Hypercore is a protocol and p2p network for distributing and replicating feeds of binary data. It is the low level component that Hyperdrive is built on top off.
npm install hypercore
It runs both in the node and in the browser using browserify.
var hypercore = require('hypercore')
var net = require('net')
var core = hypercore(db) // db is a leveldb instance
var feed = core.createFeed()
feed.append(['hello', 'world'], function () {
console.log('appended two blocks')
console.log('key is', feed.key.toString('hex'))
})
feed.on('upload', function (block, data) {
console.log('uploaded block', block, data)
})
var server = net.createServer(function (socket) {
socket.pipe(feed.replicate()).pipe(socket)
})
server.listen(10000)In another process
var core = hypercore(anotherDb)
var feed = core.createFeed(<key-printed-out-above>)
var socket = net.connect(10000)
socket.pipe(feed.replicate()).pipe(socket)
feed.on('download', function (block, data) {
console.log('downloaded block', block, data)
})Create a new hypercore instance. db should be a leveldb instance.
Create a new feed. A feed stores a list of append-only data (buffers). A feed has a .key property that you can pass in to createFeed if you want to retrieve an old feed. Per default all feeds are appendable (live).
Options include:
{
live: true,
storage: externalStorage
}If you want to create a static feed, one you cannot reappend data to, pass the {live: false} option.
The storage option allows you to store data outside of leveldb. This is very useful if you use hypercore to distribute files.
See the Storage API section for more info
Create a generic replication stream. Use the feed.join(stream) API described below to replicate specific feeds of data.
List all feed keys in the database. Optionally you can pass a callback to buffer them into an array.
Returns a feed as a writable stream.
Returns a feed as a readable stream.
As mentioned above a feed stores a list of data for you that you can replicate to other peers. It has the following API
The key of this feed. A 32 byte buffer. Other peers need this key to start replicating the feed.
A 32 byte buffer containing a public id of the feed. The public id is sha-256 hmac of the string hypercore using the feed key as the password.
You can use the public id to find other peers sharing this feed without disclosing your feed key to a third party.
The total number of known data blocks in the feed.
Call this method to ensure that a feed is opened. You do not need to call this but the .blocks property will not be populated until the feed has been opened.
Append a piece of data to the feed. If you want to append more than once piece you can pass in an array.
Retrieve a piece of data from the feed.
If you are not using a live feed you need to call this method to finalize the feed once you are ready to share it.
Finalizing will set the .key property and allow other peers to get your data.
Get a replication stream for this feed. Pipe this to another peer to start replicating this feed with another peer. If you create multiple replication streams to multiple peers you'll upload/download data to all of them (meaning the load will spread out).
Per default the replication stream encrypts all messages sent using the feed key and an incrementing nonce. This helps ensures that the remote peer also the feed key and makes it harder for a man-in-the-middle to sniff the data you are sending.
Set {encrypted: false} to disable this.
Join another replication stream. Hypercore uses a simple multiplexed protocol that allows one replication stream to be used for multiple feeds at once.
You do not need to join a replication stream that you created using feed.replicate() - you implicitly join that one.
Leave a replication stream.
Emitted when a remote feed joins the replication stream. You can use this as a signal to join the stream yourself if you want to.
Emitted when a data block has been downloaded
Emitted when a data block has been uploaded
If you want to use external storage to store the hypercore data (metadata will still be stored in the leveldb) you need to implement the following api and provide that as the storage option when creating a feed.
Some node modules that implement this interface are
- random-access-file Writes data to a file.
- random-access-memory Writes data to memory.
This API is optional. If you provide this hypercore will call .open and wait for the callback to be called before calling any other methods.
This API is required. Hypercore calls this when it wants to read data. You should return a buffer with length length that way read at the corresponding offset. If you cannot read this buffer call the callback with an error.
This API is required. Hypercore calls this when it wants to write data. You should write the buffer at the corresponding offset and call the callback afterwards. If there was an error writing you should call the callback with that error.
This API is optional. Hypercore will call this method when the feed is closing.
MIT