Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
In-process Selenium. Wraps the Selenium JAR to provide JavaScript bindings for running Selenium based tests
CoffeeScript Java
branch: master

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.
src
.gitignore
Makefile
README.md
index.js
package.json

README.md

Overview

Most Selenium solutions involve running the Selenium RC Server JAR file (eg java -jar selenium-server-standalone-2.25.0.jar), letting that listen on port 4444, and then sending commands to it over a TCP connection. This is what clients like soda do.

That works great for the most part, but has a few downsides:

  • Requires getting devs to download, setup and run yet another localhost service
  • No support for HTTP proxy (needed if you want to inject test fixture data)
  • Less flexible since only part of Selenium's true functionality is exposed through the service

This project is slighly different. Instead, it uses the same Selenium JAR file, but it loads it in-process using the robust node-java bridge. In effect, this creates "JS bindings" to Selenium based on the Java interface. So we can now do advanced stuff with Selenium without having to convert the entire project to Java.

Most notably, this means support for setting up an HTTP proxy that can be used to mock some or all of the responses from the server. It also means that I don't have to explain to other developers how to launch the Selenium service or create a launchd entry for it. There is no service. They just run the test script, and a browser magically pops open and the tests run.

For now, this project is the metaphorical "tip of the iceberg". I'm integrating Zombie.js, wrapping both that and Selenium under one roof, throwing in a configurable proxy for storing traces of the server->client traffic, replaying said traces, etc. This is the first piece of that puzzle.

Installing

npm install selenium-inproc

During install, npm will trigger the Makefile to download selenium-server-standalone-2.25.0.jar (it's ~31M). Once this finishes, you should be good to go.

Example

This is CoffeeScript, which is directly convertable to JS, so don't be intimidated if it's a new syntax to you. Same thing, just more terse.

Selenium = require('selenium-inproc')
s = new Selenium(
  url: 'http://m-local.wavii.com:3000'
  proxy: "localhost:#{PROXY_PORT}"
)
s.openBrowser () ->
  s.open '/users/auth/facebook', () ->
    s.type 'email', 'username@gmail.com'
    s.type 'pass', 'sekretPassword'
    s.click 'login', () ->

      s.open '/feed/main', () ->
        console.log "Title is #{s.getTitle()}"
        console.log "Attribute is #{s.getAttribute('css=.inline-topic@data-topic-id')}"

        s.close()
        setTimeout(() ->
          console.log "Buh-Bye"
          Proxy.close()
        , 1000)

The commands supported are just the same ones as in the Javadoc, which is pretty much just Selenese.

Here's the exaustive list: src/commands.coffee

Something went wrong with that request. Please try again.