I'm using PhantomJS 1.8.2 to run some Jasmine unit tests using JsTestDriver. The tests run fine using Chrome, but about half the time when using PhantomJS, the test result is that no test cases were found.
I've narrowed the issue down to PhantomJS failing to open the local JsTestDriver page (http://localhost:9876/capture). Here's how to reproduce this, about 50% of the times, the Loaded ... with status ... message is never shown:
Start JsTestDriver server locally
Run phantomjs phantomjs-jstd-bridge.js
The file phantomjs-jstd-bridge.js looks like this:
var page = require('webpage').create();
var url = 'http://localhost:9876/capture';
console.log('Loading ' + url);
page.open(url, function(status) {
console.log('Loaded ' + url + ' with status ' + status);
});
The first log line (Loading ...) is always displayed, but the second line coming from the callback is only printed about half the time.
What could be the cause for this? Opening the URL in question in a web browser works fine every time.
Is there any way to get more info on why PhantomJS does not call the callback?
Check some tips mentioned in the Troubleshooting wiki page. Particularly useful is tracking the network transfer activity as it may indicate whether some resources are not sent properly or other similar problems.
Related
I have a following test in Cypress:
visit first page with the header A
click on the Go to B Page button
assert that the header of the page is now B
It works fine in Chrome, but failing in Firefox, as on the page B I have some background polling requests, and when cypress switches to another test and those requests get "canceled" away, I get either TypeError: NetworkError when attempting to fetch resource or AbortError: The operation was aborted
All the requests are using fetch api, by the way.
The possibility to mute those errors through the uncaught:exception event seems a bad idea, and so does the idea to do something on the page to cancel the polling, as it is not the thing under testing.
Maybe someone has encoutnered this problem too and got some non-hacky solution?
I had a similar issue with Cypress tests in Firefox and resorted to the slightly hacky solution of using an uncaught:exception handler as you mention. It is possible to filter error messages somewhat at least:
function handleUncaughtException(err){
if (err.message.includes('Request aborted') ) {
console.log("Request aborted. Test will continue. Error:",err);
return false; // return false to make test continue
}
throw err;
}
cy.on('uncaught:exception',handleUncaughtException);
In principle you can cancel this handler when it's no longer needed. In my case though, this stopped the test working, presumably because the request started previous to or after the calls.
cy.removeListener("uncaught:exception", handleUncaughtException)
The Cypress docs have some advice on defining these: see at https://docs.cypress.io/api/events/catalog-of-events#Examples. It may be useful to put the handler in a support file, so that it is applied to all tests.
(See also https://docs.cypress.io/api/events/catalog-of-events#Event-Types and https://nodejs.org/api/events.html#events_emitter_removelistener_eventname_listener).
I'm new to cypress and love the way it is architected. However, I seem to have run into a problem early on for a very simple thing that I'm trying to do.
My workflow is:
1) Visit the site
2) Enter username and password
3) On the next screen, type a number and press submit,
4) On the next screen, select a value from a dropdown and press enter.
5) I get to the landing page of my website.
Cypress works totally fine till step 4). It seems to stall at step 5. The test runner suddenly stalls and without warning or error, shows
"Whoops, there is no test to run."
From here, when I click the "View All Tests" button, it takes me to the runner tool. There I see the indication that something is still running in the background. I tried waiting for more than 10 minutes but nothing happens until I click on the "Stop" action.
How do I debug this? Can I see what is happening via any log etc?
There could even be something wrong with my website as well, but without any log information, I'm unable to proceed further. Any help is appreciated.
To provide more context, I don't think this is a timeout based issue as if that were the case, cypress did report to me about this and stopped. I then increased the timeout.
My spec file
describe('My first test', function() {
it('Visits home page', function() {
cy.visit('https://mywebsite.com:5800', {timeout: 400000}, {pageLoadTimeout: 400000}, {defaultCommandTimeout: 400000})
cy.get('#USERNAME').type('myusername')
cy.get('#PASSWORD').type('mypassword')
cy.get('#loginbutton').click()
cy.get('#SOMELEMENT_WHERE_I_TYPE_A_UNIQUE_NUMBER').type('8056')
cy.get('#loginbutton').click()
cy.get('#loginbutton').click()
})
})
Thanks.
If you run DEBUG=cypress:* cypress open from the terminal when initially opening Cypress, there will be more debug log information printed there while you run your tests.
Also, it's always a good idea to search the issues for the project to see if anyone else has had this happen.
For some reason, the Cypress automation gets into a state where it thinks that you have no spec file. All Cypress does to determine this is to see if there is a location.hash defined on the main window -> where it usually says https://localhost:2020/__/#tests/integration/my_spec.js.
Likely this is due to security mechanisms in the app that prevent your application from being run within an iframe (which is how Cypress runs all applications under test). Maybe in your application code it is something like:
if (top !== self) {
top.location.href = self.location.href;
}
You can simply disable these checks while testing or in Cypress you can add to your test file (or a support file to have it work on every test file):
Cypress.on('window:before:load', (win) => {
Object.defineProperty(win, 'self', {
get: () => {
return window.top
}
})
})
I had the same issue. Usually this is related to the page moving out of the parent and can be solved by invoking the attribute and changing it to the current page.
cy.get('.approved-content .no-auto-submit').invoke('attr', 'target', '_self');
I was having lots of glitches with my tests using Codeception + WebDriver + Selenium + Firefox, so I decided to remove Codeception from the variables and do some tests directly using php_webdriver, to discover the problem wasn't in Codeception.
I tried to go to a page using 3 different methods:
// Click a link
$driver->findElement(WebDriverBy::id('inbox-navigation'))->click();
// Directly go to the address via URL
$driver->get('http://testing.mysite.net/#/messages');
// Click via javascript
$driver->executeScript("jQuery('#inbox-navigation').click();");
Of course none of them work and, doing it manually in my browser it works fine. The link appear as clicked, the url changes, but the related page (loaded via ajax) never gets loaded.
Here's the full source code I'm using to test this:
<?php
require __DIR__."/vendor/autoload.php";
$host = 'http://localhost:4444/wd/hub';
$driver = RemoteWebDriver::create($host, DesiredCapabilities::firefox());
$driver->manage()->window()->setSize(new \WebDriverDimension(1280, 720));
$driver->get('http://testing.mysite.net/');
$driver->findElement(WebDriverBy::id('email'))->sendKeys('john#doe.com');
$driver->findElement(WebDriverBy::id('password'))->sendKeys('secret');
$driver->findElement(WebDriverBy::id('sign-in-button'))->click();
$driver->wait(10)->until(WebDriverExpectedCondition::presenceOfElementLocated(WebDriverBy::id('inbox-navigation')));
$driver->findElement(WebDriverBy::id('inbox-navigation'))->click();
sleep(15); /// to make sure it wasn't too fast
$driver->takeScreenshot('screen.png');
var_dump($driver->getCurrentURL());
And this is the link changed to Inbox:
Also, manually executing jQuery('#inbox-navigation').click(); in console also works fine.
I'm prototyping a MVC.NET 4.0 application and am defining our Javascript test configuration. I managed to get Jasmine working in VS2012 with the Chutzpah extensions, and I am able to run pure Javascript tests successfully.
However, I am unable to load test fixture (DOM) code and access it from my tests.
Here is the code I'm attempting to run:
test.js
/// various reference paths...
jasmine.getFixtures().fixturesPath = "./";
describe("jasmine tests:", function () {
it("Copies data correctly", function () {
loadFixtures('testfixture.html');
//setFixtures('<div id="wrapper"><div></div></div>');
var widget = $("#wrapper");
expect(widget).toExist();
});
});
The fixture is in the same folder as the test file. The setFixtures operation works, but when I attempt to load the HTML from a file, it doesn't. Initially, I tried to use the most recent version of jasmine-jquery from the repository, but then fell back to the over 1 year old download version 1.3.1 because it looked like there was a bug in the newer one. Here is the message I get with 1.3.1:
Test 'jasmine tests::Copies data correctly' failed
Error: Fixture could not be loaded: ./testfixture.html (status: error, message: undefined) in file:///C:/Users/db66162/SvnProjects/MvcPrototype/MvcPrototype.Tests/Scripts/jasmine/jasmine-jquery-1.3.1.js (line 103)
When I examine the source, it is doing an AJAX call, yet I'm not running in a browser. Instead, I'm using Chutzpah, which runs a headless browser (PhantomJS). When I run this in the browser with a test harness, it does work.
Is there someone out there who has a solution to this problem? I need to be able to run these tests automatically both in Visual Studio and TeamCity (which is why I am using Chutzpah). I am open to solutions that include using another test runner in place of Chutzpah. I am also going to evaluate the qUnit testing framework in this effort, so if you know that qUnit doesn't have this problem in my configuration, I will find that useful.
I fixed the issue by adding the following setting to chutzpah.json:
"TestHarnessLocationMode": "SettingsFileAdjacent",
where chutzpah.json is in my test app root
I eventually got my problem resolved. Thank you Ian for replying. I am able to use PhantomJS in TeamCity to run the tests through the test runner. I contacted the author of Chutzpah and he deployed an update to his product that solved my problem in Visual Studio. I can now run the Jasmine test using Chutzpah conventions to reference libraries and include fixtures while in VS, and use the PhantomJS runner in TeamCity to use the test runner (html).
My solution on TeamCity was to run a batch file that launches tests. So, the batch:
#echo off
REM -- Uses the PhantomJS headless browser packaged with Chutzpah to run
REM -- Jasmine tests. Does not use Chutzpah.
setlocal
set path=..\packages\Chutzpah.2.2.1\tools;%path%;
echo ##teamcity[message text='Starting Jasmine Tests']
phantomjs.exe phantom.run.js %1
echo ##teamcity[message text='Finished Jasmine Tests']
And the Javascript (phantom.run.js):
// This code lifted from https://gist.github.com/3497509.
// It takes the test harness HTML file URL as the parameter. It launches PhantomJS,
// and waits a specific amount of time before exit. Tests must complete before that
// timer ends.
(function () {
"use strict";
var system = require("system");
var url = system.args[1];
phantom.viewportSize = {width: 800, height: 600};
console.log("Opening " + url);
var page = new WebPage();
// This is required because PhantomJS sandboxes the website and it does not
// show up the console messages form that page by default
page.onConsoleMessage = function (msg) {
console.log(msg);
// Exit as soon as the last test finishes.
if (msg && msg.indexOf("Dixi.") !== -1) {
phantom.exit();
}
};
page.open(url, function (status) {
if (status !== 'success') {
console.log('Unable to load the address!');
phantom.exit(-1);
} else {
// Timeout - kill PhantomJS if still not done after 2 minutes.
window.setTimeout(function () {
phantom.exit();
}, 10 * 1000); // NB: use accurately, tune up referring to your needs
}
});
}());
I've got exactly the same problem. AFAIK it's to do with jasmine-jquery trying to load the fixtures via Ajax when the tests are run via the file:// URI scheme.
Apparently Chrome doesn't allow this (see https://stackoverflow.com/a/5469527/1904 and http://code.google.com/p/chromium/issues/detail?id=40787) and support amongst other browsers may vary.
Edit
You might have some joy by trying to set some PhantomJS command-line options such as --web-security=false. YMMV though: I haven't tried this myself yet, but thought I'd mention it in case it's helpful (or in case anyone else know more about this option and whether it will help).
Update
I did manage to get some joy loading HTML fixtures by adding a /// <reference path="relative/path/to/fixtures" /> comment at the top of my Jasmine spec. But I still have trouble loading JSON fixtures.
Further Update
Loading HTML fixtures by adding a /// <reference path="relative/path/to/fixtures" /> comment merely loads in your HTML fixtures to the Jasmine test runner, which may or may not be suitable for your needs. It doesn't load the fixtures into the jasmine-fixtures element, and consequently your fixtures don't get cleaned up after each test.
I am running several tests with WebDriver and Firefox.
I'm running into a problem with the following command:
WebDriver.get(www.google.com);
With this command, WebDriver blocks till the onload event is fired. While this can normally takes seconds, it can take hours on websites which never finish loading.
What I'd like to do is stop loading the page after a certain timeout, somehow simulating Firefox's stop button.
I first tried execute the following JS code every time that I tried loading a page:
var loadTimeout=setTimeout(\"window.stop();\", 10000);
Unfortunately this doesn't work, probably because :
Because of the order in which scripts are loaded, the stop() method cannot stop the document in which it is contained from loading 1
UPDATE 1: I tried to use SquidProxy in order to add connect and request timeouts, but the problem persisted.
One weird thing that I found today is that one web site that never stopped loading on my machine (FF3.6 - 4.0 and Mac Os 10.6.7) loaded normally on other browsers and/or computers.
UPDATE 2: The problem apparently can be solved by telling Firefox not to load images. hopefully, everything will work after that...
I wish WebDriver had a better Chrome driver in order to use it. Firefox is disappointing me every day!
UPDATE 3: Selenium 2.9 added a new feature to handle cases where the driver appears to hang. This can be used with FirefoxProfile as follows:
FirefoxProfile firefoxProfile = new ProfilesIni().getProfile("web");
firefoxProfile.setPreference("webdriver.load.strategy", "fast");
I'll post whether this works after I try it.
UPDATE 4: at the end none of the above methods worked. I end up "killing" the threads that are taking to long to finish. I am planing to try Ghostdriver which is a Remote WebDriver that uses PhantomJS as back-end. PhantomJS is a headless WebKit scriptable, so i expect not to have the problems of a real browser such as firefox. For people that are not obligate to use firefox(crawling purposes) i will update with the results
UPDATE 5: Time for an update. Using for 5 months the ghostdriver 1.1 instead FirefoxDriver i can say that i am really happy with his performance and stability. I got some cases where we have not the appropriate behaviour but looks like in general ghostdriver is stable enough. So if you need, like me, a browser for crawling/web scraping purposes i recomend you use ghostdriver instead firefox and xvfb which will give you several headaches...
I was able to get around this doing a few things.
First, set a timeout for the webdriver. E.g.,
WebDriver wd;
... initialize wd ...
wd.manage().timeouts().pageLoadTimeout(5000, TimeUnit.MILLISECONDS);
Second, when doing your get, wrap it around a TimeoutException. (I added a UnhandledAlertException catch there just for good measure.) E.g.,
for (int i = 0; i < 10; i++) {
try {
wd.get(url);
break;
} catch (org.openqa.selenium.TimeoutException te) {
((JavascriptExecutor)wd).executeScript("window.stop();");
} catch (UnhandledAlertException uae) {
Alert alert = wd.switchTo().alert();
alert.accept();
}
}
This basically tries to load the page, but if it times out, it forces the page to stop loading via javascript, then tries to get the page again. It might not help in your case, but it definitely helped in mine, particularly when doing a webdriver's getCurrentUrl() command, which can also take too long, have an alert, and require the page to stop loading before you get the url.
I've run into the same problem, and there's no general solution it seems. There is, however, a bug about it in their bug tracking system which you could 'star' to vote for it.
http://code.google.com/p/selenium/issues/detail?id=687
One of the comments on that bug has a workaround which may work for you - Basically, it creates a separate thread which waits for the required time, and then tries to simulate pressing escape in the browser, but that requires the browser window to be frontmost, which may be a problem.
http://code.google.com/p/selenium/issues/detail?id=687#c4
My solution is to use this class:
WebDriverBackedSelenium;
//When creating a new browser:
WebDriver driver = _initBrowser(); //Just returns firefox WebDriver
WebDriverBackedSelenium backedSelenuium =
new WebDriverBackedSelenium(driver,"about:blank");
//This code has to be put where a TimeOut is detected
//I use ExecutorService and Future<?> Object
void onTimeOut()
{
backedSelenuium.runScript("window.stop();");
}
It was a really tedious issue to solve. However, I am wondering why people are complicating it. I just did the following and the problem got resolved (perhaps got supported recently):
driver= webdriver.Firefox()
driver.set_page_load_timeout(5)
driver.get('somewebpage')
It worked for me using Firefox driver (and Chrome driver as well).
One weird thing that i found today is that one web site that never stop loading on my machine (FF3.6 - 4.0 and Mac Os 10.6.7), is stop loading NORMALy in Chrome in my machine and also in another Mac Os and Windows machines of some colleague of mine!
I think the problem is closely related to Firefox bugs. See this blog post for details. Maybe upgrade of FireFox to the latest version will solve your problem. Anyway I wish to see Selenium update that simulates the "stop" button...
Basically I set the browser timeout lower than my selenium hub, and then catch the error. And then stop the browser from loading, then continue the test.
webdriver.manage().timeouts().pageLoadTimeout(55000);
function handleError(err){
console.log(err.stack);
};
return webdriver.get(url).then(null,handleError).then(function () {
return webdriver.executeScript("return window.stop()");
});
Well , the following concept worked with me on Chrome , try the same:
1) Navigate to "about:blank"
2) get element "body"
3) on the elemënt , just Send Keys Ësc
Just in case someone else might be stuck with the same forever loading annoyance, you can use simple add-ons such as Killspinners for Firefox to do the job effortlessly.
Edit : This solution doesn't work if javascript is the problem. Then you could go for a Greasemonkey script such as :
// ==UserScript==
// #name auto kill
// #namespace default
// #description auto kill
// #include *
// #version 1
// #grant none
// ==/UserScript==
function sleep1() {
window.stop();
setTimeout(sleep1, 1500);
}
setTimeout(sleep1, 5000);