Add MarkdownTextHelper sanitizer to strip control chars that crash WinUI RichTextBlock

Agent-Logs-Url: https://github.com/microsoft/PowerToys/sessions/198cc9b8-bdc8-4189-ad27-5f61e56c7da0

Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>

src/modules/cmdpal/Microsoft.CmdPal.UI.ViewModels/ContentMarkdownViewModel.cs

@@ -3,6 +3,7 @@
 // See the LICENSE file in the project root for more information.
 
 using Microsoft.CmdPal.UI.ViewModels;
+using Microsoft.CmdPal.UI.ViewModels.Helpers;
 using Microsoft.CmdPal.UI.ViewModels.Models;
 using Microsoft.CommandPalette.Extensions;
 
@@ -25,7 +26,7 @@ public partial class ContentMarkdownViewModel(IMarkdownContent _markdown, WeakRe
             return;
         }
 
-        Body = model.Body;
+        Body = MarkdownTextHelper.SanitizeMarkdown(model.Body);
         UpdateProperty(nameof(Body));
 
         model.PropChanged += Model_PropChanged;
@@ -55,7 +56,7 @@ public partial class ContentMarkdownViewModel(IMarkdownContent _markdown, WeakRe
         switch (propertyName)
         {
             case nameof(Body):
-                Body = model.Body;
+                Body = MarkdownTextHelper.SanitizeMarkdown(model.Body);
                 break;
         }
 

src/modules/cmdpal/Microsoft.CmdPal.UI.ViewModels/DetailsViewModel.cs

@@ -2,6 +2,7 @@
 // The Microsoft Corporation licenses this file to you under the MIT license.
 // See the LICENSE file in the project root for more information.
 
+using Microsoft.CmdPal.UI.ViewModels.Helpers;
 using Microsoft.CmdPal.UI.ViewModels.Models;
 using Microsoft.CommandPalette.Extensions;
 
@@ -34,7 +35,7 @@ public partial class DetailsViewModel(IDetails _details, WeakReference<IPageCont
         }
 
         Title = model.Title ?? string.Empty;
-        Body = model.Body ?? string.Empty;
+        Body = MarkdownTextHelper.SanitizeMarkdown(model.Body);
         HeroImage = new(model.HeroImage);
         HeroImage.InitializeProperties();
 

src/modules/cmdpal/Microsoft.CmdPal.UI.ViewModels/Helpers/MarkdownTextHelper.cs

@@ -0,0 +1,44 @@
+// Copyright (c) Microsoft Corporation
+// The Microsoft Corporation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using System.Text.RegularExpressions;
+
+namespace Microsoft.CmdPal.UI.ViewModels.Helpers;
+
+/// <summary>
+/// Provides helper methods for sanitizing text before passing it to the WinUI MarkdownTextBlock
+/// control, which wraps RichTextBlock internally. Certain control characters can trigger a native
+/// crash (access violation) in RichTextBlock's text-selection code path when the user double-clicks
+/// to select a word.
+/// </summary>
+/// <remarks>
+/// See <see href="https://github.com/microsoft/microsoft-ui-xaml/issues/7299"/> for the upstream
+/// WinUI bug. This is a best-effort, client-side mitigation.
+/// </remarks>
+internal static partial class MarkdownTextHelper
+{
+    // Matches characters that are known to destabilize WinUI's RichTextBlock:
+    //   - C0 control codes (U+0000–U+001F) except TAB (U+0009), LF (U+000A), CR (U+000D)
+    //   - DEL (U+007F)
+    //   - C1 control codes (U+0080–U+009F)
+    [GeneratedRegex(@"[\x00-\x08\x0B\x0C\x0E-\x1F\x7F\x80-\x9F]", RegexOptions.CultureInvariant)]
+    private static partial Regex ProblematicControlCharsRegex();
+
+    /// <summary>
+    /// Removes control characters from <paramref name="text"/> that are known to cause crashes in
+    /// the WinUI <c>RichTextBlock</c> control's double-tap word-selection code path.
+    /// Standard whitespace (TAB, LF, CR) is preserved because Markdown relies on it.
+    /// </summary>
+    /// <param name="text">The raw markdown string, possibly <see langword="null"/>.</param>
+    /// <returns>The sanitized string, or <see cref="string.Empty"/> if <paramref name="text"/> is null.</returns>
+    internal static string SanitizeMarkdown(string? text)
+    {
+        if (string.IsNullOrEmpty(text))
+        {
+            return text ?? string.Empty;
+        }
+
+        return ProblematicControlCharsRegex().Replace(text, string.Empty);
+    }
+}

src/modules/cmdpal/Tests/Microsoft.CmdPal.UI.ViewModels.UnitTests/MarkdownTextHelperTests.cs

@@ -0,0 +1,89 @@
+// Copyright (c) Microsoft Corporation
+// The Microsoft Corporation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using Microsoft.CmdPal.UI.ViewModels.Helpers;
+using Microsoft.VisualStudio.TestTools.UnitTesting;
+
+namespace Microsoft.CmdPal.UI.ViewModels.UnitTests;
+
+[TestClass]
+public partial class MarkdownTextHelperTests
+{
+    [TestMethod]
+    public void SanitizeMarkdown_NullInput_ReturnsEmpty()
+    {
+        var result = MarkdownTextHelper.SanitizeMarkdown(null);
+        Assert.AreEqual(string.Empty, result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_EmptyInput_ReturnsEmpty()
+    {
+        var result = MarkdownTextHelper.SanitizeMarkdown(string.Empty);
+        Assert.AreEqual(string.Empty, result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_CleanMarkdown_ReturnsUnchanged()
+    {
+        const string input = "# Hello\n\nThis is **bold** and _italic_.\n\n- Item 1\n- Item 2\n";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual(input, result);
+    }
+
+    [DataTestMethod]
+    [DataRow("\x00")]          // NUL
+    [DataRow("\x01")]          // SOH
+    [DataRow("\x07")]          // BEL
+    [DataRow("\x08")]          // BS
+    [DataRow("\x0B")]          // VT (Vertical Tab)
+    [DataRow("\x0C")]          // FF (Form Feed)
+    [DataRow("\x0E")]          // SO
+    [DataRow("\x1F")]          // US
+    [DataRow("\x7F")]          // DEL
+    [DataRow("\x80")]          // C1 start
+    [DataRow("\x9F")]          // C1 end
+    public void SanitizeMarkdown_SingleControlChar_IsRemoved(string controlChar)
+    {
+        var input = $"before{controlChar}after";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual("beforeafter", result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_NulBytesWithinText_AreRemoved()
+    {
+        // NUL bytes in clipboard content (the main bug trigger from the issue)
+        var input = "Hello\x00World\x00";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual("HelloWorld", result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_StandardWhitespacePreserved()
+    {
+        // TAB (0x09), LF (0x0A), CR (0x0D) must be kept — Markdown relies on them
+        const string input = "Column1\tColumn2\r\nLine2\nLine3";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual(input, result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_MixedControlAndNormalChars_OnlyControlCharsRemoved()
+    {
+        // Simulates clipboard text with embedded NUL bytes interspersed with normal text
+        var input = "This is a short \x00demo paragraph, meant \x00to include literal N\x00UL bytes.";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual("This is a short demo paragraph, meant to include literal NUL bytes.", result);
+    }
+
+    [TestMethod]
+    public void SanitizeMarkdown_UnicodeAndEmoji_NotAffected()
+    {
+        // Non-ASCII and emoji must pass through unchanged
+        const string input = "Café ☕ \U0001F600 résumé";
+        var result = MarkdownTextHelper.SanitizeMarkdown(input);
+        Assert.AreEqual(input, result);
+    }
+}