Download New File From a Website When it Becomes Available

[Peter13]

Hi,

Would greatly appreciate some help with downloading new files from a webpage when the files become available. They are usually posted once a week in a window pictured below.

I've got Power Automate to open Chrome and log into the website, but it fails when I try click on links below under View Statement and Download (there's an error that Chrome cannot download the file). I've tried desktop recorder and "Click on a link" approaches which results in Power Automate not being able to find the links.

Is there a way to handle Power Automate recognizing the new files and then download them? The new files will be at the top of the list, there will be two new files every week.

[MichaelAnnis]

Looks like you should be able to Send Keys {Tab} as many times as needed to the file and just Send Keys {Enter} when the dotted box gets to where you need.  If it's always 2 and you can run this as needed every week to get the 2, should be fairly simple to program.

Best of luck!

[Peter13]

Thank you, Michael. I've tried that. There's over a 100 {tab} to get to the link to download, but there's a text box on the way and Power Automate gets hung up on the text box and won't go past it. It even starts typing {tab} within it.

Any way around this? Or a different approach to solve this?

[MichaelAnnis]

"Move mouse to text on screen (OCR)"...seems like it is every seven days, so you should bable to calculate the Check Date (I wouldn't use settlement date as it appears in the settlement date and the previous check date.  So for this instance, you would say:

You could replace 10/29/2021 with a variable that is calculated elsewhere.  And the Offset of X where I have 300, should be however many pixels it takes to get from the center of the date field to the center of the link to click.

Whether you decide to do this in a loop or write each out separately, the second time will have to change the "Occurrence" number to "2".

Start with trying to extract the text from the page, and make sure you are getting valid OCR readings.  If not, go to OCR engine settings and start adding one to each multiplier until you get a proper OCR reading.  The higher the number, the longer it takes to OCR the screen; however, each number is increasingly more accurate.

If you are having trouble knowing how many pixels to offset, open a "Move mouse" command, and it will show the current location of your mouse.  Get the link coordinates and the date coordinates and subtract the X amounts to get your offset.

Best of Luck!

[Peter13]

Thank you, Michael.  I will try it. Appreciate your help.

[VJR]

Hi @Peter13 

Had a question that if its a web page why are you using a desktop recorder? Have you also tried the web recorder?

Also check the "Add Ui element" from the right hand side and try to detect/spy the hyperlinks.

I am sure the hyperlinks can be detected. It is a matter of choosing the correct attributes in the selectors.

[Peter13 · Accepted solution]

Hi VJR,

I actually was just able to figure out a solution. I used a combination of web and desktop recorders with UI elements. I had an issue before where UI elements were not recognized so the flow failed. Thank you for your input.

[NM2000]

Hi @Peter13,

I am faced with exactly the same problem. It would be great if you could please share a detailed description of how you solved the problem.

Thanks in advance 🙂