Home

Awesome

cabal-core

Core database, replication, swarming, and chat APIs for cabal.

Table of Contents

<details><summary>Click to expand</summary> </details>

Usage

npm install cabal-core

API

var Cabal = require('cabal-core')

var cabal = Cabal([storage][, key][, opts])

Create a cabal p2p database using storage storage, which must be either a string (filepath to directory on disk) or an instance of random-access-storage.

key is a cabal key, as a string or Buffer.

If this is a new cabal, key can be omitted and will be generated.

You can pass opts.db as a levelup or leveldown instance to use persistent storage for indexing instead of using memory. For example:

var level = require('level')
var cabal = Cabal(storage, key, { db: level('/tmp/bot.db') })

Other opts include:

cabal.getLocalKey(cb)

Returns the local user's key (as a hex string).

var ds = cabal.replicate(isInitiator[, opts])

Creates a new, live replication stream. This duplex stream can be piped into any transport expressed as a node stream (tcp, websockets, udp, utp, etc).

Ensure that isInitiator to true to one side, and false on the other. This is necessary for setting up the encryption mechanism.

opts are passed down into the underlying hypercore replication.

cabal.ready(cb)

Calls cb() when the underlying indexes are caught up.

cabal.close(cb)

Calls cb() when the cabal and its resources have been closed. This also leaves the swarm, if joined.

cabal.getMessage(key, cb)

Read a message from key, a string of feedKey@seq or an object of { key, seq } as cb(err, node) from the underlying hypercore.

Channels

cabal.channels.get(function (error, channels) {})

Retrieve a list of all channel names that exist in this cabal.

cabal.channels.events.on('add', function (channel) {})

Emitted when a new channel is added to the cabal.

Messages

var rs = cabal.messages.read(channel, opts)

Returns a readable stream of messages (most recent first) from a channel.

Pass opts.limit to set a maximum number of messages to read.

cabal.messages.events.on('message', fn)

Calls fn with every new message that arrives, regardless of channel.

cabal.messages.events.on(channel, fn)

Calls fn with every new message that arrives in channel.

Network

var swarm = require('cabal-core/swarm')

cabal.swarm(cb)

Joins the P2P swarm for a cabal. This seeks out peers who are also part of this cabal by various means (internet, local network), connects to them, and replicates cabal messages between them.

The returned object is an instance of discovery-swarm.

cabal.on('peer-added', function (key) {})

Emitted when you connect to a peer. key is a hex string of their public key.

cabal.on('peer-dropped', function (key) {})

Emitted when you lose a connection to a peer. key is a hex string of their public key.

Moderation

Cabal has a subjective moderation system.

The three roles are "admin", "moderator", and "ban/key".

Any admin/mod/ban operation can be per-channel, or cabal-wide (the @ group).

Every user sees themselves as an administrator across the entire cabal. This means they can grant admin or moderator powers to anyone, and ban anyone, but only they will see its affects on their own computer. That is, until someone adds them as an administrator or moderation from their perspective.

A cabal can be instantiated with a moderation key. This is an additional key to have your local node consider a user (the user whose key matches the moderation key) as a cabal-wide administrator (in addition to yourself).

This means that if a group of people all specify the same moderation key, they will collectively see the same set of administrators, moderators, and banned users.

cabal.moderation.listByFlag({ channel, flag })

Return a readable object stream of records for channel that for each user with flag set. Flags used by cabal-core include: "hide", "mute", "block", "admin", and "mod".

Each row object in the output stream has:

Optionally collect results into cb(err, rows).

cabal.moderation.list(cb)

Return a readable object stream of records for all moderation actions across all channels.

Each row object in the output stream has:

Optionally collect results into cb(err, rows).

cabal.moderation.listBlocks(channel, cb)

Return a readable object stream of records for the blocks in channel.

The objects in the output have the same form as listByFlag().

Optionally collect results into cb(err, rows).

cabal.moderation.listHides(channel, cb)

Return a readable object stream of records for the hides in channel.

The objects in the output have the same form as listByFlag().

Optionally collect results into cb(err, rows).

cabal.moderation.listMutes(channel, cb)

Return a readable object stream of records for the mutes in channel.

The objects in the output have the same form as listByFlag().

Optionally collect results into cb(err, rows).

cabal.moderation.listModerationBy(key, cb)

Return a readable object stream of moderation documents authored by key.

Each row object in the output is a document used for adding, removing, and setting flags.

Optionally collect results into cb(err, rows).

cabal.moderation.getFlags({ id, channel }, cb)

Get a list of flags set for the user identified by id in channel as cb(err, flags).

cabal.moderation.setFlags({ id, channel, flags }, cb)

Set an array of flags for id in channel.

cabal.moderation.addFlags({ id, channel, flags }, cb)

Add an array of flags to the existing set of flags for id in channel.

cabal.moderation.removeFlags({ id, channel, flags }, cb)

Remove an array of flags from the existing set of flags for id in channel.

cabal.moderation.events.on('update', function (update) {})

This event happens when a user's flags change with update, the log record responsible for the state change.

cabal.moderation.events.on('skip', function (skip) {})

This event happens when a moderation update was skipped with skip, the log record responsible for the state change.

Private Messages

cabal.publishPrivate(message, recipientKey, cb)

Write a message message, but encrypted so that only recipientKey (the public key of its recipient) and the sender can read it.

A timestamp field is set automatically with the current local system time.

cabal.privateMessages.list(cb)

Returns a list of strings of all users' public keys that have sent a PM to you, or who you have sent a PM to.

var rs = cabal.privateMessages.read(channel, opts)

Returns a readable stream of messages (most recent first) from a channel.

Pass opts.limit to set a maximum number of messages to read.

cabal.privateMessages.events.on('message', fn)

Calls fn with every new private message that arrives.

cabal.privateMessages.events.on(publicKey, fn)

Calls fn with every new message that arrives to or from publicKey.

Publishing

cabal.publish(message, opts, cb)

Publish message to your feed. message must have a type field set. If not, it defaults to chat/text. In general, a message is formatted as

{
  type: 'chat/text',
  content: {
    text: 'hello world',
    channel: 'cabal-dev'
  }
}

A timestamp field is set automatically with the current system time.

type is an unrestricted field: you can make up new message types and clients will happily ignore them until someone implements support for them. Well documented types include

chat/text

{
  type: 'chat/text',
  content: {
    text: 'whatever the user wants to say',
    channel: 'some channel name. if it didnt exist before, it does now!'
  }
}

Developing

Changelog

When releasing a new version the CHANGELOG.md should be updated. There are three scripts in package.json for bumping the version:

Additionally, you can do npm run changelog:fix, which will expand references to pull requests and github users to urls. For example, referencing @hackergrrl would expand to https://github.com/hackergrrl and referencing #314 would expand to https://github.com/cabal-club/cabal-core/pulls/314. Also, the table of contents in this document will be updated.

To release a new version you would typically do something like:

npm run changelog:patch
# Manually fill in details and/or tweak CHANGELOG.md
npm run changelog:fix
git add CHANGELOG.md && git commit -m 'Update changelog for x.y.z'
npm version patch
git push && git push --tags
npm publish

Try to keep the pull request titles short and descriptive and the changelog will automatically reflect this. A pull request may contain many small commits, but if it's squashed, the git history and the changelog will look clean. At the same time, all commits will still be around on GitHub if digging into details is required.

License

AGPLv3