THIS IS STILL HEAVILY IN DEVELOPMENT, USE AT YOUR OWN RISK

If you want a currently working version of puppet bridges then head over here!

Installation

• Clone matrix-puppet-bridge, use the development branch and run npm link
• Clone matrix-puppet-skype, use the development branch, run npm link matrix-puppet-bridge and then npm link
• Clone this, run npm link matrix-puppet-bridge, npm link matrix-puppet-skype and then npm install and npm start.

matrix-puppet-server

This project provides a base class for a style of Matrix bridge which primarily acts as or "puppets" a specific user (usually yourself) on your homeserver and on a third party service.

There is a common pattern in this style of bridge, such as duplicate message issues, which are dealt with by this module.

Examples

These bridges have been built using matrix-puppet-bridge:

• https://github.com/matrix-hacks/matrix-puppet-imessage
• https://github.com/matrix-hacks/matrix-puppet-groupme
• https://github.com/matrix-hacks/matrix-puppet-facebook
• https://github.com/matrix-hacks/matrix-puppet-slack
• https://github.com/matrix-hacks/matrix-puppet-hangouts
• https://github.com/matrix-hacks/matrix-puppet-signal
• https://github.com/matrix-hacks/matrix-puppet-skype

FAQ

Q: I can receive messages, I can also see messages sent by the third party client in Matrix but I cannot send messages from Matrix. Nothing appears to show up in the logs when I send a message in Matrix and nothing is actually sent. What's going on?

This is symptomatic of a homeserver that is unable to reach the bridge. If you curl the bridge URL in the yaml file that you have referenced in synapse's homeserver.yaml, does it 200 OK ? Most likely this is misconfigured. If you still have this issue ask us in #matrix-puppet-bridge:matrix.org.

Some hints:

• We use node's server.listen function to create the server. We do not pass in a hostnae, only the port that you configure. See the default behavior explained here with respect to IPv4 and IPv6: https://nodejs.org/dist/latest-v7.x/docs/api/http.html#http_server_listen_port_hostname_backlog_callback
• Use netstat and curl to see if the homeserver can truly access the bridge. It's probably the most common problem users face.

Q: My access token has changed. How can I quickly update my access token on the bridge?

Run this in your bridge directory:

node -e "new (require('matrix-puppet-bridge').Puppet)('config.json').associate()"

Q: Is this made to handle several facebook/hangouts/slack users within one bridge? In other words, can I use this for "mass hosting" of many imessage/facebook/hangouts identities with one matrix homeserver?

No, unfortunately. This is not designed for mass hosting of bridges.

1. Setup Challenges Several of the protocols we support do not lend themselves well to mass hosting. For example, the iMessage bridge must run on osx, and authentication must be handled by using the login gui of the iMessages app, and there's not a clean way of running multiple iMessages apps and automating them. Beyond that, a couple of other protocols do not support a proper oauth workflow (see facebook which definitely does not unless it's for a 'bot user', and to some extent hangouts doesn't support it either [though unsupported techniques do exist]).

2. Password Leaks Any kind of "mass hosting" setup that allowed for configuration of these bridges via a "/nickserv" type interface would require sending your facebook/hangouts password to a "man in the middle" (homeserver in this case). This is just not acceptable to the authors of this framework, so you will probably never see it implemented by us.

3. Conversation Leaks An effort to build a mass-hosted version of this would entail putting not just passwords, but also personal content and conversations on public homeservers (e.g. personal conversations over facebook, or iMessage) that you do not control, therefore we think it is better to go towards a model in which you run your own homeserver, as a prerequisite. That said, sometimes you have a more technical friend and trust them with this, but in that case we believe it's better for that friend to show you how to run your own HS if you want to use these bridges, rather than compromise on the privacy issue

In summary, this bridge framework is explicitly for bridges that are "personal" in nature. It assumes the user cares a great deal about the privacy of their facebook/hangouts/imessage/etc passwords and messages, and as such desire to run their own homeserver and all their own bridges.

That said, we are open to proposals in which we can solve 2 and 3, which would allow homeserver sharing. Such a proposal would necessarily span across the matrix ecosystem, so you may want to reference https://matrix.org/docs/spec/ if you haven't already.

Q: What about service X?

Right now I recommend you look at the examples. Right now the most complex example in terms of creating a client is imessage. The most complex example in terms of needing to make additional calls like looking up user info, check out the facebook one. For a basic middle-ground, check the groupme one.

Q: What's puppetting and why does this use it?

There are two kinds of puppetting happening here:

3rd party user puppetting

The bridge is logging into hangouts/facebook/slack/etc in such a way that it appears to send messages as you. From the perspective of other hangouts/facebook/slack users, all messages appear to come from your actual user, not from a bot or anything along those lines.

Matrix user puppetting

The bridge is logging into your matrix homeserver with your matrix username, and sometimes sends messages as your username. Why is this necessary? If you happen to send a message using a native hangouts/facebook/slack/etc client rather than using the bridge, we want to propogate the message you sent over to matrix somehow; This way you can see the entire context of your conversation within matrix, even if some of your messages were sent using a native hangouts/facebook/etc client. This means it has to appear to come from someone that represents you somehow. So why not create a ghost/bot user on the matrix side that represents your "facebook/hangouts self" for this purpose (e.g. "@YourName_facebook:your-hs.example.org")? This approach has the following limitations:

• If the bridge is not able to log in as you, it is not empowered with the ability to automatically join you to rooms. This means you must still manually accept invites for your matrix user to any newly created bridge rooms. This is especially bad if a brand new new contact messages you on facebook/hangouts/etc -- the bridge creates a new room and you may get a push notification with the room invite but no notification containing their actual message text. You would have to open the matrix client and join the room to see their message. This is an extra manual step and is not convenient compared to what we've become accustomed to with messaging apps.
• If sending from the native hangouts/facebook/etc app, and this gets shown in matrix by a virtual secondary user/bot that represents yourself rather than your real self, in matrix's eyes, this is considered a new unread message in the room, so the room state is changed to "unread" -- generally a bold room name in the matrix client. This is not really desirable; Given I'm the one that sent the new message, I've of course already read the message. Ultimately this gets pretty annoying, especially if you tend (like myself) to use these bold unread room states indicators to help quickly catch up on older missed messages.

For these reasons (and some other minor reasons I won't mention here), we settled on puppetting the matrix user.

Q: How can I prevent long push notification messages for 1 on 1 conversations?

At this time we recommend modifying sygnal. For example, see this commit: https://github.com/AndrewJDR/sygnal/commit/3813ef48a1be1b6015953974a13ee4da2b704882

The prefix seen by sygnal is that which you configure on your class:

class App extends MatrixPuppetBridgeBase {
  getServicePrefix() {
    return "__mbp__someservice";
  }
}

In the examples above, "mpb" was used as the special tag, but it can be anything you want. Keep in mind that getServicePrefix is called for creating rooms and ghost users, and also needs to match your appserver yaml file, so plan for this and expect this to be a source of problems when changing it after having run the bridge for awhile under a different service prefix.

Q: How can I add bang (!) commands to a room, such as !echo

matrix-puppet-bridge comes with a bang command processor. Simply define a method and it will be invoked instead of being forwarded to the third party service:

class App extends MatrixPuppetBridgeBase {
	handleMatrixUserBangCommand(bangCmd, matrixMsgEvent) {
		const { bangcommand, command, body } = bangCmd;
		const { room_id } = matrixMsgEvent;
		const client = this.puppet.getClient();
		const reply = (str) => client.sendNotice(room_id, str);
		if ( command === 'help' ) {
			reply([
				'Bang Commands',
				'!help .......... display this information',
				'!echo <text> ... repeat text back to you',
				'!sync .......... synchronize this room with 3rd party service',
			].join('\n'));
		} else if ( command === 'echo' ) {
			reply(body);
		} else if ( command === 'sync' ) {
			reply('command not implemented yet: '+bangcommand);
		} else {
			reply('unrecognized command: '+bangcommand);
		}
	}
}

Q: Why am I seeing duplicate messages?

We use a non-printable suffix to "tag" messages that go over the bridged network and when the message is seen on the return trip, we know to ignore it and not forward it again. This tag may be getting stripped on your network.

Try using a printable tag, which is unlikely to be stripped, by editing config.json and adding:

"deduplicationTag" : " [m]",
"deduplicationTagPattern" : " \\[m\\]"

Let us know if this doesn't work on a particular protocol! For more information, see this discussion.

Q: Where can I ask questions?

You can use GitHub issues on this or any other puppet-bridge projects.

Alternatively you can join #matrix-puppet-bridge:matrix.org