AWS CLI s3 sync for Node.js

AWS CLI s3 sync for Node.js is a modern TypeScript client to perform S3 sync operations between file systems and S3 buckets, in the spirit of the official AWS CLI command.
AWS CLI installation is NOT required by this module.

Features

• Sync from an S3 bucket to a local file system
• Sync from a local file system to an S3 bucket (with multipart uploads support)
• Sync from an S3 bucket to another S3 bucket
• Sync only new and updated objects
• Support AWS CLI options --delete, --dryrun, --size-only, --include, --exclude, --follow-symlinks, --no-follow-symlinks
• Support AWS SDK native command input options
• Monitor sync progress
• Sync any number of objects (no 1000 objects limit)
• Transfer objects concurrently
• Manage differences in folder structures easily through relocation

Why should I use this module?

1. There is no way to achieve S3 sync using the AWS SDK for JavaScript v3 alone
2. AWS CLI installation is NOT required
3. The module contains no external dependency
4. The AWS SDK peer dependency is up-to-date (AWS SDK for JavaScript v3)
5. The module overcomes a set of common limitations listed at the bottom of this README

Table of Contents

1. Getting Started
   1. Install
   2. Quick Start
      1. Client initialization
      2. Sync from file system to S3 bucket
      3. Sync from S3 bucket to file system
      4. Sync from S3 bucket to S3 bucket
      5. Monitor transfer progress
      6. Abort sync
      7. Use AWS SDK command input options
      8. Relocate objects during sync
      9. Filter source files
2. API Reference
   • Class: S3SyncClient
     • new S3SyncClient(configuration)
     • client.sync(localDir, bucketPrefix[, options])
     • client.sync(bucketPrefix, localDir[, options])
     • client.sync(sourceBucketPrefix, targetBucketPrefix[, options])
3. Change Log
4. Benchmark

Getting Started

Install

npm i s3-sync-client

Quick Start

Client initialization

S3SyncClient is a wrapper for the AWS SDK S3Client class.

1import S3Client from '@aws-sdk/client-s3';
2import { S3SyncClient } from 's3-sync-client';
3
4const s3Client = new S3Client({ /* ... */ });
5const { sync } = new S3SyncClient({ client: s3Client });

Sync from file system to S3 bucket

1// aws s3 sync /path/to/local/dir s3://mybucket2
2await sync('/path/to/local/dir', 's3://mybucket2');
3await sync('/path/to/local/dir', 's3://mybucket2', { partSize: 100 * 1024 * 1024 }); // uses multipart uploads for files higher than 100MB
4
5// aws s3 sync /path/to/local/dir s3://mybucket2/zzz --delete
6await sync('/path/to/local/dir', 's3://mybucket2/zzz', { del: true });

Sync from S3 bucket to file system

1// aws s3 sync s3://mybucket /path/to/some/local --delete
2await sync('s3://mybucket', '/path/to/some/local', { del: true });
3
4// aws s3 sync s3://mybucket2 /path/to/local/dir --dryrun
5const diff = await sync('s3://mybucket2', '/path/to/local/dir', { dryRun: true });
6console.log(diff); // log operations to perform

Sync from S3 bucket to S3 bucket

1// aws s3 sync s3://my-source-bucket s3://my-target-bucket --delete
2await sync('s3://my-source-bucket', 's3://my-target-bucket', { del: true });

Monitor sync progress

1import { TransferMonitor } from 's3-sync-client';
2
3const monitor = new TransferMonitor();
4monitor.on('progress', (progress) => console.log(progress));
5await sync('s3://mybucket', '/path/to/local/dir', { monitor });
6
7/* output:
8...
9{
10  size: { current: 11925, total: 35688 },
11  count: { current: 3974, total: 10000 }
12}
13...
14*/
15
16// to pull status info occasionally only, use monitor.getStatus():
17const timeout = setInterval(() => console.log(monitor.getStatus()), 2000);
18try {
19    await sync('s3://mybucket', '/path/to/local/dir', { monitor });
20} finally {
21    clearInterval(timeout);
22}

Abort sync

1import { AbortController } from '@aws-sdk/abort-controller';
2import { TransferMonitor } from 's3-sync-client';
3
4const abortController = new AbortController();
5setTimeout(() => abortController.abort(), 30000);
6
7await sync('s3://mybucket', '/path/to/local/dir', { abortSignal: abortController.signal });

Use AWS SDK command input options

1import mime from 'mime-types';
2/*
3commandInput properties can either be:
4  - a plain object
5  - a function returning a plain object
6*/
7
8// set ACL, fixed value
9await sync('s3://mybucket', '/path/to/local/dir', {
10    commandInput: {
11        ACL: 'aws-exec-read',
12    },
13});
14
15// set content type, dynamic value (function)
16await sync('s3://mybucket1', 's3://mybucket2', {
17    commandInput: (input) => ({
18        ContentType: mime.lookup(input.Key) || 'text/html',
19    }),
20});

Filter source files

1// aws s3 sync s3://my-source-bucket s3://my-target-bucket --exclude "*" --include "*.txt" --include "flowers/*"
2await sync('s3://my-source-bucket', 's3://my-target-bucket', {
3    filters: [
4        { exclude: () => true }, // exclude everything
5        { include: (key) => key.endsWith('.txt') }, // include .txt files
6        { include: (key) => key.startsWith('flowers/') }, // also include everything inside the flowers folder
7    ],
8});

Relocate objects during sync

1// move objects from source folder a/b/ to target folder zzz/
2await sync('s3://my-source-bucket', 's3://my-target-bucket', {
3    relocations: [ // multiple relocations can be applied
4      (currentPath) =>
5        currentPath.startsWith('a/b/')
6          ? currentPath.replace('a/b/', 'zzz/')
7          : currentPath
8    ],
9});
10
11// aws s3 sync s3://mybucket/flowers/red /path/to/local/dir
12// as in cli, folder flowers/red will be flattened during sync
13await sync('s3://mybucket/flowers/red', '/path/to/local/dir'); 

Note: relocations are applied after every other options such as filters.

Additional examples are available in the repo test directory.

API Reference

A complete API reference is available in the repo docs directory.

Class: S3SyncClient

new S3SyncClient(options)

• options <S3SyncClientConfig>
  • client <S3Client> instance of AWS SDK S3Client.

client.sync(localDir, bucketPrefix[, options])

Sync from file system to S3 bucket.
Similar to AWS CLI aws s3 sync localDir bucketPrefix [options].

• localDir <string> Local directory
• bucketPrefix <string> Remote bucket name which may contain a prefix appended with a / separator
• options <SyncBucketWithLocalOptions>
  • dryRun <boolean> Equivalent to CLI --dryrun option
  • del <boolean> Equivalent to CLI --delete option
  • deleteExcluded <boolean> Delete excluded target objects even if they match a source object. Ignored if del is false. See this CLI issue.
  • sizeOnly <boolean> Equivalent to CLI --size-only option
  • followSymlinks <boolean> Equivalent to CLI --follow-symlinks option (default true)
  • relocations <Relocation[]> Allows uploading objects to remote folders without mirroring the source directory structure. Each relocation is as a callback taking a string posix path param and returning a relocated string posix path.
  • filters <Filter[]> Almost equivalent to CLI --exclude and --include options. Filters can be specified using plain objects including either an include or exclude property. The include and exclude properties are functions that take an object key and return a boolean.
  • abortSignal <AbortSignal> Allows aborting the sync
  • commandInput <CommandInput<PutObjectCommandInput>> | <CommandInput<CreateMultipartUploadCommandInput>> Set any of the SDK <PutObjectCommandInput> or <CreateMultipartUploadCommandInput> options to uploads
  • monitor <TransferMonitor>
    • Attach progress event to receive upload progress notifications
    • Call getStatus() to retrieve progress info on demand
  • maxConcurrentTransfers <number> Each upload generates a Promise which is resolved when a local object is written to the S3 bucket. This parameter sets the maximum number of upload promises that might be running concurrently.
  • partSize <number> Set the part size in bytes for multipart uploads. Default to 5 MB.
• Returns: <Promise<SyncBucketWithLocalCommandOutput>> Fulfills with sync operations upon success.

client.sync(bucketPrefix, localDir[, options])

Sync from S3 bucket to file system.
Similar to AWS CLI aws s3 sync bucketPrefix localDir [options].

• bucketPrefix <string> Remote bucket name which may contain a prefix appended with a / separator
• localDir <string> Local directory
• options <SyncLocalWithBucketOptions>
  • dryRun <boolean> Equivalent to CLI --dryrun option
  • del <boolean> Equivalent to CLI --delete option
  • deleteExcluded <boolean> Delete excluded target objects even if they match a source object. Ignored if del is false. See this CLI issue.
  • sizeOnly <boolean> Equivalent to CLI --size-only option
  • followSymlinks <boolean> Equivalent to CLI --follow-symlinks option (default true)
  • relocations <Relocation[]> Allows downloading objects to local directories without mirroring the source folder structure. Each relocation is as a callback taking a string posix path param and returning a relocated string posix path.
  • filters <Filter[]> Almost equivalent to CLI --exclude and --include options. Filters can be specified using plain objects including either an include or exclude property. The include and exclude properties are functions that take an object key and return a boolean.
  • abortSignal <AbortSignal> Allows aborting the sync
  • commandInput <CommandInput<GetObjectCommandInput>> Set any of the SDK <GetObjectCommandInput> options to downloads
  • monitor <TransferMonitor>
    • Attach progress event to receive download progress notifications
    • Call getStatus() to retrieve progress info on demand
  • maxConcurrentTransfers <number> Each download generates a Promise which is resolved when a remote object is written to the local file system. This parameter sets the maximum number of download promises that might be running concurrently.
• Returns: <Promise<SyncLocalWithBucketCommandOutput>> Fulfills with sync operations upon success.

client.sync(sourceBucketPrefix, targetBucketPrefix[, options])

Sync from S3 bucket to S3 bucket.
Similar to AWS CLI aws s3 sync sourceBucketPrefix targetBucketPrefix [options].

• sourceBucketPrefix <string> Remote reference bucket name which may contain a prefix appended with a / separator
• targetBucketPrefix <string> Remote bucket name to sync which may contain a prefix appended with a / separator
• options <SyncBucketWithBucketOptions>
  • dryRun <boolean> Equivalent to CLI --dryrun option
  • del <boolean> Equivalent to CLI --delete option
  • deleteExcluded <boolean> Delete excluded target objects even if they match a source object. Ignored if del is false. See this CLI issue.
  • sizeOnly <boolean> Equivalent to CLI --size-only option
  • relocations <Relocation[]> Allows copying objects to remote folders without mirroring the source folder structure. Each relocation is as a callback taking a string posix path param and returning a relocated string posix path.
  • filters <Filter[]> Almost equivalent to CLI --exclude and --include options. Filters can be specified using plain objects including either an include or exclude property. The include and exclude properties are functions that take an object key and return a boolean.
  • abortSignal <AbortSignal> Allows aborting the sync
  • commandInput <CommandInput<CopyObjectCommandInput>> Set any of the SDK <CopyObjectCommandInput> options to copy operations
  • monitor <TransferMonitor>
    • Attach progress event to receive copy progress notifications
    • Call getStatus() to retrieve progress info on demand
  • maxConcurrentTransfers <number> Each copy generates a Promise which is resolved after the object has been copied. This parameter sets the maximum number of copy promises that might be running concurrently.
• Returns: <Promise<SyncBucketWithBucketCommandOutput>> Fulfills with sync operations upon success.

Change Log

See CHANGELOG.md.

Benchmark

AWS CLI s3 sync for Node.js has been developed to solve the S3 sync limitations of the existing GitHub repo and NPM modules.

Most of the existing repo and NPM modules suffer one or more of the following limitations:

• requires AWS CLI to be installed
• uses Etag to perform file comparison (Etag should be considered an opaque field and shouldn't be used)
• limits S3 bucket object listing to 1000 objects
• supports syncing local to bucket, but doesn't support syncing bucket to local
• doesn't support multipart uploads
• uses outdated dependencies
• is unmaintained

The following JavaScript modules suffer at least one of the limitations:

• https://github.com/guerrerocarlos/aws-cli-s3-sync
• https://github.com/thousandxyz/s3-lambo
• https://github.com/Quobject/aws-cli-js
• https://github.com/auth0/node-s3-client
• https://github.com/andrewrk/node-s3-client
• https://github.com/hughsk/s3-sync
• https://github.com/issacg/s3sync