CoolFace
Apppublic

alice0321/nyx

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes
sync_data.sh362 linesDownload Raw Back to root
1#!/bin/sh2 3# 添加更详细的日志4log_info() {5    echo "[$(date '+%Y-%m-%d %H:%M:%S')] INFO: $1"6}7 8log_error() {9    echo "[$(date '+%Y-%m-%d %H:%M:%S')] ERROR: $1"10}11 12# 检查环境变量13if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then14    log_error "未启用备份功能 - 缺少HF_TOKEN或DATASET_ID环境变量"15    log_info "HF_TOKEN=${HF_TOKEN:0:3}... DATASET_ID=${DATASET_ID}"16    exit 017fi18 19# 创建临时目录20TEMP_DIR="/tmp/sillytavern_backup"21DATA_DIR="/home/node/app/data"22 23# 确保目录存在并有正确权限24mkdir -p $TEMP_DIR25chmod -R 777 $TEMP_DIR26mkdir -p $DATA_DIR27chmod -R 777 $DATA_DIR28 29log_info "临时目录: $TEMP_DIR"30log_info "数据目录: $DATA_DIR"31log_info "HF_TOKEN: ${HF_TOKEN:0:5}..."32log_info "DATASET_ID: $DATASET_ID"33 34# 安装python和huggingface_hub35if ! command -v python3 > /dev/null 2>&1; then36    log_info "正在安装Python..."37    apk add --no-cache python3 py3-pip38else39    log_info "Python3已安装: $(python3 --version)"40fi41 42# 确保pip已安装43if ! command -v pip3 > /dev/null 2>&1; then44    log_info "正在安装pip..."45    apk add --no-cache py3-pip46else47    log_info "Pip3已安装: $(pip3 --version)"48fi49 50# 安装或更新huggingface_hub51log_info "正在安装/更新huggingface_hub..."52pip3 install --no-cache-dir --upgrade huggingface_hub53log_info "huggingface_hub安装完成"54 55# 测试huggingface_hub是否正常工作56if ! python3 -c "import huggingface_hub; print('huggingface_hub版本:', huggingface_hub.__version__)" > /dev/null 2>&1; then57    log_error "huggingface_hub导入失败,正在重试安装..."58    pip3 install --no-cache-dir huggingface_hub59fi60 61# 测试权限是否正常62touch "${TEMP_DIR}/test_file" && rm "${TEMP_DIR}/test_file"63if [ $? -ne 0 ]; then64    log_error "临时目录权限测试失败,正在修复权限..."65    chmod -R 777 $TEMP_DIR66fi67 68# 测试与HuggingFace API的连接69log_info "正在测试与HuggingFace API的连接..."70python3 -c "71from huggingface_hub import HfApi72try:73    api = HfApi(token='$HF_TOKEN')74    user_info = api.whoami()75    print(f'成功连接到HuggingFace API,用户: {user_info}')76except Exception as e:77    print(f'连接HuggingFace API失败: {str(e)}')78    exit(1)79"80if [ $? -ne 0 ]; then81    log_error "HuggingFace API连接测试失败,请检查令牌是否有效"82else83    log_info "HuggingFace API连接测试成功"84fi85 86# 生成唯一的测试文件名87TEST_FILE_NAME="test_file_$(date +%s)"88 89# 测试数据集权限90log_info "正在测试Dataset权限..."91python3 -c "92from huggingface_hub import HfApi93try:94    api = HfApi(token='$HF_TOKEN')95    96    # 创建本地测试文件97    with open('$TEMP_DIR/test_file', 'w') as f:98        f.write('test')99    100    # 上传测试文件101    test_file_name = '$TEST_FILE_NAME'102    print(f'正在上传测试文件: {test_file_name}')103    104    api.upload_file(105        path_or_fileobj='$TEMP_DIR/test_file',106        path_in_repo=test_file_name,107        repo_id='$DATASET_ID',108        repo_type='dataset'109    )110    print('成功上传测试文件到Dataset')111    112    # 删除已上传的测试文件113    print('正在删除测试文件...')114    api.delete_file(115        path_in_repo=test_file_name,116        repo_id='$DATASET_ID',117        repo_type='dataset'118    )119    print('已成功删除测试文件')120    121except Exception as e:122    print(f'Dataset权限测试失败: {str(e)}')123    exit(1)124"125if [ $? -ne 0 ]; then126    log_error "Dataset权限测试失败,请检查DATASET_ID是否正确且有写入权限"127else128    log_info "Dataset权限测试成功,测试文件已清理"129fi130 131# 确保本地测试文件被删除132rm -f "$TEMP_DIR/test_file"133 134# 上传备份135upload_backup() {136    file_path="$1"137    file_name="$2"138    139    if [ ! -f "$file_path" ]; then140        log_error "备份文件不存在: $file_path"141        return 1142    fi143    144    log_info "开始上传备份: $file_name ($(du -h $file_path | cut -f1))"145    146    python3 -c "147from huggingface_hub import HfApi148import sys149import os150import time151 152def manage_backups(api, repo_id, max_files=10):153    try:154        files = api.list_repo_files(repo_id=repo_id, repo_type='dataset')155        backup_files = [f for f in files if f.startswith('sillytavern_backup_') and f.endswith('.tar.gz')]156        backup_files.sort()157        158        if len(backup_files) >= max_files:159            files_to_delete = backup_files[:(len(backup_files) - max_files + 1)]160            for file_to_delete in files_to_delete:161                try:162                    api.delete_file(path_in_repo=file_to_delete, repo_id=repo_id, repo_type='dataset')163                    print(f'已删除旧备份: {file_to_delete}')164                except Exception as e:165                    print(f'删除 {file_to_delete} 时出错: {str(e)}')166    except Exception as e:167        print(f'管理备份文件时出错: {str(e)}')168 169token='$HF_TOKEN'170repo_id='$DATASET_ID'171 172try:173    api = HfApi(token=token)174    175    # 检查文件大小176    file_size = os.path.getsize('$file_path')177    print(f'备份文件大小: {file_size / (1024*1024):.2f} MB')178 179    # 确认Dataset存在180    try:181        dataset_info = api.dataset_info(repo_id=repo_id)182        print(f'Dataset信息: {dataset_info.id}')183    except Exception as e:184        print(f'获取Dataset信息失败: {str(e)}')185    186    start_time = time.time()187    print(f'开始上传: {start_time}')188    189    # 上传文件190    api.upload_file(191        path_or_fileobj='$file_path',192        path_in_repo='$file_name',193        repo_id=repo_id,194        repo_type='dataset'195    )196    197    end_time = time.time()198    print(f'上传完成,耗时: {end_time - start_time:.2f} 秒')199    print(f'成功上传 $file_name')200    201    # 管理备份202    manage_backups(api, repo_id)203except Exception as e:204    print(f'上传文件时出错: {str(e)}')205    sys.exit(1)206"207    if [ $? -ne 0 ]; then208        log_error "备份上传失败"209        return 1210    else211        log_info "备份上传成功"212        return 0213    fi214}215 216# 下载最新备份217download_latest_backup() {218    log_info "开始下载最新备份..."219    220    python3 -c "221from huggingface_hub import HfApi222import sys223import os224import tarfile225import tempfile226import time227 228try:229    api = HfApi(token='$HF_TOKEN')230    print('已创建API实例')231    232    # 列出仓库文件233    try:234        files = api.list_repo_files(repo_id='$DATASET_ID', repo_type='dataset')235        print(f'仓库文件数量: {len(files)}')236    except Exception as e:237        print(f'列出仓库文件失败: {str(e)}')238        sys.exit(1)239    240    backup_files = [f for f in files if f.startswith('sillytavern_backup_') and f.endswith('.tar.gz')]241    print(f'找到备份文件数量: {len(backup_files)}')242    243    if not backup_files:244        print('未找到备份文件')245        sys.exit(0)246    247    # 按名称排序(实际上是按时间戳排序)248    latest_backup = sorted(backup_files)[-1]249    print(f'最新备份文件: {latest_backup}')250    251    with tempfile.TemporaryDirectory() as temp_dir:252        print(f'创建临时目录: {temp_dir}')253        254        start_time = time.time()255        print(f'开始下载: {start_time}')256        257        # 下载文件258        try:259            filepath = api.hf_hub_download(260                repo_id='$DATASET_ID',261                filename=latest_backup,262                repo_type='dataset',263                local_dir=temp_dir264            )265            print(f'文件下载到: {filepath}')266        except Exception as e:267            print(f'下载文件失败: {str(e)}')268            sys.exit(1)269        270        end_time = time.time()271        print(f'下载完成,耗时: {end_time - start_time:.2f} 秒')272        273        if filepath and os.path.exists(filepath):274            # 确保目标目录存在275            os.makedirs('$DATA_DIR', exist_ok=True)276            277            # 检查文件权限278            print(f'文件权限: {oct(os.stat(filepath).st_mode)[-3:]}')279            280            # 解压文件281            try:282                with tarfile.open(filepath, 'r:gz') as tar:283                    print('开始解压文件...')284                    tar.extractall('$DATA_DIR')285                    print('文件解压完成')286            except Exception as e:287                print(f'解压文件失败: {str(e)}')288                sys.exit(1)289            290            print(f'成功从 {latest_backup} 恢复备份')291        else:292            print('下载的文件路径无效')293            sys.exit(1)294except Exception as e:295    print(f'下载备份过程中出错: {str(e)}')296    sys.exit(1)297"298    if [ $? -ne 0 ]; then299        log_error "备份下载失败"300        return 1301    else302        log_info "备份下载成功"303        return 0304    fi305}306 307# 首次启动时下载最新备份308log_info "正在从HuggingFace下载最新备份..."309download_latest_backup310 311# 同步函数312sync_data() {313    log_info "数据同步服务已启动"314    315    while true; do316        log_info "开始同步进程,时间: $(date)"317        318        if [ -d "$DATA_DIR" ]; then319            timestamp=$(date +%Y%m%d_%H%M%S)320            backup_file="sillytavern_backup_${timestamp}.tar.gz"321            backup_path="${TEMP_DIR}/${backup_file}"322            323            log_info "创建备份文件: $backup_path"324            325            # 检查数据目录内容326            file_count=$(find "$DATA_DIR" -type f | wc -l)327            log_info "数据目录文件数量: $file_count"328            329            if [ "$file_count" -eq 0 ]; then330                log_info "数据目录为空,跳过备份"331            else332                # 压缩数据目录333                tar -czf "$backup_path" -C "$DATA_DIR" .334                if [ $? -ne 0 ]; then335                    log_error "创建压缩文件失败"336                else337                    log_info "压缩文件创建成功: $(du -h $backup_path | cut -f1)"338                    339                    # 上传备份340                    log_info "正在上传备份到HuggingFace..."341                    upload_backup "$backup_path" "$backup_file"342                    343                    # 删除临时备份文件344                    rm -f "$backup_path"345                    log_info "已删除临时备份文件"346                fi347            fi348        else349            log_error "数据目录不存在: $DATA_DIR"350            mkdir -p "$DATA_DIR"351            chmod -R 777 "$DATA_DIR"352        fi353        354        # 设置同步间隔355        SYNC_INTERVAL=${SYNC_INTERVAL:-3600}356        log_info "下次同步将在 ${SYNC_INTERVAL} 秒后进行..."357        sleep $SYNC_INTERVAL358    done359}360 361# 启动同步进程362sync_data