4 weeks ago - last edited 4 weeks ago
4 weeks ago
Hi, this looks pretty standard to me. In my opinion, the root cause of the difficulty of automating desktop applications in Blue Prism comes down to that it does not have the concept of selectors that span multiple elements and/or the concept of anchors such as being able to target an ancestor/parent and then click on a child of it. Browser mode, surface automation, and Smart Vision don't have this issue as they do have that capability. I imagine Blue Prism intends to improve this in the future, but I've never come across a real solution to this. There are some workarounds that you could try.
Here are some things that I and other developers have tried in scenarios like this:
4 weeks ago
Hi David,
Thank you for your reply.
Wednesday - last edited Wednesday
Adding to what David has already mentioned, based on your update it seems that the issue is no longer only about identifying or navigating to the element, but specifically about the fact that the application is not exposing the displayed value through UIA or AA.
Since the parent attributes are also empty, AA gives the same result, and Match Index only helps with navigation but not with reading the value, this suggests that the value may be visually rendered on the screen without being published as an accessible property. In that case, Blue Prism can locate the control, but it cannot reliably read what the application itself does not expose through the automation layer.
At this point, I would separate the problem into two parts:
1. Interaction with the field
If Match Index can reliably move from one field to another, it may still be useful for interacting with the screen, provided that additional validation is added before the action. For example, checking that the expected labels, window title, number of child elements, or screen layout are still consistent before proceeding.
2. Reading the displayed value
If the value is not available in UIA, AA, parent attributes, or child attributes, then reading it through standard spying may not be feasible. In that case, the alternatives would be to retrieve the value from another source such as an export, clipboard behavior, application report, database, API, log file, or to use Region Mode/OCR where it is stable enough.
If Region Mode works for some elements, it may be worth standardizing the screen as much as possible before reading: fixed resolution, fixed zoom level, maximized window, consistent theme, and controlled RDP/session settings. This can make image-based extraction more reliable, although it should still be treated as a fallback rather than the preferred approach.
The most robust long-term solution would be to involve the application team and ask whether these controls can expose proper accessibility properties, such as Name, Value, Automation ID, or accessible text. Without that, Blue Prism may not have a dependable technical handle to read those values directly.
So in short, I agree with David’s points, but based on your findings, this looks less like a Blue Prism spying configuration issue and more like an application accessibility/exposure limitation. Blue Prism can only consume the properties made available by the target application; if the display value is only visually painted and not exposed through UIA/AA, then a workaround or an application-side change would likely be required.
Wednesday
Thank you, Omar-Abu-Snineh, for understanding the issue and for sharing your suggestions. We are currently following the same approach you outlined in part 1. For part 2, we are using the Clipboard alongside send keys, and we are also working on implementing region mode wherever possible by standardizing the screen.
We are coordinating with the application team on this issue, as there are several automation opportunities lined up for this application. Thanks again.