Unable to select UI element on Google Trends page (Chrome, Edge, FireFox)

[Anonymous]

Hi, I am using Launch New Browser (Chrome, Edge, FF) to navigate to this below URL.

https://trends.google.com/trends/explore?date=2020-01-01 2020-12-31,2019-01-01 2019-12-31,2018-01-01 2018-12-31&geo=US,US,US&q=how to cook ramps,how to cook ramps,how to cook ramps

On the page, I am trying to click this down arrow "download" icon so that I can initiate the download of the CSV file.  I tried using Click UI element in window, Click link on web page, Click download link on web page but none of them seems to identify that specific element (highlighting in red box).  In fact, it only puts red box on the whole page as "Document" and won't let me select anything else on the page.  I've tried recorders as well.  

How can I select the UI element on this Google Trends page?

Thank you so much for your help!!

[Anonymous]

@DamoBird365   Thank you so much for your help.  This is an awesome forum!!  Thank you!!

[Anonymous]

Not sure why, but it does not work anymore.  Maybe the selector is getting confused with other download elements?  

Robin.Core.ActionException: Click failed (failed to get UI element)
at Robin.Modules.UIAutomation.Actions.ClickElementInWindow.Execute(ActionContext context)
at Robin.Runtime.Engine.ActionRunner.RunAction(String action, Dictionary`2 inputArguments, Dictionary`2 outputArguments, IActionStatement statement)

[Anonymous]

JavaScript worked if I use the existing instance of Chrome.  If I do new instance, it throws an error.  I need to find the right and reliable combination I suppose.  Thanks.

[fraenK]

The selector might be wrong, strange ordinals. Don't use UI automation when you can use web automation. 

[Anonymous]

Thank you @fraenK .  I don't see Click element action under Web automation.  Which action should I use?

[Anonymous]

Oh, Press button on web page!!??  trying that now

[Anonymous]

Oh are these all UI element actions??  going back to the earlier posts.  Maybe I completely missed the nomenclature...

[burque505]

hi @Anonymous , I think the problem here is that you're launching Chrome again each time through the loop and navigating to 'CurrentItem'. You might try moving the Chrome launch back up out of the loop and navigating instead to

https://trends.google.com/trends/explore?date=2020-01-01 2020-12-31,2019-01-01 2019-12-31,2018-01-01 2018-12-31&geo=US,US,US&q=how to cook ramps,how to cook ramps,how to cook ramps

just like before. Seems to me it should work then.

As @fraenK pointed out before, there's more than one button with that title. If you use this

function ExecuteScript() {document.querySelector("div[class='fe-multi-heat-map-generated fe-atoms-generic-container'] button[title='CSV'] i[class='material-icons-extended gray']").click(); }

there's only one such button.

Also (my 2 cents' worth), although others may feel differently, there is no reason in the world for you not to use Javascript, VBScript, Powershell, Python or any other scripting language that PAD can handle, as long as a) it's easier than one of the alternatives and b) it works and c) it works repeatably and reliably.  Scripting capability is there for a reason.

Microsoft recognizes this. See PAD scripting. 

[fraenK]

Just add a new UI element (browser) and click the button and then modify the selector, liked mentioned way earlier: https://powerusers.microsoft.com/t5/Power-Automate-Desktop/Unable-to-select-UI-element-on-Google-Tre...

I would advise you to go trough the Power Automate Desktop basic training: https://docs.microsoft.com/en-us/learn/modules/pad-power-automate-desktop-development-essentials/

@burque505 In RPA you try to avoid 3rd party scripting as much as possible, because it adds complexity, which is very hard to manage. Scripting support is there to reuse existing scripts, when you migrate from earlier solutions or if there is something you can't handle with the built-in functionality/without scripting.

This challenge here is so **bleep** easy to solve with the built-in functionality.

[Anonymous]

Thank you both, @fraenK and @burque505 .  I appreciate both point of views equally and respectfully.  I use UiPath Studio and KNIME and others and I'm getting into PAD just now and I truly appreciate the help and support.  Thank you!!

[Anonymous]

I have all the Chrome instances closed and am still getting this error.  I'll try it on another machine if that makes a difference.  

Robin.Core.ActionException: Failed to assume control of Chrome (Internal error or communication failure). ---> System.NullReferenceException: Object reference not set to an instance of an object.
at Robin.Shared.Runtime.WebUtilities.ChromeLauncher.CreateWebBrowser(IntPtr handle, TabInfo tab)
at Robin.Shared.Runtime.WebUtilities.WebExtensionsBrowserLauncher.LaunchNew(ProcessWindowStyle processWindowStyle, String initialUrl)
at Robin.Shared.Runtime.WebUtilities.Chrome.TryStartNew(ProcessWindowStyle processWindowStyle, String initialUrl)
at Robin.Modules.WebAutomation.Actions.WebAutomationActions.LaunchNewChrome(Variant initialUrl, Variant chromeTabTitle, Variant chromeTabUrl, Variant dialogButtonToPress, Variant& webBrowserInstance, Int32 operation, Int32 windowState, Int32 attachMode, Boolean waitForWebPageToLoad, Int32 courseOfActionIfDialogAppears, Boolean clearCache, Boolean clearCookies)
--- End of inner exception stack trace ---
at Robin.Modules.WebAutomation.Actions.LaunchChromeBase.Execute(ActionContext context)
at Robin.Runtime.Engine.ActionRunner.RunAction(String action, Dictionary`2 inputArguments, Dictionary`2 outputArguments, IActionStatement statement)

[burque505]

@fraenK, with all respect, speaking from my 5+ years working with RPA, you are flat out wrong.

A. The scripting involved is not "3rd party." It's actions built into PAD. Powershell is Microsoft. IronPython is Microsoft. VBScript is Microsoft. Javascript is of course not MS but it's everywhere.

B. I fail to see how using simple JS on a web page is more complex than the convoluted editing of selectors.

C. You say "[s]cripting support is there to reuse existing scripts, when you migrate from earlier solutions or if there is something you can't handle with the built-in functionality/without scripting." If you can supply some authority for this, it would be helpful. In fact, on almost all platforms (e.g. UiPath, Automation Anywhere, Intellibot, Blue Prism, openrpa, ElectroNeek, G1ANT) you had better be prepared to do some scripting or you'll never get hired, if that's your goal.

The problem to be dealt with here is at least multiple.

1) The particular target page has active anti-automation measures. Try banging it long enough and you'll hit a captcha. This also means if you're trying to automate it (rotating proxy considerations aside) you had better be prepared to maintain it, because the page IS GOING TO CHANGE.

2) Sometimes UI Automation works (especially with Chrome, as here, with its "Chrome Legacy Window") even when it shouldn't. This UI Automation works reliably (so far, although it is certain to crash and burn for the reasons stated above and is unlikely to work in other browsers). Web Automation for this is a poor substitute for either JS or UI Automation for this particular case.

[burque505]

@Anonymous, I think I have misled you here, my apologies. I would try just launching a new Chrome instance, as you're going to be navigation to new URLs for each iteration of the loop, right? But that shouldn't stop it from working, I don't think. The error about not being able to communicate with Chrome is a head-scratcher. That said, if Chrome fails to shut down correctly I sometimes find orphaned instances of it still running in Task Manager. Would you mind checking to see if there are any such instances still running? You might also restarting (I know, I hate it when people tell me that, especially when I call call up Comcast for technical support 😀).

[Anonymous]

Thank you @burque505 .  I am googling to look for information.  Meantime, I tried Edge and it works with your JavaScript and can loop thru fine except that it's downloading the geoMap.csv files.  It's a progress!!

[Anonymous]

I tried the same workflow in another Windows machine and I am getting that same error.  I may stick with Edge since it worked now, just need to download the right file but it would actually be good if I can download these different files from the page.  I have to step away but thank you so much!!!!

[burque505]

Hi, @Anonymous, I've noticed that too (geoMap.csv, I mean). What is the name of the file you're wanting? If I use the UI Automation flow I posted earlier it downloads multiTimeline.csv. This is an interesting situation! Fun (well for me at least 😀).

As @fraenK quite rightly pointed out earlier, there are multiple potential download buttons. I know you're not going to be back to the forum for a while, but I'm going to to experiment with the different buttons (JS - sorry @fraenK 😀) to see if I can get a reliable, repeatable unique file name.

[Anonymous]

This is the file I was originally attempting to download multiTimeline.csv, but it's interesting that now the script is downloading the map too.  🙂  Added benefit!!!

[fraenK]

@burque505 Nope, worked for and with most of the leading vendors you've mentioned in/with their largest projects and customers.

A. "3rd party" was more about the technology, additional scripting environments bring additional knowledge, development, infrastructure, testing and maintenance requirements as well as their own problems. Because you mentioned it:

• PowerShell is not executed by PAD, but by Windows' PowerShell environment = dependency
• IronPython support is based on an older, just bundled DLL, which is not directly developed by Microsoft but an Open Source community. It "_aims_ to be a fully compatible implementation of the Python language". The PAD documentation is not very transparent about that, which other users already led to a full blown Python environment assumption
• VBScript is not executed by PAD as well, but by the Windows Scripting Host = dependency
• JavaScript (browser) is not executed by PAD as well, but by the chosen browser and we all know how standardized JavaScript support is... = dependency
• JavaScript (Windows) is not really JavaScript but Microsoft's JScript, which is a dialect and executed by Windows' Scripting Host (PAD documentation is not very transparent about that, which other users already led to a full blown JavaScript support assumption) = dependency

B. PAD is meant as a low code environment. You go to the selector builder and deselect unnecessary attributes and select necessary ones. Just a few clicks, no code and JS knowledge required, no additional dependencies. Why replacing an easy-to-use and working built-in functionality with additional code? 

C. RPA vendors added scripting support because customers requested it, because they wanted to reuse scripts or were skilled in those languages. And there were/are requirements which you can't meet with the built-in functionality. I was and am managing those discussions between customers and R&D. It's of cause always good to have some scripting skills, but for automation on scale, it's technical debt, which means additional maintenance efforts.

[fraenK]

Just use this code, it works fine in every browser, 30seconds to create:

WebAutomation.AttachToEdgeByUrl TabUrl: $'''https://trends.google.com/trends/explore?''' BrowserInstance=> Browser
WebAutomation.FormFilling.PressButton BrowserInstance: Browser Control: appmask['Web Page \'Google Trends\'']['1st CSV download button']

Add a UI element by clicking on the download button and strip down the selector in the selector builder down to the code, or switch to custom selector and paste the code:

button[Class="widget-actions-item export"]:eq(0)  

[burque505]

WebAutomation.AttachToEdgeByUrl TabUrl: $'''https://trends.google.com/trends/explore?''' BrowserInstance=> Browser

That won't work in every browser (that's for Edge only), and the URL should be

https://trends.google.com/trends/explore?date=2020-01-01 2020-12-31,2019-01-01 2019-12-31,2018-01-01 2018-12-31&geo=US,US,US&q=how to cook ramps,how to cook ramps,how to cook ramps

Also,

WebAutomation.FormFilling.PressButton BrowserInstance: Browser Control: appmask['Web Page \'Google Trends\'']['1st CSV download button']

will require the creation of an appmask first, and will definitely crash for anybody (at least using the free version, I think) who pastes that code in. Also, there is no form there, so Robin's placement of this action

WebAutomation.FormFilling.PressButton

in the subcategory it's in is sure to generate confusion.

The selector you supply

button[Class="widget-actions-item export"]:eq(0)  

doesn't work for me in Edge.