Label with no ClassNN. How to find it or read text from Label another way?

[FalseShepard]

I'm writing a script to automate my work with a specific circuit emulator.

Everything is fine, except one window. When measuring chart values, I can't read a numeric value with AHK because the numeric value has no ClassNN.
It is important to clarify that this text cannot be selected by the mouse, it is something like a "label", not a "textbox"

I have tried Window Spy and a script to determine the object properties under the mouse pointer - the ClassNN remains empty. (See pic.)

Code used to show ToolTip with window info:

Code: Select all

    MouseGetPos(&x, &y, &w, &c)
    MouseGetPos(,,,&h, 2)
    ToolTip("x: " x "`ny: " y "`nwindow: " w "`nclassNN: " c "`nHWND: " h)

What I get with it and Window Spy:

image.png (45.23 KiB) Viewed 469 times

I tried using Easy OCR but it makes errors, especially with numbers 6 and 9. I need to copy this text accurately, so I want to get direct access to this text field.

I'd appreciate any solutions provided.

[boiler]

You may be able to use UIA.

[FalseShepard]

You may be able to use UIA.

Good lib, but doesn't help.

For example, I can get this label (it cant be selected by mouse) by its "Name" property

image.png (28.07 KiB) Viewed 296 times

UIATreeInspector_002085.png (7.65 KiB) Viewed 296 times

But as for "Линейка" window from what I need to extract value - there is no way to do it, even with UIA - it doesn't have a property that contains its value.

1.scm _ - FASTMEAN_002086.png (90.66 KiB) Viewed 296 times

Maybe I do something wrong?

[boiler]

Good lib, but doesn't help.

It's possible that it's just not available there either.

Maybe I do something wrong?

I don't have much experience with that library to know.

[Descolada]

@FalseShepard in that case the text is most likely just not accessible. You could try reading the values directly from the programs memory, which would probably be rather difficult.

Alternatively try to optimize OCR performance. For example, with the OCR.ahk library you should set the zoom factor to 2 and make sure the language is correct (probably should be "ru"). Also, the built-in Windows OCR which OCR.ahk uses isn't the most accurate, other OCR engines such as Tesseract might give better results.

Another option would be to implement OCR yourself by using the FindText library: you would need to capture images for all individual numbers from 0-9, do the FindText call, and then use FindText's OCR function.