Project Description

Main goal of this implementation is to run 2 independent Edge Agents (clients), running on mobile or desktop, which are able to make a connection and send basic messages to each other via Routing Agent with these restrictions:

Edge Agent is independent on underlying communication layer. It can communicate via either HTTP request-response, WebSockets or Push Notifications.
Edge Agent can go offline and still receive its messages when goes back to online.
There should be an option to connect more clients (Edge Agents) to one Routing Agent.
Prevent correlation.

Basic Explanation of Implementation

Agent class has method receiveMessage which unpacks incoming inboud message and then pass it to the dispatch method. This method just tries to find particular handler according to message @type attribute. Handler then process the message, calls services if needed and also creates outbound message to be send by sender, if it's required by protocol.

If handler returns an outbound message then method sendMessage packs the message with defined recepient and routing keys. This method also creates forwardMessage when routing keys are available. The way an outbound message is send depends on the implementation of MessageSender interface. Outbound message just need to contain all information which is needed for given comminucation (e. g. HTTP endpoint for HTTP protocol).

Install dependencies

yarn

Running tests

Run e2e tests with in memory messaging

You don't have to start agencies for these tests. Communication is done via RxJS subscriptions.

yarn test -t "agents"

Run e2e tests with HTTP based routing agencies

You have to start agencies first.

Open terminal and start Alice's agency:

./run.sh agency01

Open new terminal and start Bob's agency:

./run.sh agency02

Run tests:

yarn test -t "with agency"

Run all tests

You have to start agencies first, because it runs all tests togheter.

yarn test