq 1 ]; then
    mkdir -p "$_stage_scope_root"
    chown "$STAGE_USER" "$_stage_scope_root"
    chmod 0755 "$_stage_scope_root"
  fi
}

run_init_sync() {
  if [ "$INIT_EXECUTE" -eq 1 ]; then
    require_root
  fi

  RESOLVED_LIVE_ROOT=$(resolve_existing_path "$LIVE_ROOT")

  if [ "$INIT_EXECUTE" -eq 1 ]; then
    mkdir -p "$INCOMING_ROOT"
    db_backup_existing
    : > "$PERMS_DB"
  fi

  run_for_selected_scopes run_init_scope
}

apply_metadata_to_temp() {
  if ! user_exists "$2"; then
    fail "owner '$2' from permissions database does not exist"
  fi

  if ! group_exists "$3"; then
    fail "group '$3' from permissions database does not exist"
  fi

  chown "$2:$3" "$1"
  chmod "$4" "$1"
}

promote_staged_file() {
  _stage_file=$1
  _relpath=$2
  _dest_file=$3
  _resolved_dest=$(resolve_existing_path "$_dest_file")

  if ! is_under_root "$_resolved_dest" "$RESOLVED_LIVE_ROOT"; then
    log_failure "$_relpath" "outside-live-root"
    return
  fi

  if [ ! -f "$PERMS_DB" ]; then
    log_failure "$_relpath" "perms-db-missing"
    return
  fi

  if ! _db_line=$(db_lookup_line "$_relpath"); then
    log_skip "$_relpath" "destination-missing-perms"
    return
  fi

  _owner=$(db_extract_field "$_db_line" "owner")
  _group=$(db_extract_field "$_db_line" "group")
  _mode=$(db_extract_field "$_db_line" "mode")

  if [ "$PROMOTE_EXECUTE" -eq 0 ]; then
    log_line "$(timestamp_utc) DRY-RUN $_relpath -> $_dest_file owner=$_owner group=$_group mode=$_mode"
    return
  fi

  _tmp_file=$(mktemp "$(dirname -- "$_dest_file")/.promote-sync.XXXXXX")
  cp "$_stage_file" "$_tmp_file"
  apply_metadata_to_temp "$_tmp_file" "$_owner" "$_group" "$_mode"
  mv -f "$_tmp_file" "$_dest_file"
  log_update "APPLY" "$_relpath" "$_dest_file"
}

run_promote_scope() {
  _scope=$1
  _stage_scope_root="$INCOMING_ROOT/$_scope"

  if [ ! -d "$_stage_scope_root" ]; then
    log_skip "$_scope" "staging-scope-missing"
    return
  fi

  _entry_list=$(mktemp "${TMPDIR:-/tmp}/promote-sync.promote.XXXXXX")
  find "$_stage_scope_root" \( -type f -o -type l \) > "$_entry_list"

  while IFS= read -r _entry || [ -n "$_entry" ]; do
    _relpath=${_entry#"$INCOMING_ROOT/"}

    if ! validate_path_policy "$_relpath"; then
      log_skip "$_relpath" "excluded-or-unsafe"
      continue
    fi

    if [ -L "$_entry" ]; then
      log_skip "$_relpath" "symlink-excluded"
      continue
    fi

    _dest_file="$LIVE_ROOT/$_relpath"

    if [ ! -e "$_dest_file" ]; then
      log_skip "$_relpath" "destination-missing"
      continue
    fi

    if ! ensure_parent_directory_exists "$_dest_file"; then
      log_skip "$_relpath" "destination-parent-missing"
      continue
    fi

    promote_staged_file "$_entry" "$_relpath" "$_dest_file"
  done < "$_entry_list"

  rm -f "$_entry_list"
}

run_promote() {
  if [ "$PROMOTE_EXECUTE" -eq 1 ]; then
    require_root
  fi

  RESOLVED_LIVE_ROOT=$(resolve_existing_path "$LIVE_ROOT")
  run_for_selected_scopes run_promote_scope
}

print_summary() {
  log_line "$(timestamp_utc) SUMMARY updated=$UPDATED_COUNT skipped=$SKIPPED_COUNT failed=$FAILED_COUNT mode=${MODE:-unknown}"
}

parse_args() {
  while [ "$#" -gt 0 ]; do
    case "$1" in
      --dry-run)
        [ -z "$MODE" ] || fail "choose only one primary mode"
        MODE=promote
        PROMOTE_EXECUTE=0
        ;;
      --apply)
        [ -z "$MODE" ] || fail "choose only one primary mode"
        MODE=promote
        PROMOTE_EXECUTE=1
        ;;
      --init-sync)
        [ -z "$MODE" ] || fail "choose only one primary mode"
        MODE=init
        ;;
      --apply-init)
        INIT_EXECUTE=1
        ;;
      --only)
        shift
        [ "$#" -gt 0 ] || fail "--only requires a scope"
        ONLY_SCOPE=$1
        is_allowed_scope "$ONLY_SCOPE" || fail "unsupported scope '$ONLY_SCOPE'"
        ;;
      --perms-db)
        shift
        [ "$#" -gt 0 ] || fail "--perms-db requires a path"
        PERMS_DB=$1
        ;;
      --log)
        shift
        [ "$#" -gt 0 ] || fail "--log requires a path"
        LOG_FILE=$1
        ;;
      --stage-root)
        shift
        [ "$#" -gt 0 ] || fail "--stage-root requires a path"
        STAGE_ROOT=$1
        INCOMING_ROOT=$STAGE_ROOT/incoming
        ;;
      --live-root)
        shift
        [ "$#" -gt 0 ] || fail "--live-root requires a path"
        LIVE_ROOT=$1
        ;;
      --stage-user)
        shift
        [ "$#" -gt 0 ] || fail "--stage-user requires a user"
        STAGE_USER=$1
        ;;
      --help|-h)
        usage
        exit 0
        ;;
      *)
        fail "unknown argument: $1"
        ;;
    esac
    shift
  done
}

main() {
  parse_args "$@"

  [ -n "$MODE" ] || fail "choose one of --dry-run, --apply, or --init-sync"

  if [ "$INIT_EXECUTE" -eq 1 ] && [ "$MODE" != "init" ]; then
    fail "--apply-init requires --init-sync"
  fi

  if [ "$MODE" = "init" ] && [ "$INIT_EXECUTE" -eq 0 ]; then
    log_line "$(timestamp_utc) INFO init-sync-preview only; add --apply-init to write staging files and permissions db"
  fi

  case "$MODE" in
    init)
      run_init_sync
      ;;
    promote)
      run_promote
      ;;
    *)
      fail "unsupported mode '$MODE'"
      ;;
  esac

  print_summary
}

main "$@"