cloudron-box/src/backupformat/tgz.js

'use strict';

exports = module.exports = {
    getBackupFilePath,
    download,
    upload
};

const assert = require('assert'),
    async = require('async'),
    BoxError = require('../boxerror.js'),
    DataLayout = require('../datalayout.js'),
    debug = require('debug')('box:backupformat/tgz'),
    { DecryptStream, EncryptStream } = require('../hush.js'),
    once = require('../once.js'),
    path = require('path'),
    ProgressStream = require('../progress-stream.js'),
    promiseRetry = require('../promise-retry.js'),
    stream = require('stream/promises'),
    storage = require('../storage.js'),
    tar = require('tar-fs'),
    tar2 = require('tar'),
    zlib = require('zlib');

function getBackupFilePath(backupConfig, remotePath) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof remotePath, 'string');

    const rootPath = backupConfig.rootPath;
    const fileType = backupConfig.encryption ? '.tar.gz.enc' : '.tar.gz';

    // we don't have a rootPath for noop
    if (backupConfig.provider === 'noop') return remotePath + fileType;

    return path.join(rootPath, remotePath + fileType);
}

function tarPack(dataLayout, encryption) {
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof encryption, 'object');

    const pack = tar.pack('/', {
        dereference: false, // pack the symlink and not what it points to
        entries: dataLayout.localPaths(),
        ignoreStatError: (path, err) => {
            debug(`tarPack: error stat'ing ${path} - ${err.code}`);
            return err.code === 'ENOENT'; // ignore if file or dir got removed (probably some temporary file)
        },
        map: function(header) {
            header.name = dataLayout.toRemotePath(header.name);
            // the tar pax format allows us to encode filenames > 100 and size > 8GB (see #640)
            // https://www.systutorials.com/docs/linux/man/5-star/
            if (header.size > 8589934590 || header.name > 99) header.pax = { size: header.size };
            return header;
        },
        strict: false // do not error for unknown types (skip fifo, char/block devices)
    });

    const gzip = zlib.createGzip({});
    const ps = new ProgressStream({ interval: 10000 }); // emit 'progress' every 10 seconds

    pack.on('error', function (error) {
        debug('tarPack: tar stream error. %o', error);
        ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    gzip.on('error', function (error) {
        debug('tarPack: gzip stream error. %o', error);
        ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
    });

    if (encryption) {
        const encryptStream = new EncryptStream(encryption);
        encryptStream.on('error', function (error) {
            debug('tarPack: encrypt stream error. %o', error);
            ps.emit('error', new BoxError(BoxError.EXTERNAL_ERROR, error.message));
        });

        pack.pipe(gzip).pipe(encryptStream).pipe(ps);
    } else {
        pack.pipe(gzip).pipe(ps);
    }

    return ps;
}

async function tarExtract(inStream, dataLayout, encryption, progressCallback) {
    assert.strictEqual(typeof inStream, 'object');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof encryption, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    const ps = new ProgressStream({ interval: 10000 }); // display a progress every 10 seconds

    const extract = tar2.extract({
        cwd: '/tmp',
        dmode: 500, // ensure directory is writable
        preserveOwner: false, // use uid/gid of current process
        strict: true, // error on warnings,
        preservePaths: true, // allow absolute paths. otherwise will sandbox to cwd

        // to map paths, we cannot use transform hook because it is only used for files and not directories
        // we can use filter hook as well - https://github.com/isaacs/node-tar/issues/357#issuecomment-1416491212
        onReadEntry(entry) {
            // debug(entry.header.path, entry.header.type, entry.header.size);
            entry.path = dataLayout.toLocalPath(entry.header.path);
        },
        onwarn(code, message /*, data */) {
            debug(`extract warning:${message} ${code}`);
        },
    });

    ps.on('progress', function (progress) {
        const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
        if (!transferred && !speed) return progressCallback({ message: 'Downloading backup' }); // 0M@0MBps looks wrong
        progressCallback({ message: `Downloading ${transferred}M@${speed}MBps` });
    });

    if (encryption) {
        const decrypt = new DecryptStream(encryption);
        await stream.pipeline(inStream, ps, decrypt, extract);
    } else {
        await stream.pipeline(inStream, ps, extract);
    }

    return ps;
}

async function download(backupConfig, remotePath, dataLayout, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof remotePath, 'string');
    assert(dataLayout instanceof DataLayout, 'dataLayout must be a DataLayout');
    assert.strictEqual(typeof progressCallback, 'function');

    debug(`download: Downloading ${remotePath} to ${dataLayout.toString()}`);

    const backupFilePath = getBackupFilePath(backupConfig, remotePath);

    await promiseRetry({ times: 5, interval: 20000, debug }, async () => {
        progressCallback({ message: `Downloading backup ${backupFilePath}` });
        const sourceStream = await storage.api(backupConfig.provider).download(backupConfig, backupFilePath);
        await tarExtract(sourceStream, dataLayout, backupConfig.encryption, progressCallback);
    });
}

async function upload(backupConfig, remotePath, dataLayout, progressCallback) {
    assert.strictEqual(typeof backupConfig, 'object');
    assert.strictEqual(typeof remotePath, 'string');
    assert.strictEqual(typeof dataLayout, 'object');
    assert.strictEqual(typeof progressCallback, 'function');

    debug(`upload: Uploading ${dataLayout.toString()} to ${remotePath}`);

    return new Promise((resolve, reject) => {
        async.retry({ times: 5, interval: 20000 }, function (retryCallback) {
            retryCallback = once(retryCallback); // protect again upload() erroring much later after tar stream error

            const tarStream = tarPack(dataLayout, backupConfig.encryption);

            tarStream.on('progress', function (progress) {
                const transferred = Math.round(progress.transferred/1024/1024), speed = Math.round(progress.speed/1024/1024);
                if (!transferred && !speed) return progressCallback({ message: 'Uploading backup' }); // 0M@0MBps looks wrong
                progressCallback({ message: `Uploading backup ${transferred}M@${speed}MBps` });
            });
            tarStream.on('error', retryCallback); // already returns BoxError

            storage.api(backupConfig.provider).upload(backupConfig, getBackupFilePath(backupConfig, remotePath), tarStream, retryCallback);
        }, (error) => {
            if (error) return reject(error);
            resolve();
        });
    });
}