Skip to content

Latest commit

 

History

History
396 lines (283 loc) · 13.7 KB

README.md

File metadata and controls

396 lines (283 loc) · 13.7 KB

Node Test Helper

This test-helper module makes the node unit test framework from the Node-RED core available for node contributors.

Using the test-helper, your tests can start the Node-RED runtime, load a test flow, and receive messages to ensure your node code is correct.

Adding to your node project dependencies

Node-RED is required by the helper as a peer dependency, meaning it must be installed along with the helper itself. To create unit tests for your node project, add this test helper and Node-RED as follows:

npm install node-red-node-test-helper node-red --save-dev

This will add the helper module to your package.json file:

...
  "devDependencies": {
    "node-red":"^0.18.4",
    "node-red-node-test-helper": "^0.1.8"
  }
...

Using a local Node-RED install for tests

If you already have Node-RED installed for development, you can create a symbolic link to your local installation. For example, if Node-RED is cloned in your ~/projects folder use:

npm install ~/projects/node-red --no-save

Adding test script to package.json

To run your tests you can add a test script to your package.json file in the scripts section. To run all of the files with the _spec.js prefix in the test directory for example:

  ...
  "scripts": {
    "test": "mocha \"test/**/*_spec.js\""
  },
  ...

This will allow you to use npm test on the command line.

Creating unit tests

We recommend putting unit test scripts in the test/ folder of your project and using the *_spec.js (for specification) suffix naming convention.

Example unit test

Here is an example test for testing the lower-case node in the Node-RED documentation. Here we name our test script test/lower-case_spec.js.

test/lower-case_spec.js:

var should = require("should");
var helper = require("node-red-node-test-helper");
var lowerNode = require("../lower-case.js");

helper.init(require.resolve('node-red'));

describe('lower-case Node', function () {

  beforeEach(function (done) {
      helper.startServer(done);
  });

  afterEach(function (done) {
      helper.unload();
      helper.stopServer(done);
  });

  it('should be loaded', function (done) {
    var flow = [{ id: "n1", type: "lower-case", name: "lower-case" }];
    helper.load(lowerNode, flow, function () {
      var n1 = helper.getNode("n1");
      try {
        n1.should.have.property('name', 'lower-case');
        done();
      } catch(err) {
        done(err);
      }
    });
  });

  it('should make payload lower case', function (done) {
    var flow = [
      { id: "n1", type: "lower-case", name: "lower-case",wires:[["n2"]] },
      { id: "n2", type: "helper" }
    ];
    helper.load(lowerNode, flow, function () {
      var n2 = helper.getNode("n2");
      var n1 = helper.getNode("n1");
      n2.on("input", function (msg) {
        try {
          msg.should.have.property('payload', 'uppercase');
          done();
        } catch(err) {
          done(err);
        }
      });
      n1.receive({ payload: "UpperCase" });
    });
  });
});

In this example, we require should for assertions, this helper module, as well as the lower-case node we want to test, located in the parent directory.

We then have a set of mocha unit tests. These tests check that the node loads correctly, and ensures it makes the payload string lower case as expected.

Note how the assertion failures are caught explicitly and passed to the done() call. Node-RED swallows exceptions that are raised in the flow, so we make sure the test framework is aware of them. Not doing so would simply lead to a test timeout because done() is never called in case of an assertion failure.

Initializing Helper

To get started, we need to tell the helper where to find the node-red runtime. This is done by calling helper.init(require.resolve('node-red')) as shown.

The helper takes an optional userSettings parameter which is merged with the runtime defaults.

helper.init(require.resolve('node-red'), { 
    functionGlobalContext: { os:require('os') }
});

Getting nodes in the runtime

The asynchronous helper.load() method calls the supplied callback function once the Node-RED server and runtime is ready. We can then call the helper.getNode(id) method to get a reference to nodes in the runtime. For more information on these methods see the API section below.

Receiving messages from nodes

The second test uses a helper node in the runtime connected to the output of our lower-case node under test. The helper node is a mock node with no functionality. By adding "input" event handlers as in the example, we can check the messages received by the helper.

To send a message into the lower-case node n1 under test we call n1.receive({ payload: "UpperCase" }) on that node. We can then check that the payload is indeed lower case in the helper node input event handler.

Working with Spies

A Spy (docs) helps you collect information about how many times a function was called, with what, what it returned, etc.

This helper library automatically creates spies for the following functions on Node.prototype (these are the same functions as mentioned in the "Creating Nodes" guide):

  • trace()
  • debug()
  • warn()
  • log()
  • status()
  • send()

Warning: Don't try to spy on these functions yourself with sinon.spy(); since they are already spies, Sinon will throw an exception!

Synchronous Example: Initialization

The FooNode Node will call warn() when it's initialized/constructed if somethingGood isn't present in the config, like so:

// /path/to/foo-node.js
module.exports = function FooNode (config) {
  RED.nodes.createNode(this, config);

  if (!config.somethingGood) {
    this.warn('badness');
  }
}

You can then assert:

// /path/to/test/foo-node_spec.js
const FooNode = require('/path/to/foo-node');

it('should warn if the `somethingGood` prop is falsy', function (done) {
  const flow = {
    name: 'n1',
    somethingGood: false,
    /* ..etc.. */
  };
  helper.load(FooNode, flow, function () {
    n1.warn.should.be.calledWithExactly('badness');
    done();
  });
});

Synchronous Example: Input

When it receives input, FooNode will immediately call error() if msg.omg is true:

// somewhere in FooNode constructor
this.on('input', msg => {
  if (msg.omg) {
    this.error('lolwtf');
  }
  // ..etc..
});

Here's an example of how to make that assertion:

describe('if `omg` in input message', function () {
  it('should call `error` with "lolwtf" ', function (done) {
    const flow = {
      name: 'n1',
      /* ..etc.. */
    };
    helper.load(FooNode, flow, function () {
      const n1 = helper.getNode('n1')
      n1.receive({omg: true});
      n1.on('input', () => {
        n1.warn.should.be.calledWithExactly('lolwtf');
        done();
      });
    });
  });
});

Asynchronous Example

Later in FooNode's input listener, warn() may asynchronously be called, like so:

// somewhere in FooNode constructor function
this.on('input', msg => {
  if (msg.omg) {
    this.error('lolwtf');
  }
  // ..etc..

  Promise.resolve()
    .then(() => {
      if (msg.somethingBadAndWeird) {
        this.warn('bad weirdness');
      }
    });
});

The strategy in the previous example used for testing behavior of msg.omg will not work! n1.warn.should.be.calledWithExactly('bad weirdness') will throw an AssertionError, because warn() hasn't been called yet; EventEmitters are synchronous, and the test's input listener is called directly after the input listener in FooNode's function finished--but before the Promise is resolved!

Since we don't know when exactly warn() will get called (short of the slow, race-condition-prone solution of using a setTimeout and waiting n milliseconds, then checking), we need a different way to inspect the call. Miraculously, this helper module provides a solution.

The helper will cause the FooNode to asynchronously emit an event when warn is called (as well as the other methods in the above list). The event name will be of the format call:<methodName>; in this case, methodName is warn, so the event name is call:warn. The event Will pass a single argument: a Spy Call object (docs) corresponding to the latest method call. You can then make an assertion against this Spy Call argument, like so:

describe('if `somethingBadAndWeird` in input msg', function () {
  it('should call "warn" with "bad weirdness" ', function (done) {
    const flow = {
      name: 'n1',
      /* ..etc.. */
    };
    helper.load(FooNode, flow, function () {
      const n1 = helper.getNode('n1')
      n1.receive({somethingBadAndWeird: true});
      // because the emit happens asynchronously, this listener
      // will be registered before `call:warn` is emitted.
      n1.on('call:warn', call => {
        call.should.be.calledWithExactly('bad weirdness');
        done();
      });
    });
  });
});

As you can see, looks very similar to the synchronous solution; the only differences are the event name and assertion target.

Note: The "asynchronous" strategy will also work if and only if a synchronous call to the spy is still the most recent when we attempt to make the assertion. This can lead to subtle bugs when refactoring, so exercise care when choosing which strategy to use.

Running your tests

To run your tests:

npm test

Producing the following output (for this example):

> [email protected] test /dev/work/node-red-contrib-lower-case
> mocha "test/**/*_spec.js"

lower-case Node
  ✓ should be loaded
  ✓ should make payload lower case

2 passing (50ms)

Creating test flows with the editor

To create a test flow with the Node-RED editor, export the test flow to the clipboard, and then paste the flow into your unit test code. One helpful technique to include helper nodes in this way is to use a debug node as a placeholder for a helper node, and then search and replace "type":"debug" with "type":"helper" where needed.

Using catch and status nodes in test flows

To use catch and status or other nodes that depend on special handling in the runtime in your test flows, you will often need to add a tab to identify the flow, and associated z properties to your nodes to associate the nodes with the flow. For example:

var flow = [{id:"f1", type:"tab", label:"Test flow"},
  { id: "n1", z:"f1", type: "lower-case", name: "test name",wires:[["n2"]] },
  { id: "n2", z:"f1", type: "helper" }

Additional examples

For additional test examples taken from the Node-RED core, see the .js files supplied in the test/examples folder and the associated test code at test/nodes in the Node-RED repository.

API

Work in progress.

load(testNode, testFlows, testCredentials, cb)

Loads a flow then starts the flow. This function has the following arguments:

  • testNode: (object|array of objects) Module object of a node to be tested returned by require function. This node will be registered, and can be used in testFlows.
  • testFlow: (array of objects) Flow data to test a node. If you want to use flow data exported from Node-RED editor, export the flow to the clipboard and paste the content into your test scripts.
  • testCredentials: (object) Optional node credentials.
  • cb: (function) Function to call back when testFlows has been started.

unload()

Return promise to stop all flows, clean up test runtime.

getNode(id)

Returns a node instance by id in the testFlow. Any node that is defined in testFlows can be retrieved, including any helper node added to the flow.

clearFlows()

Stop all flows.

request()

Create http (supertest) request to the editor/admin url.

Example:

helper.request().post('/inject/invalid').expect(404).end(done);

settings(userSettings)

Merges any userSettings with the defaults returned by RED.settings. Each invocation of this method will overwrite the previous userSettings to prevent unexpected problems in your tests.

This will enable you to replicate your production environment within your tests, for example where you're using the functionGlobalContext to enable extra node modules within your functions.

// functions can now access os via global.get('os')
helper.settings({ functionGlobalContext: { os:require('os') } });

// reset back to defaults
helper.settings({ });

startServer(done)

Starts a Node-RED server for testing nodes that depend on http or web sockets endpoints like the debug node. To start a Node-RED server before all test cases:

before(function(done) {
    helper.startServer(done);
});

stopServer(done)

Stop server. Generally called after unload() complete. For example, to unload a flow then stop a server after each test:

afterEach(function(done) {
    helper.unload().then(function() {
        helper.stopServer(done);
    });
});

url()

Return the URL of the helper server including the ephemeral port used when starting the server.

log()

Return a spy on the logs to look for events from the node under test. For example:

var logEvents = helper.log().args.filter(function(evt {
    return evt[0].type == "batch";
});

Running helper examples

npm run examples

This runs tests on an included lower-case node (as above) as well as snaphots of some of the core nodes' Javascript files to ensure the helper is working as expected.