tools/hf-backup.sh

#!/bin/sh
# shellcheck shell=sh

set -e

TS(){ date "+%Y-%m-%d %H:%M:%S"; }
log(){ echo "[$(TS)] [hf-backup] $*" >&2; }  # 日志走 stderr，避免污染命令替换

HF_TOKEN="${HF_TOKEN:-}"
DATASET_ID="${DATASET_ID:-}"

# 备份配置（相对路径更稳，和 -C / 配合）
SYNC_INTERVAL="${SYNC_INTERVAL:-3600}"
MAX_BACKUPS="${MAX_BACKUPS:-10}"
BACKUP_PREFIX="${BACKUP_PREFIX:-crs_backup}"

# 建议默认：数据目录 +（如有）redis 数据目录
BACKUP_PATHS="${BACKUP_PATHS:-app/data,app/data/redis}"
# 公开/半公开 dataset 可以更激进地排除敏感文件
BACKUP_EXCLUDE="${BACKUP_EXCLUDE:-app/data/init.json,app/data/users.json,*.tmp,*.cache,**/.DS_Store}"

HF_PY="/app/tools/hf_backup.py"
TMP_DIR="/tmp/crs_backup"
mkdir -p "${TMP_DIR}"

need_hf() {
  if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
    log "HF_TOKEN/DATASET_ID 缺失，无法使用备份功能"; exit 1
  fi
}

mk_archive() {
  ts="$(date +%Y%m%d_%H%M%S)"
  archive="${TMP_DIR}/${BACKUP_PREFIX}_${ts}.tar.gz"

  # 构造 tar 参数（安全，不受空格/通配影响）
  set -- -czf "$archive"

  # --exclude
  IFS=','; for x in $BACKUP_EXCLUDE; do
    [ -z "$x" ] && continue
    case "$x" in /*) x=${x#/} ;; esac
    set -- "$@" --exclude="$x"
  done; unset IFS

  # 切到根目录
  set -- "$@" -C /

  # 打包路径
  IFS=','; for p in $BACKUP_PATHS; do
    [ -z "$p" ] && continue
    case "$p" in /*) p=${p#/} ;; esac
    set -- "$@" "$p"
  done; unset IFS

  log "创建归档：$archive"
  tar "$@"
  echo "$archive"
}

cmd_upload_once() {
  need_hf
  archive="$(mk_archive)"          # 只捕获 stdout 的“文件路径”这一行
  [ -f "$archive" ] || { log "归档不存在：$archive"; return 1; }
  log "上传：$archive"
  python3 "$HF_PY" upload \
    --token "$HF_TOKEN" \
    --repo  "$DATASET_ID" \
    --file  "$archive" \
    --prefix "$BACKUP_PREFIX" \
    --max   "$MAX_BACKUPS"
  rm -f "$archive" || true
  log "上传完成"
}

cmd_restore() {
  need_hf
  out="$(python3 "$HF_PY" restore --token "$HF_TOKEN" --repo "$DATASET_ID" --prefix "$BACKUP_PREFIX" 2>/dev/null || true)"
  if [ -z "$out" ]; then
    log "未找到可恢复的备份，跳过。"
    exit 0
  fi
  latest_file="$out"
  if [ ! -f "$latest_file" ]; then
    log "本地文件不存在：$latest_file"; exit 1
  fi
  log "解压：$latest_file -> /"
  tar -xzf "$latest_file" -C /
  log "恢复完成"
}

cmd_daemon() {
  need_hf
  while :; do
    log "开始周期备份（间隔 ${SYNC_INTERVAL}s，前缀 ${BACKUP_PREFIX}，保留 ${MAX_BACKUPS}）"
    if ! cmd_upload_once; then
      log "本轮备份失败（忽略，等待下次循环）"
    fi
    sleep "$SYNC_INTERVAL"
  done
}

case "${1:-}" in
  restore) cmd_restore ;;
  once)    cmd_upload_once ;;
  daemon)  cmd_daemon ;;
  *) echo "用法: $0 {restore|once|daemon}" >&2; exit 1 ;;
esac