Friday, June 29, 2007

Saturday, June 16, 2007

Copying Files from the Device to the Desktop using .NET

Recently, I’ve been working on a tool that creates a backup of a SQL Server Compact Edition database on the device to the desktop. To accomplish this, I used the Remote API (RAPI). Unfortunately, the Remote API is not yet available in managed code. In this article I would like to demonstrate how to P/Invoke methods from the Remote API for copying files from the device to the desktop using managed code.

First, we’ll need some P/Invokes to rapi.dll

[DllImport("rapi.dll")]
static extern int CeRapiInit();

[DllImport("rapi.dll")]
static extern int CeRapiUninit();

[DllImport("rapi.dll")]
static extern int CeCloseHandle(IntPtr hObject);

[DllImport("rapi.dll", CharSet = CharSet.Unicode)]
static extern IntPtr CeCreateFile(
  string lpFileName,
  uint dwDesiredAccess,
  int dwShareMode,
  int lpSecurityAttributes,
  int dwCreationDisposition,
  int dwFlagsAndAttributes,
  int hTemplateFile);

[DllImport("rapi.dll", CharSet = CharSet.Unicode)]
static extern int CeReadFile(
  IntPtr hFile,
  byte[] lpBuffer,
  int nNumberOfbytesToRead,
  ref int lpNumberOfbytesRead,
  int lpOverlapped);

const int ERROR_SUCCESS = 0;
const int OPEN_EXISTING = 3;
const int INVALID_HANDLE_VALUE = -1;
const int FILE_ATTRIBUTE_NORMAL = 0x80;
const uint GENERIC_READ = 0x80000000;


Now let’s create a method called CopyFromDevice(string remote_file, string local_file). The remote_file parameter is the source file on the device that you wish to copy. The local_file parameter is the destination filename on the desktop.

public static void CopyFromDevice(string remote_file, string local_file)
{
  bool rapi = CeRapiInit() == ERROR_SUCCESS;
  if (!rapi) {
    return;
  }

  IntPtr remote_file_ptr = CeCreateFile(
    remote_file,
    GENERIC_READ,
    0,
    0,
    OPEN_EXISTING,
    FILE_ATTRIBUTE_NORMAL,
    0);

  if (remote_file_ptr.ToInt32() == INVALID_HANDLE_VALUE) {
    return;
  }

  FileStream local_file_stream = new FileStream(
    local_file,
    FileMode.Create,
    FileAccess.Write);

  int read = 0;
  int size = 1024 * 4;
  byte[] data = new byte[size];

  CeReadFile(remote_file_ptr, data, size, ref read, 0);

  while (read > 0) {
    local_file_stream.Write(data, 0, read);
    if (CeReadFile(remote_file_ptr, data, size, ref read, 0) == 0) {
      CeCloseHandle(remote_file_ptr);
      local_file_stream.Close();
      return;
    }
  }

  CeCloseHandle(remote_file_ptr);
  local_file_stream.Flush();
  local_file_stream.Close();

  if (rapi) {
    CeRapiUninit();
  }

  if (!File.Exists(local_file)) {
    throw new FileNotFoundException("The file was not copied to the desktop");
  }
}

To use the code above you will have to know the full path of the file on the device. The way I did it was to read the registry on the device and check if my application was installed, if it was then I get the path of the application and pass as the path in my remote_file parameter.

Tuesday, June 12, 2007

Retrieving the Icon Image within the System Image List in .NETCF 2.0

Here's a nice trick for retrieving the icon image of a file or folder from the system image list. All we actually need is to P/Invoke SHGetFileInfo and use Icon.FromHandle() to get the Icon.

First, we need to declare our P/Invokes.

[StructLayout(LayoutKind.Sequential)]
struct SHFILEINFO
{
  public IntPtr hIcon;
  public IntPtr iIcon;
  public uint dwAttributes;
  [MarshalAs(UnmanagedType.ByValTStr, SizeConst = 260)]
  public string szDisplayName;
  [MarshalAs(UnmanagedType.ByValTStr, SizeConst = 80)]
  public string szTypeName;
}

const uint SHGFI_ICON = 0x000000100;
const uint SHGFI_LARGEICON = 0x000000000;
const uint SHGFI_SMALLICON = 0x000000001;
const uint SHGFI_SELECTICON = 0x000040000;

[DllImport("coredll.dll")]
static extern IntPtr SHGetFileInfo(string pszPath, uint dwFileAttributes,
  ref SHFILEINFO psfi, uint cbSizeFileInfo, uint uFlags);


To get an instance of System.Drawing.Icon for the small icon of a file

Icon GetSystemIconSmall(string file)
{
  SHFILEINFO shinfo = new SHFILEINFO();
  IntPtr i = SHGetFileInfo(file, 0, ref shinfo,
    (uint)Marshal.SizeOf(shinfo), SHGFI_ICON | SHGFI_SMALLICON);

  return Icon.FromHandle(shinfo.hIcon);
}


For the large icon of a file

Icon GetSystemIconLarge(string file)
{
  SHFILEINFO shinfo = new SHFILEINFO();
  IntPtr i = SHGetFileInfo(file, 0, ref shinfo,
    (uint)Marshal.SizeOf(shinfo), SHGFI_ICON | SHGFI_LARGEICON);

  return Icon.FromHandle(shinfo.hIcon);
}


For the small icon of a file when it is selected

Icon GetSystemIconSmallSelected(string file)
{
  SHFILEINFO shinfo = new SHFILEINFO();
  IntPtr i = SHGetFileInfo(file, 0, ref shinfo,
    (uint)Marshal.SizeOf(shinfo), SHGFI_ICON | SHGFI_SMALLICON | SHGFI_SELECTICON);

  return Icon.FromHandle(shinfo.hIcon);
}


And last for the large icon of a file when it is selected

Icon GetSystemIconLargeSelected(string file)
{
  SHFILEINFO shinfo = new SHFILEINFO();
  IntPtr i = SHGetFileInfo(file, 0, ref shinfo,
    (uint)Marshal.SizeOf(shinfo), SHGFI_ICON | SHGFI_LARGEICON | SHGFI_SELECTICON);

  return Icon.FromHandle(shinfo.hIcon);
}

Ok, now how is this helpful? Well if you want to implement a File Explorer-ish control, then wouldn't have to include Icons and other images in your application. You can just use the icons in the system image list

Saturday, June 9, 2007

Logging Unhandled Exceptions

Bugs are sometimes unavoidable. They're best caught during the development or testing phase. There might be some cases where the developer forgot to handle possible exceptions in a function. It could be possible that this exception isn't handled anywhere at all. But even so, it is still possible to catch this exception. To do this we handle the UnhandledException event of the current AppDomain. We should do this in our static void Main() before calling Application.Run([Main Form])

Here's a small snippet to accomplish this task.

[C# CODE]

static void Main()
{
  AppDomain.CurrentDomain.UnhandledException +=
    new UnhandledExceptionEventHandler(CurrentDomain_UnhandledException);
  Application.Run(new MainForm());
}

static void CurrentDomain_UnhandledException(object sender, UnhandledExceptionEventArgs e)
{
  Exception exception = (Exception)e.ExceptionObject;
  if (exception != null) {
    Error.Append(exception.Message, exception.StackTrace);
  }
}

Now we need a mechanism for saving to a Error log file. Let's create a simple class called Error() and add a function called Append(string message, string stacktrace)

[C# CODE]

public class Error
{
  internal static void Append(string message, string stacktrace)
  {
    string file = string.Format("{0}\\Errors.txt",
      Path.GetDirectoryName(Assembly.GetExecutingAssembly().GetName().CodeBase));
    if (File.Exists(file)) {
      FileInfo fi = new FileInfo(file);
      if (fi.Length > 100 * 1024) {
        fi.Delete();
      }
    }
    StreamWriter sw = new StreamWriter(file, true, Encoding.UTF8);
    sw.WriteLine(string.Format(
      "-=-=-=-=-=--=-=-\n{0}\nMESSAGE:\n{1}\nSTACK TRACE:\n{3}\n",
        DateTime.Now, message, stacktrace));
    sw.Close();
    }
  }
}

The Error() class should be put in a namespace that is accessible throughout the application. This will be very helpful tool for finding those nasty almost impossible to reproduce bugs.

Thursday, June 7, 2007

Integrating with TomTom Navigator

PDA's are used for pretty much everything these days. From the bunch of devices the I work (play) with, I took a great liking to devices that have built in GPS receivers. These devices are usually bundled with really cool navigation software from various vendors. Some of these navigation software have SDK's that you can buy separately. By using these SDK's, you can fully integrate navigation features to your mobile solutions.

In this article, I would like to discuss how to integrate a .NET Compact Framework application with TomTom Navigator. I will also demonstrate an example of making a generic navigator wrapper so your application is not just bound to one kind of navigation software.

Before we get started, we need to have the TomTom Navigator SDK. Unfortunately this is not free, but can be easily purchased from the TomTom Pro website.

Before we dig into more detail, let's go through our software requirements. We need the following:

    1. Visual Studio 2005
    2. Windows Mobile 5.0 SDK for Pocket PC
    3. A device running Windows Mobile 5.0 with TomTom Navigator 5 installed
    4. The TomTom Navigator SDK
    5. ActiveSync 4.2 or higher (for Vista, the Mobile Device Center)


Now, Lets get started...


Here is what we need to make:

    1. native wrapper for the TomTom SDK (native dll)
    2. generic navigator wrapper in .NET CF
    3. managed TomTom wrapper
    4. device application that will call TomTom SDK wrapper methods

Sounds pretty simple doesn't it?


1. Native Wrapper for the TomTom SDK

We will first need a little help from native code to access the TomTom SDK. We cannot access the TomTom SDK directly from .NET due to the architecture of the SDK. We have to wrap around the TomTom SDK C++ classes and methods and expose them as C type functions.

In your native wrapper, lets say we want to wrap the following TomTom SDK functions:
  - GetApplicationVersion(TError* err, TVersion* ver)
  - FlashMessage(TError* err, char* msg, int ms)
  - NavigateToAddress(TError* aError, char* aCity, char* aStreet, char* aHouseNr, char* aPostcode)


[C++ CODE]

#include "sdkconstants.h"
#include "TomTomAPI.h"
#include "TomTomGoFileLayer.h"

#define CLIENT_NAME "client"

CTomTomAPI::TError err;
int res = 0;

BOOL APIENTRY DllMain(
  HANDLE hModule, DWORD ul_reason_for_call, LPVOID lpReserved )
{
  switch (ul_reason_for_call)
  {
  case DLL_PROCESS_ATTACH:
  case DLL_THREAD_ATTACH:
  case DLL_THREAD_DETACH:
  case DLL_PROCESS_DETACH:
    break;
  }
  return TRUE;
}

extern "C" __declspec(dllexport) int TTN_GetApplicationVersion(
  int* iError, LPTSTR szVersion, int *iBuildNumber )
{
  MTomTomCommunicationLayerInterface *comms =
    DEFAULT_TRANSPORTATION_LAYER(CLIENT_NAME,2005,TOMTOM_TCPIP_PORT);

  CTomTomAPI api(*comms);
  CTomTomAPI::TVersion version;
  res = api.GetApplicationVersion(&err, &version);
  *iError = err.iError;

  TCHAR str[16];
  _stprintf(str, TEXT("%S"), version.iVersion);
  lstrcpy( szVersion, (LPTSTR)str );
  *iBuildNumber = version.iBuildNumber;

  delete comms;
  return res;
}

extern "C" __declspec(dllexport) int TTN_FlashMessage(
   int* iError, char* aMessage, int aMilliSeconds )
{
  char message[256];
  sprintf(message, "%S", aMessage);

  MTomTomCommunicationLayerInterface *comms =
    DEFAULT_TRANSPORTATION_LAYER(CLIENT_NAME,2005,TOMTOM_TCPIP_PORT);

  CTomTomAPI api(*comms);
  res = api.FlashMessage(&err, message, aMilliSeconds);
  *iError = err.iError;

  delete comms;
  return res;
}

extern "C" __declspec(dllexport) int TTN_NavigateToAddress(
   int* iError, char* aCity, char* aStreet, char* aHouseNr, char* aPostcode )
{
  char city[256];
  char street[256];
  char houseNr[16];
  char postcode[32];

  sprintf(city, "%S", aCity);
  sprintf(street, "%S", aStreet);
  sprintf(houseNr, "%S", aHouseNr);
  sprintf(postcode, "%S", aPostcode);

  MTomTomCommunicationLayerInterface *comms =
    DEFAULT_TRANSPORTATION_LAYER(CLIENT_NAME,2005,TOMTOM_TCPIP_PORT);

  CTomTomAPI api(*comms);
  res = api.NavigateToAddress(&err, city, street, houseNr, postcode);
  *iError = err.iError;

  delete comms;
  return res;
}

Let's set the output of the project to be called TTSDK.dll

2. Generic Navigator Wrapper in .NET CF

Once we've gotten our native wrapper up and running, we create a generic navigator wrapper. We start off by creating a smart device class library project. Once the project is created, add the following classes: INavigator.cs, Navigator.cs, and Common.cs

Lets go and define the common objects we want to use in Common.cs

[C# CODE]

public struct NVersion {
  string Version;
  int BuildNumber;
}

INavigator.cs will be an interface defining the how the wrapper will look like. Lets add methods for the 3 TomTom SDK methods we want to use.

[C# CODE]

public interface INavigator
{
  NVersion GetApplicationVersion();
  void FlashMessage(string text, int duration);
  void NavigateToAddress(string city, string street, string houseno, string zipcode);
}

Navigator.cs will be the class your application will call. This will load the managed TomTom wrapper as an instance of INavigator. Navigator itself will implement INavigator and will return calls from the TomTom wrapper.

[C# CODE]

public class Navigator : INavigator
{
  private INavigator instance;

  public Navigator(string typeName)
  {
    Type type = Type.GetType(typeName);
    if (type == null) {
      throw new TypeLoadException();
    } else {
      instance = (INavigator)Activator.CreateInstance(type);
      if (instance == null) {
        throw new TypeLoadException();
      }
    }
  }

  public TVersion GetApplicationVersion()
  {
    return instance.GetApplicationVersion();
  }

  public void FlashMessage(string text, int duration)
  {
    instance.FlashMessage(text, duration);
  }

  public void NavigateToAddress(string city, string street, string houseno, string zipcode)
  {
    instance.NavigateToAddress(city, street, houseno, zipcode);
  }
}

The default constructor for Navigator accepts a type name. The format for type name is "[Namespace].[ClassName], [AssemblyName]"

3. Managed TomTom Wrapper

Here we create a new smart device class library. Once the project is created, add a reference to the generic navigator wrapper since we will implement the INavigator interface and add a class called TomTom.cs

Lets implement TomTom.cs as INavigator

[C# CODE]

[DllImport("TTSDK.dll", EntryPoint="TTN_GetApplicationVersion")]
internal static extern int TTN_GetApplicationVersion(
  ref int iError, StringBuilder szVersion, ref int iBuildNumber);

[DllImport("TTSDK.dll", EntryPoint="TTN_FlashMessage")]
internal static extern int TTN_FlashMessage(
  ref int iError, string aMessage, int aMilliseconds);

[DllImport("TTSDK.dll", EntryPoint="TTN_NavigateToAddress")]
internal static extern int TTN_NavigateToAddress(
  ref int iError, string aCity, string aStreet, string aHouseNo, string aPostcode);

public void FlashMessage(string aMessage, int aMilliseconds)
{
  if (0 != TTN_FlashMessage(ref iError, aMessage, aMilliseconds)) {
    throw new InvalidOperationException();
  }
}

public NVersion GetApplicationVersion()
{
  NVersion version = new TVersion();
  StringBuilder szVersion = new StringBuilder();
  int iBuildNumber = 0;
  int iError = 0;

  if (0 != TTN_GetApplicationVersion(ref iError, szVersion, ref iBuildNumber)) {
    throw new InvalidOperationException();
  } else {
    version.iVersion = szVersion.ToString();
    version.iBuildNumber = iBuildNumber;
  }
}

public void NavigateToAddress(string sCity, string sStreet, string sHouseNo, string sPostcode)
{
  int iError = 0;

  if (0 != TTN_NavigateToAddress(ref iError, sCity, sStreet, sHouseNo, sPostcode)) {
    throw new InvalidOperationException();
  }
}

Now our TomTom wrapper is pretty much ready

4. Device Application

In our application, we want to integrate with TomTom Navigator for navigating to a specific address. The address can could be retrieved from a web service, or stored in your pocket outlook. For this article, we're going to retrieve address information of a customer from Pocket Outlook.

In order to do this, we will need the Windows Mobile 5.0 SDK for Pocket PC to be installed. Let's start off by creating a Windows Mobile 5.0 device application project. Once the project is created, add a reference to the Navigator wrapper and the TomTom wrapper. Next we have to build the Native wrapper project, and add the output file TTSDK.dll to our project. Set TTSDK.dll to be "Copied if Newer". To retrieve address information from contacts, we must add a reference to Microsoft.WindowsMobile.PocketOutlook.dll.

Once the references and files are in place, we can start adding some code to Form1.cs. No need to change the name of the main form since this is only a small demo. We need to have a control that can contain the contacts, lets use the ComboBox control for now. Add a ComboBox control to the form and call it cbContacts. Lets add a "Navigate to" button to the form as well and call it btnNavigate.

To retrieve a list of contacts we need to create a global instance of Microsoft.WindowsMobile.PocketOutlook.OutlookSession and Microsoft.WindowsMobile.PocketOutlook.ContactsCollection, once we instantiate our OutlookSession, we can then retrieve a list of Contacts through OutlookSession.Contacts.Items.

To communicate with TomTom, we create an instance of Navigator(). The default constructor for Navigator will need a typeName for loading the TomTom wrapper as INavigator. It would be a smart idea to store the typeName in a seperate file, text or xml would be perfect. Once again, we do this so that if we want our application to integrate with different navigation software, we don't have to re-write everything. In this demo, the typeName will just be a hard coded string constant.

[C# CODE]

private Microsoft.WindowsMobile.PocketOutlook.OutlookSession session;
private Microsoft.WindowsMobile.PocketOutlook.ContactsCollection contacts;

private const string TYPENAME="[The namespace].[The class name], [The assembly name]";
private Navigator navigator;

public Form1()
{
  InitializeComponent();

  btnNavigate.Click += new EventHandler(btnNavigate_Click);
  Closing += new CancelEventHandler(Form1_Closing);

  navigator = new Navigator(TYPENAME);

  string restriction = "[BusinessAddressStreet] <> \" \" OR [HomeAddressStreet] <> \" \"";
  session = new OutlookSession();
  contacts = session.Contacts.Items.Restrict(restriction);
  cbContacts.DataSource = contacts;
}

private void Form1_Closing(object sender, CancelEventArgs e)
{
  contacts.Dispose();
  contacts = null;

  session.Dispose();
  session = null;
}

private void btnNavigate_Click(object sender, EventArgs e)
{
  Contact contact = cbContacts.SelectedItem as Contact;
  navigator.FlashMessage("Navigating...", 1500);
  navigator.NavigateToAddress(contact.BusinessAddressCity,
    contact.BusinessAddressStreet,
    "PARSE THE HOUSE NUMBER...",
    contact.BusinessAddressPostalCode);
}

When the application launches, your pocket outlook contacts that have a valid address will be loaded into our ComboBox control. Let's select an item in the ComboBox. Once you click on the Navigate button it will launch TomTom Navigator and display the message "Navigating" for 1.5 seconds, after that it will start calculating the route from your current location to your destination (in this case, the selected contact).

That wasn't too hard was it?

Wednesday, June 6, 2007

Programmatically Minimize an Application in .NET CF 2.0

I once made a solution that runs on full screen. The solution was written completely in managed code (except for the CE Setup and other small stuff..). Since I took over the screen completely, I don't have access to the (X) button in the upper right corner of the screen. I wanted the application running at all times, but I also wanted the user to be able to get in and out of the application. Since the user won't be able to access the Start button, I added a "Close" button to my application. This "Close" button won't exit the application, instead it will just Minimize the application.

In the .NET Compact Framework 2.0, you can't just set the form's WindowState to WindowState.Minimized since the WindowState enum only contains Normal and Maximized. Currently, the only way you can programmatically minimize an application is by doing a P/Invoke to ShowWindow and passing SW_MINIMIZE to specify how the window will be displayed. It is also required that your Form has the Taskbar visible, this is done by setting the following properties:

  FormBorderStyle = FormBorderStyle.FixedDialog;
  WindowState = FormWindowState.Normal;
  ControlBox = true;
  MinimizeBox = true;
  MaximizeBox = true;

Here's a small code snippet of how to minimize your application

[DllImport("coredll.dll")]
static extern int ShowWindow(IntPtr hWnd, int nCmdShow);

const int SW_MINIMIZED = 6;

void Minimize() {
  // The Taskbar must be enabled to be able to do a Smart Minimize
  this.FormBorderStyle = FormBorderStyle.FixedDialog;
  this.WindowState = FormWindowState.Normal;
  this.ControlBox = true;
  this.MinimizeBox = true;
  this.MaximizeBox = true;

  // Since there is no WindowState.Minimize, we have to P/Invoke ShowWindow
  ShowWindow(this.Handle, SW_MINIMIZED);
}

Programmatically Refreshing the Today Screen

A simple trick for forcing the today screen to re-read from the registry (or refresh) is by sending the message WM_WININICHANGE with the parameter 0x000000F2 to a window called the DesktopExplorerWindow.

Here's a small code snippet on how to accomplish this programmatically:

void RefreshTodayScreen() {
  HWND hWnd = FindWindow(_T("DesktopExplorerWindow"), _T("Desktop"));
  SendMessage(hWnd, WM_WININICHANGE, 0x000000F2, 0);
}

and in managed code...

[DllImport("coredll.dll")]
static extern IntPtr FindWindow(string class_name, string caption);

[DllImport("coredll.dll")]
static extern int SendMessage(IntPtr hWnd, uint Msg, int wParam, int lParam);

const uint WM_WININICHANGE = 0x1a;

void RefreshTodayScreen() {
  IntPtr hWnd = FindWindow("DesktopExplorerWindow", "Desktop");
  SendMessage(hWnd, WM_WININICHANGE, 0x000000F2, 0);
}

Monday, June 4, 2007

Querying Overridden Check-in Policies

In the Team Foundation Server, you can enable certain policies for checking in files. The default install will contain policies that can verify that the check-in is associated with a work item, or that unit tests were created for the changes made, etc etc. But even though certain rules were made for checking in, the user is still given the possibility to override these policies, if the user decides to override the policy then the user is prompted with a dialog, where the user can input their "reason" for ignoring such policies. This action is logged to the TFS databases.

In this article I will show you how to query the database "TfsVersionControl" and "TfsWarehouse" to get more information for the overridden check-in policy.

let's start off with openning a query to the database server that TFS uses then type in the following query:

SELECT
    'Changeset ID'=p.ChangeSetId,
    'Creation Date'=cs.CreationDate,
    'Check-in Comment'=cs.Comment,
    'Override Reason'=p.Comment,
    'Owner'=pr.Person,
    'Email'=pr.Email
FROM
    TfsVersionControl..tbl_PolicyOverride p
INNER JOIN
    TfsVersionControl..tbl_ChangeSet cs ON p.ChangeSetId=cs.ChangeSetId
INNER JOIN
    TfsVersionControl..tbl_Identity i ON cs.OwnerId=i.IdentityId
INNER JOIN
    TfsWarehouse..Person pr ON i.DisplayName=(pr.Domain+'\'+pr.Alias)
ORDER BY
    cs.ChangeSetId DESC

This query will provide you with the change set number, date, check-in comment, override reason, change set owner and email address for every overridden check-in policy.

My first ever blog post

After being asked numerous times why I don't have a blog, I finally decided that maybe I should have one.

For a short introduction:

My name is Christian Resma Helle, I was born and raised in the Philippines. I currently reside and work in Copenhagen, Denmark. I work as a Developer/Consultant specializing in Windows Mobile based solutions and have taken an interest in the Visual Studio Team System.

My blog would probably contain technical articles regarding Windows Mobile development and the VSTS.