I'm building a webcrawler using Selenium Webdriver with Chrome. It uses
links = self.browser.find_elements_by_xpath("//*[#href]")
gets all the links on a page and then loops through the links to filter them. However, on every site I've tested, once it gets to a certain link on a certain page, it stalls at this line:
url = link.get_attribute("href")
It's not an error, it just stops, as if it hit an unending loop. For a given site, it's always on the same link on the same page for every test run. The stalling does not occur with Firefox - everything works fine. However, I need it to do this using Chrome for other reasons (trust me on this).
The version of Chrome I'm using is 28.0.1500.52, and I'm using the latest Chrome driver, which is compatible with that version. What could be going on here?
EDIT (IGNORE THIS EDIT)
After looking into it some more, I've found what's causing the problem: There's a part further down in the code that mouses over every link that's found.
hov = ActionChains(self.browser).move_to_element(link)
hov.perform()
When I comment that part out, it works. Now I have to figure out why this should have any effect on getting the href attribute from a Selenium link object (especially when it works fine for the first arbitrary number of link objects)...
EDIT 2
Ignore the first edit.
Related
Suddenly I can no longer call up variable values when hovering in the Source Code tab of Google Chrome's Developer Tools. Nothing happens. A few days ago everything was working fine. Current Chrome Version (105.0.5195.102)
What's weird: if I deploy a version and then try, everything works. It just doesn't work on localhost
If I use res in the callback function, I don't see a pop-up
I'm using 105.0.5195.128, and it will not work with my bundled TS - code. It has always worked as it should.
But look.... I took a Demo Angular: https://ng-devtools.netlify.app/ , and somehow it works... but the popup is very different from what I'm used to.
i'm struggling with the classic HTML input of type file. I was using it on my angular application and i had freeze sometimes, the whole browser became unresponsive, and the only thing that works is the scroll.
You can't click any buttons, select any text or even change tab on the browser for about 5 seconds. I tried a lot of different file type and sizes and it doesn't change any of this, it even freeze sometimes when i don't select any file in the explorer and i click the close button directly.
So I thought it was my implementation that was bad and i went to the Mozilla developper website where you have an example and I have exactly the same behaviour. I also tried to take only the HTML from the Mozilla website and put it alone on a .html file and again, same behaviour.
This does not occur 100% of the time, but I would argue that on my end it does it about 60/70 % of the times.
I tried it on chrome and Firefox (both up to date), on multiple computer (all running Windows though) and i have the same behaviour on all of them.
I don't have any error or anyting in the consoles.
I don't know what to do with this, i'm pretty sure it can't be the file's input on the whole web that are bugged, but i tried a lot of various things with always the same results. Should i report it somewhere? ( if so, where? ) Or what would you suggest to do to investigate this further?
I couldn't find anyone discussing this issue on the internet apart for this thread that had no solution, they also say in this thread that the bug is not reproducible in Edge, but i just tried and it does the same thing.
Thanks for reading me and for any help about this.
Chrome freezes for few seconds when after any use of file field.
It was because I had a shortcut in "Quick Access" menu in windows explorer. This shortcut has been linked with a folder shared by network. I've removed this shortcut and everything is good now.
Same here, I have a mapped network drive that is not responding, this make a 5 seconds delay. After unmapping the drive, no more delay. The delay is the same in firefox or chrome.
Same here, if you want to remove it, on Windows 10, click on the icon Quick access (blue star), then right click on the dead link appearing in the "Frequent folders" panel on the right panel and choose "Unpin/Remove from Quick access". There should be no error message.
Hello dear people of Stackoverflow,
currently I am writing acceptance tests with Codeception using Selenium as WebDriver module. In my test to check if our sub-navigation exists, is complete and works I struggle with the following piece of code:
$I->see('Partners');
$I->click('Partners');
all the see calls are working perfectly, but the click call fails with an ElementNotVisibleException stating that the element to be clicked is not visible, which I don't understand since the see call worked.
To make sure we're not seeing any old "Partners" string on the page but it really is the link I want to click, I changed the call by adding the a selector.
$I->see('Partners', 'a');
$I->click('Partners');
Still, I'm getting the ElementNotVisibleException from before. In the next step, I added a context to the click call to this:
$I->click('Partners', 'a');
Which made the exception disappear. Still, the click never happens, the test simply fails with the message:
Link or Button or CSS or XPath element with 'Partner' was not found.
I even used XPath in the click call like this: //a[text()='Partners'] and get the same message.
What bothers me the most with this is that the see call seems to see the link I am trying to click and moreover, even the page source and screenshot provided by Codeception after a fail contain this very link, visible and in valid HTML.
I tried to click on some other elements within the sub-navigation with the same result, while - which is even more strange - clicks on links in the main navigation seem to work just fine.
I have no idea why this doesn't work but any help is greatly appreciated.
I don't understand exactly why it happens in some cases and not others, but I've found a solution that works for me: Use CSS selectors only.
Basically in some cases (I think it has to do with clicking on div/span rather than a/button) ->see() will work with the fuzzy text string as first argument, but ->click() requires the explicit CSS selector.
So this will have the "I can see but not click" problems:
// BROKEN
$I->see("All topics", ".header-taxonomy .taxonomy-list-title");
$I->click("All topics", ".header-taxonomy .taxonomy-list-title");
But this will work:
// WORKS
$I->see("All topics", ".header-taxonomy .taxonomy-list-title");
$I->click(".header-taxonomy .taxonomy-list-title");
Obviously it's not ideal at all, partly because it requires a unique selector, but for me mostly just because it makes the tests and testing output less consistent and harder to read.
Either way, I just wanted to get this testing script finished, and now it's working, and because I'm still using ->see() with the fuzzy text locator, I'm still confident that the text is showing before I click.
Notably, this issue doesn't happen when using the PHPBrowser module, which is what tripped me up. I "upgraded" to WebDriver and found my scripts were broken in all these places due to the different behavior of ->click() (in addition to the expected need to open menus before using them, which wasn't needed in PHPBrowser)
When a site gets stuck in chrome with the 'loading www.site.com' status, Is there anyway to know which of the elements on the DOM is delaying the load? for example, it can be an external js file or image that is taking long time to load. I tried using the network tab in the DeveloperTools, but it only shows the elements already loaded and the time they took. I couldn't find a way to see which element is stuck.
I found a way that I'm not sure it works in all cases, but it did for me. I used the Timeline tab (now shown in the Performance tab), it's one of the tabs in Chrome Dev Tools.
Open Chrome Developer tools (F12 or CMD + Alt + I in Mac) and click the Performance tab
Make sure you drag the grey timeline bar all the way to the right, you want to see the most recent section of events so you can find what's getting stuck:
Check just down below the records sidebar. If you don't see this like in the capture below, try toggling the icons, there are different view modes. Now, going all the way down I found these. It doesn't look like that's going well. Obviously some kind of infinite loop in some random code:
I couldn't find out if it was an extension itself or some other bit of code. Clicking through I only could find // Copyright 2014 The Chromium Authors. All rights reserved..
Checking in Incognito mode clear of extensions, the website loads just fine.
I'd also advise to check the Console tab to see if there are were any errors that might cause the infinite loading thing. You'll often find clues there.
Got an odd problem, only persistent in Chrome browser. Can't imagine what would be the problem. Chrome's "Back" button doesn't work, nor does history.go(-1), but works fine on all other browsers. To see the problem simply click this link, which will asynchronously initiate searches on two separate services, or sites if you will. Upon click on any search result list item, we traverse to another page, but "Back" doesn't work for Chrome.
http://vps-net.com/MSSMine/?search=some
Any suggestions or ideas are welcome.
I found the problem with the page you have linked, I don't believe it to be an issue with Chrome.
When you load the page, it also loads two iFrames as the request to the site finishes. Chrome takes a somewhat different approach to history, allowing you to navigate not simply from changes in the URL displayed, but through every new individual browser-initiated request. When the iFrames load, Chrome adds history for each step of the page loading process, creating 4 separate history items. When you go back 1 "page load", it takes you back to the site you linked to, at a different step in the process. You aren't seeing anything change because as soon as that page loads, any unloaded iFrame now loads again.
If you do history.go(-8) or some other large number, you should see the page at that many pageloads ago. However, if the link is opened "in a new tab" then the history begins for that tab at the URL you told it to open, effectively limiting the history to the pages that occurred during that tab's lifespan.
This was tested on Chrome 34.0.1847.116 running under Ubuntu 13.10.
My back button was not working as well, but it would not work with any browser. I went into my add/delete software and deleted what software was added the night before - BOOM! All is fine now.