Firefox 引导扩展：获取浏览器窗口的本机 HWND 句柄

Question

I have an external application and I want it to display some information on top of the browser window. My bootstrapped extension needs to pass the browser window handle (native HWND) to my application, along with some other useful information about the window. I'm able to do the communication between them, the only thing that is missing is a way to get the native HWND of the Firefox window.

I read a lot about it and although I belive it's possible, I couldn't find a working solution. Here's what I've tried so far:

This one should give me nsIBaseWindow , so I could get nsIBaseWindow.nativeHandle or nsIBaseWindow.ParentNativeWindow , but no success:

var window = SomeDOMWindow; // Informative
var baseWindow = window.QueryInterface(Components.interfaces.nsIInterfaceRequestor)
                        .getInterface(Components.interfaces.nsIWebNavigation)
                        .QueryInterface(Components.interfaces.nsIDocShellTreeItem)
                        .treeOwner
                        .QueryInterface(Components.interfaces.nsIInterfaceRequestor)
                        .getInterface(Components.interfaces.nsIXULWindow)
                        .docShell
                        .QueryInterface(Components.interfaces.nsIBaseWindow);

The above code is widely spread on forums, but I couldn't get it to work for me.

The other one does not seem to be much accurate since it gets the HWND based on the window's class and title, which can lead to wrong results:

Components.utils.import("resource://gre/modules/ctypes.jsm");
var lib = ctypes.open("user32.dll");
var fww = lib.declare("FindWindowW", ctypes.winapi_abi,
  ctypes.voidptr_t, ctypes.jschar.ptr, ctypes.jschar.ptr);
var sfw = lib.declare("SetForegroundWindow", ctypes.winapi_abi,
  ctypes.int32_t, ctypes.voidptr_t);
var hwnd = fww("MozillaWindowClass", document.title);
setTimeout(function() {
  sfw(hwnd);
  lib.close();
}, 3000);

Any help would be appreciated.

Answer 1

window must be a root one (ie an instance of ChromeWindow )

The following code should work

var win = Cc["@mozilla.org/appshell/window-mediator;1"].getService(Ci.nsIWindowMediator).getMostRecentWindow("navigator:browser");
var basewindow = win.QueryInterface(Ci.nsIInterfaceRequestor)
                 .getInterface(Ci.nsIWebNavigation)
                 .QueryInterface(Ci.nsIDocShellTreeItem)
                 .treeOwner
                 .QueryInterface(Ci.nsIInterfaceRequestor)
                 .nsIBaseWindow;
var nativehandle = basewindow.nativeHandle;

Answer 2

The problem was that I was querying the wrong interface from the subject param in the xul-window-registered observer. I need to get an nsIDOMWindow instead of an nsIXULWindow so the first code mentioned in my question works. So now I'm doing the following, with some piece of code @Noit suggested:

observe: function(subject, topic, data) {
    var newWindow  = subject.QueryInterface(Ci.nsIInterfaceRequestor).getInterface(Ci.nsIDOMWindowInternal || Ci.nsIDOMWindow);
    var basewindow = newWindow.QueryInterface(Ci.nsIInterfaceRequestor)
                     .getInterface(Ci.nsIWebNavigation)
                     .QueryInterface(Ci.nsIDocShellTreeItem)
                     .treeOwner
                     .QueryInterface(Ci.nsIInterfaceRequestor)
                     .nsIBaseWindow;
    var nativehandle = basewindow.nativeHandle;
}

And it works!

Thank you very much for your help.

Answer 3

I also just came across this, it might be nice:

Cu.import("resource://gre/modules/ctypes.jsm");

/*start getcursorpos*/
var lib = ctypes.open("user32.dll");

/*foreground window stuff*/
var FindWindowA = lib.declare('FindWindowA', ctypes.winapi_abi, ctypes.uint32_t, ctypes.jschar.ptr, ctypes.jschar.ptr)
var GetForegroundWindow = lib.declare('GetForegroundWindow', ctypes.winapi_abi, ctypes.uint32_t)
function doFindWindow() {
    var wm = Cc['@mozilla.org/appshell/window-mediator;1'].getService(Ci.nsIWindowMediator);
    var title = wm.getMostRecentWindow('navigator:browser').gBrowser.contentDocument.title;
    Cu.reportError('title=' + title)
    var ret = FindWindowA('', title + ' - Mozilla Firefox');
    //var ret = GetForegroundWindow();

    Cu.reportError(ret);
}
/*end foreground window stuff*/

Answer 4

The code in the answer of user 'paa' worked until Firefox version 69. If you execute it in Firefox 70 you will get an exception:

TypeError: win.QueryInterface is not a function

This is strange because the variable win has the same content in Firefox 69 and 70.

When I execute alert(win) I get: "[object ChromeWindow]" in both browsers.

And alert(win.document.title) displays correctly the title of the document in both browsers.

I downloaded the sourcecode of both Firefox versions to compare them and possibly find the cause. But the source code of Firefox is huge (2 Gigabyte) and nearly completely free of comments. I found that I'm wasting my time with that approach.

It is extremely difficult to understand sourcecode of Firefox which runs spread over multiple processes which communicate with each other. It seems that the content of the variable win corresponds to the C++ class mozIDOMWindowProxy or nsChromeOuterWindowProxy . But these seem to be only wrapper classes for other classes. Finally I gave up trying to understand Firefox sourcecode.

But playing around for some hours I finally found a solution by try and error.

It is even simpler:

var baseWindow = win.docShell
                    .treeOwner
                    .nsIBaseWindow; 

It works on Firefox 70 up to 79 (which is currently the latest version). However this new code does not run on Firefox versions <= 62. On Firefox 62 or older you get the error

TypeError: win.docShell is undefined

So the Firefoxes from 63 to 69 allow both versions of code. Maybe in version 70 the QueryInterface() has been removed because it is not needed anymore and has become legacy?

NOTE: In Firefox 68 they made another change. Now there are 2 native windows: The toplevel 'MozillaWindowClass' now has a child window 'MozillaCompositorWindowClass ' which runs in another process and draws the web content.