I am creating a tool on an application that opens some windows forms to get information from users, my tool should deal with these windows forms by itself without users' interactions.
I have initiated an event to get the opened form's process when it opens by the following code:
mgmtWtch = new ManagementEventWatcher("Select * From Win32_ProcessStartTrace");
mgmtWtch.EventArrived += WatchManagementEvent;
mgmtWtch.Start();
The shown window has OK button which i want to click, and i don't know how to make this action. while the parameter that i can get from this event is
EventArrivedEventArgs e
my question is how can i click the OK button through this event handler?
thanks in advance.
Have you looked at .Net's GUI automation API?
You'll want the UIAutomationClient and UIAutomationTypes assemblies.
I've used this API to drive installers, UIs during testing etc.
I found this link useful initially.
http://blogs.msdn.com/b/oldnewthing/archive/2013/04/08/10409196.aspx
e.g. assuming you have the parent window (i.e. the Form) for the button and you know the button's ID:
using System.Windows.Automation;
....
static AutomationElement FindById(AutomationElement root, string id, bool directChild)
{
Assert(root != null, "Invalid input: ParentWindow element 'root' is null.");
Condition conditions = new PropertyCondition(AutomationElement.AutomationIdProperty, id);
return root.FindFirst(directChild ? TreeScope.Children : TreeScope.Descendants, conditions);
}
....
AutomationElement button = FindById(containerWindow, id.ToString(), true);
InvokePattern invokePattern = null;
try
{
invokePattern = button.GetCurrentPattern(InvokePattern.Pattern) as InvokePattern;
}
catch (InvalidOperationException)
{
MessageBox.Show("The UI element named " + button.GetCurrentPropertyValue(AutomationElement.NameProperty) + " is not a button");
return false;
}
invokePattern.Invoke();
If you don't know the button's ID but do know it's name i.e. the text on the button then replace AutomationElement.AutomationIdProperty with AutomationElement.NameProperty in FindById (and rename the method appropriately)
Assuming the button is in the top-level Form window and you know the title displayed in this Form window, the following code will get the button's parent window:
bool ignoreCase = true; // or false if preferred
Condition conditions = new PropertyCondition(
AutomationElement.NameProperty,
windowTitle,
ignoreCase ? PropertyConditionFlags.IgnoreCase : PropertyConditionFlags.None
);
AutomationElement myForm =
AutomationElement.RootElement.FindFirst(
TreeScope.Children,
conditions );
The Window Title can be retrieved from the process you already have via the process' MainWindowTitle property.
Related
I would like to test to Outlook application and send an email to me by using White Stack framework. I implemented code who click on 'New Item' and after that the new window appears. I want to type a my mail to TextBox 'To' but I don't know how get to access to second window 'Untitled - Message (HTML)'. Photo
[TestMethod]
public void mail()
{
var application = Application.Launch(appPatch);
Thread.Sleep(2000);
var window = application.GetWindow(appTitle, InitializeOption.NoCache);
SearchCriteria searchCriteriaNewEmail = SearchCriteria.ByText("New Email");
Button buttonNewEmail = window.Get<Button>(searchCriteriaNewEmail);
buttonNewEmail.Click();
Thread.Sleep(1000);
string windowTitle = "Untitled‬ - Message(HTML)";
var window2 = application.GetWindow(windowTitle, InitializeOption.NoCache);
SearchCriteria searchCriteriaTo = SearchCriteria.ByText("To");
TextBox tbxTo = window2.Get<TextBox>(searchCriteriaTo);
tbxTo.BulkText = "mymail#gmail.com";
Thread.Sleep(2000);
window.Close();
}
The "second window" is in the same process id, so with that you can't simply use the .GetWindow() method.
For MDI applications you should use MdiChild(), or if it is a modal window, use .ModalWindow() on the parent window, or go back to the application then use Window() to get the window.
I would add that getting the window by it's title is not recommended, since that will change as you type the new title of the e-mail.
I'm trying to click a button on an external windows application. The following code successfully finds the element, brings the parent window into focus and then "manually" clicks the button
This works okay...
Process tProcess = Process.GetProcesses().FirstOrDefault(x => x.MainWindowTitle.StartsWith("MainWindowName"));
if (tProcess != null)
{
TestStack.White.Application application = TestStack.White.Application.Attach(tProcess.Id);
var tWindow = application.GetWindow(SearchCriteria.ByAutomationId("SubWindowName"), InitializeOption.NoCache);
SearchCriteria searchCriteria = SearchCriteria.ByAutomationId("btnCalibrate");
var calibrateBtn = tWindow.Get<TestStack.White.UIItems.Button>(searchCriteria);
tWindow.Focus();
var clickablePoint = calibrateBtn.AutomationElement.GetClickablePoint();
Mouse.Instance.Click(clickablePoint);
}
The problem with this is that Mouse.Instance.Click(clickablePoint); moves the cursor, ideally I don't want the cursor moved.
My initial code tried to click the button directly using the following
Process tProcess = Process.GetProcesses().FirstOrDefault(x => x.MainWindowTitle.StartsWith("MainWindowName"));
if (tProcess != null)
{
TestStack.White.Application application = TestStack.White.Application.Attach(tProcess.Id);
var tWindow = application.GetWindow(SearchCriteria.ByAutomationId("SubWindowName"), InitializeOption.NoCache);
SearchCriteria searchCriteria = SearchCriteria.ByAutomationId("btnCalibrate");
var calibrateBtn = tWindow.Get<TestStack.White.UIItems.Button>(searchCriteria);
tWindow.Focus();
calibrateBtn.Click();
}
but this gives the following error every time
TestStack.White.AutomationException
HResult=0x80131500
Message=Cannot perform action on Button. AutomationId:btnCalibrate, Name:Calibrate, ControlType:button, FrameworkId:WinForm,
Source=TestStack.White
StackTrace:
at TestStack.White.UIItems.UIItem.PerformIfValid(Action action) in c:\TeamCity\buildAgent\work\89a20b30302799e\src\TestStack.White\UIItems\UIItem.cs:line 254
at TestStack.White.UIItems.UIItem.Click() in c:\TeamCity\buildAgent\work\89a20b30302799e\src\TestStack.White\UIItems\UIItem.cs:line 231
at BetfairStreamingAPI.RadForm1.radLabelBetTime_Click(Object sender, EventArgs e) in D:
Does anyone know why the second method is throwing this error and if it's possible to fix this so that the button can be clicked without manually moving the cursor?
Edit: Screenshot of attempt to set togglestate
The solution to this particular problem appears to be use .RaiseClickEvent() instead of .Click()
The following code works
Process tProcess = Process.GetProcesses().FirstOrDefault(x => x.MainWindowTitle.StartsWith("MainWindowName"));
if (tProcess != null)
{
TestStack.White.Application application = TestStack.White.Application.Attach(tProcess.Id);
var tWindow = application.GetWindow(SearchCriteria.ByAutomationId("SubWindowName"), InitializeOption.NoCache);
SearchCriteria searchCriteria = SearchCriteria.ByAutomationId("btnCalibrate");
var calibrateBtn = tWindow.Get<TestStack.White.UIItems.Button>(searchCriteria);
calibrateBtn.RaiseClickEvent();
}
It's not entirely clear from the White docs when/why this is preferred. I found method RaiseClickEvent this on this link https://github.com/TestStack/White/commit/7b6d4dbc0008c3375e2ebf8810c55cb1abf91b60
EDIT2
I think you might have found something interesting. Since your button state is Indeterminate, it could be worth turning it on before clicking it:
calibrateBtn.State = ToggleState.On;
EDIT1
Alright, let's sort this out together.
There are only two reasons for that action to fail:
The button is not enabled, which I guess can't be the case
The button is OffScreen
If you do something like
Console.WriteLine(calibrateBtn.IsOffScreen.ToString());
You should see
true
If so, try this before you click it:
var pattern = calibrateBtn.AutomationElement.GetCurrentPattern(System.Windows.Automation.InvokePattern.Pattern);
(pattern as System.Windows.Automation.InvokePattern).Invoke();
Thanks in advance.
Is it possible to activate a tab in another program using an IntPtr? If so, how?
SendKeys is not an option.
Perhaps what I need is a fishing lesson. I have exhausted Google and my lead developer.
I would appreciate an outright solution OR a recommendation to continue my Google efforts.
basic process is:
I drag a shortcut icon to the launcher
This opens the target application (Notepad++) and grabs IntPtr, etc.
I would like to programmatically select various items in Notepad++ such as Edit, menu items under Edit, or a doc tab.
The basic code I am running is:
the 'blob'
item 1: IntPtr of item
item 2: IntPtr of itemsChild
item 3: control text of item 1
item 4: is rectangle parameters of item 1
root contains similar info:
As others pointed out, the standard way of doing this is to use UI Automation. Notepad++ does support UI Automation (to some extent, as it's somehow automatically provided by the UI Automation Windows layers).
Here is a sample C# console app that demonstrates the following sceanrio (you need to reference UIAutomationClient.dll, UIAutomationProvider.dll and UIAutomationTypes.dll):
1) get the first running notepad++ process (you must start at least one)
2) open two files (note there may be already other opened tabs in notepad++)
3) selects all tabs in an infinite loop
class Program
{
static void Main(string[] args)
{
// this presumes notepad++ has been started somehow
Process process = Process.GetProcessesByName("notepad++").FirstOrDefault();
if (process == null)
{
Console.WriteLine("Cannot find any notepad++ process.");
return;
}
AutomateNpp(process.MainWindowHandle);
}
static void AutomateNpp(IntPtr handle)
{
// get main window handle
AutomationElement window = AutomationElement.FromHandle(handle);
// display the title
Console.WriteLine("Title: " + window.Current.Name);
// open two arbitrary files (change this!)
OpenFile(window, #"d:\my path\file1.txt");
OpenFile(window, #"d:\my path\file2.txt");
// selects all tabs in sequence for demo purposes
// note the user can interact with n++ (for example close tabs) while all this is working
while (true)
{
var tabs = GetTabsNames(window);
if (tabs.Count == 0)
{
Console.WriteLine("notepad++ process seems to have gone.");
return;
}
for (int i = 0; i < tabs.Count; i++)
{
Console.WriteLine("Selecting tab:" + tabs[i]);
SelectTab(window, tabs[i]);
Thread.Sleep(1000);
}
}
}
static IList<string> GetTabsNames(AutomationElement window)
{
List<string> list = new List<string>();
// get tab bar
var tab = window.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Tab));
if (tab != null)
{
foreach (var item in tab.FindAll(TreeScope.Children, PropertyCondition.TrueCondition).OfType<AutomationElement>())
{
list.Add(item.Current.Name);
}
}
return list;
}
static void SelectTab(AutomationElement window, string name)
{
// get tab bar
var tab = window.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Tab));
// get tab
var item = tab.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.NameProperty, name));
if (item == null)
{
Console.WriteLine("Tab item '" + name + "' has been closed.");
return;
}
// select it
((SelectionItemPattern)item.GetCurrentPattern(SelectionItemPattern.Pattern)).Select();
}
static void OpenFile(AutomationElement window, string filePath)
{
// get menu bar
var menu = window.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.MenuBar));
// get the "file" menu
var fileMenu = menu.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.NameProperty, "File"));
// open it
SafeExpand(fileMenu);
// get the new File menu that appears (this is quite specific to n++)
var subFileMenu = fileMenu.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Menu));
// get the "open" menu
var openMenu = subFileMenu.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.NameProperty, "Open..."));
// click it
((InvokePattern)openMenu.GetCurrentPattern(InvokePattern.Pattern)).Invoke();
// get the new Open dialog (from root)
var openDialog = WaitForDialog(window);
// get the combobox
var cb = openDialog.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.ComboBox));
// fill the filename
((ValuePattern)cb.GetCurrentPattern(ValuePattern.Pattern)).SetValue(filePath);
// get the open button
var openButton = openDialog.FindFirst(TreeScope.Children, new AndCondition(
new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Button),
new PropertyCondition(AutomationElement.NameProperty, "Open")));
// press it
((InvokePattern)openButton.GetCurrentPattern(InvokePattern.Pattern)).Invoke();
}
static AutomationElement WaitForDialog(AutomationElement element)
{
// note: this should be improved for error checking (timeouts, etc.)
while(true)
{
var openDialog = element.FindFirst(TreeScope.Children, new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Window));
if (openDialog != null)
return openDialog;
}
}
static void SafeExpand(AutomationElement element)
{
// for some reason, menus in np++ behave badly
while (true)
{
try
{
((ExpandCollapsePattern)element.GetCurrentPattern(ExpandCollapsePattern.Pattern)).Expand();
return;
}
catch
{
}
}
}
}
If you wonder how this has been made, then you must read about UI Automation. The mother of all tools is called Inspect: https://msdn.microsoft.com/library/windows/desktop/dd318521.aspx
Make sure you get version at least 7.2.0.0. Note there is another one called UISpy but inspect is better.
Note, unfortunately, notepad++ tab text content - because it's based on the custom scintilla editor control - does not properly supports automation (we can't read from it easily, I suppose we'd have to use scintilla Windows messages for this), but it could be added to it (hey, scintilla guys, if you read this ... :).
In addition to the answer from Garath, you might also want to investigate the Windows automation API's i.e. the technology used to implement coded UI tests for GUI applications. As part of regular functional testing, I routinely control an external application from a set of NUnit tests using these API's.
Tools like UIAVerify will give you an indication of what controls are available in the application and you can use the Invoke Pattern (and many others) to interact with the controls at run-time.
If you want a detailed example of how to use the automation API's, the open source TestStack White project is pretty handy.
It is almost not possible if SendKeys is not an option but read more
Now more important part of the question- why:
We have to look how win32 application works: it has a WndProc/WindowProc method which is resposible for processing "events" form the UI.
So every event in the windows application must go through above method. SendKeys method is a special of SendMessage (MSDN), so you can use SendMessage to control other exe than your.
Simple code could look like:
IntPtr hwnd = FindWindow("Notepad++", null);
SendMessageA(hwnd, WM_COMMAND, SOMETHING1, SOMETHING2);
There is already on StackOverflow example how to do that with chrome: C# - Sending messages to Google Chrome from C# application , but this is only a start. You will have to find out what exactly message you want to send.
In exactly situation which you described I will try to send WM_MOUSE and WM_KEYBORD events to Notepad++ events, but it is only an idea :)
I have a C# Form application that has a TabControl in the main form. This TabControl is used to display multilple TabPages that contain a CustomControl. This CustomControl is just a Panel with a few buttons and a PictureBox.
Here is a picture of my app when it starts up. As you can see the tab control (the white area) is empty:
If the user clicks the "Add Image" button they are presented with an OpenFileDialog to select the image then the addImage method is called with the selected file:
private void doAddImage()
{
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.Filter = Constants.Global.IMAGE_FILE_FILTER();
if (openFileDialog.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
string imageFileName = openFileDialog.FileName;
addImage(imageFileName);
}
}
private void addImage(string imageFileName)
{
// Create a new bitmap and image preview custom control. Then create a new tab
// page and add the custom control to the tab page.
Bitmap bitMap = new Bitmap(imageFileName);
ImagePreviewCustomControl previewControl = new ImagePreviewCustomControl(bitMap);
previewControl.Dock = DockStyle.Fill;
TabPage tabPage = new TabPage(Path.GetFileNameWithoutExtension(imageFileName));
tabPage.Controls.Add(previewControl);
// Insert the new tab page to the right of the currently selected tab page
int selectedTabIndex = imagesTabControl.SelectedIndex;
imagesTabControl.TabPages.Insert(selectedTabIndex + 1, tabPage);
imagesTabControl.SelectedIndex = selectedTabIndex + 1;
}
As you can see, in the addImage method I create the Bitmap, CustomControl, and TabPage and then insert it into the TabControl.
I start my application, click the "Add Image" button, everything works just fine.
Here is a picture with a tab page added:
While I am testing my app I don't want to have to add an image manually using the OpenFileDialog every time so in my constructor I just call addImage with some fixed image file name that I want to test with:
public ImageViewerApp()
{
InitializeComponent();
addImage(#"C:\MyImages\Calculator-3-icon.png");
}
The problem I am having is that when I try to add the image in my constructor it doesn't show up in the TabControl. The application starts up blank (like the first picture).
As stated above when the application is already running and I click the "Add Image" button it gets added just fine.
I found a property in the TabControl class called Created which states:
"Gets a value indicating whether the control has been created"
So to try and figure out what's going on I write the value of Created to the console just before I call addImage in the constructor. (I have a custom console for debugging my Form applications.)
public ImageViewerApp()
{
InitializeComponent();
TestConsole.WriteLine(imagesTabControl.Created);
addImage(#"D:\Development\Work\Other\Stardock\Start8\_downloaded\Calculator-3-icon.png");
}
The value of Created just before the call to addImage in the constructor is:
False
I put another console output inside the addImage method:
private void doAddImage()
{
TestConsole.WriteLine(imagesTabControl.Created);
OpenFileDialog openFileDialog = new OpenFileDialog();
...
...
}
The value of Created after the app has started and the user presses the "Add Image" button is:
True
Why is it that the TabControl is not Created inside my constructor (even after the InitializeComponent() call) and the once the application is running it is Created?
=UPDATE========================================================================
Based on the suggestion by Hans Passant I have added the following code to my addImage method:
int selectedTabIndex = -1;
if (imagesTabControl.TabCount > 0)
{
selectedTabIndex = imagesTabControl.SelectedIndex;
}
else
{
selectedTabIndex = imagesTabControl.SelectedIndex + 1;
}
imagesTabControl.TabPages.Insert(selectedTabIndex, tabPage);
imagesTabControl.SelectedIndex = selectedTabIndex;
This doesn't work.
===============================================================================
=UPDATE2=======================================================================
int selectedTabIndex = imagesTabControl.SelectedIndex;
if (imagesTabControl.TabCount == 0) selectedTabIndex = -1;
imagesTabControl.TabPages.Insert(selectedTabIndex, tabPage);
imagesTabControl.SelectedIndex = selectedTabIndex;
This causes the following Exception:
{"InvalidArgument=Value of '-1' is not valid for 'index'.\r\nParameter name: index"}
===============================================================================
=UPDATE3=======================================================================
I tried the folllowing code:
int selectedTabIndex = imagesTabControl.SelectedIndex;
if (imagesTabControl.TabCount == 0) selectedTabIndex = -1;
imagesTabControl.TabPages.Insert(selectedTabIndex + 1, tabPage);
imagesTabControl.SelectedIndex = selectedTabIndex + 1;
This one doesn't throw an exception but again no tab page added after calling
addImage in the constructor.
===============================================================================
=UPDATE4=======================================================================
I have kindof given up on adding an image in the constructor. So instead I am using an enum RunMode and a variable RUN_MODE of that type. Then, if RUN_MODE == RunMode.TESTI call a method to add a random image when I click the button. (The OpenFileDialog is not used. I just parse through all the image files in the fixed directory IMAGE_DIRECTORY.
enum RunMode { NORMAL, TEST }
private static string IMAGE_DIRECTORY = #"D:\\Work\Images";
...
...
private void doAddImage()
{
if (RUN_MODE == RunMode.TEST)
{
addRandomImage();
return;
}
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.Filter = Constants.Global.IMAGE_FILE_FILTER();
if (openFileDialog.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
string imageFileName = openFileDialog.FileName;
addImage(imageFileName);
}
}
private void addRandomImage()
{
string[] allFiles = Directory.GetFiles(IMAGE_DIRECTORY);
List<string> imageFileNames = new List<string>();
foreach (string file in allFiles)
{
bool isImageFile = Constants.Global.IMAGE_EXTENSIONS.Contains(Path.GetExtension(file));
if (isImageFile)
{
imageFileNames.Add(file);
}
}
int randomIndex = new Random().Next(imageFileNames.Count);
addImage(imageFileNames.ElementAt(randomIndex));
}
This works. Now when I click the "Add Image" button during TEST_MODE I skip the
OpenFileDialog and just add a random image.
I would like to understand the issues with TabControl but at this point I just
need to continue development. My current solution works great.
As I person who like to understand everything I would like to use other people's
suggestions so I will keep monitoring this question for a solution.
===============================================================================
JonP's answer gave me the idea to just wait for the window handle to be created before inserting the tab, so I tried some events occuring between Form construction and Tab Control display.
I found it to work with both the Load or Shown events:
Right-click on the Form (the root, not child controls) in the Designer view > Properties > Events (flash icon) > Behavior > enter a method name for the Load or Shown event and confirm. To generate a Load event callback you can also double-click on the Form itself. This should generate something like this:
this.Load += new System.EventHandler(this.Form1_Load);
// or
this.Shown += new System.EventHandler(this.Form1_Shown);
Setup the tabs in the callback:
private void Form1_Load(object sender, EventArgs e)
{
// Add image (this will call imagesTabControl.TabPages.Insert(selectedTabIndex + 1, tabPage))
// This must be done on Load event because Insert requires
// the window handle, which is not ready in the constructor
addImage(#"path_to_image.png");
}
I have had this problem too and have found a workaround; I think it must be a bug with Insert():
Don't use Insert(), it usually does nothing, use Add() instead; this reliably adds a TabPage to the end of the collection.
After adding it swap it with the tab position where you actually want it.
imagesTabControl.TabPages.Add(tabPage);
// Now swap the two tabs:
imagesTabControl.TabPages[imagesTabControl.TabCount - 1] = imagesTabControl.TabPages[selectedTabIndex + 1];
imagesTabControl.TabPage[selectedTabIndex + 1] = tabPage;
Your mileage may vary of course :-)
Stop Press! An even better fix is to read the class's Handle member before calling Insert():
var handle = imagesTabControl.Handle;
Insert() works perfectly after you do that. Obvious isn't it???? The help page for Handle has this possible relevant Remark showing that the object actually does something when you read Handle:
The value of the Handle property is a Windows HWND. If the handle has not yet been created, referencing this property will force the handle to be created.
You could remove the TabControl from the designer and then instead just manually create the TabControl programmatically and add it to the Form immediately after InitializeComponent(). Then after you create the TabControl, call addImage(). Something like:
InitializeComponent();
TabControl tc = new TabControl();
tc.Location = new Point(10, 10);
tc.Size = new Size(100, 100);
tc.Visible = true;
tc.Anchor = (AnchorStyles.Bottom | AnchorStyles.Right | AnchorStyles.Left | AnchorStyles.Top);
this.Controls.Add(tc)
addImage("c:\pathToImage\image.bmp");
I'm trying to test file download with Watin 2.1.0 against IE9. I used the suggested code from the accepted answer to the question Downloading a file with Watin in IE9, like this:
var downloadHandler = new FileDownloadHandler(fname);
WebBrowser.Current.AddDialogHandler(downloadHandler);
link.ClickNoWait();
downloadHandler.WaitUntilFileDownloadDialogIsHandled(15);
downloadHandler.WaitUntilDownloadCompleted(200);
However, the downloadHandler.WaitUntilFileDownloadDialogIsHandled(15) call times out. What should I do?
File download dialog doesn't work in IE9 (Windows7) NetFramework 4.0.
Following code snippet might help you resolve the issue:
First you must add references UIAutomationClient and UIAutomationTypes to your test project.
After In Ie9 Tools -> View Downloads -> Options define path to your save folder.
The next method extends Browser class
public static void DownloadIEFile(this Browser browser)
{
// see information here (http://msdn.microsoft.com/en-us/library/windows/desktop/ms633515(v=vs.85).aspx)
Window windowMain = new Window(WatiN.Core.Native.Windows.NativeMethods.GetWindow(browser.hWnd, 5));
System.Windows.Automation.TreeWalker trw = new System.Windows.Automation.TreeWalker(System.Windows.Automation.Condition.TrueCondition);
System.Windows.Automation.AutomationElement mainWindow = trw.GetParent(System.Windows.Automation.AutomationElement.FromHandle(browser.hWnd));
Window windowDialog = new Window(WatiN.Core.Native.Windows.NativeMethods.GetWindow(windowMain.Hwnd, 5));
// if doesn't work try to increase sleep interval or write your own waitUntill method
Thread.Sleep(1000);
windowDialog.SetActivate();
System.Windows.Automation.AutomationElementCollection amc = System.Windows.Automation.AutomationElement.FromHandle(windowDialog.Hwnd).FindAll(System.Windows.Automation.TreeScope.Children, System.Windows.Automation.Condition.TrueCondition);
foreach (System.Windows.Automation.AutomationElement element in amc)
{
// You can use "Save ", "Open", ''Cancel', or "Close" to find necessary button Or write your own enum
if (element.Current.Name.Equals("Save"))
{
// if doesn't work try to increase sleep interval or write your own waitUntil method
// WaitUntilButtonExsist(element,100);
Thread.Sleep(1000);
System.Windows.Automation.AutomationPattern[] pats = element.GetSupportedPatterns();
// replace this foreach if you need 'Save as' with code bellow
foreach (System.Windows.Automation.AutomationPattern pat in pats)
{
// '10000' button click event id
if (pat.Id == 10000)
{
System.Windows.Automation.InvokePattern click = (System.Windows.Automation.InvokePattern)element.GetCurrentPattern(pat);
click.Invoke();
}
}
}
}
}
if you want click 'Save As' replace foreach code with this
System.Windows.Automation.AutomationElementCollection bmc = element.FindAll(System.Windows.Automation.TreeScope.Children, System.Windows.Automation.Automation.ControlViewCondition);
System.Windows.Automation.InvokePattern click1 = (System.Windows.Automation.InvokePattern)bmc[0].GetCurrentPattern(System.Windows.Automation.AutomationPattern.LookupById(10000));
click1.Invoke();
Thread.Sleep(10000);
System.Windows.Automation.AutomationElementCollection main = mainWindow.FindAll(System.Windows.Automation.TreeScope.Children
,System.Windows.Automation.Condition.TrueCondition);
foreach (System.Windows.Automation.AutomationElement el in main)
{
if (el.Current.LocalizedControlType == "menu")
{
// first array element 'Save', second array element 'Save as', third second array element 'Save and open'
System.Windows.Automation.InvokePattern clickMenu = (System.Windows.Automation.InvokePattern)
el.FindAll(System.Windows.Automation.TreeScope.Children, System.Windows.Automation.Condition.TrueCondition) [1].GetCurrentPattern(System.Windows.Automation.AutomationPattern.LookupById(10000));
clickMenu.Invoke();
//add ControlSaveDialog(mainWindow, filename) here if needed
break;
}
}
Edit:
Also if you need to automate the save as dialog specifying a path and clicking save you can do it by adding this code just before break;
private static void ControlSaveDialog(System.Windows.Automation.AutomationElement mainWindow, string path)
{
//obtain the save as dialog
var saveAsDialog = mainWindow
.FindFirst(TreeScope.Descendants,
new PropertyCondition(AutomationElement.NameProperty, "Save As"));
//get the file name box
var saveAsText = saveAsDialog
.FindFirst(TreeScope.Descendants,
new AndCondition(
new PropertyCondition(AutomationElement.NameProperty, "File name:"),
new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Edit)))
.GetCurrentPattern(ValuePattern.Pattern) as ValuePattern;
//fill the filename box
saveAsText.SetValue(path);
Thread.Sleep(1000);
//find the save button
var saveButton =
saveAsDialog.FindFirst(TreeScope.Descendants,
new AndCondition(
new PropertyCondition(AutomationElement.NameProperty, "Save"),
new PropertyCondition(AutomationElement.ControlTypeProperty, ControlType.Button)));
//invoke the button
var pattern = saveButton.GetCurrentPattern(InvokePattern.Pattern) as InvokePattern;
pattern.Invoke();
}
IE9 no longer uses a dialog window for saving files. Instead, it uses the notification bar to prevent focus from being removed from the web site. See http://msdn.microsoft.com/en-us/ie/ff959805.aspx under "Download Manager" for reference.
Unfortunately, this means that the current FileDownloadHandler in WatiN will not work. It instantiates a "DialogWatcher" class per browser instance that is a basic message pump for any kind of child window. When child windows are encountered, the DialogWatcher checks to see if the window is specifically a dialog (which the notification bar is not). If it is a dialog, it then iterates over the registered IDialogHandler instances calling "CanHandleDialog." Even if the notification bar were a dialog, it is of a different Window Style (http://msdn.microsoft.com/en-us/library/windows/desktop/ms632600(v=vs.85).aspx), which is how WatiN detects the type of dialog.
From what I can see, there is no support yet for detecting the IE 9 notification bar and its prompts in WatiN. Until that support is added, you will not be able to automate downloading files in IE9.