Using watir-webdriver, how do I wait for a page to load after I click a link?
At the moment I am using:
sleep n
But this is not ideal as the page response varies so much.
Is there a way to test whether the page is ready or whether there is a certain element in the page? I understand in the normal Watir gem there is Watir::Waiter.wait_until or something similar, but I don't see this in the webdriver version.
I don't know if they're the best way, but this is how I'm handling this for waiting for an updating div to clear:
while browser.div(:id=>"updating_div").visible? do sleep 1 end
This is how I handle waiting for something to display:
until browser.div(:id=>"some_div").exists? do sleep 1 end
Today's release adds an optional require that brings in some helpers for waiting for elements. These are not (at the moment) available in Watir 1.6, so be aware if you use both libraries side by side.
Check "AJAX and waiting for elements" in the Watir-webdriver Wiki for more information.
The best summary is found in "Waiting".
This is it in a nutshell:
require 'watir-webdriver'
b = Watir::Browser.start 'bit.ly/watir-webdriver-demo'
b.select_list(:id => 'entry_1').wait_until_present
b.text_field(:id => 'entry_0').when_present.set 'your name'
b.button(:value => 'Submit').click
b.button(:value => 'Submit').wait_while_present
Watir::Wait.until { b.text.include? 'Thank you' }
This is how I wait for AJAX in my project:
ajax_loader = $b.element(:xpath => "//*[#id='spinner-modal-transparent' and #aria-hidden='true']/div/div/div/div/img[#alt='Ajax transparent loader']")
if ajax_loader.exists?
ajax_loader.wait_while_present(timeout=350)
else
puts "The AJAX loader was not present."
end
browser.wait_until can be used.
It's more helpful because you can define what to wait for in the parameters (()), as in:
browser.wait_until(browser.text.include("some text"))
You can use the wait_until or waituntilExists methods.
I had the same problem, and I tried to fix it by combining wait_until_present and
until browser.div(:id=>"some_div").exists? do sleep 1 end
tricks povided by #marc:
some_div = browser.div(:id => 'some_div')
begin
Watir::Wait.until
some_div.visible?
end
rescue Watir::Wait::TimeoutError
until some_div.visible?
sleep 1
end
end
Notice that it is your own responsibility to make sure that
div(:id => "some_div")
does exist.
Related
While trying to set up some simple end-to-end tests with Jest and Puppeteer, I've found that any test I write will inexplicably fail with a timeout.
Here's a simple example test file, which deviates only slightly from Puppeteer's own example:
import puppeteer from 'puppeteer';
describe('Load Google Puppeteer Test', () => {
test('Load Google', async () => {
const browser = await puppeteer.launch({
headless: false
});
const page = await browser.newPage();
await page.goto('https://google.co.uk');
await expect(page).toMatch("I'm Feeling Lucky");
await browser.close();
});
});
And the response it produces:
TimeoutError: Text not found "I'm Feeling Lucky"
waiting for function failed: timeout 500ms exceeded
I have tried adding in custom timeouts to the goto line, the test clause, amongst other things, all with no effect. Any ideas on what might be causing this? Thanks.
What I would say is happening here is that using toMatch expects text to be displayed. However, in your case, the text you want to verify is text associated with a button.
You should try something like this:
await expect(page).toMatchElement('input[value="I\'m Feeling Lucky"]');
Update 1:
Another possibility (and it's one you've raised yourself) is that the verification is timing out before the page has a chance to load. This is a common issue, from my experience, with executing code in headless mode. It's very fast. Sometimes too fast. Statements can be executed before everything in the UI is ready.
In this case you're better off adding some waitForSelector statements throughout your code as follows:
await page.waitForSelector('input[value="I\'m Feeling Lucky"]');
This will ensure that the selector you want is displayed before carrying on with the next step in your code. By doing this you will make your scripts much more robust while maintaining efficiency - these waits won't slow down your code. They'll simply pause until puppeteer registers the selector you want to interact with / verify as being displayed. Most of the time you won't even notice the pause as it will be so short (I'm talking milliseconds).
But this will make your scripts rock solid while also ensuring that things won't break if the web page is slower to respond for any reason during test execution.
You're probably using 'expect-puppeteer' package which does the toMatch expect. This is not a small deviation. The weird thing is that your default timeout isn't 30 seconds as the package's default, check that.
However, to fix your issue:
await expect(page).toMatch("I'm Feeling Lucky", { timeout: 6000 });
Or set the default timeout explicitly using:
page.setDefaultTimeout(timeout)
See here.
How can I let the user know when they are getting a hot code push?
At the moment the screen will go blank during the push, and the user will feel it's rather weird. I want to reassure them the app is updating.
Is there a hook or something which I can use?
Here's the shortest solution I've found so far that doesn't require external packages:
var ALERT_DELAY = 3000;
var needToShowAlert = true;
Reload._onMigrate(function (retry) {
if (needToShowAlert) {
console.log('going to reload in 3 seconds...');
needToShowAlert = false;
_.delay(retry, ALERT_DELAY);
return [false];
} else {
return [true];
}
});
You can just copy that into the client code of your app and change two things:
Replace the console.log with an alert modal or something informing the user that the screen is about to reload.
Replace ALERT_DELAY with some number of milliseconds that you think are appropriate for the user to read the modal from (1).
Other notes
I'd recommend watching this video on Evented Mind, which explains what's going on in a little more detail.
You can also read the comments in the reload source for further enlightenment.
I can image more complex reload logic, especially around deciding when to allow a reload. Also see this pacakge for one possible implementation.
You could send something on Meteor.startup() in your client-side code. I personally use Bert to toast messages.
I've been looking a bit at Selenium, and I'm beginning to like it, since I know some Java programming and find both Java and C# pretty straight-forward for simple things like this.
However, I'm struggling with a test that creates a new post in Wordpress, from the Dashboard page:
This is the Selenium code (in C#):
(The Driver instance is, obviously, a driver class I've created - for starting the browser and connecting to the wordpress site.)
1: Driver.Instance.FindElement(By.Id("title)).SendKeys(title);
2: Thread.Sleep(1000);
3:
4: Instance.SwitchTo().Frame("content_ifr");
5: Thread.Sleep(1000);
6:
7: Driver.Instance.SwitchTo().ActiveElement().SendKeys("something");
Now, what happens is that the title is easily found (by ID, so I wouldn't expect problems there), and I can easily insert the title text (line 1).
But the inline frame for the post body is causing problems. When running the test, after the topic is filled in, the cursor changes to the body area (line 4) - as planned. However, nothing more happens. The SendKeys("string") method (ine 7) doesn't seem to work there.
Any ideas?
EDIT: Of course - an important piece of information is that the iframe in Wordpress simply loads a TinyMCE editor. So, in the page source, there's only a body tag with the javascript loading of the editor.
EDIT2: Of course, something suddenly changed. Without ANY change to the wordpress page, the "content_ifr" is now suddenly missing (?!!!!!?) The Selenium test fails with "unable to locate frame...", and it's also suddenly missing from the page source.
EDIT3: I also noticed something:
Driver.Instance.SwitchTo().Frame(iframe);
Driver.Instance.FindElement(By.Id("tinymce")).SendKeys("message body");
It's the SECOND line that makes the cursor switch to the mce field, not the line with the .SwitchTo(). However, I need the first line - the second line does nothing on its own. This is approaching something really stupid. I've been looking for a solution to this for a week - this doesn't exactly bode well for Selenium. The Selenium user group doesn't even want to answer when I ask them.
Also - if I skip the SendKeys() method in the second line, nothing happens. So, it seems that the two lines does ALLMOST what it should, right up to and including placing the cursor in the correct spot. But it never sends any text.
EDIT4 (last): After actually figuring out how to use IJavaScriptExecutor, it works using the solution(s) below.
Java method to handle TinyMCE editor would look like:
public void entersTopicOfBody(String textToBeTyped, WebDriver driver) {
driver.switchTo().frame("content_ifr");
WebElement body = driver.findElement(By.xpath("//body"));
body.click();
JavascriptExecutor executor = (JavascriptExecutor)driver;
executor.executeScript("arguments[0].innerHTML = '"+ textToBeTyped+"'", body);
driver.switchTo().defaultContent();
}
Below is some C# code that publishes a post. I think the main issues you have are due to timing issues.
I've done a bit of Selenium recently and I favour implicit waits: it waits for a maximum time period for the item to be available, but returns as soon as possible. So you can specify a max wait of 100 seconds, but if it finds it in 1 second, it will only wait 1 second. Much more efficient vs sleeping for an arbitrary length of time. See this post about Implicit and Explicit waits
But even with implicit waits, it may not solve all issues. When coding the sample below, I ran into the issue where the "Publish" button was disabled and re-enabled after some time. And that's when you have to look at the code to see what it is doing as well. It's times such as these where sleeps can help you fix the problem for a quick fix if you do not wish to debug too much: just be sure to set a large enough sleep time and be wary that it could be inconsistent in the future.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using OpenQA.Selenium;
using OpenQA.Selenium.Remote;
using OpenQA.Selenium.Support.UI;
using OpenQA.Selenium.Support.Events;
namespace SeleniumTest
{
class Program
{
static void Main(string[] args)
{
IWebDriver driver = new OpenQA.Selenium.Firefox.FirefoxDriver();
driver.Manage().Timeouts().ImplicitlyWait(TimeSpan.FromSeconds(30));
// enter your configurations here
driver.Navigate().GoToUrl("http://localhost/wordpress/wp-admin/post-new.php");
driver.FindElement(By.Id("user_login")).SendKeys("admin");
driver.FindElement(By.Id("user_pass")).SendKeys("yourpassword");
driver.FindElement(By.Id("wp-submit")).Click();
driver.FindElement(By.Id("title")).SendKeys("the title");
var iframe = driver.FindElement(By.Id("content_ifr"));
driver.SwitchTo().Frame(iframe);
// your solution which works in my instance
//driver.SwitchTo().ActiveElement().SendKeys("hello tiny mce from selenium");
// send keys with exact element
//driver.FindElement(By.Id("tinymce")).SendKeys("hello tiny mce from selenium");
// javascript - 1
IJavaScriptExecutor js = driver as IJavaScriptExecutor;
var tinymce = driver.FindElement(By.Id("tinymce"));
IJavaScriptExecutor executor = (IJavaScriptExecutor)driver;
executor.ExecuteScript("arguments[0].innerHTML = 'hello tiny mce via javascript'", tinymce);
// javascript - 0
driver.SwitchTo().DefaultContent();
var wait = new WebDriverWait(driver, TimeSpan.FromSeconds(11));
wait.Until((d) => { return !d.FindElement(By.Id("publish")).GetAttribute("class").Contains("disabled"); }); // wait for publish button to be enabled
driver.FindElement(By.Id("publish")).Click();
driver.FindElement(By.Id("message")); // wait for message on next page to verify it is posted
driver.Close();
}
}
}
A PHP version of olyv solution:
$content = 'my text';
$this->frame( 'content_ifr' );
$body = $this->byXPath( '//body' );
$body->click();
$script = 'arguments[0].innerHTML = "" + arguments[1] + ""; ';
$this->execute( [
'script' => $script,
'args' => [ $body->toWebDriverObject(), $content ],
]
);
$this->frame( null );
I know that I am a bit late to the party, but I just found a solution that is (I believe) much simpler than the answers given so far. So I decided to post it here in case it could help someone else.
There is no need to switch frames here. What you wanna do is 'click' on the button in the top right corner of the text editor that says "Text", which has id = "content-html". Now, you can 'send keys" to the textarea, which has id = "content".
Here is some Python code that does just this:
driver.find_element_by_id("content-html").click()
driver.find_element_by_id("content").send_keys("Some text...")
Hope it helps
This is giving me quite some headache. I have an page-tab-application, where DB-interaction uses the facebook-user-id to assign and save data and also to check user permissions. Until a weak ago everything was working fine, but now with the upcoming december-changes this setup doesnt work anymore:
config.php:
$facebook = new Facebook( array(
'appId' => $app_id,
'secret' => $app_secret,
'cookie' => true
));
index.php:
includes config.php and gets the signed request (not important for the question
javascript.js:
calls the read-user-status.php and handles the data
read-user-status.php:
gives json-response, includes config.php and calls the $facebook -> getUser()-function to get the uid
Even when called from the index.php directly after page-load, I sometimes get the uid and sometimes I don't. Strangly enough I usually have to wait a little until I reload the page and then it works again. But this isn't always the case. This all is just very strange to me.
EDIT: Should have mentioned that this call:
$uid = $facebook -> getUser();
if ($uid) {
try {
// Proceed knowing you have a logged in user who's authenticated.
$user_profile = $facebook -> api('/me');
} catch (FacebookApiException $e) {
error_log($e);
$uid = FALSE;
echo "EXCEPTION $e";
}
}
gives out "EXCEPTION An active access token must be used to query information about the current user".
I know there quite a lot of similar questions out there, but none of the answers were helpful to my particular (and probably to the new breaking changes relied) problem.
EDIT2: I now suppose that it is a sdk-bug (https://developers.facebook.com/bugs/238039849657148 , thanks to CBroe). Any recommendations for a work-around are of course very welcome.
EDIT 3, TEMPORARY SOLUTION
Everytime you make an ajax request, you post the token you get from the FB.getLoginStatus or FB.login and read it out in the php file and set it via $facebook -> setAccessToken. Not suitable in all circumstances (you definately need to use post), is slower and brings some security issues, but still works.
Sounds like you are affected by the bug I reported beginning of November, https://developers.facebook.com/bugs/238039849657148
They’ve confirmed it and say they’re working on a fix – but since the change is only a few days away now, they should hurry up a little …
I got this working by doing the following...
if(!$user){
$loginUrl = $facebook->getLoginUrl(array(
'scope' => 'email',
'redirect_uri' => $app_url
));
header('Location: ' . $loginUrl);
}
I also added my app to be integrated with:
Website with Facebook login
App on Facebook
Page Tab
try by adding access token to request.
$accessToken = $facebook->getAccessToken();
$user_profile = $facebook->api('/me?access_token=' . $accessToken);
I found a work-around for this, until it is fixed (which it seems like, wont be in time until the changes take place).
Everytime you make an ajax request, you post the token you get from the FB.getLoginStatus or FB.login and read it out in the php file and set it via $facebook -> setAccessToken. Not suitable in all circumstances (you definately need to use post), is slower and brings some security issues, but still works.
if it you are lucky and your version of php sdk still registers session variables than right after _graph method declaration:
//find this method below
protected function _graph ($path, $method = 'GET', $params = array ())
{
//paste right after _graph method declaration code below:
if (isset($_SESSION["fb_".$this->getAppId()."_access_token"]))
{
$this->setAccessToken($_SESSION["fb_".$this->getAppId()."_access_token"]);
}
//till here
//and you are good to go
//remember: your version of sdk must be registering access token variable in session
//right after ajax call
//i used git to get version before last commit of sdk published on github
I am using Ruby on Rails with Cucumber and Capybara.
How would I go about testing a simple confirm command ("Are you sure?")?
Also, where could I find further documentation on this issue?
The selenium driver now supports this
From Capybara you would access it like this:
page.driver.browser.switch_to.alert.accept
or
page.driver.browser.switch_to.alert.dismiss
or
page.driver.browser.switch_to.alert.text
Seems like there's no way to do it in Capybara, unfortunately. But if you're running your tests with the Selenium driver (and probably other drivers that support JavaScript), you can hack it. Just before performing the action that would bring up the confirm dialog, override the confirm method to always return true. That way the dialog will never be displayed, and your tests can continue as if the user had pressed the OK button. If you want to simulate the reverse, simply change it to return false.
page.evaluate_script('window.confirm = function() { return true; }')
page.click('Remove')
I've implemented these two web steps in /features/step_definitions/web_steps.rb:
When /^I confirm popup$/ do
page.driver.browser.switch_to.alert.accept
end
When /^I dismiss popup$/ do
page.driver.browser.switch_to.alert.dismiss
end
If you want to specifically test the message being displayed, here's a particularly hacky way to do so. I don't endorse it as beautiful code, but it gets the job done. You'll need to load http://plugins.jquery.com/node/1386/release, or change it to do cookies natively if you don't want jQuery.
Use this sort of story:
Given I am on the menu page for the current booking
And a confirmation box saying "The menu is £3.50 over budget. Click Ok to confirm anyway, or Cancel if you want to make changes." should pop up
And I want to click "Ok"
When I press "Confirm menu"
Then the confirmation box should have been displayed
And these steps
Given /^a confirmation box saying "([^"]*)" should pop up$/ do |message|
#expected_message = message
end
Given /^I want to click "([^"]*)"$/ do |option|
retval = (option == "Ok") ? "true" : "false"
page.evaluate_script("window.confirm = function (msg) {
$.cookie('confirm_message', msg)
return #{retval}
}")
end
Then /^the confirmation box should have been displayed$/ do
page.evaluate_script("$.cookie('confirm_message')").should_not be_nil
page.evaluate_script("$.cookie('confirm_message')").should eq(#expected_message)
page.evaluate_script("$.cookie('confirm_message', null)")
end
Updating this for current releases of Capybara. Most Capybara drivers today support the modal API. To accept a confirm modal you would do
accept_confirm do # dismiss_confirm if not accepting
click_link 'delete' # whatever action triggers the modal to appear
end
This can be used in Cucumber with something like
When /^(?:|I )press "([^"]*)" and confirm "([^"]*)"$/ do |button, msg|
accept_confirm msg do
click_button(button)
end
end
which will click the named button and then accept a confirm box with text matching msg
The capybara-webkit driver supports this as well.
Scenario: Illustrate an example has dialog confirm with text
#
When I confirm the browser dialog with tile "Are you sure?"
#
=====================================================================
my step definition here:
And(/^I confirm the browser dialog with title "([^"]*)"$/) do |title|
if page.driver.class == Capybara::Selenium::Driver
page.driver.browser.switch_to.alert.text.should eq(title)
page.driver.browser.switch_to.alert.accept
elsif page.driver.class == Capybara::Webkit::Driver
sleep 1 # prevent test from failing by waiting for popup
page.driver.browser.confirm_messages.should eq(title)
page.driver.browser.accept_js_confirms
else
raise "Unsupported driver"
end
end
Prickle adds some handy convenience methods for working with popups in selenium and webkit
This gist has steps to test a JS confirm dialog in Rails 2 and 3 with any Capybara driver.
It's an adaptation of a previous answer, but doesn't need the jQuery Cookie plugin.
Tried the above answers with no luck. In the end this worked for me:
#browser.alert.ok