What do test automation frameworks provide that my approach of scripting with Selenium Webdriver doesn't?

Question

I am a newbie to software test automation and have written the following test script in Selenium Webdriver and Ruby binding. It performs user actions (clicks, enters, fills up values etc.). I have put basic asserts that match the screen text value with value I provide. Below is the code:

puts "Test Run 1 has started ""["+Time.now.strftime('%H:%M:%S CST')+"]"
require "selenium-webdriver"
require "colorize"
profile = Selenium::WebDriver::Firefox::Profile.new
profile['browser.cache.disk.enable'] = false
browser = $browser = Selenium::WebDriver.for :firefox, :profile => profile

browser.manage().window().maximize();
browser.get "https://cameleon-6945--dev.cs11.cloudforce.com"
main_window = browser.window_handle
browser.find_element(name:"username").clear()
browser.find_element(name:"username").send_keys "abcd@vertex.com"
browser.find_element(name:"pw").send_keys "1234"
browser.find_element(name:"Login").click
browser.find_element(link_text:"Cameleon Quotes").click
wait = Selenium::WebDriver::Wait.new(:timeout => 45)
#Open Test Run 1 created quote
wait.until {browser.find_element(:css,"#bodyCell > div.bRelatedList > div.hotListElement > div > div.pbBody > table > tbody > tr.dataRow.even.first > th > a")}
browser.find_element(:css,"#bodyCell > div.bRelatedList > div.hotListElement > div > div.pbBody > table > tbody > tr.dataRow.even.first > th > a").click


wait.until {browser.find_element(:xpath,"/html/body/div[3]/div[3]/div[2]/div[3]/div[2]/div[2]/div[2]/form/div[3]/div/div[2]/div[2]/div/div/div/table")}
#browser.save_screenshot "Cart Overview - RegressionRun @ "+Time.now.strftime('%Y-%m-%d %H%M%S')+".jpeg"

#wait.until {browser.find_element(:css,"body > div.mainPartBox > div.boxBody > div.main > div.processBar > div.backgroundProcessBarMiddle > a:nth-child(7) > div.processBarElement.noSelected > div")}
#browser.find_element(:css,"body > div.mainPartBox > div.boxBody > div.main > div.processBar > div.backgroundProcessBarMiddle > a:nth-child(7) > div.processBarElement.noSelected > div").click

#wait.until {browser.find_element(:css,"body > div.CombinedBox > div.boxBody > div.main > div:nth-child(8) > iframe")}
#wait.until {browser.find_element(:xpath,"/html/body/div[3]/div[3]/div[2]/div[3]/iframe")}
browser.manage.timeouts.page_load = 35

#browser.switch_to.frame(cart_frame) 
puts "\n\n"
puts "Assertions to verify cart content values\n\n".yellow
element_value1 = browser.find_element(:css,"#total > tbody > tr > td:nth-child(4) > span").text
if element_value1 == "$314,507.30"
  puts 'Contract Sales Price = ' +element_value1
  puts 'Value as expected in the cart, Test Passed'
else
  puts 'Test failed, Contract Sales price value does not match the expected value'
end

puts "\n"
element_value2 = browser.find_element(:css,"#total > tbody > tr > td:nth-child(6) > span").text
if element_value2 == "$157,253.65"
    puts 'Contract Cost Price = ' +element_value2 
    puts 'Value as expected in the cart, Test Passed'
else
    puts 'Test failed, Contract Cost price value does not match the expected value'.red
end

puts "\n"
element_value3 = browser.find_element(:css,"#total > tbody > tr > td:nth-child(8) > span").text
if element_value3 == "50.00"
    puts 'Contract gross margin = ' +element_value3
    puts 'Value as expected in the cart, Test Passed'
else    
    puts 'Test failed, Gross margin value does not match the expected value'
end

puts "\n\n\t\t\t\t\tTest Case 1 passed successfully, proceeding to Test Case 2 - Generate Document\n".green

It's not integrated with any framework, it's just a test script.

My questions are:

Is this test totally worthless if compared to tools like Cucumber/Capybara etc.?

I have captured most of the elements using XPaths and this way the only way since it lacked classes, ids etc. Now if there's a minor change in page structure, like a new div is introduced, this script will fail with NoElemenFoundError. Is there any way we can avoid it? Or this is the only way test scripts are written and we need to update them regularly with new developments?

Answer 1

That exercise is worth doing once for your own experience. A couple of ways that using rspec for example would be an improvement are that rspec gives you a way to organize multiple tests, and that it boils down the results of your tests to a single indicator of success or failure, so that you can just look at the last line of output (or have your CI server look at the exit status) instead of reading pages of messages to see whether all your tests passed. Cucumber differs from rspec in that it allows you to read your entire test in English without reading any code in between, which is extremely valuable for thinking through requirements and might allow you to collaborate on your tests with non-programmers. Capybara provides methods that do much what you're doing in your script but are more succinct.

Sensitivity to page structure is an issue even with more sophisticated tools. One way to minimize the issue is to not assert any more of the page structure than you have to. If there is only one table on the page and only one row of results then td:nth-child(6) is all the selector you need. Another way to deal with that issue is to add IDs or classes to your pages to support testing.

Answer 2

I would say it is worth to have such scripts, but as soon as your number of scripts grows you need to manage them in a way where minimum maintenance is required.

So you can use various strategies to design a framework where you can have central object repository and test data. Refer here! for more details on designs.

I agree with Dave, to avoid assert on page structure. And if you maintain a central repository then there would be little modification needed if there is changes in structure of a page.