Walkthrough: Displaying Statement Completion


The new home for Visual Studio documentation is Visual Studio 2017 Documentation on docs.microsoft.com.

The latest version of this topic can be found at Walkthrough: Displaying Statement Completion.

You can implement language-based statement completion by defining the identifiers for which you want to provide completion and then triggering a completion session. You can define statement completion in the context of a language service, define your own file name extension and content type and then display completion for just that type, or you can trigger completion for an existing content type—for example, "plaintext". This walkthrough shows how to trigger statement completion for the "plaintext" content type, which is the content type of text files. The "text" content type is the ancestor of all other content types, including code and XML files.

Statement completion is typically triggered by typing certain characters—for example, by typing the beginning of an identifier such as "using". It is typically dismissed by pressing the Spacebar, Tab, or Enter key to commit a selection. You can implement the IntelliSense features that are triggered by typing a character by using a command handler for the keystrokes (the IOleCommandTarget interface) and a handler provider that implements the IVsTextViewCreationListener interface. To create the completion source, which is the list of identifiers that participate in completion, implement the ICompletionSource interface and a completion source provider (the ICompletionSourceProvider interface). The providers are Managed Extensibility Framework (MEF) component parts. They are responsible for exporting the source and controller classes and importing services and brokers—for example, the ITextStructureNavigatorSelectorService, which enables navigation in the text buffer, and the ICompletionBroker, which triggers the completion session.

This walkthrough shows how to implement statement completion for a hard-coded set of identifiers. In full implementations, the language service and the language documentation are responsible for providing that content.

Starting in Visual Studio 2015, you do not install the Visual Studio SDK from the download center. It is included as an optional feature in Visual Studio setup. You can also install the VS SDK later on. For more information, see Installing the Visual Studio SDK.

To create a MEF project

  1. Create a C# VSIX project. (In the New Project dialog, select Visual C# / Extensibility, then VSIX Project.) Name the solution CompletionTest.

  2. Add an Editor Classifier item template to the project. For more information, see Creating an Extension with an Editor Item Template.

  3. Delete the existing class files.

  4. Add the following references to the project and make sure that CopyLocal is set to false:







The completion source is responsible for collecting the set of identifiers and adding the content to the completion window when a user types a completion trigger, such as the first letters of an identifier. In this example, the identifiers and their descriptions are hard-coded in the AugmentCompletionSession method. In most real-world uses, you would use your language’s parser to get the tokens to populate the completion list.

To implement the completion source

  1. Add a class file and name it TestCompletionSource.

  2. Add these imports:

    using System;
    using System.Collections.Generic;
    using System.Linq;
    using System.Text;
    using System.ComponentModel.Composition;
    using Microsoft.VisualStudio.Language.Intellisense;
    using Microsoft.VisualStudio.Text;
    using Microsoft.VisualStudio.Text.Operations;
    using Microsoft.VisualStudio.Utilities;

  3. Modify the class declaration for TestCompletionSource so that it implements ICompletionSource:

    internal class TestCompletionSource : ICompletionSource

  4. Add private fields for the source provider, the text buffer, and a list of Completion objects (which correspond to the identifiers that will participate in the completion session):

        private TestCompletionSourceProvider m_sourceProvider;
        private ITextBuffer m_textBuffer;
        private List<Completion> m_compList;

  5. Add a constructor that sets the source provider and buffer. The TestCompletionSourceProvider class is defined in later steps:

        public TestCompletionSource(TestCompletionSourceProvider sourceProvider, ITextBuffer textBuffer)
            m_sourceProvider = sourceProvider;
            m_textBuffer = textBuffer;

  6. Implement the AugmentCompletionSession method by adding a completion set that contains the completions you want to provide in the context. Each completion set contains a set of Completion completions, and corresponds to a tab of the completion window. (In Visual Basic projects, the completion window tabs are named Common and All.) The FindTokenSpanAtPosition method is defined in the next step.

        void ICompletionSource.AugmentCompletionSession(ICompletionSession session, IList<CompletionSet> completionSets)
            List<string> strList = new List<string>();
            m_compList = new List<Completion>();
            foreach (string str in strList)
                m_compList.Add(new Completion(str, str, str, null, null));
            completionSets.Add(new CompletionSet(
                "Tokens",    //the non-localized title of the tab
                "Tokens",    //the display title of the tab

  7. The following method is used to find the current word from the position of the cursor:

        private ITrackingSpan FindTokenSpanAtPosition(ITrackingPoint point, ICompletionSession session)
            SnapshotPoint currentPoint = (session.TextView.Caret.Position.BufferPosition) - 1;
            ITextStructureNavigator navigator = m_sourceProvider.NavigatorService.GetTextStructureNavigator(m_textBuffer);
            TextExtent extent = navigator.GetExtentOfWord(currentPoint);
            return currentPoint.Snapshot.CreateTrackingSpan(extent.Span, SpanTrackingMode.EdgeInclusive);

  8. Implement the Dispose() method:

        private bool m_isDisposed;
        public void Dispose()
            if (!m_isDisposed)
                m_isDisposed = true;

The completion source provider is the MEF component part that instantiates the completion source.

To implement the completion source provider

  1. Add a class named TestCompletionSourceProvider that implements ICompletionSourceProvider. Export this class with a ContentTypeAttribute of "plaintext" and a NameAttribute of "test completion".

    [Name("token completion")]
    internal class TestCompletionSourceProvider : ICompletionSourceProvider

  2. Import a ITextStructureNavigatorSelectorService, which is used to find the current word in the completion source.

        internal ITextStructureNavigatorSelectorService NavigatorService { get; set; }

  3. Implement the TryCreateCompletionSource method to instantiate the completion source.

        public ICompletionSource TryCreateCompletionSource(ITextBuffer textBuffer)
            return new TestCompletionSource(this, textBuffer);

The completion command handler provider is derived from a IVsTextViewCreationListener, which listens for a text view creation event and converts the view from an IVsTextView—which enables the addition of the command to the command chain of the Visual Studio shell—to an ITextView. Because this class is a MEF export, you can also use it to import the services that are required by the command handler itself.

To implement the completion command handler provider

  1. Add a file named TestCompletionCommandHandler.

  2. Add these using statements:

    using System;
    using System.ComponentModel.Composition;
    using System.Runtime.InteropServices;
    using Microsoft.VisualStudio;
    using Microsoft.VisualStudio.Editor;
    using Microsoft.VisualStudio.Language.Intellisense;
    using Microsoft.VisualStudio.OLE.Interop;
    using Microsoft.VisualStudio.Shell;
    using Microsoft.VisualStudio.Text;
    using Microsoft.VisualStudio.Text.Editor;
    using Microsoft.VisualStudio.TextManager.Interop;
    using Microsoft.VisualStudio.Utilities;

  3. Add a class named TestCompletionHandlerProvider that implements IVsTextViewCreationListener. Export this class with a NameAttribute of "token completion handler", a ContentTypeAttribute of "plaintext", and a TextViewRoleAttribute of Editable.

    [Name("token completion handler")]
    internal class TestCompletionHandlerProvider : IVsTextViewCreationListener

  4. Import the IVsEditorAdaptersFactoryService, which enables conversion from a IVsTextView to a ITextView, a ICompletionBroker, and a SVsServiceProvider that enables access to standard Visual Studio services.

        internal IVsEditorAdaptersFactoryService AdapterService = null;
        internal ICompletionBroker CompletionBroker { get; set; }
        internal SVsServiceProvider ServiceProvider { get; set; }

  5. Implement the VsTextViewCreated method to instantiate the command handler.

        public void VsTextViewCreated(IVsTextView textViewAdapter)
            ITextView textView = AdapterService.GetWpfTextView(textViewAdapter);
            if (textView == null)
            Func<TestCompletionCommandHandler> createCommandHandler = delegate() { return new TestCompletionCommandHandler(textViewAdapter, textView, this); };

Because statement completion is triggered by keystrokes, you must implement the IOleCommandTarget interface to receive and process the keystrokes that trigger, commit, and dismiss the completion session.

To implement the completion command handler

  1. Add a class named TestCompletionCommandHandler that implements IOleCommandTarget:

    internal class TestCompletionCommandHandler : IOleCommandTarget

  2. Add private fields for the next command handler (to which you pass the command), the text view, the command handler provider (which enables access to various services), and a completion session:

        private IOleCommandTarget m_nextCommandHandler;
        private ITextView m_textView;
        private TestCompletionHandlerProvider m_provider;
        private ICompletionSession m_session;

  3. Add a constructor that sets the text view and the provider fields, and adds the command to the command chain:

        internal TestCompletionCommandHandler(IVsTextView textViewAdapter, ITextView textView, TestCompletionHandlerProvider provider)
            this.m_textView = textView;
            this.m_provider = provider;
            //add the command to the command chain
            textViewAdapter.AddCommandFilter(this, out m_nextCommandHandler);

  4. Implement the QueryStatus method by passing the command along:

        public int QueryStatus(ref Guid pguidCmdGroup, uint cCmds, OLECMD[] prgCmds, IntPtr pCmdText)
            return m_nextCommandHandler.QueryStatus(ref pguidCmdGroup, cCmds, prgCmds, pCmdText);

  5. Implement the Exec method. When this method receives a keystroke, it must do one of these things:

    • Allow the character to be written to the buffer, and then trigger or filter completion. (Printing characters do this.)

    • Commit the completion, but do not allow the character to be written to the buffer. (Whitespace, Tab, and Enter do this when a completion session is displayed.)

    • Allow the command to be passed on to the next handler. (All other commands.)

    Because this method may display UI, call IsInAutomationFunction to make sure that it is not called in an automation context:

        public int Exec(ref Guid pguidCmdGroup, uint nCmdID, uint nCmdexecopt, IntPtr pvaIn, IntPtr pvaOut)
            if (VsShellUtilities.IsInAutomationFunction(m_provider.ServiceProvider))
                return m_nextCommandHandler.Exec(ref pguidCmdGroup, nCmdID, nCmdexecopt, pvaIn, pvaOut);
            //make a copy of this so we can look at it after forwarding some commands
            uint commandID = nCmdID;
            char typedChar = char.MinValue;
            //make sure the input is a char before getting it
            if (pguidCmdGroup == VSConstants.VSStd2K && nCmdID == (uint)VSConstants.VSStd2KCmdID.TYPECHAR)
                typedChar = (char)(ushort)Marshal.GetObjectForNativeVariant(pvaIn);
            //check for a commit character
            if (nCmdID == (uint)VSConstants.VSStd2KCmdID.RETURN
                || nCmdID == (uint)VSConstants.VSStd2KCmdID.TAB
                || (char.IsWhiteSpace(typedChar) || char.IsPunctuation(typedChar)))
                //check for a a selection
                if (m_session != null && !m_session.IsDismissed)
                    //if the selection is fully selected, commit the current session
                    if (m_session.SelectedCompletionSet.SelectionStatus.IsSelected)
                        //also, don't add the character to the buffer
                        return VSConstants.S_OK;
                        //if there is no selection, dismiss the session
            //pass along the command so the char is added to the buffer
            int retVal = m_nextCommandHandler.Exec(ref pguidCmdGroup, nCmdID, nCmdexecopt, pvaIn, pvaOut);
            bool handled = false;
            if (!typedChar.Equals(char.MinValue) && char.IsLetterOrDigit(typedChar))
                if (m_session == null || m_session.IsDismissed) // If there is no active session, bring up completion
                else    //the completion session is already active, so just filter
                handled = true;
            else if (commandID == (uint)VSConstants.VSStd2KCmdID.BACKSPACE   //redo the filter if there is a deletion
                || commandID == (uint)VSConstants.VSStd2KCmdID.DELETE)
                if (m_session != null && !m_session.IsDismissed)
                handled = true;
            if (handled) return VSConstants.S_OK;
            return retVal;

  6. This code is a private method that triggers the completion session:

        private bool TriggerCompletion()
            //the caret must be in a non-projection location 
            SnapshotPoint? caretPoint =
            textBuffer => (!textBuffer.ContentType.IsOfType("projection")), PositionAffinity.Predecessor);
            if (!caretPoint.HasValue)
                return false;
            m_session = m_provider.CompletionBroker.CreateCompletionSession
                caretPoint.Value.Snapshot.CreateTrackingPoint(caretPoint.Value.Position, PointTrackingMode.Positive),
            //subscribe to the Dismissed event on the session 
            m_session.Dismissed += this.OnSessionDismissed;
            return true;

  7. The next example is a private method that unsubscribes from the Dismissed event:

        private void OnSessionDismissed(object sender, EventArgs e)
            m_session.Dismissed -= this.OnSessionDismissed;
            m_session = null;

To test this code, build the CompletionTest solution and run it in the experimental instance.

To build and test the CompletionTest solution

  1. Build the solution.

  2. When you run this project in the debugger, a second instance of Visual Studio is instantiated.

  3. Create a text file and type some text that includes the word "add".

  4. As you type first "a" and then "d", a list that contains "addition" and "adaptation" should be displayed. Notice that addition is selected. When you type another "d", the list should contain only "addition", which is now selected. You can commit "addition" by pressing the Spacebar, Tab, or Enter key, or dismiss the list by typing Esc or any other key.

Walkthrough: Linking a Content Type to a File Name Extension