I have an application that I am trying to automate on Windows. I need to find the location of a window that is running inside the application, and then automate a couple of mouse events on the application.
In a previous incarnation of the software that I am automating, I was able to search for child windows of the process which were named using the GetWindowText WinAPI function from C# (in combination with GetWindowTextLength).
The software manufacturers have now updated the software and updated the way that the child windows are drawn. Now each window lacks a caption, and has a class name of QWidget. I can no longer use my old strategy to find the child window location. I presume the use of QWidget means the windowing system uses the Qt framework.
Is there any way of pulling any data from the QWidget using PInvoke that I might be able to identify my windows with?
There's a couple problems here. One is that you can't get "unshared" data from another process. You can get at window data by pinvoking methods like GetWindowLong; but unless you know specific data about what QWidget does in that data (the other problem), there's not much you can do with the data.
Another problem is if you want to use most QT objects in a managed application (you can do this with C++/CLI and IJW) you need to initialize A QT Application object in your application... I'm not sure how this would impact what you want to do.
Related
I have been charged with the task to send data from COM 1 to COM 2 in Windows CE 4.2. A running application takes data from COM 1 and displays it to the user in a form (textbox I assume). I then have to take that data and send it out COM 2. COM 1 is being used by the program displaying the data and I know of no way to hijack COM 1.
I figure trying to do a screen scrap would be the next step. Unfortunately this is compact framework and an old version at that and from a lot of research it seems managed code is out of the question . Many of the API functions I would use are not available; FindWindowEx for example.
Here is where I am at now. I have created two projects. One runs with a TextBox and some wording. A separate application runs and tries to read the text in that TextBox. I have been able to find the running process based off the name of the form using FindWindow API. Using code I have found on this site I have even been able to enumerate through the controls of the form. However my TextBox is never found and many of the controls that are found where never placed on the form by myself (listbox, button). I assume those are the form's initial controls.
Does anyone have any experience with this? Currently this is in C# but VB or Visual C++ will be fine. Even if you have any ideas on a third party application. BTW I am not given the option to upgrade to a higher version of compact framework.
Thank you.
An update I just found out about. It does not look like I only screen scrape only new data but instead have to screen scrape the entire screen and send it out COM2. Someone will scan a barcode and I will send out all screen data through COM2. The data may include a picture etc.
For a native C application I have one sceleton that enumerates all child windows and controls inside a dialog: http://code.google.com/p/rdp-auto-login/source/browse/trunk/rdp-auto-login/tscDialog.cpp. See ScanTSCWindow and the results found in the comment "TSC dialog elements".
I started with remote spy and looked thru the RDM window to find the CtrlID values. There is also a nice tool called zDump (http://www.hjgode.de/wp/2009/06/11/zdump-take-a-look-inside-windows-ce/) that runs on device and enables you to look at window elements.
The theory is that every element in Windows (either Mobile or desktop) is a window. Windows are accesible by there window handle. The handle is assign by the OS during CreateWindow/Ex. Inside dialogs, elements can be identified by there control ID (a resource value), the window class (ie "EDIT", "LISTBOX") and window text and internally by the window handle at creation.
The problem with Compact Framework apps is that they hide many of these basics and dialog (Form) elements can not be easily identified from another process.
As you say you are not able to capture COM1, what is, if you stop the application and then open COM1? As knonw, normally on one application can access a serial port at the same time. You can then read the serial data directly and do not need to access a foreign window.
There are also drivers that enable port mirroring or multiple access. Even for Windows CE based OS (ie http://www.virtual-serial-port.org/products/serial-splitter-mobile/).
If the application is a compact framework one you can take a look inside the code using .Net Reflector or similar .Net decompilers. I use that often to mimik or learn from other apps.
You say "I have been charged with the task to send data from COM 1 to COM 2 in Windows CE 4.2. A running application takes data from COM 1 and displays it to the user in a form (textbox I assume). I then have to take that data and send it out COM 2. COM 1 is being used by the program displaying the data and I know of no way to hijack COM 1." and if you do not start the other app you can write your own and do not need to parse the foreign app.
Possibly you can post the other app or more details of what it does what you can not do.
EDIT/UPDATE:
as we now know it is Intermec TE2000 (terminal emulation) the answer is to use the XMLRPC interface provided by TE2000. The interface is able to call back a function hosted by an xmlrpc server and send all screen content (text, fields and attributes) on screen changes. I have working c++ stl windowsce code for that.
If the device is connected via network, the xmlrpc server can even run on a PC.
As TE2000 does use native drwastring API you will not success in reading texts from the window. If you screen capture the window, you will have to do OCR on the image. XMLRPC does avoid all this.
UPDATE2:
I finished a class lib to get async screen updates using TE2000 xmlrpc: see https://github.com/hjgode/ITE_xml_rpc/tree/master/XmlRpcCS/XmlRpcCF
and http://community.intermec.com/t5/Thin-Client-Based-Development/Printing-CV60-Screen-Windows-CE-4-2/m-p/28663/highlight/false#M473
Hello all i have a rather specific Question and please consider my apologies if this is not a standard stack-overflow question ,
i need to know is it possible to get all control names of another exe e.g.
i have a c# application and i want to get all control names of another standalone application running on my computer say b.exe so if b.exe have a textbox and a button in it i want to have names of textbox and button in my c# exe and also i want to have a listener for click events of b.exe
till now whatever i have researched is nothing special actually i dont want full code snaps i just want a guideline can i do it thorugh pinvoke? or winapi? please explain what is pinvoke and winapi and is it possible through application hooking ? once again i need a guideline a way to follow so please help me out regarding this, i know i can get active windows name through pinvoke or winapi but my requirements are little high
1 . Access to control names:
Unfortunately I do not .NET (C#) but under Visual C++ this would be impossible.
The reason is: The controls do not have names in the executable any longer but the names are simply converted to numbers.
Example "myTextControl" would be 1234 in the executable. The information about the (original) name of the control gets completely lost while compiling.
2 . Accessing controls in other executables:
I think under .NET (C#) it is not as easy as under native programming languages (directly accessing the Windows API) but it is possible to access controls of other executables as long as you know the number (e.g. 1234).
Therefore you'll have to find out the window handle of the dialog window containing the control and then you can send messages to the control. Unfortunately many messages containing pointers will require some very tricky hacks. Messages without pointer work well.
3 . Creating a listener for a control in other executables:
Creating a listener (e.g. a click listener) on a control in another EXE is possible however very tricky: This would require writing a native (this means: not .NET / C#) DLL file which is then combined with the Windows API function "SetWindowsHookEx".
I do not think that a .NET API (required for C#) already exists that will do this because there are only few use cases for this.
I was playing around with Microsoft Spy++ and noticed that not only does it find the open processes, but can find the individual components running in each process. For example there is this application that allows you to open a window in which there is a textbox for an IP address and textbox for a port. Spy++ can detect these components. Knowing that Spy++ can detect them, is there anyway possible to find them in a separate c# application and go on to MODIFY their contents and otherwise interact with the program? (such as firing a click event on a button)
This is feasible. Try use PInvoke (InterOp) or AutomationElement, or AutomationPeer (for WPF applications) to automate all you wish to do.
Also you might wish to try Inspect and UISpy application as well.
Automation elements/peer is a non-intrusive mechanism to control UI using accessibility framework. One of the weaknesses in windows is its lack of defence against code injection. Put simply:
As a privileged user,
- You can Open and Modify a running Process image
- Make it load your OWN DLL
- Make it run your OWN thread (that potentially listens to commands from your process) and
- allows you to read any bits of memory you want.
Look at detours (http://research.microsoft.com/en-us/projects/detours/) for how to do it with Managed Processes.. Unfortunately, Microsoft removed the inject at runtime features.
Also look at http://msdn.microsoft.com/en-us/magazine/cc163617.aspx for doing things in the managed world (Apps like Snoop utilise that)
I am looking for a way to embed Windows Forms applications written in C# in a C++ windows application. The native application main window is subdivided into several panes. The C# app is supposed to appear within one of those panes, i.e. the root window of the C# component (the outermost form) must be a child window of the main application.
Can this be done? If so, how?
Some additional context: To my knowledge, there are two ways to go about this. First, host the CLR in the native app by using the .net hosting APIs (ICLRRuntimeHost, etc.). Second, host the CLR by putting the windows form in an ActiveX control.
Regarding the first approach, I have managed to get the CLR started up and to load a C# assembly (thanks largely to Mattias Högström). Where I am hitting a road block is that I see no way how to tell the component I am running in the CLR that it needs to be a child of a window passed in from the C++ side.
I have also experimented with the second method (using ActiveX and thanks to Daniel Yanovsky). It almost, but only almost, works for my purposes. I can let arbitrary Windows Forms components run in a child pane of the native app. BUT they always run on the main thread of the main app. This means they use the windows message loop of the main app. MSDN says this will not work reliably since the standard Windows message loops don't meet the requirements of Windows Forms (I wanted to post the link to MSDN here but have already used up my new-user-two-link-allotment).
The exceptions to the message loop issue are, according to MSDN, Internet Explorer and MFC apps. The native app I am using as a host is definitely not Internet Explorer. Also, it uses the windows API as wrapped by wxWidgets so MFC is not (or at least not a welcome) option.
The solutions Microsoft proposes involve letting the C# components run in their own message loops on their own threads. This, at least as far as I can tell, necessarily leads back to the first approach mentioned above. So I am back to the question of getting a Windows Form to work under a passed-in parent window.
Again, I am interested in any input that clarifies the child window issue, independent of the approaches I have mentioned here. But in light of the context I could reduce the general question to two specific questions (and I would need an answer to only one of them):
Given a Windows Form hosted in an ActiveX control, how can I allow the form to run in its own message loop on its own thread?
or
Given a Windows Form running in a CLR hosted by a native app, how can I make the form be a child window of a window in the native app?
yes, it can be done. We did the same thing years ago. This is our approach: .NET control also has native window handle, we get these handles through C++/CLI and pass them to Win32, and add these handles as the children of native windows. So the .NET controls run on the main thread of the main app, the problematic part is the message loop, as you have mentioned in your question. We need route the message between the nation and .NET windows if needed. As I remembered, we fixed a lot of bugs related to this, and still there were some mysterious problem we didn't figure out.
There's anothe way: using WPF interop. http://msdn.microsoft.com/en-us/library/ms742522.aspx . According to MS, this should fix the message problems, but we didn't try this approach. You can just use Win32 to host WPF, then use WPF to host Winform.
So for your last 2 questions:
You can't. Only one message loop.
Use handle. This article talks about the handle of Windows Forms: http://blogs.msdn.com/b/jfoscoding/archive/2004/11/24/269416.aspx
We have an old Windows 32 bit app written in C++ which does some stuff and displays the results in what resembles a textbox.
I have been ask to write an application in C# that reads the data out of the old app and then further process the data.
The issue is how do I read the textbox in the old application?
Someone told me I can get the “handle “of the application using windows API and step through the controls and then read each ones data! Is this true and if so how would I do it from C#?
This is to be a .Net 4 Windows forms Application.
Many thanks
You're probably going to have to use some Interop calls to accomplish this, specifically using a combination of FindWindow / FindWindowEx and SendMessage & WM_GETTEXT / WM_GETTEXTLENGTH.
Here is an article on the subject (in c++, however the same concepts will just need to be ported to use P/Invoke), it's a bit dated but I believe is should be very relevant to your situation.
You can use Spy++ application (comes with VisualStudio) to inspect your native application and find class name of control you are looking for.
Having that, you can get your native application's main window handle, which is easy if you are responsible for launching that application from your c# app:
...
var proc = Process.Start();
var mainWndHandle = proc.MainWindowHandle;
Otherwise, you will have to find other means of finding that window, fe. you can use function that I've described below to look for that window on your desktop (see msdn page for more info)
If you have all that, you can then get handle to the textbox control, using FindWindowEx function (as well as you can use it to find main window, by passing NULL as a hwndParent).
And when you have handle to this textbox, you can read it contents, by sending message of WM_GETTEXT.
If you haven't been PInvoke'ing much, you can check http://pinvoke.net/ for reference on how to call WINAPI function from your c# program.
Do you really need the UI of the C++ application? Because it will be very unconfortable to have 2 separate UIs with different message pumps. You'll need to syncronize this mess on the update of the value.
Why don't you properly strip the logic code from the C++ application and expose it in a Interop CLR project (C++/CLI)? If you don't need to add more features to the C++ app, it seems very straightforward to me