StaleElementException when Clicking on a TableRow

问题:

<div id="crm" class="row gridrow clickable ng-scope" ng-repeat="customer in customerList" ng-click="gotoRecord(customer.id)">
     <i class="col m1 s1 tiny fa fa-male"></i>
     <div class="col m3 s11 ng-binding"> Allard</div>
     <div class="col m2 s12 ng-binding"></div>
</div>

I have this snippet of HTML, it's one row displaying as a result of a search action for a Customer with nameCustomer 'Allard'. I want to click on this customer to continue to the next page but most of the time this results in a StaleElementException.

I tried it two different ways, using Protractor and without Protractor.

First way:

IWebElement elem = driver.FindElement(By.XPath("//*[contains(text(),'" + nameCustomer + "')]//parent::div[contains(@id,'crm')]"));
        ExplicitWait.WaitAndClick(driver, elem);

Second way:

var customers = driver.FindElements(NgBy.Repeater("customer in customerList"));
        foreach (var customer in customers)
        {
            if (elem.Text.Equals(nameCustomer))
            {
                elem.Click();
            }
        }

回答1:

The problem (I think)

With StaleReferenceExceptions, an IWebElement that was previously created is no longer attached to the DOM (you probably already know this). What's most likely happening is this:
1: You click search.
2: Selenium executes driver.FindElement(...) and finds a matching element.
3: Then the search function finishes and the DOM updates. The old IWebElement found previously is gone.
4: Then Selenium tries to click on the element (which no longer exists, causing the StaleElementException. There is an element that matches the one that was there before, but it's not the same element in Selenium's eyes.)

Your statement that this happens "most of the time" makes me suspect this is the case even more, because the exception would depend on the order of events, which would vary depending on the relative speeds of Selenium vs. the web-page.

How to resolve (if this is your problem)

You need to find something on the page that will indicate to Selenium that the search action is done. This is where the creativity of writing GUI automation scripts comes in. If there is something on the page that you know will change as a result of the load, craft an explicit wait to ensure that is complete. Maybe there is a loading bar that shows up, or a message that appears when the search is done. You could grab an element that doesn't match your search before clicking search, then do an explicit wait to make sure it disappears before going on to look for the result you do expect to be there.

It would look something like this below.

# Search action performed before this.
WebDriverWait wait= new WebDriverWait(driver, TimeSpan.FromSeconds(secondsToWait));
wait.Until(ExpectedConditions.InvisibilityOfElementLocated(By.XPath( expressionForElementThatWillDissapear )));


IWebElement elem = driver.FindElement(By.XPath("//*[contains(text(),'" + nameCustomer + "')]//parent::div[contains(@id,'crm')]"));
            ExplicitWait.WaitAndClick(driver, elem);

I'd recommend making the method to implement the explicit wait above. These situations will come up often.