adding-editor-classifiers

Custom Text Classification (Syntax Highlighting) in Visual Studio Extensions

Classifiers assign classification types (keyword, comment, string, etc.) to spans of text. Visual Studio maps these types to colors and styles. Common scenarios:

• Syntax-highlight a custom language or DSL
• Add extra coloring to existing languages (e.g., highlight specific identifiers)
• Mark diagnostic regions with distinct formatting

Classifiers are the foundation of syntax highlighting in VS. They run on every editor render pass, so performance is critical — a slow classifier makes typing feel laggy across the entire editor. Classifiers also integrate with the VS Fonts & Colors system, allowing users to customize your classification colors through Tools > Options.

When to use classifiers vs. alternatives:

• Syntax coloring for a custom language or DSL → classifier (this skill)
• Simple keyword/syntax coloring via grammar file (no code) → TextMate grammar (see vs-textmate-grammar)
• Visual decorations, highlights, or overlays on text → adornments (see vs-editor-adornment)
• Custom fonts/colors settings that users can configure → vs-fonts-and-colors

---

MEF Asset Type Requirement

Any extension that uses MEF editor exports must declare the MEF asset type in the .vsixmanifest file. Without this, Visual Studio will not discover your MEF components and your classifier will not load.

Add this inside the <Assets> element of source.extension.vsixmanifest:

<Asset Type="Microsoft.VisualStudio.MefComponent"
       d:Source="Project"
       d:ProjectName="%CurrentProject%"
       Path="|%CurrentProject%|" />

If you are using a multi-project solution, replace %CurrentProject% with the project name that contains the MEF exports.

---

1. VisualStudio.Extensibility (out-of-process, recommended)

The VisualStudio.Extensibility SDK does not currently support custom classifiers or classification types. Text classification relies on MEF-exported IClassifier components which must run in-process.

If you need custom syntax highlighting, use the VSSDK (in-process) MEF approach.

---

2. VSIX Community Toolkit (in-process)

The Community Toolkit does not add a separate classification API — classifiers use the same MEF-based VSSDK APIs. The code below is the standard VSSDK approach, which works in a Community Toolkit extension.

---

3. VSSDK (in-process, legacy)

Custom classification involves three MEF exports:

1. Classification type definition — defines a new classification type name
2. Classification format definition — maps the type to visual styling (color, bold, etc.)
3. Classifier provider — creates IClassifier instances for each text buffer

NuGet packages: Microsoft.VisualStudio.SDK, Microsoft.VisualStudio.Editor, Microsoft.VisualStudio.Language.StandardClassification Key namespaces: Microsoft.VisualStudio.Text.Classification, Microsoft.VisualStudio.Utilities

Step 1: Define a classification type

using System.ComponentModel.Composition;
using Microsoft.VisualStudio.Text.Classification;
using Microsoft.VisualStudio.Utilities;

namespace MyExtension;

internal static class ClassificationTypes
{
    [Export(typeof(ClassificationTypeDefinition))]
    [Name("MyKeyword")]
    internal static ClassificationTypeDefinition MyKeywordType = null;

    [Export(typeof(ClassificationTypeDefinition))]
    [Name("MyComment")]
    internal static ClassificationTypeDefinition MyCommentType = null;
}

Step 2: Define the format (colors / styles)

using System.ComponentModel.Composition;
using System.Windows.Media;
using Microsoft.VisualStudio.Text.Classification;
using Microsoft.VisualStudio.Utilities;

namespace MyExtension;

[Export(typeof(EditorFormatDefinition))]
[ClassificationType(ClassificationTypeNames = "MyKeyword")]
[Name("MyKeyword")]
[UserVisible(true)]
[Order(Before = Priority.Default)]
internal sealed class MyKeywordFormat : ClassificationFormatDefinition
{
    public MyKeywordFormat()
    {
        DisplayName = "My Keyword";
        ForegroundColor = Colors.CornflowerBlue;
        IsBold = true;
    }
}

[Export(typeof(EditorFormatDefinition))]
[ClassificationType(ClassificationTypeNames = "MyComment")]
[Name("MyComment")]
[UserVisible(true)]
[Order(Before = Priority.Default)]
internal sealed class MyCommentFormat : ClassificationFormatDefinition
{
    public MyCommentFormat()
    {
        DisplayName = "My Comment";
        ForegroundColor = Colors.Green;
        IsItalic = true;
    }
}

Step 3: Implement the classifier provider

using System.ComponentModel.Composition;
using Microsoft.VisualStudio.Text;
using Microsoft.VisualStudio.Text.Classification;
using Microsoft.VisualStudio.Utilities;

namespace MyExtension;

[Export(typeof(IClassifierProvider))]
[ContentType("text")] // or your custom content type
internal sealed class MyClassifierProvider : IClassifierProvider
{
    [Import]
    internal IClassificationTypeRegistryService ClassificationRegistry { get; set; }

    public IClassifier GetClassifier(ITextBuffer textBuffer)
    {
        return textBuffer.Properties.GetOrCreateSingletonProperty(
            () => new MyClassifier(textBuffer, ClassificationRegistry));
    }
}

Step 4: Implement the classifier

using System;
using System.Collections.Generic;
using Microsoft.VisualStudio.Text;
using Microsoft.VisualStudio.Text.Classification;

namespace MyExtension;

internal sealed class MyClassifier : IClassifier
{
    private readonly ITextBuffer _buffer;
    private readonly IClassificationType _keywordType;
    private readonly IClassificationType _commentType;

    // Keywords to highlight
    private static readonly HashSet<string> Keywords = new(StringComparer.OrdinalIgnoreCase)
    {
        "SELECT", "FROM", "WHERE", "INSERT", "UPDATE", "DELETE", "JOIN", "ON"
    };

    public MyClassifier(ITextBuffer buffer, IClassificationTypeRegistryService registry)
    {
        _buffer = buffer;
        _keywordType = registry.GetClassificationType("MyKeyword");
        _commentType = registry.GetClassificationType("MyComment");
    }

    public event EventHandler<ClassificationChangedEventArgs> ClassificationChanged;

    public IList<ClassificationSpan> GetClassificationSpans(SnapshotSpan span)
    {
        var classifications = new List<ClassificationSpan>();
        string text = span.GetText();
        int lineStart = span.Start.Position;

        // Simple line-based classification
        foreach (var line in span.Snapshot.Lines)
        {
            if (line.Start < span.Start || line.Start >= span.End)
                continue;

            string lineText = line.GetText();

            // Classify comments (lines starting with --)
            if (lineText.TrimStart().StartsWith("--"))
            {
                classifications.Add(new ClassificationSpan(
                    line.Extent, _commentType));
                continue;
            }

            // Classify keywords
            var words = lineText.Split(new[] { ' ', '\t', ',', '(', ')' },
                StringSplitOptions.RemoveEmptyEntries);
            int pos = 0;
            foreach (string word in words)
            {
                int idx = lineText.IndexOf(word, pos, StringComparison.OrdinalIgnoreCase);
                if (idx >= 0 && Keywords.Contains(word))
                {
                    var wordSpan = new SnapshotSpan(
                        span.Snapshot, line.Start + idx, word.Length);
                    classifications.Add(new ClassificationSpan(wordSpan, _keywordType));
                }
                pos = idx + word.Length;
            }
        }

        return classifications;
    }
}

Defining a custom content type

If you're creating a classifier for a new file type, define a content type and file extension:

internal static class MyContentType
{
    [Export]
    [Name("myLanguage")]
    [BaseDefinition("code")]
    internal static ContentTypeDefinition MyLanguageContentType = null;

    [Export]
    [FileExtension(".mylang")]
    [ContentType("myLanguage")]
    internal static FileExtensionToContentTypeDefinition MyLanguageFileExtension = null;
}

Then change your classifier provider's [ContentType("text")] to [ContentType("myLanguage")].

Augmenting existing classifiers

To add classifications on top of an existing language (e.g., highlight custom identifiers in C#), use [ContentType("CSharp")] on your provider. Your classifier's spans will merge with the built-in C# classifier.

Key points

• GetClassificationSpans is called frequently — keep it fast. Avoid allocations and complex parsing.
• Fire ClassificationChanged when your classification data changes (e.g., after a background parse completes).
• Use IClassificationTypeRegistryService to look up classification types.
• [UserVisible(true)] on the format definition makes the classification appear in Tools > Options > Fonts and Colors.
• Use [Order] on the format to control priority when multiple classifiers produce overlapping spans.
• Remember to add the MEF asset type to your .vsixmanifest — see the top of this document.

---

Key guidance

• VisualStudio.Extensibility does not support custom text classifiers.
• VSSDK / Community Toolkit — Export ClassificationTypeDefinition, ClassificationFormatDefinition, and IClassifierProvider via MEF. Implement IClassifier.GetClassificationSpans to return classified text spans.
• Always declare the MEF component asset type in source.extension.vsixmanifest.
• For new languages, also define a content type and file extension mapping.
• Keep GetClassificationSpans fast — it runs on every render pass.

Troubleshooting

• Classifier doesn't load / text isn't colored: Check the MEF asset type in .vsixmanifest. Also verify [ContentType] matches the file type you're testing with.
• Classification format definition doesn't appear in Fonts & Colors settings: Ensure [UserVisible(true)] is set on the ClassificationFormatDefinition. Without it, the classification exists internally but isn't exposed in Tools > Options.
• Colors look wrong or disappear in Dark theme: You're hard-coding RGB colors. Use ClassificationFormatDefinition properties that respect VS theme tokens, or define separate colors for Light/Dark/High Contrast. See vs-theming.
• Editor becomes sluggish after adding classifier: GetClassificationSpans is doing too much parsing. Pre-parse on a background thread (triggered by ITextBuffer.Changed) and cache results; return from cache in GetClassificationSpans.
• ClassificationChanged event causes infinite loop: Your ClassificationChanged handler is triggering a re-classification. Guard against re-entrancy.

What NOT to do

Do NOT do heavy parsing or I/O in GetClassificationSpans — it runs synchronously on every render pass. Parse on a background thread and cache results.

Do NOT hard-code colors in ClassificationFormatDefinition without considering Dark and High Contrast themes — use VS theme-aware color tokens or per-theme overrides.

Do NOT forget the MefComponent asset type in .vsixmanifest — without it, your classifier provider and format definitions are silently ignored.

Do NOT forget to fire ClassificationChanged when cached parse results update — without it, stale highlighting persists until the user scrolls.

See also

• vs-textmate-grammar
• vs-editor-tagger
• vs-editor-adornment
• vs-fonts-and-colors
• vs-theming

References

• Walkthrough: Highlighting Text (VSSDK)
• IClassifier Interface
• ClassificationFormatDefinition
• Defining Content Types and File Extensions