I am using ruby and capybara(which leverages selenium) to automate walking through a website. After navigating to a new page I verify that the new page URL is what i'm expecting. My issue comes when I walk through an order funnel that is a single page but loads different views.
Some code...
I create my session instance then have additional code opening the browser and walking to a certain point in the website that I wont include
$session = Capybara::Session.new(:selenium)
My line for checking the browser URL without search params ie: everything after '?'
if url == $session.current_url.to_s.split("?")[0]
urlCorrect = true
end
This code works fine when my URL is
https://www.homepage.com
Then I click on a link that takes me to my order funnel ... https://www.homepage.com/order#/orderpage1?option1=something&option2=somethingelse
My function still matches the expected URL. But the issue comes when I move to the second page of the order funnel :
https://www.homepage.com/order#/orderpage2?option1=something&option2=somethingelse
My capybara code to get current url still returns the URL from orderpage1. Im guessing its because there is no postback when moving from orderpage1 to orderpage2 but i dont know how to force a postback or tell capybara to re-grab the url
Any advice will be appreciated.
Thanks
Quick Edit: I forgot to mention this behavior is only in IE. Chrome and Firefox both work correctly with the exact same code
Capybara grabs the current_url by querying the browser - it doesn't cache it. The issue you're probably running into is that clicking the link to move to the next page doesn't wait for the page change to happen, so if you call current_url before the page load has happened you'll still get the original url. There are two solutions for that - 1. use capybara to look for content that doesn't appear until the new page is loaded ( have_content ), 2. use the has_current_path? method which will wait/retry for a period of time until the current_path/url match
$session.has_current_path?('expected path')
There are options if you want to match against the full url, and you can use a regex to match as well - http://www.rubydoc.info/gems/capybara/Capybara/SessionMatchers#has_current_path%3F-instance_method
Thanks to Tom Walpole for finding the bug report for this issue. This link sums up the root of the issue and provides a few workarounds if anyone else is encountering this issue.
https://github.com/angular/protractor/issues/132
Related
I've already tried to find anything about that problem but I guess I either was not sure how to shortly describe the problem to find a solution or nobody else had that before which I can't think of. Maybe my thinking is wrong, too.
I have a stateless wicket 1.6 form with an ajax supporting panel (WebMarkupContainer with output id). The panel holds a dataview with paging navigator. The dataview itself is filled by a DataProvider.
The panel shows some entries from the database and below that is the navigator. by clicking any page on the navigator, the panel is refreshed (ajax) and shows content from that page. The page itself is not re-rendered by the browser.
When I now leave the page by navigating to another internal page (so basically when leaving the dataview-panel-page in any way) to open a detail page or so and then return to that dataview-page the navigator is resetted (because it's stateless I guess). The navigator can't remember which page to show and begins at the top of the first page again.
The question is: How can I solve this? I would like to i.ex. navigate to page 2 and then temporary leave the page for another internal page. When returning I want to be on page 2, focussed on the record where I clicked the link to "details" before. This also happens when I just open a new page in a new Browser tab.
Thank you!
Here's some code:
final WebMarkupContainer gamesPanel = new AjaxContainer("gamesPanel");
final DataView<Game> dataView =
new GameDataView("gameOverview", targetCurrencyModel, searchTextModel, gameFilterModel,
new GameDataProvider(searchTextModel, gameFilterModel, targetCurrencyModel));
dataView.setItemsPerPage(ITEMS_PER_PAGE);
gamesPanel.add(dataView);
final XNPagingNavigator navigator = new XNPagingNavigator("navigator", dataView);
navigator.setOutputMarkupId(true);
add(navigator);
You guys can try what I mean: The page I'm talking about is http://www.xbox-now.de. Just navigate to page 2, then click on details and return to main page.
I think you might use History API to push a new state when you click on navigation bar. In the new URL for the state you can include a parameter that indicates the index of the current navigator page.
You can customize your AJAX link in order to perform this operations when user click on it.
For more details on History API see Updating address bar with new URL without hash or reloading the page
I solved this in the NoWicket framework by introducing a model aware page cache which reuses page instances when hashcode/equals matches in the page model. You can see the implementation in the GitHub repo, try this implementation of the IPageFactory wrapper as a starting point, but it is more code involved there, just check out the code, debug the example and navigate the code to understand it better in order to apply it to your application. See your use case in action by trying this data table example in the documentation website (which you can also debug locally): http://invesdwin.de/nowicket/ajaxdatatable
Try changing the paging index, navigate to a different page and navigate back to the data table example page. You will still see the paging index that you left there.
Thank you guys for your replies. I've done it now another way. Used the base code from here and modified it in some ways. Added some nice css AttributeModifiers to indicate the actual page:
item.add(new AttributeModifier("class", new PageLinkCssModel(pageable, pageIndex, "active")));
Then I modified some code to add or reset the page parameter, that it's 1) used only once and 2) keeps all the actual page parameters which were there before adding own ones. So I am just appending the page number now. This way I can keep my initially mount path like www.foo.bar/path/path/path. Complete URL would now look like: www.foo.bar/path/path?page=123.
To pass my entered page (i.e. page=5) to the data provider I just had to override the providers iterator. It should start with the page I entered. Due to the ugly generated navigator URLs (which are extremly bad for SEO) I now have nice looking urls which are working independently what wasn't possible before. And that's also the reason why I could not get back to the correct page. The navigator URL was not lookup-able by wicket.
new DataView<GamePrice>("gamePriceOverview", new GameDetailDataProvider(gameId, targetRegion) {
#Override
public Iterator<GamePrice> iterator(final long first, final long count) {
return super.iterator(ITEMS_PER_PAGE * getCurrentPage(), count);
}
getCurrentPage() comes from the base template and gets the actual page number entered (if one is entered):
public long getCurrentPage() {
// -1 weil zero based
return getRequest().getQueryParameters().getParameterValue("page").toString() != null
? (getRequest().getQueryParameters().getParameterValue("page").toLong() - 1)
: 0;
}
So instead of having ugly SEO-unfriendly URLs which are also not compatible to work independant (directly enter the generated url) I now have the same URL I expect with an added page-parameter.
URL now would looks like:
http://localhost:8080/game/4249/de/doom-preorder?page=2
URL before was:
localhost:8080/game/4249/DE/doom-preorder?0-1.ILinkListener-gamePrices-navigator-navigation-2-pageLink
If I now go back from the detail page to the main index with active "Bookmarkable-Navigator", I correctly come back to the page and position where I left it (because of bookmarkable page links).
That's how I achieved this problem with a nice bonus: User- and SEO-friendly URLs.
I clicked on a button on some page which redirects me to some other page.but between these pages there is one page which comes only for 1 or 2 seconds. I have to verify that page.I am using selenium.
Any suggestions?
What you could do here, is do a simple waitForElement subroutine.
Ideally, when your test framework cannot find an element to operate with, it will fail.
So we can assume, that when your test gets past waitForElement(theInterimElement) then your interim element has appeared, and we can continue.
Check the url of the page is the url of the desired page and wait while this is not true.
while (!webDriver.Url.Contains(desiredUrl.ToString()))
Thread.Sleep(50);
I have a REST Url http://localhost/issues presenting me a search form for querying issues of a bug tracker system. The search form has a textfield with name="f". The action of the form is action="/issues" and the method=GET. So if I type "foobar" into the textfield, I am getting redirected to http://localhost/issues?f=foobar.
To prevent page reload, I use AJAX with jquery. So I used $(document).on('submit',...) to bind an AJAX query function to the submit function of the form. I used firebug to proove, that jquery sends the request and no page reload is taken. I also have a function to fill results of the response into my site presented in the browser. Everything works fine for the first search attempt.
But if I type another search string "anotherfoobar" into the textfield and resend the request, the request URL is http://localhost/issues?f=foobar&f=anotherfoobar, so the search parameter is only appended to the URL, not updatedas I thought it would happen.
I've read about "hashing" the URL parameters to prevent this behaviour here on stackoverflow, but do not fully understand what and why this is happening and what "hashing" means exactly. Perhaps it's not the right solution for me. So I kindly ask someone to explain me how I can solve this.
I've solved it! Don't ask how exactly, what I did was the following:
The AJAX response is the whole Web page as I would get it with a page reload. with jquery I select a div which includes the search form AND the results.
I simply defined a div only for the results, and update only the content of the results div. After doing so, everything works fine, and the results div updates with multiple search attempts.
Another case of not knowing what I do, but I do it fine ;)
Perhaps someone still can explain me my error so wisdom can rise.
I am implementing a non-Javascript page for a client. The page contains a number of third party grids. When JS is turned off and a sort is performed the page i,s posted back to the server , which is fine. However, I want the page to navigate to the grid that was sorted.
I can do this by using anchors, and I have tested it. However, as I am new to MVC I don't know how to append the anchor to the outgoing URL.
For example I may get a URL like Team/User/42?SortGrid1-field-asc
After it has been processed by the controller I need the URL to be sent to the client as
Team/User/42?SortGrid1-field-asc#Grid1
Any ideas?
I have tried to append the anchor internally using lines like
return Redirect(Url.Action("User", "Team", new { Id = Id }) + "#Grid1");
But fail to see how to stop it going off in an infinite loop. If I redirect to another action then I go down a whole new path , which, code wise, will go on forever.
Have a look at the answer to a question I posted: How can I add an anchor tag to my URL?
I think you may have to go through the controller and use generateURL to get the anchor into the url
This is probably going to get a resounding no, but I am wondering if it possible to have the URl change dynamically with using hashing, and without invoking a http request from the browser?
My client is keen on using AJAX for main navigation. This is fine, when the end user goes to the front page first, but when they want to use the deep linking, despite it working, it forces an extra load time as the page loads the front page, then invokes the AJAX from the hash.
UPDATE: Could it be possible, given that what I want to avoid is the page reload (the reason is that it looks bad) to stem the reload by catching the hash with PHP before the headers are sent, and redirecting before the page load. This way only one page loads, and the redirect is all but invisible to the user. Not sure how to do this, but seems like it is possible?
Yes, this is possible. I often do this to store state in the hash part of the URL. The result is that the page doesn't reload, but if the user does reload, they're taken to the right page.
Using this method, the URL will look like: "/index#page=home" or "/index#page=about"
You'll need to write a JavaScript function that handles navigation, and you'll need a containing div that gets rewritten with the contents fetched from AJAX.
Home
About
Questions
<div id="content"></div>
<script type="text/javascript">
function link(page) {
location.hash = "page="+page;
loadPage(page);
}
// NOTE: This is using MooTools. Use the AJAX method in whatever
// JavaScript framework you're using.
function loadPage(page) {
new Request.HTML({
url: "/ajax/"+page+".html",
onSuccess: function(tree, elements, html) {
document.id('content').setProperty('html', html);
}
}).get();
}
</script>
Now, you'll also need to have something that checks the hash on page load to load the right content initially. Again, this is using MooTools, but use whatever onLoad method your JavaScript framework provides.
<script type="text/javascript">
document.addEvent('domready', function() {
parts = location.hash.split('=');
loadPage(parts[1]);
}
</script>
Ok, the problem is that opening an AJAX link of the form http://example.com/#xyz results in a full page being downloaded to the browser, and then the AJAX-altered content is changed once the page has loaded and checked the hash part of its URL. The user has a diconcerting experience.
You can hugely improve this by making a page that just contains the static elements - menus, etc. - and a loading GIF in the content area. This page checks its URL upon loading and dynamically fetches the content specified by the hash part. The page can have any URL you want; we'll use http://example.com/a. Links to this page (http://example.com/a#xyz) now provide a good user experience for users with scripting enabled.
However, new users won't come to the site by fetching http://example.com/a; they'll fetch http://example.com. This is fine - serve the full page, including the home page content and links that don't require scripting to work (e.g., http://example.com/xyz). A script run on loading this page should alter the href of AJAXable links to their AJAX form (http://example.com/a#xyz); thus the first link a user clicks on will result in a full page load but subsequent ones won't.
The only remaining problem is is a no-script user gets sent an AJAX link. You can add a noscript block to the AJAX page that contains a message explaining the problem and provides a link back to the homepage; you could include instructions on how to enable scripting or even how to modify the link by removing a# and pressing enter.
It's not a great answer, but you can offer a different link in the page itself; e.g., if the address bar shows /#xyz you include a link to /xyz somewhere in the page. You could also add a link or button that uses script to bookmark the page, which would again use the non-AJAX form of the link.