Selenium Architecture

Architecture

The WebDriver architecture does not follow the same approach as Selenium RC, which

was written purely in JavaScript for all the browser automation. The JavaScript, in Selenium

RC, would then emulate user actions. This JavaScript would automate the browser from

within the browser. WebDriver on the other hand tries to control the browser from outside

the browser. It uses accessibility API to drive the browser. The accessibility API is used by

a number of applications for accessing and controlling applications when they are used by

disabled users and is common to web browsers.

WebDriver uses the most appropriate way to access the accessibility API. If we look at

Firefox, it uses JavaScript to access the API. If we look at Internet Explorer, it uses C++. This

approach means we can control browsers in the best possible way but has the downside

that new browsers entering the market will not be supported straight away like we can with

Selenium RC.

Where that approach doesn't work we will then inject JavaScript into the page. Examples of

this are found in the new HTML5.                             

   

WebDriver API

  

WebDriver  SPI

JSON  Wire Protocol

Selenium Server

The system is made up of four different sections.

WebDriver API

The WebDriver API is the part of the system that you interact with all the time. Things

have changed from the 140 line long API that the Selenium RC API had. This is now more

manageable and can actually fit on a normal screen. You will see this when you start using

WebDriver in the next chapter. This is made up of the WebDriver and the WebElement objects.

                driver.findElement(By.name("q"))

           and

              element.sendKeys("I love cheese");

WebDriver SPI

When code enters the Stateless Programming Interface or SPI, it is then called to a

mechanism that breaks down what the element is, by using a unique ID, and then calling a

command that is relevant. All of the API calls above then call down.

Using the example in the previous section would be like the following code, once it was

in the SPI:

              findElement(using="name", value="q")

             sendKeys(element="webdriverID", value="I love cheese")

From there we call the JSON Wire protocol. We still use HTTP as the main transport

mechanism. We communicate to the browsers and have a simple client server transport

architecture the WebDriver developers created the JSON Wire Protocol.

 JSON Wire protocol

The WebDriver developers created a transport mechanism called the JSON Wire Protocol.

This protocol is able to transport all the necessary elements to the code that controls it. It

uses a REST like API as the way to communicate.

Selenium server

The Selenium server, or browser, depending on what is processing, uses the JSON Wire

commands to break down the JSON object and then does what it needs to. This part of the

code is dependent on which browser it is running on.As mentioned earlier, it could be done in the browser via C++; if it's in IE or if not available we inject Selenium.