HAdes
HAdes

Reputation: 16993

How do I properly clean up Excel interop objects?

I'm using the Excel interop in C# (ApplicationClass) and have placed the following code in my finally clause:

while (System.Runtime.InteropServices.Marshal.ReleaseComObject(excelSheet) != 0) { }
excelSheet = null;
GC.Collect();
GC.WaitForPendingFinalizers();

Although this kind of works, the Excel.exe process is still in the background even after I close Excel. It is only released once my application is manually closed.

What am I doing wrong, or is there an alternative to ensure interop objects are properly disposed of?

Upvotes: 808

Views: 354200

Answers (30)

Dietrich Baumgarten
Dietrich Baumgarten

Reputation: 704

My answer is late and its only purpose is to support the solution proposed by Govert.

Automating Excel or other COM objects from the COM-agnostic .NET world is a “tough nut,” as we say in German, and you can easily go nuts. I rely on the following steps:

  1. Native COM functions return an integer of type HRESULT whose value indicates success or failure of the function. In C# the HRESULT value is omitted and the retval parameter of the native COM function is the return value. A HRESULT that indicates a failure will result in an exception. Therefore, critical COM functions must be enclosed in a try/catch block.

  2. COM objects are represented in C# by smart variables that manage their own loading and uploading. Therefore, don’t Marshall.ReleaseComObject anything. That is the job of the Garbage Collection. References to COM objects behave in C# as smart pointers in C++ which aren’t released either.

  3. For each interaction with Excel, get one and only one instance ExcelApp of the Application interface and create a scope within which ExcelApp lives. This is necessary, because the CLR won’t free the resources of Excel before any reference to Excel goes out of scope. A new Excel process is started in the background. The best scope is the try part of a try/catch block since the loading of Excel may fail in which case an exception is thrown.

  4. Generate via Collection properties new objects like Workbook(s), Worksheet(s), and Cell(s). Don’t care for the voodoo one-dot-good, two-dot-bad rule, use as many references as you like, and as said before don’t Marshall.ReleaseComObject anything.

  5. When you are done with Excel, call the Quit function within the scope made for Excel handling. This should be the last statement within this scope. The Quit function is necessary to quit Excel but the process is not unloaded before all reference variables to this instance leave the scope. While your Excel instance is loaded, don’t allow any user actions that would bypass the Quit function.

Before you run my app, open the task manager and watch in the Processes tab the entries in background processes. When you start the program, a new Excel process entry appears in the list and stays there until the thread wakes up again after 5 seconds. After the call to the Quit function the variable ExcelApp goes out of scope unloading the Excel process, that gracefully vanishes from the list of background processes.

using System;
using System.Threading;
using Excel = Microsoft.Office.Interop.Excel;

class Program
{
  private static void DoSomething(Excel.Workbook Wb)
  {
    Excel.Worksheet NewWs = Wb.Worksheets.Add();
                        
    for (int i = 1; i < 10; i++)
    {
      NewWs.Cells[i, 1] = i;
    }

    Wb.Save();
    Wb.Close();
    //NewWs goes out of scope and will be released by the GC.
  }

  static void Main(string[] args)
  {
    try
    {
      Excel.Application ExcelApp = new Excel.Application();
      Excel.Workbook Wb = ExcelApp.Workbooks.Open(@"D:\ExcelFiles\\CSharpExample.xlsx");
      DoSomething(Wb);
      ExcelApp?.Quit(); //Don't forget!
      Thread.Sleep(5000);
      //ExcelApp and Wb go out of scope and will be released by the GC.
    }     
    catch (Exception ex)
    {
      Console.WriteLine(ex.Message);
    }
  }
}

Upvotes: 2

br3nt
br3nt

Reputation: 9566

I really like when things clean up after them selves... So I made some wrapper classes that do all the cleanup for me! These are documented further down.

The end code is quite readable and accessible. I haven't yet found any phantom instances of Excel running after I Close() the workbooks and Quit() the application (besides where I debug and close the app mid process).

function void OpenCopyClose() {
  var excel = new ExcelApplication();
  var workbook1 = excel.OpenWorkbook("C:\Temp\file1.xslx", readOnly: true);
  var readOnlysheet = workbook1.Worksheet("sheet1");

  var workbook2 = excel.OpenWorkbook("C:\Temp\file2.xslx");
  var writeSheet = workbook.Worksheet("sheet1");

  // do all the excel manipulation

  // read from the first workbook, write to the second workbook.
  var a1 = workbook1.Cells[1, 1];
  workbook2.Cells[1, 1] = a1

  // explicit clean-up
  workbook1.Close(false);
  workbook2 .Close(true);
  excel.Quit();
}

Note: You can skip the Close() and Quit() calls but if you are writing to an Excel document you will at least want to Save(). When the objects go out of scope (the method returns) the class finalizers will automatically kick in and do any cleanup. Any references to COM objects from the Worksheet COM object will automatically be managed and cleaned up as long as you are careful with the scope of your variables, eg keep variables local to the current scope only when storing references to COM objects. You can easily copy values you need to POCOs if you need, or create additional wrapper classes as discussed below.

To manage all this, I have created a class, DisposableComObject, that acts as a wrapper for any COM object. It implements the IDisposable interface and also contains a finalizer for those that don't like using.

The Dispose() method calls Marshal.ReleaseComObject(ComObject) and then sets the ComObjectRef property to null.

The object is in a disposed state when the private ComObjectRef property is null.

If the ComObject property is accessed after being disposed, a ComObjectAccessedAfterDisposeException exception is thrown.

The Dispose() method can be called manually. It is also called by the finalizer, at the conclusion of a using block, and for using var at the conclusion of the scope of that variable.

The top level classes from Microsoft.Office.Interop.Excel, Application, Workbook, and Worksheet, get their own wrapper classes where each are subclasses of DisposableComObject

Here is the code:

/// <summary>
/// References to COM objects must be explicitly released when done.
/// Failure to do so can result in odd behavior and processes remaining running after the application has stopped.
/// This class helps to automate the process of disposing the references to COM objects.
/// </summary>
public abstract class DisposableComObject : IDisposable
{
    public class ComObjectAccessedAfterDisposeException : Exception
    {
        public ComObjectAccessedAfterDisposeException() : base("COM object has been accessed after being disposed") { }
    }

    /// <summary>The actual COM object</summary>
    private object ComObjectRef { get; set; }

    /// <summary>The COM object to be used by subclasses</summary>
    /// <exception cref="ComObjectAccessedAfterDisposeException">When the COM object has been disposed</exception>
    protected object ComObject => ComObjectRef ?? throw new ComObjectAccessedAfterDisposeException();

    public DisposableComObject(object comObject) => ComObjectRef = comObject;

    /// <summary>
    /// True, if the COM object has been disposed.
    /// </summary>
    protected bool IsDisposed() => ComObjectRef is null;

    public void Dispose()
    {
        Dispose(true);
        GC.SuppressFinalize(this); // in case a subclass implements a finalizer
    }

    /// <summary>
    /// This method releases the COM object and removes the reference.
    /// This allows the garbage collector to clean up any remaining instance.
    /// </summary>
    /// <param name="disposing">Set to true</param>
    protected virtual void Dispose(bool disposing)
    {
        if (!disposing || IsDisposed()) return;
        Marshal.ReleaseComObject(ComObject);
        ComObjectRef = null;
    }

    ~DisposableComObject()
    {
        Dispose(true);
    }
}

There is also a handy generic subclass which makes usage slightly easier.

public abstract class DisposableComObject<T> : DisposableComObject
{
    protected new T ComObject => (T)base.ComObject;

    public DisposableComObject(T comObject) : base(comObject) { }
}

Finally, we can use DisposableComObject<T> to create our wrapper classes for the Excel interop classes.

The ExcelApplication subclass has a reference to a new Excel application instance and is used to open workbooks.

OpenWorkbook() returns an ExcelWorkbook which is also a subclass of DisposableComObject.

Dispose() has been overridden to quit the Excel application before calling the base Dispose() method. Quit() is an alias of Dispose().

public class ExcelApplication : DisposableComObject<Application>
{
    public class OpenWorkbookActionCancelledException : Exception
    {
        public string Filename { get; }

        public OpenWorkbookActionCancelledException(string filename, COMException ex) : base($"The workbook open action was cancelled. {ex.Message}", ex) => Filename = filename;
    }

    /// <summary>The actual Application from Interop.Excel</summary>
    Application App => ComObject;

    public ExcelApplication() : base(new Application()) { }

    /// <summary>Open a workbook.</summary>
    public ExcelWorkbook OpenWorkbook(string filename, bool readOnly = false, string password = null, string writeResPassword = null)
    {
        try
        {
            var workbook = App.Workbooks.Open(Filename: filename, UpdateLinks: (XlUpdateLinks)0, ReadOnly: readOnly, Password: password, WriteResPassword: writeResPassword, );

            return new ExcelWorkbook(workbook);
        }
        catch (COMException ex)
        {
            // If the workbook is already open and the request mode is not read-only, the user will be presented
            // with a prompt from the Excel application asking if the workbook should be opened in read-only mode.
            // This exception is raised when when the user clicks the Cancel button in that prompt.
            throw new OpenWorkbookActionCancelledException(filename, ex);
        }
    }

    /// <summary>Quit the running application.</summary>
    public void Quit() => Dispose(true);

    /// <inheritdoc/>
    protected override void Dispose(bool disposing)
    {
        if (!disposing || IsDisposed()) return;
        App.Quit();
        base.Dispose(disposing);
    }
}

ExcelWorkbook also subclasses DisposableComObject<Workbook> and is used to open worksheets.

The Worksheet() methods returns ExcelWorksheet which, you guessed it, is also an subclass of DisposableComObject<Workbook>.

The Dispose() method is overridden and fist closes the worksheet before calling the base Dispose().

NOTE: I've added some extension methods which is uses to iterate over Workbook.Worksheets. If you get compile errors, this is why. Ill add the extension methods at the end.

public class ExcelWorkbook : DisposableComObject<Workbook>
{
    public class WorksheetNotFoundException : Exception
    {
        public WorksheetNotFoundException(string message) : base(message) { }
    }

    /// <summary>The actual Workbook from Interop.Excel</summary>
    Workbook Workbook => ComObject;

    /// <summary>The worksheets within the workbook</summary>
    public IEnumerable<ExcelWorksheet> Worksheets => worksheets ?? (worksheets = Workbook.Worksheets.AsEnumerable<Worksheet>().Select(w => new ExcelWorksheet(w)).ToList());
    private IEnumerable<ExcelWorksheet> worksheets;

    public ExcelWorkbook(Workbook workbook) : base(workbook) { }

    /// <summary>
    /// Get the worksheet matching the <paramref name="sheetName"/>
    /// </summary>
    /// <param name="sheetName">The name of the Worksheet</param>
    public ExcelWorksheet Worksheet(string sheetName) => Worksheet(s => s.Name == sheetName, () => $"Worksheet not found: {sheetName}");

    /// <summary>
    /// Get the worksheet matching the <paramref name="predicate"/>
    /// </summary>
    /// <param name="predicate">A function to test each Worksheet for a macth</param>
    public ExcelWorksheet Worksheet(Func<ExcelWorksheet, bool> predicate, Func<string> errorMessageAction) => Worksheets.FirstOrDefault(predicate) ??  throw new WorksheetNotFoundException(errorMessageAction.Invoke());

    /// <summary>
    /// Returns true of the workbook is read-only
    /// </summary>
    public bool IsReadOnly() => Workbook.ReadOnly;

    /// <summary>
    /// Save changes made to the workbook
    /// </summary>
    public void Save()
    {
        Workbook.Save();
    }

    /// <summary>
    /// Close the workbook and optionally save changes
    /// </summary>
    /// <param name="saveChanges">True is save before close</param>
    public void Close(bool saveChanges)
    {
        if (saveChanges) Save();
        Dispose(true);
    }

    /// <inheritdoc/>
    protected override void Dispose(bool disposing)
    {
        if (!disposing || IsDisposed()) return;
        Workbook.Close();
        base.Dispose(disposing);
    }
}

Finally, the ExcelWorksheet.

UsedRows() simply returns an enumerable of unwrapped Microsoft.Office.Interop.Excel.Range objects. I haven't yet encountered a situation where COM objects accessed from properties of the Microsoft.Office.Interop.Excel.Worksheet object need to manually wrapped like was needed with Application, Workbook, and Worksheet. These all seem to clean them selves up automatically. Mostly, I was just iterating over Ranges and getting or setting values, so my particular use-case isn't as advanced as the available functionality.

There is no override of Dispose() in this case as no special action needs to take place for worksheets.

public class ExcelWorksheet : DisposableComObject<Worksheet>
{
    /// <summary>The actual Worksheet from Interop.Excel</summary>
    Worksheet Worksheet => ComObject;

    /// <summary>The worksheet name</summary>
    public string Name => Worksheet.Name;

    // <summary>The worksheets cells (Unwrapped COM object)</summary>
    public Range Cells => Worksheet.Cells;

    public ExcelWorksheet(Worksheet worksheet) : base(worksheet) { }

    /// <inheritdoc cref="WorksheetExtensions.UsedRows(Worksheet)"/>
    public IEnumerable<Range> UsedRows() => Worksheet.UsedRows().ToList();
}

It is possible to add even more wrapper classes. Just add additional methods to ExcelWorksheet as needed and return the COM object in a wrapper class. Just copy what we did when wrapping the workbook via ExcelApplication.OpenWorkbook() and ExcelWorkbook.WorkSheets.

Some useful extension methods:

public static class EnumeratorExtensions
{
    /// <summary>
    /// Converts the <paramref name="enumerator"/> to an IEnumerable of type <typeparamref name="T"/>
    /// </summary>
    public static IEnumerable<T> AsEnumerable<T>(this IEnumerable enumerator)
    {
        return enumerator.GetEnumerator().AsEnumerable<T>();
    }

    /// <summary>
    /// Converts the <paramref name="enumerator"/> to an IEnumerable of type <typeparamref name="T"/>
    /// </summary>
    public static IEnumerable<T> AsEnumerable<T>(this IEnumerator enumerator)
    {
        while (enumerator.MoveNext()) yield return (T)enumerator.Current;
    }

    /// <summary>
    /// Converts the <paramref name="enumerator"/> to an IEnumerable of type <typeparamref name="T"/>
    /// </summary>
    public static IEnumerable<T> AsEnumerable<T>(this IEnumerator<T> enumerator)
    {
        while (enumerator.MoveNext()) yield return enumerator.Current;
    }
}

public static class WorksheetExtensions
{
    /// <summary>
    /// Returns the rows within the used range of this <paramref name="worksheet"/>
    /// </summary>
    /// <param name="worksheet">The worksheet</param>
    public static IEnumerable<Range> UsedRows(this Worksheet worksheet) =>
        worksheet.UsedRange.Rows.AsEnumerable<Range>();
}

Upvotes: 0

Anton Shepelev
Anton Shepelev

Reputation: 1011

Of the three general strategies considered in other answers, killing the excel process is clearly a hack, whereas invoking the garbage collector is a brutal shotgun approach meant to compensate for incorrect deallocation of COM-objects. After lots of experimentation and rewriting the management of COM objects in my version-agnostic and late-bound wrapper, I have come to the conclusion that accurate and timely invocations of Marshal.ReleaseComObject() is the most efficient and elegant strategy. And no, you do not ever need FinalReleaseComObject(), because in a well-writtin program each COM acquired on once and therefore requires a single decrement of the reference counter.

One shall make sure to release every single COM object, preferably as soon as it is no longer needed. But it is perfectly possible to release everything right after quitting the Excel application, at the only expense of higher memory usage. Excel will close as expected as long as one does not loose or forget to release a COM object.

The simplest and most obvious aid in the process is wrapping every interop object into a .NET class implementing IDisposable, where the Dispose() method invokes ReleaseComObject() on its interop object. Doing it in the destructor, as proposed in here, makes no sense because destructors are non-deterministic.

Show below is our wrapper's method that obtains a cell from WorkSheet bypassing the intermediate Cells member. Notice the way it disposes of the intermediate object after use:

public ExcelRange XCell( int row, int col)
{   ExcelRange anchor, res;
    using( anchor = Range( "A1") )
    {   res = anchor.Offset( row - 1, col - 1 );  }
    return res;
}

The next step may be a simple memory manager that will keep track of every COM object obtained and make sure to release it after Excel quits if the user prefers to trade some RAM usage for simpler code.

Futher reading

  1. How to properly release Excel COM objects,
  2. Releasing COM objects: Garbage Collector vs. Marshal.RelseaseComObject.

Upvotes: 0

Philip Fourie
Philip Fourie

Reputation: 116827

This worked for a project I was working on:

excelApp.Quit();
Marshal.ReleaseComObject (excelWB);
Marshal.ReleaseComObject (excelApp);
excelApp = null;

We learned that it was important to set every reference to an Excel COM object to null when you were done with it. This included Cells, Sheets, and everything.

Upvotes: 47

MushroomSoda
MushroomSoda

Reputation: 380

I had this same problem getting PowerPoint to close after newing up the Application object in my VSTO AddIn. I tried all the answers here with limited success.

This is the solution I found for my case - DONT use 'new Application', the AddInBase base class of ThisAddIn already has a handle to 'Application'. If you use that handle where you need it (make it static if you have to) then you don't need to worry about cleaning it up and PowerPoint won't hang on close.

Upvotes: 0

tjarrett
tjarrett

Reputation: 21

Here is a really easy way to do it:

[DllImport("User32.dll")]
static extern uint GetWindowThreadProcessId(IntPtr hWnd, out int lpdwProcessId);
...

int objExcelProcessId = 0;

Excel.Application objExcel = new Excel.Application();

GetWindowThreadProcessId(new IntPtr(objExcel.Hwnd), out objExcelProcessId);

Process.GetProcessById(objExcelProcessId).Kill();

Upvotes: 2

nightcoder
nightcoder

Reputation: 13509

Preface: my answer contains two solutions, so be careful when reading and don't miss anything.

There are different ways and advice of how to make Excel instance unload, such as:

  • Releasing EVERY com object explicitly with Marshal.FinalReleaseComObject() (not forgetting about implicitly created com-objects). To release every created com object, you may use the rule of 2 dots mentioned here:
    How do I properly clean up Excel interop objects?

  • Calling GC.Collect() and GC.WaitForPendingFinalizers() to make CLR release unused com-objects * (Actually, it works, see my second solution for details)

  • Checking if com-server-application maybe shows a message box waiting for the user to answer (though I am not sure it can prevent Excel from closing, but I heard about it a few times)

  • Sending WM_CLOSE message to the main Excel window

  • Executing the function that works with Excel in a separate AppDomain. Some people believe Excel instance will be shut, when AppDomain is unloaded.

  • Killing all excel instances which were instantiated after our excel-interoping code started.

BUT! Sometimes all these options just don't help or can't be appropriate!

For example, yesterday I found out that in one of my functions (which works with excel) Excel keeps running after the function ends. I tried everything! I thoroughly checked the whole function 10 times and added Marshal.FinalReleaseComObject() for everything! I also had GC.Collect() and GC.WaitForPendingFinalizers(). I checked for hidden message boxes. I tried to send WM_CLOSE message to the main Excel window. I executed my function in a separate AppDomain and unloaded that domain. Nothing helped! The option with closing all excel instances is inappropriate, because if the user starts another Excel instance manually, during execution of my function which works also with Excel, then that instance will also be closed by my function. I bet the user will not be happy! So, honestly, this is a lame option (no offence guys). So I spent a couple of hours before I found a good (in my humble opinion) solution: Kill excel process by hWnd of its main window (it's the first solution).

Here is the simple code:

[DllImport("user32.dll")]
private static extern uint GetWindowThreadProcessId(IntPtr hWnd, out uint lpdwProcessId);

/// <summary> Tries to find and kill process by hWnd to the main window of the process.</summary>
/// <param name="hWnd">Handle to the main window of the process.</param>
/// <returns>True if process was found and killed. False if process was not found by hWnd or if it could not be killed.</returns>
public static bool TryKillProcessByMainWindowHwnd(int hWnd)
{
    uint processID;
    GetWindowThreadProcessId((IntPtr)hWnd, out processID);
    if(processID == 0) return false;
    try
    {
        Process.GetProcessById((int)processID).Kill();
    }
    catch (ArgumentException)
    {
        return false;
    }
    catch (Win32Exception)
    {
        return false;
    }
    catch (NotSupportedException)
    {
        return false;
    }
    catch (InvalidOperationException)
    {
        return false;
    }
    return true;
}

/// <summary> Finds and kills process by hWnd to the main window of the process.</summary>
/// <param name="hWnd">Handle to the main window of the process.</param>
/// <exception cref="ArgumentException">
/// Thrown when process is not found by the hWnd parameter (the process is not running). 
/// The identifier of the process might be expired.
/// </exception>
/// <exception cref="Win32Exception">See Process.Kill() exceptions documentation.</exception>
/// <exception cref="NotSupportedException">See Process.Kill() exceptions documentation.</exception>
/// <exception cref="InvalidOperationException">See Process.Kill() exceptions documentation.</exception>
public static void KillProcessByMainWindowHwnd(int hWnd)
{
    uint processID;
    GetWindowThreadProcessId((IntPtr)hWnd, out processID);
    if (processID == 0)
        throw new ArgumentException("Process has not been found by the given main window handle.", "hWnd");
    Process.GetProcessById((int)processID).Kill();
}

As you can see I provided two methods, according to Try-Parse pattern (I think it is appropriate here): one method doesn't throw the exception if the Process could not be killed (for example the process doesn't exist anymore), and another method throws the exception if the Process was not killed. The only weak place in this code is security permissions. Theoretically, the user may not have permissions to kill the process, but in 99.99% of all cases, user has such permissions. I also tested it with a guest account - it works perfectly.

So, your code, working with Excel, can look like this:

int hWnd = xl.Application.Hwnd;
// ...
// here we try to close Excel as usual, with xl.Quit(),
// Marshal.FinalReleaseComObject(xl) and so on
// ...
TryKillProcessByMainWindowHwnd(hWnd);

Voila! Excel is terminated! :)

Ok, let's go back to the second solution, as I promised in the beginning of the post. The second solution is to call GC.Collect() and GC.WaitForPendingFinalizers(). Yes, they actually work, but you need to be careful here!
Many people say (and I said) that calling GC.Collect() doesn't help. But the reason it wouldn't help is if there are still references to COM objects! One of the most popular reasons for GC.Collect() not being helpful is running the project in Debug-mode. In debug-mode objects that are not really referenced anymore will not be garbage collected until the end of the method.
So, if you tried GC.Collect() and GC.WaitForPendingFinalizers() and it didn't help, try to do the following:

1) Try to run your project in Release mode and check if Excel closed correctly

2) Wrap the method of working with Excel in a separate method. So, instead of something like this:

void GenerateWorkbook(...)
{
  ApplicationClass xl;
  Workbook xlWB;
  try
  {
    xl = ...
    xlWB = xl.Workbooks.Add(...);
    ...
  }
  finally
  {
    ...
    Marshal.ReleaseComObject(xlWB)
    ...
    GC.Collect();
    GC.WaitForPendingFinalizers();
  }
}

you write:

void GenerateWorkbook(...)
{
  try
  {
    GenerateWorkbookInternal(...);
  }
  finally
  {
    GC.Collect();
    GC.WaitForPendingFinalizers();
  }
}

private void GenerateWorkbookInternal(...)
{
  ApplicationClass xl;
  Workbook xlWB;
  try
  {
    xl = ...
    xlWB = xl.Workbooks.Add(...);
    ...
  }
  finally
  {
    ...
    Marshal.ReleaseComObject(xlWB)
    ...
  }
}

Now, Excel will close =)

Upvotes: 232

SamSar
SamSar

Reputation: 1

Tested with Microsoft Excel 2016

A really tested solution.

To C# Reference please see: https://stackoverflow.com/a/1307180/10442623

To VB.net Reference please see: https://stackoverflow.com/a/54044646/10442623

1 include the class job

2 implement the class to handle the apropiate dispose of excel proces

Upvotes: 0

Aloha
Aloha

Reputation: 1

There i have an idea,try to kill the excel process you have opened:

  1. before open an excelapplication,get all the process ids named oldProcessIds.
  2. open the excelapplication.
  3. get now all the excelapplication process ids named nowProcessIds.
  4. when need to quit,kill the except ids between oldProcessIds and nowProcessIds.

    private static Excel.Application GetExcelApp()
         {
            if (_excelApp == null)
            {
                var processIds = System.Diagnostics.Process.GetProcessesByName("EXCEL").Select(a => a.Id).ToList();
                _excelApp = new Excel.Application();
                _excelApp.DisplayAlerts = false;
    
                _excelApp.Visible = false;
                _excelApp.ScreenUpdating = false;
                var newProcessIds = System.Diagnostics.Process.GetProcessesByName("EXCEL").Select(a => a.Id).ToList();
                _excelApplicationProcessId = newProcessIds.Except(processIds).FirstOrDefault();
            }
    
            return _excelApp;
        }
    
    public static void Dispose()
        {
            try
            {
                _excelApp.Workbooks.Close();
                _excelApp.Quit();
                System.Runtime.InteropServices.Marshal.ReleaseComObject(_excelApp);
                _excelApp = null;
                GC.Collect();
                GC.WaitForPendingFinalizers();
                if (_excelApplicationProcessId != default(int))
                {
                    var process = System.Diagnostics.Process.GetProcessById(_excelApplicationProcessId);
                    process?.Kill();
                    _excelApplicationProcessId = default(int);
                }
            }
            catch (Exception ex)
            {
                _excelApp = null;
            }
    
        }
    

Upvotes: 0

VVS
VVS

Reputation: 19604

Excel does not quit because your application is still holding references to COM objects.

I guess you're invoking at least one member of a COM object without assigning it to a variable.

For me it was the excelApp.Worksheets object which I directly used without assigning it to a variable:

Worksheet sheet = excelApp.Worksheets.Open(...);
...
Marshal.ReleaseComObject(sheet);

I didn't know that internally C# created a wrapper for the Worksheets COM object which didn't get released by my code (because I wasn't aware of it) and was the cause why Excel was not unloaded.

I found the solution to my problem on this page, which also has a nice rule for the usage of COM objects in C#:

Never use two dots with COM objects.


So with this knowledge the right way of doing the above is:

Worksheets sheets = excelApp.Worksheets; // <-- The important part
Worksheet sheet = sheets.Open(...);
...
Marshal.ReleaseComObject(sheets);
Marshal.ReleaseComObject(sheet);

POST MORTEM UPDATE:

I want every reader to read this answer by Hans Passant very carefully as it explains the trap I and lots of other developers stumbled into. When I wrote this answer years ago I didn't know about the effect the debugger has to the garbage collector and drew the wrong conclusions. I keep my answer unaltered for the sake of history but please read this link and don't go the way of "the two dots": Understanding garbage collection in .NET and Clean up Excel Interop Objects with IDisposable

Upvotes: 711

Breealzibub
Breealzibub

Reputation: 8095

I've traditionally followed the advice found in VVS's answer. However, in an effort to keep this answer up-to-date with the latest options, I think all my future projects will use the "NetOffice" library.

NetOffice is a complete replacement for the Office PIAs and is completely version-agnostic. It's a collection of Managed COM wrappers that can handle the cleanup that often causes such headaches when working with Microsoft Office in .NET.

Some key features are:

  • Mostly version-independent (and version-dependant features are documented)
  • No dependencies
  • No PIA
  • No registration
  • No VSTO

I am in no way affiliated with the project; I just genuinely appreciate the stark reduction in headaches.

Upvotes: 12

'This sure seems like it has been over-complicated. From my experience, there are just three key things to get Excel to close properly:

1: make sure there are no remaining references to the excel application you created (you should only have one anyway; set it to null)

2: call GC.Collect()

3: Excel has to be closed, either by the user manually closing the program, or by you calling Quit on the Excel object. (Note that Quit will function just as if the user tried to close the program, and will present a confirmation dialog if there are unsaved changes, even if Excel is not visible. The user could press cancel, and then Excel will not have been closed.)

1 needs to happen before 2, but 3 can happen anytime.

One way to implement this is to wrap the interop Excel object with your own class, create the interop instance in the constructor, and implement IDisposable with Dispose looking something like

That will clean up excel from your program's side of things. Once Excel is closed (manually by the user or by you calling Quit) the process will go away. If the program has already been closed, then the process will disappear on the GC.Collect() call.

(I'm not sure how important it is, but you may want a GC.WaitForPendingFinalizers() call after the GC.Collect() call but it is not strictly necessary to get rid of the Excel process.)

This has worked for me without issue for years. Keep in mind though that while this works, you actually have to close gracefully for it to work. You will still get accumulating excel.exe processes if you interrupt your program before Excel is cleaned up (usually by hitting "stop" while your program is being debugged).'

Upvotes: 1

Hermes Monteiro
Hermes Monteiro

Reputation: 101

This is the only way that really works for me

        foreach (Process proc in System.Diagnostics.Process.GetProcessesByName("EXCEL"))
        {
            proc.Kill();
        }

Upvotes: -1

Antoine Meltzheim
Antoine Meltzheim

Reputation: 9924

¨°º¤ø„¸ Shoot Excel proc and chew bubble gum ¸„ø¤º°¨

public class MyExcelInteropClass
{
    Excel.Application xlApp;
    Excel.Workbook xlBook;

    public void dothingswithExcel() 
    {
        try { /* Do stuff manipulating cells sheets and workbooks ... */ }
        catch {}
        finally {KillExcelProcess(xlApp);}
    }

    static void KillExcelProcess(Excel.Application xlApp)
    {
        if (xlApp != null)
        {
            int excelProcessId = 0;
            GetWindowThreadProcessId(xlApp.Hwnd, out excelProcessId);
            Process p = Process.GetProcessById(excelProcessId);
            p.Kill();
            xlApp = null;
        }
    }

    [DllImport("user32.dll")]
    static extern int GetWindowThreadProcessId(int hWnd, out int lpdwProcessId);
}

Upvotes: 7

Amit Mittal
Amit Mittal

Reputation: 2726

"Never use two dots with COM objects" is a great rule of thumb to avoid leakage of COM references, but Excel PIA can lead to leakage in more ways than apparent at first sight.

One of these ways is subscribing to any event exposed by any of the Excel object model's COM objects.

For example, subscribing to the Application class's WorkbookOpen event.

Some theory on COM events

COM classes expose a group of events through call-back interfaces. In order to subscribe to events, the client code can simply register an object implementing the call-back interface and the COM class will invoke its methods in response to specific events. Since the call-back interface is a COM interface, it is the duty of the implementing object to decrement the reference count of any COM object it receives (as a parameter) for any of the event handlers.

How Excel PIA expose COM Events

Excel PIA exposes COM events of Excel Application class as conventional .NET events. Whenever the client code subscribes to a .NET event (emphasis on 'a'), PIA creates an instance of a class implementing the call-back interface and registers it with Excel.

Hence, a number of call-back objects get registered with Excel in response to different subscription requests from the .NET code. One call-back object per event subscription.

A call-back interface for event handling means that, PIA has to subscribe to all interface events for every .NET event subscription request. It cannot pick and choose. On receiving an event call-back, the call-back object checks if the associated .NET event handler is interested in the current event or not and then either invokes the handler or silently ignores the call-back.

Effect on COM instance reference counts

All these call-back objects do not decrement the reference count of any of the COM objects they receive (as parameters) for any of the call-back methods (even for the ones that are silently ignored). They rely solely on the CLR garbage collector to free up the COM objects.

Since GC run is non-deterministic, this can lead to the holding off of Excel process for a longer duration than desired and create an impression of a 'memory leak'.

Solution

The only solution as of now is to avoid the PIA’s event provider for the COM class and write your own event provider which deterministically releases COM objects.

For the Application class, this can be done by implementing the AppEvents interface and then registering the implementation with Excel by using IConnectionPointContainer interface. The Application class (and for that matter all COM objects exposing events using callback mechanism) implements the IConnectionPointContainer interface.

Upvotes: 6

Govert
Govert

Reputation: 16907

First - you never have to call Marshal.ReleaseComObject(...) or Marshal.FinalReleaseComObject(...) when doing Excel interop. It is a confusing anti-pattern, but any information about this, including from Microsoft, that indicates you have to manually release COM references from .NET is incorrect. The fact is that the .NET runtime and garbage collector correctly keep track of and clean up COM references. For your code, this means you can remove the whole `while (...) loop at the top.

Second, if you want to ensure that the COM references to an out-of-process COM object are cleaned up when your process ends (so that the Excel process will close), you need to ensure that the garbage collector runs. You do this correctly with calls to GC.Collect() and GC.WaitForPendingFinalizers(). Calling this twice is safe, and ensures that cycles are definitely cleaned up too (though I'm not sure it's needed, and would appreciate an example that shows this).

Third, when running under the debugger, local references will be artificially kept alive until the end of the method (so that local variable inspection works). So GC.Collect() calls are not effective for cleaning object like rng.Cells from the same method. You should split the code doing the COM interop from the GC cleanup into separate methods. (This was a key discovery for me, from one part of the answer posted here by @nightcoder.)

The general pattern would thus be:

Sub WrapperThatCleansUp()

    ' NOTE: Don't call Excel objects in here... 
    '       Debugger would keep alive until end, preventing GC cleanup

    ' Call a separate function that talks to Excel
    DoTheWork()

    ' Now let the GC clean up (twice, to clean up cycles too)
    GC.Collect()    
    GC.WaitForPendingFinalizers()
    GC.Collect()    
    GC.WaitForPendingFinalizers()

End Sub

Sub DoTheWork()
    Dim app As New Microsoft.Office.Interop.Excel.Application
    Dim book As Microsoft.Office.Interop.Excel.Workbook = app.Workbooks.Add()
    Dim worksheet As Microsoft.Office.Interop.Excel.Worksheet = book.Worksheets("Sheet1")
    app.Visible = True
    For i As Integer = 1 To 10
        worksheet.Cells.Range("A" & i).Value = "Hello"
    Next
    book.Save()
    book.Close()
    app.Quit()

    ' NOTE: No calls the Marshal.ReleaseComObject() are ever needed
End Sub

There is a lot of false information and confusion about this issue, including many posts on MSDN and on Stack Overflow (and especially this question!).

What finally convinced me to have a closer look and figure out the right advice was blog post Marshal.ReleaseComObject Considered Dangerous together with finding the issue with references kept alive under the debugger that was confusing my earlier testing.

Upvotes: 45

akirakonenu
akirakonenu

Reputation: 71

So far it seems all answers involve some of these:

  1. Kill the process
  2. Use GC.Collect()
  3. Keep track of every COM object and release it properly.

Which makes me appreciate how difficult this issue is :)

I have been working on a library to simplify access to Excel, and I am trying to make sure that people using it won't leave a mess (fingers crossed).

Instead of writing directly on the interfaces Interop provides, I am making extension methods to make live easier. Like ApplicationHelpers.CreateExcel() or workbook.CreateWorksheet("mySheetNameThatWillBeValidated"). Naturally, anything that is created may lead to an issue later on cleaning up, so I am actually favoring killing the process as last resort. Yet, cleaning up properly (third option), is probably the least destructive and most controlled.

So, in that context I was wondering whether it wouldn't be best to make something like this:

public abstract class ReleaseContainer<T>
{
    private readonly Action<T> actionOnT;

    protected ReleaseContainer(T releasible, Action<T> actionOnT)
    {
        this.actionOnT = actionOnT;
        this.Releasible = releasible;
    }

    ~ReleaseContainer()
    {
        Release();
    }

    public T Releasible { get; private set; }

    private void Release()
    {
        actionOnT(Releasible);
        Releasible = default(T);
    }
}

I used 'Releasible' to avoid confusion with Disposable. Extending this to IDisposable should be easy though.

An implementation like this:

public class ApplicationContainer : ReleaseContainer<Application>
{
    public ApplicationContainer()
        : base(new Application(), ActionOnExcel)
    {
    }

    private static void ActionOnExcel(Application application)
    {
        application.Show(); // extension method. want to make sure the app is visible.
        application.Quit();
        Marshal.FinalReleaseComObject(application);
    }
}

And one could do something similar for all sorts of COM objects.

In the factory method:

    public static Application CreateExcelApplication(bool hidden = false)
    {
        var excel = new ApplicationContainer().Releasible;
        excel.Visible = !hidden;

        return excel;
    }

I would expect that every container will be destructed properly by the GC, and therefore automatically make the call to Quit and Marshal.FinalReleaseComObject.

Comments? Or is this an answer to the question of the third kind?

Upvotes: 0

craig.tadlock
craig.tadlock

Reputation: 1918

I followed this exactly... But I still ran into issues 1 out of 1000 times. Who knows why. Time to bring out the hammer...

Right after the Excel Application class is instantiated I get a hold of the Excel process that was just created.

excel = new Microsoft.Office.Interop.Excel.Application();
var process = Process.GetProcessesByName("EXCEL").OrderByDescending(p => p.StartTime).First();

Then once I've done all the above COM clean-up, I make sure that process isn't running. If it is still running, kill it!

if (!process.HasExited)
   process.Kill();

Upvotes: 8

D.G.
D.G.

Reputation: 169

After trying

  1. Release COM objects in reverse order
  2. Add GC.Collect() and GC.WaitForPendingFinalizers() twice at the end
  3. No more than two dots
  4. Close workbook and quit application
  5. Run in release mode

the final solution that works for me is to move one set of

GC.Collect();
GC.WaitForPendingFinalizers();

that we added to the end of the function to a wrapper, as follows:

private void FunctionWrapper(string sourcePath, string targetPath)
{
    try
    {
        FunctionThatCallsExcel(sourcePath, targetPath);
    }
    finally
    {
        GC.Collect();
        GC.WaitForPendingFinalizers();
    }
}

Upvotes: 9

Shivam Srivastava
Shivam Srivastava

Reputation: 4586

Use:

[DllImport("user32.dll")]
private static extern uint GetWindowThreadProcessId(IntPtr hWnd, out uint lpdwProcessId);

Declare it, add code in the finally block:

finally
{
    GC.Collect();
    GC.WaitForPendingFinalizers();
    if (excelApp != null)
    {
        excelApp.Quit();
        int hWnd = excelApp.Application.Hwnd;
        uint processID;
        GetWindowThreadProcessId((IntPtr)hWnd, out processID);
        Process[] procs = Process.GetProcessesByName("EXCEL");
        foreach (Process p in procs)
        {
            if (p.Id == processID)
                p.Kill();
        }
        Marshal.FinalReleaseComObject(excelApp);
    }
}

Upvotes: 0

Blaz Brencic
Blaz Brencic

Reputation: 674

As some have probably already written, it's not just important how you close the Excel (object); it's also important how you open it and also by the type of the project.

In a WPF application, basically the same code is working without or with very few problems.

I have a project in which the same Excel file is being processed several times for different parameter value - e.g. parsing it based on values inside a generic list.

I put all Excel-related functions into the base class, and parser into a subclass (different parsers use common Excel functions). I didn't want that Excel is opened and closed again for each item in a generic list, so I've opened it only once in the base class and close it in the subclass. I had problems when moving the code into a desktop application. I've tried many of the above mentioned solutions. GC.Collect() was already implemented before, twice as suggested.

Then I've decided that I will move the code for opening Excel to a subclass. Instead of opening only once, now I create a new object (base class) and open Excel for every item and close it at the end. There is some performance penalty, but based on several tests Excel processes are closing without problems (in debug mode), so also temporary files are removed. I will continue with testing and write some more if I will get some updates.

The bottom line is: You must also check the initialize code, especially if you have many classes, etc.

Upvotes: 1

Porkbutts
Porkbutts

Reputation: 964

A great article on releasing COM objects is 2.5 Releasing COM Objects (MSDN).

The method that I would advocate is to null your Excel.Interop references if they are non-local variables, and then call GC.Collect() and GC.WaitForPendingFinalizers() twice. Locally scoped Interop variables will be taken care of automatically.

This removes the need to keep a named reference for every COM object.

Here's an example taken from the article:

public class Test {

    // These instance variables must be nulled or Excel will not quit
    private Excel.Application xl;
    private Excel.Workbook book;

    public void DoSomething()
    {
        xl = new Excel.Application();
        xl.Visible = true;
        book = xl.Workbooks.Add(Type.Missing);

        // These variables are locally scoped, so we need not worry about them.
        // Notice I don't care about using two dots.
        Excel.Range rng = book.Worksheets[1].UsedRange;
    }

    public void CleanUp()
    {
        book = null;
        xl.Quit();
        xl = null;

        GC.Collect();
        GC.WaitForPendingFinalizers();
        GC.Collect();
        GC.WaitForPendingFinalizers();
    }
}

These words are straight from the article:

In almost all situations, nulling the RCW reference and forcing a garbage collection will clean up properly. If you also call GC.WaitForPendingFinalizers, garbage collection will be as deterministic as you can make it. That is, you'll be pretty sure exactly when the object has been cleaned up—on the return from the second call to WaitForPendingFinalizers. As an alternative, you can use Marshal.ReleaseComObject. However, note that you are very unlikely to ever need to use this method.

Upvotes: 5

Saber
Saber

Reputation: 5660

You should be very careful using Word/Excel interop applications. After trying all the solutions we still had a lot of "WinWord" process left open on server (with more than 2000 users).

After working on the problem for hours, I realized that if I open more than a couple of documents using Word.ApplicationClass.Document.Open() on different threads simultaneously, IIS worker process (w3wp.exe) would crash leaving all WinWord processes open!

So I guess there is no absolute solution to this problem, but switching to other methods such as Office Open XML development.

Upvotes: 2

quixver
quixver

Reputation: 574

Excel is not designed to be programmed via C++ or C#. The COM API is specifically designed to work with Visual Basic, VB.NET, and VBA.

Also all the code samples on this page are not optimal for the simple reason that each call must cross a managed/unmanaged boundary and further ignore the fact that the Excel COM API is free to fail any call with a cryptic HRESULT indicating the RPC server is busy.

The best way to automate Excel in my opinion is to collect your data into as big an array as possible / feasible and send this across to a VBA function or sub (via Application.Run) which then performs any required processing. Furthermore - when calling Application.Run - be sure to watch for exceptions indicating excel is busy and retry calling Application.Run.

Upvotes: -1

bill_the_loser
bill_the_loser

Reputation: 195

I think that some of that is just the way that the framework handles Office applications, but I could be wrong. On some days, some applications clean up the processes immediately, and other days it seems to wait until the application closes. In general, I quit paying attention to the details and just make sure that there aren't any extra processes floating around at the end of the day.

Also, and maybe I'm over simplifying things, but I think you can just...

objExcel = new Excel.Application();
objBook = (Excel.Workbook)(objExcel.Workbooks.Add(Type.Missing));
DoSomeStuff(objBook);
SaveTheBook(objBook);
objBook.Close(false, Type.Missing, Type.Missing);
objExcel.Quit();

Like I said earlier, I don't tend to pay attention to the details of when the Excel process appears or disappears, but that usually works for me. I also don't like to keep Excel processes around for anything other than the minimal amount of time, but I'm probably just being paranoid on that.

Upvotes: 1

MagicKat
MagicKat

Reputation: 9821

Anything that is in the Excel namespace needs to be released. Period

You can't be doing:

Worksheet ws = excel.WorkBooks[1].WorkSheets[1];

You have to be doing

Workbooks books = excel.WorkBooks;
Workbook book = books[1];
Sheets sheets = book.WorkSheets;
Worksheet ws = sheets[1];

followed by the releasing of the objects.

Upvotes: 31

Tom Brearley
Tom Brearley

Reputation: 11

I am currently working on Office automation and have stumbled across a solution for this that works every time for me. It is simple and does not involve killing any processes.

It seems that by merely looping through the current active processes, and in any way 'accessing' an open Excel process, any stray hanging instance of Excel will be removed. The below code simply checks for processes where the name is 'Excel', then writes the MainWindowTitle property of the process to a string. This 'interaction' with the process seems to make Windows catch up and abort the frozen instance of Excel.

I run the below method just before the add-in which I am developing quits, as it fires it unloading event. It removes any hanging instances of Excel every time. In all honesty I am not entirely sure why this works, but it works well for me and could be placed at the end of any Excel application without having to worry about double dots, Marshal.ReleaseComObject, nor killing processes. I would be very interested in any suggestions as to why this is effective.

public static void SweepExcelProcesses()
{           
            if (Process.GetProcessesByName("EXCEL").Length != 0)
            {
                Process[] processes = Process.GetProcesses();
                foreach (Process process in processes)
                {
                    if (process.ProcessName.ToString() == "excel")
                    {                           
                        string title = process.MainWindowTitle;
                    }
                }
            }
}

Upvotes: 1

dicksters
dicksters

Reputation: 131

Just to add another solution to the many listed here, using C++/ATL automation (I imagine you could use something similar from VB/C#??)

Excel::_ApplicationPtr pXL = ...
  :
SendMessage ( ( HWND ) m_pXL->GetHwnd ( ), WM_DESTROY, 0, 0 ) ;

This works like a charm for me...

Upvotes: 0

Mike Rosenblum
Mike Rosenblum

Reputation: 12157

You can actually release your Excel Application object cleanly, but you do have to take care.

The advice to maintain a named reference for absolutely every COM object you access and then explicitly release it via Marshal.FinalReleaseComObject() is correct in theory, but, unfortunately, very difficult to manage in practice. If one ever slips anywhere and uses "two dots", or iterates cells via a for each loop, or any other similar kind of command, then you'll have unreferenced COM objects and risk a hang. In this case, there would be no way to find the cause in the code; you would have to review all your code by eye and hopefully find the cause, a task that could be nearly impossible for a large project.

The good news is that you do not actually have to maintain a named variable reference to every COM object you use. Instead, call GC.Collect() and then GC.WaitForPendingFinalizers() to release all the (usually minor) objects to which you do not hold a reference, and then explicitly release the objects to which you do hold a named variable reference.

You should also release your named references in reverse order of importance: range objects first, then worksheets, workbooks, and then finally your Excel Application object.

For example, assuming that you had a Range object variable named xlRng, a Worksheet variable named xlSheet, a Workbook variable named xlBook and an Excel Application variable named xlApp, then your cleanup code could look something like the following:

// Cleanup
GC.Collect();
GC.WaitForPendingFinalizers();

Marshal.FinalReleaseComObject(xlRng);
Marshal.FinalReleaseComObject(xlSheet);

xlBook.Close(Type.Missing, Type.Missing, Type.Missing);
Marshal.FinalReleaseComObject(xlBook);

xlApp.Quit();
Marshal.FinalReleaseComObject(xlApp);

In most code examples you'll see for cleaning up COM objects from .NET, the GC.Collect() and GC.WaitForPendingFinalizers() calls are made TWICE as in:

GC.Collect();
GC.WaitForPendingFinalizers();
GC.Collect();
GC.WaitForPendingFinalizers();

This should not be required, however, unless you are using Visual Studio Tools for Office (VSTO), which uses finalizers that cause an entire graph of objects to be promoted in the finalization queue. Such objects would not be released until the next garbage collection. However, if you are not using VSTO, you should be able to call GC.Collect() and GC.WaitForPendingFinalizers() just once.

I know that explicitly calling GC.Collect() is a no-no (and certainly doing it twice sounds very painful), but there is no way around it, to be honest. Through normal operations you will generate hidden objects to which you hold no reference that you, therefore, cannot release through any other means other than calling GC.Collect().

This is a complex topic, but this really is all there is to it. Once you establish this template for your cleanup procedure you can code normally, without the need for wrappers, etc. :-)

I have a tutorial on this here:

Automating Office Programs with VB.Net / COM Interop

It's written for VB.NET, but don't be put off by that, the principles are exactly the same as when using C#.

Upvotes: 288

Martin
Martin

Reputation: 3472

The accepted answer did not work for me. The following code in the destructor did the job.

if (xlApp != null)
{
    xlApp.Workbooks.Close();
    xlApp.Quit();
}

System.Diagnostics.Process[] processArray = System.Diagnostics.Process.GetProcessesByName("EXCEL");
foreach (System.Diagnostics.Process process in processArray)
{
    if (process.MainWindowTitle.Length == 0) { process.Kill(); }
}

Upvotes: 1

Related Questions