How to handle a popup in selenium webdriver?

What can be the best design strategy to follow for designing an automation frame work for a Web App using webdriver (python)?

• I have been given a task of automating an enterprise website. I am starting the project from scratch. I have zeroed in on using Selenium WebDriver with Python for this. Since last few days i am going through various blogs on the internet suggesting to use Page Object model. If people who have experience on this give inputs on the best model to follow keeping in mind code re-usability and robustnus, will be highly appreciated. Any pointers to any resource suggesting the same are also welcome.

• Answer:

  I have developed a testing framework based on Selenium WebDriver for a client, to test a web application, and can share the lessons learned. The framework and the tests were written in Java, but mostly the same principles should apply. I will use pseudo code below to illustrate my points. Usage of the Page (or View) Objects pattern The basis of the page objects is to separate between the test code and the definitions of how to access the element (i.e. the element selectors). Those shouldn't be limited to actual web pages, but any view, or part of the view in the web application. For example: The Login Page itself, A navigation bar that appears in multiple pages, A date picker control. Take each view component like that and wrap the element access. Allow composition and inheritence of For example: class BlogPost extends BlogPage TagList postTags Searchbox postsSearch ShareDialog clickShareButton: .... Have parametrized view objects for reusable components, that in each container view can have different settings and selectors: DropDownMenu (itemsList, parentDivSelector) Log simulated user interactions in the components, to understand the executed script scenario, and where the tests failed. (more about reporting coming up...) class Button (name, selector) click: testContext.log("Button {name} clicked") Element(selector).click() A test context to log additional data toThose are not unit tests, but complex user scenarios. The tests should be able to save additional data to it. You should then be able to display it in the reports (below) under the correct test. Some examples of kind of information is used to save under the executing test context: Screen captures on explicit calls to save them implicit screen captures in the end of or in errors Logging messages, in different levels (info, traces, etc) Time stamped information for all the above Test environment metadata: Browser, OS, etc. Web automation tests reportsAs stated, the kind of information needed to understand what went wrong with the test scenarios, a different report than a simple unit test report is needed. It needs to report the data you logged in the text context in a scenario like fashion. For example: Login test Browser: Chrome 30 on Windows 00:00 Starting new browser session 00:00 Opening home page 00:01 Clicked login 00:05 Navigation to login page completed 00:05 entered "amitay" in username textbox 00:06 enter "secret" in password. 00:06 [screen shot] 00:08 clicked submit button 00:10 Navigation to dashoard page completed. 00:11 ***Error: expected url to be .... 00:12 [screenshot] Additional test runner capabilitiesYour test framework (either your own implementation or the one you are leveraging) might need to support additional capabilities: Parallel test execution (web automation using selenium can be slow...) Dependent tests (run testC if and when testB and testC are completed Parametrized test (for example, to run the same tests on different WebDrivers (i.e. different browsers) Creating a new browser session, and continuing one from a completed test. Configuration The tests might run in different configurations: Continuous integration server, development, nightly tests, per commit tests. Some configuration that you might need: Web driver instance creation settings 3rd party (such as sauce labs) authentication Which tests to run Which tests to run on which browsers Timeouts Some other suggestions Consider using a 3rd party "Remote Web Driver as a service", such as browserstack or saucelabs. The main advantage is saving the huge pain which is maintaing a stable multiple browser environments. Additional advantage is the additional features they add (such as recorded videos, selenium logging). Write some tests as you write the framework. The issues you'll encounter will teach you a lot about what are the actual needs. Don't write too complex test scenarios. Web drivers test in real-world can be quite fragile, and too complex tests may lead to many false negatives, when it's the test which failed. Consider when it is better to write javascript unit tests instead of browser automation tests. They tend to be much more robust. A tip for java developers: Give testNG a chance. I found it very powerful and extendable, and is a great test harness for more than unit tests. This ended up rather lengthy, more of a post than an answer. If you've read it this far, it might be time to go write some selenium tests! Good luck!