A streaming upload tool for Amazon S3
A streaming upload tool for Amazon S3, taking input from a
readdirp stream, and outputting the
s3-sync is also optionally backed by a level database to use as a local cache for file uploads. This way, you can minimize the frequency you have to hit S3 and speed up the whole process considerably.
You can use this to sync complete directory trees with S3 when deploying static websites. It's a work in progress, so expect occasional API changes and additional features.
npm install s3-sync
Creates an upload stream. Passes its options to knox, so at a minimum you'll need:
key: Your AWS access key.
secret: Your AWS secret.
bucket: The bucket to upload to.
The following are also specific to s3-sync:
concurrency: The maximum amount of files to upload concurrently.
retries: The maximum number of times to retry uploading a file before failing. By default the value is 7.
headers: Additional headers to include on each file.
hashKey: By default, file hashes are stored based on the file's absolute path. This doesn't work very nicely with temporary files, so you can pass this function in to map the file object to a string key for the hash.
acl: Use a custom ACL header. Defaults to
force: Force s3-sync to overwrite any existing files.
You can also store your local cache in S3, provided you pass the following
options, and use
putCache (see below) before/after uploading:
cacheDest: the path to upload your cache backup to in S3.
cacheSrc: the local, temporary, text file to stream to before uploading to S3.
If you want more control over the files and their locations that you're uploading, you can write file objects directly to the stream, e.g.:
var stream = s3synckey: processenvAWS_ACCESS_KEYsecret: processenvAWS_SECRET_KEYbucket: 'sync-testing'streamwritesrc: __filenamedest: '/uploader.js'streamendsrc: __dirname + '/README.md'dest: '/README.md'
src is the absolute local file path, and
dest is the location to
upload the file to on the S3 bucket.
db is an optional argument - pass it a level database and it'll keep a
local cache of file hashes, keeping S3 requests to a minimum.
Uploads your level cache, if available, to the S3 bucket. This means that your cache only needs to be populated once.
Streams a previously uploaded cache from S3 to your local level database.
Emitted when a file has failed to upload. This will be called each time the file is attempted to be uploaded.
Here's an example using
readdirp to upload a local directory to
an S3 bucket:
var level = require'level's3sync = require's3-sync'readdirp = require'readdirp'// To cache the S3 HEAD results and speed up the// upload process. Usage is optional.var db = level__dirname + '/cache'var files = readdirproot: __dirnamedirectoryFilter: '!.git' '!cache'// Takes the same options arguments as `knox`,// plus some additional options listed abovevar uploader = s3syncdbkey: processenvAWS_ACCESS_KEYsecret: processenvAWS_SECRET_KEYbucket: 'sync-testing'concurrency: 16prefix : 'mysubfolder/' //optional prefix to files on S3on'data'console.logfilefullPath + ' -> ' + fileurlfilespipeuploader
You can find another example which includes remote cache storage at example.js.