Java app

[pgriffin]

I need to write testing scripts for a Java based app. I cannot use Webrecorder to 'see' the window names/objects OR Veiw System Windows.

I am no java expert. How do I best navigate these screens, edit fields, combo boxes, etc...?

[Captive]

Depending on the applets that you will be working with, you may be able to use something as simple as MouseMoveRel to click on the same part of the window at each stage.

You can use GetRectCheckSum (GetPixelColor if it's reaaaaaly simple) to figure out what's being shown at a certain part.

Or, there's that fancy new "Image Recognition Library" you can investigate.
( http://www.mjtnet.com/imagerecognition.htm ).

It all depends on what exactly you need to do, and what the situation is.

[pgriffin]

I am familiar with GetRectChecksum (way too slow). I can use GetPixelcolor (just chain a few of them to together) to do what I call "fingerprinting". The objects are generally going to be located in the same locations, I was just wondering if there is a tool I can use to actually use functions like (but not identical to) GetControlText, SetControlText, Pushbutton, etc....

[Rain]

//Depending on system speed this script adds 10000 pixels in 6 seconds
//Desired_Color returns the value of all 10000 pixels combined from a 100 by 100 pixel area
//Adjust the X and Y positions to your needs
//let>X_Position= and let>Y_Position= are the starting positions of the search
//if>X_Position= and if>Y_Position= are the ending positions of the search

Label>Pre_Get_Color_Loop
let>X_Position=0
let>Y_Position=0
let>Desired_Color=0
GetTime>Start_Time1
Label>Get_Color_Loop
GetPixelColor>X_Position,Y_Position,Pixel_Color
add>Desired_Color,%Pixel_Color%
add>X_Position,1
if>X_Position=100
let>X_Position=0
add>Y_Position,1
if>Y_Position=100,What_Is_The_Color
endif
GoTo>Get_Color_Loop
Label>What_Is_The_Color
MessageModal>Desired_Color
if>Desired_Color=1111111,Color_Correct_Do_Something,Color_Incorrect_Do_Something_Else
Label>Color_Incorrect_Do_Something_Else
Label>Color_Correct_Do_Something


.

I agree with SkunkWorks GetRectCheckSum> is way too slow.
It takes GetRectCheckSum>0,0,100,100,Pixel_Color 20 seconds to return the value of the same area

See for your self by running the script below

Label>Pre_Get_Color_Loop
let>X_Position=0
let>Y_Position=0
let>Desired_Color=0
GetTime>Start_Time1
Label>Get_Color_Loop
GetPixelColor>X_Position,Y_Position,Pixel_Color
add>Desired_Color,%Pixel_Color%
add>X_Position,1
if>X_Position=100
let>X_Position=0
add>Y_Position,1
if>Y_Position=100,What_Is_The_Color
endif
GoTo>Get_Color_Loop
Label>What_Is_The_Color
GetTime>End_Time1
GetTime>Start_Time2
GetRectCheckSum>0,0,100,100,Pixel_Color
GetTime>End_Time2
MessageModal>Search 1 started at: %Start_Time1%%crlf%Search 1 completed at: %End_Time1%%crlf%GetRectCheckSum times%crlf%Search 2 started at: %Start_Time2%%crlf%Search 2 completed at: %End_Time2%

[Marcus Tettmar]

GetRectCheckSum checks every single pixel, and uses it's position and colour, to calculate a unique checksum. That's why it's slow. It should only be used for small rectangles.

For larger areas, for image comparison or to find an image on a screen (or wait for a screen area to change) you'll get better results from the Image Recognition Library which is really fast:
http://www.mjtnet.com/imagerecognition.htm

[Rain]

10000 Pixels was an extreme example to point out how much faster a simple add loop is. I normally search a much smaller area that is roughly 50 pixels and by using my add loop I keep my scripts fast.

I have played it with the Image Recognition Library and had good results.
But here is my question. If I want to use it in an exe on a client’s computer does my client have to install the Image Recognition Library? If that’s the case the Image Recognition Library is not helping me at all. It would be a hassle to tell my clients they have to download and install another piece of software to run my programs.

Thanks

[Marcus Tettmar]

But here is my question. If I want to use it in an exe on a client’s computer does my client have to install the Image Recognition Library? If that’s the case the Image Recognition Library is not helping me at all. It would be a hassle to tell my clients they have to download and install another piece of software to run my programs.

No, just the DLL. They don't have to download anything. You can use the Image Recognition Library functions in your exe and then you just have to send your client the EXE and the DLL. With WinZip you could create a simple self extracting exe which puts both your exe and the dll into the same folder. your exe would load the DLL from the script dir:

LibLoad>%SCRIPT_DIR%\imglib.dll,ilh

[Rain]

Thanks for the quick reply. I guess I should have asked a little sooner.

[pgriffin]

this thread has been a big help. I will quickly ramp up on using the Image Lib.

I had considered the "pixel by pixel" search shown earlier, but was concerned about speed. I also thought there must be something better. One thing I have used very effectively is what I call "FingerPrint". That is pick set number of pixels (50 or so) to test for an image. Just as when a person is 'finger printed', you don't need to analyze the entire image, just need enough points to be sure of uniqueness. I think if a fingerprint system finds 10 'matches' in peculiarities of your print, it considers that a good match.

FingerPrint is super fast, can be written easily into a subroutine, and I've never mis-identified an image. (And you don't need to understand DLL's or include them in your zip, etc...)

SkunkWorks

[Marcus Tettmar]

One thing I have used very effectively is what I call "FingerPrint". That is pick set number of pixels (50 or so) to test for an image. Just as when a person is 'finger printed', you don't need to analyze the entire image, just need enough points to be sure of uniqueness. I think if a fingerprint system finds 10 'matches' in peculiarities of your print, it considers that a good match.

The image recognition library uses a similar approach.