Nemo-core
Nemo-core provides a simple way to add selenium automation to your NodeJS web projects. With a powerful configuration ability provided by krakenjs/confit, and plugin architecture, Nemo-core is flexible enough to handle any browser/device automation need.
Nemo-core is built to easily plug into any task runner and test runner. But in this README we will only cover setup and architecture of Nemo-core as a standalone entity.
Getting started
Install
npm install --save-dev nemo-core
Pre-requisites
Webdriver
Please see here for more information about setting up a webdriver. As long as you have the appropriate browser or browser driver (selenium-standalone, chromedriver) on your PATH, the rest of this should work fine.
Nemo-core and Confit in 90 seconds
Nemo-core uses confit - a powerful, expressive and intuitive configuration system - to elegantly expose the Nemo-core and selenium-webdriver APIs.
Direct configuration
If you install this repo you'll get the following in examples/setup.js
. Note the NemoCore()
constructor is directly accepting the needed configuration,
along with a callback function.
NemoCore({
"driver": {
"browser": "firefox"
},
'data': {
'baseUrl': 'https://www.paypal.com'
}
}, function (err, nemo) {
//always check for errors!
if (!!err) {
console.log('Error during Nemo-core setup', err);
}
nemo.driver.get(nemo.data.baseUrl);
nemo.driver.getCapabilities().
then(function (caps) {
console.info("Nemo-core successfully launched", caps.caps_.browserName);
});
nemo.driver.quit();
});
you can use async/await as well
const Nemo = require('nemo-core');
(async () => {
const nemo = await Nemo({
driver: {
browser: 'firefox'
},
data: {
baseUrl: 'https://www.paypal.com'
}
});
await nemo.driver.get(nemo.data.baseUrl);
const caps = await nemo.driver.getCapabilities();
console.log(`nemo-core successfully launched with ${caps.get('browserName')}`);
await nemo.driver.quit();
})();
make sure to handle errors when using async/await
You can modify the example above to handle errors. Make sure to define nemo outside of your try block, so you can "quit" the driver if an error occurs after nemo has been successfully created
const Nemo = require('nemo-core');
let nemo;
(async () => {
try {
nemo = await Nemo({
driver: {
browser: 'firefox'
},
data: {
baseUrl: 'https://www.paypal.com'
}
});
await nemo.driver.get(nemo.data.baseUrl);
throw new Error('artificial error to test error handling!');
nemo.driver.quit(); // this is not called, but we can call it in our catch block
} catch (err) {
console.log(err);
// attempt to quit the driver in error scenarios
if (nemo) {
nemo.driver.quit();
}
}
})();
Run it:
$ node examples/setup.js
Nemo-core successfully launched firefox
Using config files
Look at examples/setupWithConfigFiles.js
//passing __dirname as the first argument tells confit to
//look in __dirname + '/config' for config files
NemoCore(__dirname, function (err, nemo) {
//always check for errors!
if (!!err) {
console.log('Error during Nemo-core setup', err);
}
nemo.driver.get(nemo.data.baseUrl);
nemo.driver.getCapabilities().
then(function (caps) {
console.info("Nemo-core successfully launched", caps.caps_.browserName);
});
nemo.driver.quit();
});
Note the comment above that passing a filesystem path as the first argument to NemoCore()
will tell confit to look in that directory + /config
for config files.
Look at examples/config/config.json
{
"driver": {
"browser": "config:BROWSER"
},
"data": {
"baseUrl": "https://www.paypal.com"
},
"BROWSER": "firefox"
}
That is almost the same config as the first example. But notice "config:BROWSER"
. Yes, confit will resolve that to the config property "BROWSER"
.
Run this and it will open the Firefox browser:
$ node examples/setup.js
Nemo-core successfully launched firefox
Now run this command:
$ node examples/setupWithConfigDir.js --BROWSER=chrome
Nemo-core successfully launched chrome
Here, confit resolves the --BROWSER=chrome
command line argument and overrides the BROWSER
value from config.json
Now this command:
$ BROWSER=chrome node examples/setupWithConfigDir.js
Nemo-core successfully launched chrome
Here, confit resolves the BROWSER
environment variable and overrides BROWSER
from config.json
What if we set both?
$ BROWSER=chrome node examples/setupWithConfigDir.js --BROWSER=phantomjs
Nemo-core successfully launched chrome
You can see that the environment variable wins.
Now try this command:
$ NODE_ENV=special node examples/setupWithConfigDir.js
Nemo-core successfully launched phantomjs
Note that confit uses the value of NODE_ENV to look for an override config file. In this case config/special.json
:
{
"driver": {
"browser": "phantomjs"
},
"data": {
"baseUrl": "https://www.paypal.com"
}
}
Hopefully this was an instructive dive into the possibilities of Nemo-core + confit. There is more to learn but hopefully this is enough to whet your appetite for now!
Nemo-core and Plugins in 60 Seconds
Look at the example/setupWithPlugin.js
file:
NemoCore(basedir, function (err, nemo) {
//always check for errors!
if (!!err) {
console.log('Error during Nemo-core setup', err);
}
nemo.driver.getCapabilities().
then(function (caps) {
console.info("Nemo-core successfully launched", caps.caps_.browserName);
});
nemo.driver.get(nemo.data.baseUrl);
nemo.cookie.deleteAll();
nemo.cookie.set('foo', 'bar');
nemo.cookie.getAll().then(function (cookies) {
console.log('cookies', cookies);
console.log('=======================');
});
nemo.cookie.deleteAll();
nemo.cookie.getAll().then(function (cookies) {
console.log('cookies', cookies);
});
nemo.driver.quit();
});
Notice the nemo.cookie
namespace. This is actually a plugin, and if you look at the config for this setup:
{
"driver": {
"browser": "firefox"
},
"data": {
"baseUrl": "https://www.paypal.com"
},
"plugins": {
"cookie": {
"module": "path:./nemo-cookie"
}
}
}
You'll see the plugins.cookie
section, which is loading examples/plugin/nemo-cookie.js
as a plugin:
'use strict';
module.exports = {
"setup": function (nemo, callback) {
nemo.cookie = {};
nemo.cookie.delete = function (name) {
return nemo.driver.manage().deleteCookie(name);
};
nemo.cookie.deleteAll = function () {
return nemo.driver.manage().deleteAllCookies();
};
nemo.cookie.set = function (name, value, path, domain, isSecure, expiry) {
return nemo.driver.manage().addCookie(name, value, path, domain, isSecure, expiry)
};
nemo.cookie.get = function (name) {
return nemo.driver.manage().getCookie(name);
};
nemo.cookie.getAll = function () {
return nemo.driver.manage().getCookies();
};
callback(null);
}
};
Running this example:
$ node examples/setupWithPlugin.js
Nemo-core successfully launched firefox
cookies [ { name: 'foo',
value: 'bar',
path: '',
domain: 'www.paypal.com',
secure: false,
expiry: null } ]
=======================
cookies []
$
This illustrates how you can create a plugin, and the sorts of things you might want to do with a plugin.
API
Nemo-core
var nemo = NemoCore([[nemoBaseDir, ][config, ][callback]] | [Confit object]);
@argument nemoBaseDir {String}
(optional) - If provided, should be a filesystem path to your test suite. Nemo-core will expect to find a /config
directory beneath that.
<nemoBaseDir>/config/config.json
should have your default configuration (described below). nemoBaseDir
can alternatively be set as an environment variable. If it is
not set, you need to pass your configuration as the config
parameter (see below).
@argument config {Object}
(optional) - Can be a full configuration (if nemoBaseDir
not provided) or additional/override configuration to what's in your config files.
@argument callback {Function}
(optional) - This function will be called once the nemo
object is fully resolved. It may be called with an error as the first argument which has important debugging information. So make sure to check for an error. The second argument is the resolved nemo
object.
@argument Confit object {Object}
(optional) - If a Confit object is passed, the configuration step is skipped and the passed object is used directly.
@returns nemo {Object|Promise}
- Promise returned if no callback provided. Promise resolves with the same nemo object as would be given to the callback.
The nemo object has the following properties:
-
driver The live
selenium-webdriver
API. See WebDriver API Doc: http://seleniumhq.github.io/selenium/docs/api/javascript/module/selenium-webdriver/index_exports_WebDriver.html -
wd This is a reference to the
selenium-webdriver
module: http://seleniumhq.github.io/selenium/docs/api/javascript/module/selenium-webdriver/index.html -
data Pass through of any data included in the
data
property of the configuration -
_config This is the confit configuration object. Documented here: https://github.com/krakenjs/confit#config-api
-
plugin namespace(s) Plugins are responsible for registering their own namespace under nemo. The convention is that the plugin should register the same namespace as is specified in the plugin configuration. E.g. the
nemo-view
module registers itself asnemo.view
and is configured like this:
{
"driver": ...
"plugins": {
"view": {
"module": "nemo-view"
}
}
You could also have a config that looks like this, and nemo-view
will still register itself as nemo.view
{
"driver": ...
"plugins": {
"cupcakes": {
"module": "nemo-view"
}
}
But that's confusing. So please stick to the convention.
Typical usage of Nemo-core constructor
A typical pattern would be to use mocha
as a test runner, resolve nemo
in the context of the mocha before
function, and use
the mocha done
function as the callback:
var nemo;
describe('my nemo suite', function () {
before(function (done) {
NemoCore(config, function (err, resolvedNemo-core) {
nemo = resolvedNemo-core;
done(err)
});
});
it('will launch browsers!', function (done) {
nemo.driver.get('https://www.paypal.com');
nemo.driver.quit().then(function () {
done();
});
});
});
Configure
Calling Configure
will return a promise which resolves as a Confit object. This is the same method Nemo-core calls internally in the basic use case. You might want to call Configure
if
you are interested in the resolved configuration object but not yet ready to start the webdriver. An example would be if you want to make further
changes to the configuration based on what gets resolved, prior to starting the webdriver.
function Configure([nemoBaseDir, ][configOverride])
@argument nemoBaseDir {String}
(optional) - If provided, should be a filesystem path. There should be a /config
directory beneath that.
<nemoBaseDir>/config/config.json
should have your default configuration. nemoBaseDir
can alternatively be set as an environment variable. If it is
not set, you need to pass your configuration as the config
parameter (see below).
@argument config {Object}
(optional) - Can be a full configuration (if nemoBaseDir
not provided) or additional/override configuration to what's in your config files.
@returns {Promise}
- Promise resolves as a confit object:
Configuration Input
{
"driver": { /** properties used by Nemo-core to setup the driver instance **/ },
"plugins": { /** plugins to initialize **/},
"data": { /** arbitrary data to pass through to nemo instance **/ }
}
This configuration object is optional, as long as you've got nemoData
set as an environment variable (see below).
driver
Here are the driver
properties recognized by Nemo-core. This is ALL of them. Please be aware that you really only need to supply "browser" to get things working initially.
browser (optional)
Browser you wish to automate. Make sure that your chosen webdriver has this browser option available. While this is "optional" you must choose a browser. Either use this property or the builders.forBrowser
option (see below).
If both are specified, builders.forBrowser
takes precedence.
local (optional, defaults to false)
Set local to true if you want Nemo-core to attempt to start a standalone binary on your system (like selenium-standalone-server) or use a local browser/driver like Chrome/chromedriver or PhantomJS.
server (optional)
Webdriver server URL you wish to use. This setting will be overridden if you are using builders.usingServer
serverProps (optional/conditional)
Additional server properties required of the 'targetServer'
You can also set args and jvmArgs to the selenium jar process as follows:
"serverProps": {
"port": 4444,
"args": ["-firefoxProfileTemplate","/Users/medelman/Desktop/ffprofiles"],
"jvmArgs": ["-someJvmArg", "someJvmArgValue"]
}
jar (optional/conditional)
Path to your webdriver server Jar file. Leave unset if you aren't using a local selenium-standalone Jar (or similar).
serverCaps (optional)
serverCaps would map to the capabilities here: http://seleniumhq.github.io/selenium/docs/api/javascript/module/selenium-webdriver/index_exports_Capabilities.html
Some webdrivers (for instance ios-driver, or appium) would have additional capabilities which can be set via this variable. As an example, you can connect to saucelabs by adding this serverCaps:
"serverCaps": {
"username": "medelman",
"accessKey": "b38e179e-079a-417d-beb8-xyz", //not my real access key
"name": "Test Suite Name", //sauce labs session name
"tags": ['tag1','tag2'] //sauce labs tag names
}
proxyDetails (optional)
If you want to run test by setting proxy in the browser, you can use 'proxyDetails' configuration. Following options are available: direct, manual, pac and system. Default is 'direct'. For more information refer : http://seleniumhq.github.io/selenium/docs/api/javascript/module/selenium-webdriver/proxy.html
"proxyDetails" : {
"method": "manual",
"args": [{"http": "localhost:9001","ftp":"localhost:9001","https":"localhost:9001"}]
}
builders (optional)
This is a JSON interface to any of the Builder methods which take simple arguments and return the builder. See the Builder class here: http://seleniumhq.github.io/selenium/docs/api/javascript/module/selenium-webdriver/index_exports_Builder.html
Useful such functions are:
- forBrowser (can take the place of "browser", "local" and "jar" properties above)
- withCapabilities (can take the place of "serverCaps" above)
The nemo setup routine will prefer these "builder" properties over other abstracted properties above, if there is a conflict.
selenium.version (optional)
Since nemo requires a narrow range of versions of selenium-webdriver, you may have a need to upgrade selenium-webdriver (or downgrade) outside of the supported versions that nemo uses.
You can do that by using selenium.version
. E.g.
"driver": {
"browser": "firefox",
"selenium.version": "^2.53.1"
}
Nemo-core will upgrade its internal dependency to what is set in this property. The npm install
will only run if the version specified is not already installed.
custom driver
You can also provide a module, which exports a function that returns a fully formed WebDriver
object. To do so, follow
this example: #177 (comment)
plugins
Plugins are registered with JSON like the following (will vary based on your plugins)
{
"plugins": {
"samplePlugin": {
"module": "path:plugin/sample-plugin",
"arguments": [...],
"priority": 99
},
"view": {
"module": "nemo-view"
}
}
}
Plugin.pluginName parameters:
-
module {String} - Module must resolve to a require'able module, either via name (in the case it is in your dependency tree) or via path to the file or directory.
-
arguments {Array} (optional, depending on plugin) - Your plugin will be called via its setup method with these arguments:
[configArg1, configArg2, ..., ]nemo, callback
. Please note that the braces there indicate "optional". The arguments will be applied viaFunction.apply
-
priority {Number} (optional, depending on plugin) - A
priority
value of < 100 will register this plugin BEFORE the selenium driver object is created. This means that such a plugin can modifyconfig
properties prior to driver setup. Leavingpriority
unset will register the plugin after the driver object is created.
data
Data will be arbitrary stuff that you might like to use in your tests. In a lot of the examples we set data.baseUrl
but again, that's arbitrary and not required. You could
pass and use data.cupcakes
if you want. Cupcakes are awesome.
Shortstop handlers
Shortstop handlers are data processors that key off of directives in the JSON data. Ones that are enabled in nemo are:
path
use path to prepend the nemoBaseDir
(or process.cwd()
) to a value. E.g. if nemoBaseDir
is .../myApp/tests
then
a config value of 'path:plugin/myPlugin'
will resolve to .../myApp/tests/plugin/myPlugin
env
use env to reference environment variables. E.g. a config value of 'env:PATH'
will resolve to
/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:...
argv
use argv to reference argv variables. E.g. a config value of 'argv:browser'
will resolve to firefox
if you
set --browser firefox
as an argv on the command line
config
Use config to reference data in other parts of the JSON configuration. E.g. in the following config.json:
{
"driver": {
...
},
"plugins": {
"myPlugin": {
"module": "nemo-some-plugin",
"arguments": ["config:data.someProp"]
}
},
"data": {
"someProp": "someVal"
}
}
The value of plugins.myPlugin.arguments[0]
will be someVal
file
Please see https://github.com/krakenjs/shortstop-handlers#handlersfilebasedir-options
base64
Please see https://github.com/krakenjs/shortstop-handlers#handlersbase64
require
Please see https://github.com/krakenjs/shortstop-handlers#handlersrequirebasedir
exec
Please see https://github.com/krakenjs/shortstop-handlers#handlersexecbasedir
glob
Please see https://github.com/krakenjs/shortstop-handlers#handlersglobbasediroptions
Plugins
Authoring a plugin, or using an existing plugin, is a great way to increase the power and usefulness of your Nemo-core installation. A plugin should add
its API to the nemo
object it receives and passes on in its constructor (see "plugin interface" below)
plugin interface
A plugin should export a setup function with the following interface:
module.exports.setup = function myPlugin([arg1, arg2, ..., ]nemo, callback) {
...
//add your plugin to the nemo namespace
nemo.myPlugin = myPluginFactory([arg1, arg2, ...]); //adds myMethod1, myMethod2
//error in your plugin setup
if (err) {
callback(err);
return;
}
//continue
callback(null);
};
When nemo initializes your plugin, it will call the setup method with any arguments supplied in the config plus the nemo object, plus the callback function to continue plugin initialization.
Then in your module where you use Nemo-core, you will be able to access the plugin functionality:
var Nemo-core = require('nemo');
NemoCore({
'driver': {
'browser': 'firefox',
'local': true,
'jar': '/usr/local/bin/selenium-server-standalone.jar'
},
'data': {
'baseUrl': 'https://www.paypal.com'
},
'plugins': {
'myPlugin': {
'module': 'path:plugin/my-plugin',
'arguments': [...]
'priority': 99
},
}
}, function (err, nemo) {
nemo.driver.get(nemo.data.baseUrl);
nemo.myPlugin.myMethod1();
nemo.myPlugin.myMethod2();
nemo.driver.sleep(5000).
then(function() {
console.info('Nemo-core was successful!!');
nemo.driver.quit();
});
});
Logging and debugging
Nemo-core uses the debug module for console logging and error output. There are two classes of logging, nemo:log
and nemo:error
If you want to see both classes of output, simply use the appropriate value of the DEBUG environment variable when you run nemo:
$ DEBUG=nemo:* <nemo command>
To see just one:
$ DEBUG=nemo:error <nemo command>
Why Nemo-core?
Because we NEed MOre automation testing!
Unit Tests
-
Unit tests run by default using headless browser PhantomJS. To run unit tests out of box, You must have PhantomJS installed on your system and must be present in the path
-
If you want to run unit tests on your local browser, like lets say Firefox/Chrome (make sure ChromeDriver is in current path), you need to update browser in unit test configuration, for example the browser section under
test/config/config.json
like here -
How to run unit tests?
npm test
will run unit tests as well as lint taskgrunt simplemocha
will just run unit testsgrunt
- default grunt task will run linting as well as unit tests- To run directly using mocha assuming its globally installed on your system
mocha -t 60s
- Or a specific test,
mocha --grep @allArgs@ -t 60s
- Or post
npm install
on nemo module, you can runnode_modules/.bin/mocha --grep @allArgs@ -t 60s