I have inherited some GEB tests that are testing logging into a site (and various error cases/validation warnings).
The test runs through some validation failures and then it attempts to re-navigate to the same page (just to refresh the page/dom) and attempts a valid login. Using GEB's to() method, it detects that you are attempting to navigate to the page you are on, it just calls refresh - the problem here is that attempts to refresh the last POST request, and the driver displays the
"To display this page, Firefox must send information that will repeat any action (such as a search or order confirmation) that was performed earlier"
message - as the test is not expecting this popup, it hangs and the tests timeout.
Is there a way to turn off these warnings in Firefox webdriver? or to auto-ignore/accept them via Selenium or GEB?
GEB Version: 0.9.2,
Selenium Version: 2.39.0
(Also tried with minor version above: 0.9.3 & 2.40.0)
Caveats:
I know about the POST/Re-direct/GET pattern - but am not at liberty to change the application code in this case
The warning message only causes an issue intermittently (maybe 1 in 5 times) - I have put this down to speed/race conditions whereby the test completes the next actions before the message appears - I know a possible solution is to update tests to wait for message to appear and then accept, but my question is, is there a global setting that can just avoid these being triggered/displayed?
That refresh() is there to work around an issue with IE driver which ignores calls to driver.get() with the same url as the current one.
Instead of monkey patching Browser class (which might bite you somewhere down the line or might not) I would change the url of your login page class. You might for example add an insignificant query string - I think that simply a ? at the end should suffice. The driver.currentUrl == newUrl condition will evaluate to false and you will not see that popup anymore.
If I understand you issue properly this might help. In Groovy you can modify a class on the fly.
We use Spock with Geb and I placed this in a Super class which all Spock Spec inherit from. Eg: QSpec extends GebSpec.
It is the original method slightly modified with the original code commented out so you know what has been changed. I use this technique in several required places to alter Geb behaviour.
static {
Browser.metaClass.go = { Map params, String url ->
def newUrl = calculateUri(url, params)
// if (driver.currentUrl == newUrl) {
// driver.navigate().refresh()
// } else {
// driver.get(newUrl)
// }
driver.get(newUrl)
if (!page) {
page(Page)
}
}
}
Related
On a page in a web app, a loading screen/widget continue to appear after the user leaves the text field.
The app is using version 1.6.0.3. I looked at the most recent version of Prototype 1.7.3 and I did not find that function.
I also tested other instances when this method is called. If there is no user input, the widget does not hang-up.
This error is displayed in the console of Chrome’s developer tools.
at E (framework.pack.js:1699)
at Function.stopObserving (framework.pack.js:1732)
at A.hide (ui.pack.js:13410)
at A.render (ui.pack.js:13591)
at A.updateChoices (ui.pack.js:13650)
at A.onComplete (ui.pack.js:13786)
at Object.oncomplete (framework.pack.js:76)
at framework.pack.js:2748
The specific method in questions seems to be in the Prototype.js file this =>
if (element._prototypeEventID) return element._prototypeEventID[0];
arguments.callee.id = arguments.callee.id || 1;
return element._prototypeEventID = [++arguments.callee.id];
}
I expect the loading widget to disappear after the save is done, but it is still on the page. The console of Chrome's developer tools also has a second error:
validateFormCheck # common.js:1031
It looks like the getEventID method is being called where the undefined warning/error triggers.
In version 1.6.0.3 getEventID is only called in createWrapper and stopObserving, I see stopObserving is in the call stack that you posted so let's go with that one.
stopObserving() takes 1 required parameter and 2 optional parameters (element, eventName, handler) if you only pass the element to the function it looks it up and then deletes all the PrototypeJS observers attached to that element. If you pass eventName and/or handler as well stopObserving will only specifically delete the observer you tell it to.
That being said, if the element is removed from the DOM before stopObserving is called this could cause the error you are seeing.
2 fixes that could work
move the call to stopObserving() above the call to remove()
comment out the call to stopObserving() and see if page behaves like you want it to
Is it possible to use JGiven (with or without Spring support) to retrieve the statements before / during execution? For example, if we had a fairly typical login acceptance test i.e.
public class LoginFeatureTest extends SpringScenarioTest<GivenIAmAtTheLoginPage, WhenILogin, ThenTheLoginActionWillBeSuccessful> {
#Test
public void my_login_test() {
given().I_am_a_new_user()
.and().I_am_at_the_login_page();
when().I_login_with_username_$_and_password_$("dave", "dave123");
then().the_home_page_is_visible();
}
}
Is it possible to get something access to the following information?
My Login Test (start)
Given I am a new user
and I am at the login page
When I login with username dave and password dave123
Then the home page is visible
My Login Test (end)
i.e. what i'm looking for is: -
The name of a scenario method + all it's given, when, then and and statement calls _(formatted via JGiven formatting).
When each scenario method starts at run-time.
When each given, when, then and and executes at run-time.
When the scenario method ends.
This will give me the ability to visually show in a UI (a) exactly what is going to execute and (2) it's current position during execution (with durations).
12:00:01.012 [ My Login Test (start) ]
12:00:02.035 23ms Given I am a new user
12:00:02.051 16ms and I am at the login page
----> When I login with username dave and password dave123
Then the home page is visible
[ end ]
I'm thinking Spring AOP might come to the rescue here? Or does JGiven provide anything useful buried in it's code?
At the moment there is no possibility to do that. I have created an issue for that: https://github.com/TNG/JGiven/issues/328.
It should not be too difficult to implement this as there is internally already a listener concept. If you are still interested you could propose an API and integrate the mechanism in JGiven.
I have a filter set up as follow to control users login status.
class SecurityFilters {
def filters = {
login(controller:'login|logout|proxy|API|error', action:'*', invert: true) {
before = {
if (!session.isLoggedIn){
switch(controllerName){
case "enroll":
switch(actionName){
...
default:
log.warn "Permission Denied. Default action for enroll."
render(view: '/permissionDenied', model: [message: "You must be logged in to access the enroll system. If you are a consumer, please contact your agent for more information."])
break
}
break
...
}
else {
switch(controllerName){
case "agent":
if (!session.user.isAgent) {
render view: "/permissionDenied", model: [message: 'This portion of the site is only available to agents.']
return false
}
break
....
}// switch
}// else
}// before
...
}// login
}// filters
The problem I am having is that when I run this in development it works fine but when I run it on our QA system it works fine for a while and then suddenly it stops working correctly.
I added logging and I can see that the session information is available in the filter and the session variable (session.user.isAgent) is set correctly (true) but the code inside the if(!session.user.isAgent) gets executed regardless.
I can's seem to find the cause for odd behaviour.
My question is has anyone seen this behaviour before and how did they solve it or have any ideas of where to look for probable cause for the sudden change in the way the filter is working.
Thanks in advance.
UPDATE (02/19/2014):
After adding more logging in an effort to hunt down the cause the filter did not execute the code in the if(!session.user.isAgent) as it was doing before. Now it runs normally and then executes only the render line for when the user is not logged in. The logging still show that the user is logged in and that (s)he is an agent but then it runs the render but not the lines of code above it. It is as if there was a "goto" the render line after it completes checking if the agent is logged in.
Again any information or solution would be appreciated
I've had a few issues with filters and Groovy truth. The problem I was seeing was that no errors were logged, even with aggressive exception catching (i.e. catching Throwable) and the only output in the browser was a blank page. This seems to happen only in Filters- everywhere else the errors get logged.
In my case, the issue was down to Groovy truth. I was trying to set a Boolean attribute on the session, but every time I did this it failed. In the end I had to convert the value to a String, and then set it, and it worked.
I know this isn't a direct answer, but I've been bitten a few times by the above and for example, lazy GString evaluation.
If you're still debugging, I'd log some output showing the underlying Class type of what you think you're dealing with. It may be that when your boolean conditions are evaluated above an exception is being thrown and swallowed. I'd log the actual values of each of your conditional statements to see what they are. Also remove each line one by one, to see if the failure goes away. And/or replace your conditionals with absolute values i.e. true/false to see if the code gets executed. If it does, it points to an error in the current conditional evaluation.
I recorded some test cases with CUIT in VS2010. Everything worked fine the day before. So, today I run again, all the test failed, with the warning: The following element is no longer available ... and I got the exception : Can't perform "Click" on the hidden control, which is not true because all the controls are not hidden. I tried on the other machine, and they failed as well.
Does anyone know why it happens? Is it because of the web application for something else? Please help, thanks.
PS: So I tried to record a new test with the same controls that said "hidden controls", and the new test worked!? I don't understand why.
EDIT
The warning "The following element blah blah ..." appears when I tried to capture an element or a control while recording. The source code of the button is said 'hidden'
public HtmlImage UIAbmeldenImage
{
get
{
if ((this.mUIAbmeldenImage == null))
{
this.mUIAbmeldenImage = new HtmlImage(this);
#region Search Criteria
this.mUIAbmeldenImage.SearchProperties[HtmlImage.PropertyNames.Id] = null;
this.mUIAbmeldenImage.SearchProperties[HtmlImage.PropertyNames.Name] = null;
this.mUIAbmeldenImage.SearchProperties[HtmlImage.PropertyNames.Alt] = "abmelden";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.AbsolutePath] = "/webakte-vnext/content/apps/Ordner/images/logOut.png";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.Src] = "http://localhost/webakte-vnext/content/apps/Ordner/images/logOut.png";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.LinkAbsolutePath] = "/webakte-vnext/e.consult.9999/webakte/logout/index";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.Href] = "http://localhost/webakte-vnext/e.consult.9999/webakte/logout/index";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.Class] = null;
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.ControlDefinition] = "alt=\"abmelden\" src=\"http://localhost/web";
this.mUIAbmeldenImage.FilterProperties[HtmlImage.PropertyNames.TagInstance] = "1";
this.mUIAbmeldenImage.WindowTitles.Add("Akte - Test Akte Coded UI VS2010");
#endregion
}
return this.mUIAbmeldenImage;
}
}
Although I am running Visual Studio 2012, I find it odd that we started experiencing the same problem on the same day, I can not see any difference in the DOM for the Coded UI Tests I have for my web page, but for some reason VS is saying the control is hidden and specifies the correct ID of the element it is looking for (I verified that the ID is still the same one). I even tried to re-record the action, because I assumed that something must have changed, but I get the same error.
Since this sounds like the same problem, occurring at the same time I am thinking this might be related to some automatic update? That's my best guess at the moment, I am going to look into it, I will update my post if I figure anything out.
EDIT
I removed update KB2870699, which removes some voulnerability in IE, this fixed the problems I was having with my tests. This update was added on the 12. september, so it fits. Hope this helps you. :)
https://connect.microsoft.com/VisualStudio/feedback/details/800953/security-update-kb2870699-for-ie-breaks-existing-coded-ui-tests#tabs
Official link to get around the problem :
http://blogs.msdn.com/b/visualstudioalm/archive/2013/09/17/coded-ui-mtm-issues-on-internet-explorer-with-kb2870699.aspx
The problem is more serious than that! In my case I can't even record new Coded UI Tests. After I click in any Hyper Link of any web page of my application the coded UI test builder cannot record that click "The following element is no longer available....".
Apparently removing the updates, as said by AdrianHHH do the trick!
Shut down VS2010, launch it again "Run as administrator".
There may be a field in the SearchProperties (or possible the FilterProperties) that has a value set by the web site, or that represents some kind of window ID on your desktop. Another possibility is that the web page title changes from day to day or visit to visit. Different executions of the browser or different visits to the web page(s) create different values. Removing these values from the SearchProperties (or FilterProperties) or changing the check for the title from an equals to a contains for a constant part of the title should fix the problem. Coded UI often searches for more values than the minimum set needed.
Compare the search properties etc for the same control in the two recorded tests.
Update based extra detail given in the comments:
I solved a similar problem as follows. I copied property code similar to that shown in your question into a method that called FindMatchingControls. I checked how many controls were returned, in my case up to 3. I examined various properties of the controls found, by writing lots of text to a debug file. In my case I found that the Left and Top properties were negative for the unwanted, ie hidden, controls.
For your code rather than just using the UIAbmeldenImage property, you might call the method below. Change an expression such as
HtmlImage im = UIMap.abc.def.UIAbmeldenImage;
to be
HtmlImage im = FindHtmlHyperLink(UIMap.abc.def);
Where the method is:
public HtmlImage FindHtmlHyperLink(HtmlDocument doc)
{
HtmlImage myImage = new HtmlImage(doc);
myImage.SearchProperties[HtmlImage.PropertyNames.Id] = null;
myImage.SearchProperties[HtmlImage.PropertyNames.Name] = null;
myImage.SearchProperties[HtmlImage.PropertyNames.Alt] = "abmelden";
myImage.FilterProperties[HtmlImage.PropertyNames.AbsolutePath] = "/webakte-vnext/content/apps/Ordner/images/logOut.png";
myImage.FilterProperties[HtmlImage.PropertyNames.Src] = "http://localhost/webakte-vnext/content/apps/Ordner/images/logOut.png";
myImage.FilterProperties[HtmlImage.PropertyNames.LinkAbsolutePath] = "/webakte-vnext/e.consult.9999/webakte/logout/index";
myImage.FilterProperties[HtmlImage.PropertyNames.Href] = "http://localhost/webakte-vnext/e.consult.9999/webakte/logout/index";
myImage.FilterProperties[HtmlImage.PropertyNames.Class] = null;
myImage.FilterProperties[HtmlImage.PropertyNames.ControlDefinition] = "alt=\"abmelden\" src=\"http://localhost/web";
myImage.FilterProperties[HtmlImage.PropertyNames.TagInstance] = "1";
myImage.WindowTitles.Add("Akte - Test Akte Coded UI VS2010");
UITestControlCollection controls = myImage.FindMatchingControls();
if (controls.Count > 1)
{
foreach (UITestControl con in controls)
{
if ( con.Left < 0 || con.Top < 0 )
{
// Not on display, ignore it.
}
else
{
// Select this one and break out of the loop.
myImage = con as HtmlImage;
break;
}
}
}
return myImage;
}
Note that the above code has not been compiled or tested, it should be taken as ideas not as the final code.
I had the same problem on VS 2012. As a workaround, you can remove that step, and re-record it again. That usually works.
One of the biggest problem while analyzing the Coded UI test failures is that the error stack trace indicates the line of code which might be completely unrelated to the actual cause of failure.
I would suggest you to enable HTML logging in your tests - this will display step by step details of how Coded UI tried to execute the tests - with screenshots of your application. It will also highlight the control in red which Coded UI is trying to search/operate upon.This is very beneficial in troubleshooting the actual cause of test failures.
To enable tracing you can just add the below code to your app.config file --
I'm having an strange behaviour using grails event-push plugin with IE8.
I'm using grails 2.2, event-push 1.0M7 and also AngularJS (but this shouldn't be the problem).
I have defined the event in MyEvents.groovy in conf dir
events = {
'newActivity' browser:true
}
Here is the method of a service that sends the events to the browser
#Listener(namespace='browser')
public void sendEvents(UserLevel level, Map<String,List<UserAchievement>> achievements, Map<String,List<UserMission>> missions){
def activities = []
if (level)
activities << new RecentActivity(level)
achievements["user"]?.each{activities << new RecentActivity(it)}
achievements["team"]?.each{ activities << new RecentActivity(it)}
missions["user"]?.each{activities << new RecentActivity(it)}
missions["team"]?.each{activities << new RecentActivity(it)}
if (activities && activities.size()>0)
event(topic:"newActivity",data:activities)
}
And here the js event listener
grailsEvents.on("newActivity",function(data){
for (var i=0;i<data.length;i++){//Add
changeUserImage(data[i]);
$scope.$apply(function(){$scope.activities.unshift(data[i]);});
}
while ($scope.activities.length>$scope.maxElements){ //remove
$scope.$apply(function(){$scope.activities.pop();});
}
});
Edit: These are the grailsEvents inicializations I've tried all with the same result:
window.grailsEvents = new grails.Events($jostalan.urls.base);
window.grailsEvents = new grails.Events($jostalan.urls.base, {transport:"websocket"});
window.grailsEvents = new grails.Events($jostalan.urls.base, {transport:"websocket",shared:true});
I know that sometimes the event is called twice and other none because I've tried putting a console.log, but I've deleted it just to prevent problems with IE8. Also, in Chrome and Firefox everything works as expected, just one call to the method when expected.
The only thing that I've seen that sounds strange for me (as I don't really understand what it means), it's that sometimes I see in grails console the following trace:
grails> cpr.AtmosphereFramework If you have more than one Connector enabled, make sure they both use the same protocol, e.g NIO/APR or HTTP for all. If not, org.atmosphere.container.BlockingIOCometSupport will be used and cannot be changed.
Anyone knows why this strange behaviour is happening and how to solve it?
Edit: I've also tested it with plain javascript and the same strange behaviour happens. Sometimes the event is called, sometimes not and sometime twice
Edit: I've also tried deleting the #Listener annotation from service method as I understand that's not necessary.