Update xunit.analyzers and group SkiaSharp dependency updates

.github/dependabot.yml

@@ -27,6 +27,22 @@ updates:
     directory: "/"
     schedule:
       interval: "daily"
+    # SkiaSharp and its native assets must always have the same version - separate PRs each fail
+    # the build until the other one is merged, so update them together in one PR. SkiaSharpCompare
+    # (test dependency) is built against a specific SkiaSharp version and joins the same PR.
+    groups:
+      skiasharp:
+        patterns:
+          - "SkiaSharp"
+          - "SkiaSharp.*"
+          - "Codeuctivity.SkiaSharpCompare"
+    # Wait a few days before picking up a new SkiaSharp, so SkiaSharpCompare has time to publish a
+    # release built against it and both land in the same grouped PR.
+    cooldown:
+      default-days: 3
+      include:
+        - "SkiaSharp"
+        - "SkiaSharp.*"
     ignore:
       - dependency-name: "SixLabors.ImageSharp"
         versions: ["3.x", "4.x"]

OpenXmlPowerTools.Tests/OpenXmlPowerTools.Tests.csproj

@@ -6,14 +6,13 @@
   </PropertyGroup>
   <ItemGroup>
     <PackageReference Include="Codeuctivity.HtmlRenderer" Version="4.0.529" />
-    <PackageReference Include="Codeuctivity.SkiaSharpCompare" Version="4.0.331" />
-    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="18.7.0" />
-    <PackageReference Include="xunit.analyzers" Version="2.0.0" />
-    <PackageReference Include="xunit.runner.visualstudio" Version="3.1.5">
+    <PackageReference Include="Codeuctivity.SkiaSharpCompare" Version="4.0.342" />
+    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="18.11.0" />
+    <PackageReference Include="xunit.runner.visualstudio" Version="4.0.1">
       <PrivateAssets>all</PrivateAssets>
       <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
     </PackageReference>
-    <PackageReference Include="xunit.v3" Version="4.0.0" />
+    <PackageReference Include="xunit.v3" Version="4.0.2" />
   </ItemGroup>
 
   <ItemGroup>

OpenXmlPowerTools.Tests/OpenXmlRegexTests.cs

@@ -405,5 +405,26 @@ public void CanMatchDespiteInvisibleLayoutMarkers()
             }
         }
 
+        [Theory]
+        [InlineData(null)]
+        [InlineData("default")]
+        [InlineData("preserve")]
+        public void CanReplaceSplitMatchesWithNonBreakingSpaces(string space)
+        {
+            XElement Text(string value) => new XElement(W.t,
+                space == null ? null : new XAttribute(XNamespace.Xml + "space", space), value);
+
+            var paragraph = new XElement(W.p,
+                new XElement(W.r, Text("Prefix\u00A0{[if")),
+                new XElement(W.r, Text("\u00A0Ready]}\u00A0Middle\u00A0{[if")),
+                new XElement(W.r, Text("\u00A0Ready]}\u00A0Suffix")));
+            var count = OpenXmlRegex.Replace(new[] { paragraph },
+                new Regex(Regex.Escape("{[if\u00A0Ready]}")), "Selected", null);
+
+            Assert.Equal(2, count);
+            Assert.Equal("Prefix\u00A0Selected\u00A0Middle\u00A0Selected\u00A0Suffix",
+                string.Concat(paragraph.Descendants(W.t).Select(t => t.Value)));
+        }
+
     }
 }

OpenXmlPowerTools.Tests/UnicodeMapperTests.cs

@@ -69,6 +69,20 @@ public void CanCreateCoalescedRuns()
             Assert.Equal("Third", mixedRuns.Elements(W.t).Skip(2).First().Value);
         }
 
+        [Theory]
+        [InlineData("\u00A0")]
+        [InlineData("\u00A0text")]
+        [InlineData("text\u00A0")]
+        [InlineData("\u00A0text\u00A0")]
+        public void CanRoundTripNonBreakingSpaces(string text)
+        {
+            var runs = UnicodeMapper.StringToCoalescedRunList(text, null);
+
+            // NBSP is content; the mapper's own writer does not require xml:space for it.
+            Assert.All(runs.Elements(W.t), t => Assert.Null(t.Attribute(XNamespace.Xml + "space")));
+            Assert.Equal(text, runs.Select(UnicodeMapper.RunToString).StringConcatenate());
+        }
+
         [Fact]
         public void CanMapSymbols()
         {
@@ -201,32 +215,40 @@ public void HonorsXmlSpace()
         // into a canonical form when displaying or saving the document. These tests
         // compare RunToString’s output against Word’s canonicalized output to ensure
         // that whitespace is treated as semantic content in the same way Word treats it.
-        [Fact]
-        public void TreatsXmlSpaceLikeWord()
+        [Theory]
+        [InlineData("UM-Whitespace", 107)]
+        [InlineData("UM-Whitespace-Unicode", 43)]
+        public void TreatsXmlSpaceLikeWord(string fixture, int caseCount)
         {
             var sourceDir = new System.IO.DirectoryInfo("../../../../TestFiles/");
             // Test document: crafted to include many whitespace patterns that Word accepts as valid input
-            var testDoc = new System.IO.FileInfo(System.IO.Path.Combine(sourceDir.FullName, "UM-Whitespace-test.docx"));
+            var testDoc = new System.IO.FileInfo(System.IO.Path.Combine(sourceDir.FullName, fixture + "-test.docx"));
             var testWmlDoc = new WmlDocument(testDoc.FullName);
             var testParagraphs = testWmlDoc.MainDocumentPart
                             .Element(W.body)
                             .Elements(W.p).ToList();
             // Canonical document: the same test document after being opened and saved by Word,
             // representing Word's own normalized interpretation of that whitespace
-            var expectedDoc = new System.IO.FileInfo(System.IO.Path.Combine(sourceDir.FullName, "UM-Whitespace-Word-saved.docx"));
+            var expectedDoc = new System.IO.FileInfo(System.IO.Path.Combine(sourceDir.FullName, fixture + "-Word-saved.docx"));
             var expectedWmlDoc = new WmlDocument(expectedDoc.FullName);
             var expectedParagraphs = expectedWmlDoc.MainDocumentPart
                             .Element(W.body)
                             .Elements(W.p).ToList();
+            // Each fixture ends with one blank paragraph after its name/content pairs.
+            Assert.Equal(caseCount * 2 + 1, testParagraphs.Count);
+            Assert.Equal(testParagraphs.Count, expectedParagraphs.Count);
+            Assert.Empty(testParagraphs.Last().Descendants(W.t));
+            Assert.Empty(expectedParagraphs.Last().Descendants(W.t));
             // Iterate through pairs of paragraphs (test name, test content, expected result)
-            for (int i = 0; i < testParagraphs.Count - 1; i += 2)
+            for (int i = 0; i < caseCount * 2; i += 2)
             {
                 var testNameParagraph = testParagraphs[i];
                 var testContentParagraph = testParagraphs[i + 1];
                 // Get the test name from the first paragraph
                 var testName = testNameParagraph.Descendants(W.t)
                     .Select(t => (string)t)
                     .StringConcatenate();
+                Assert.Equal(testName, ExtractExpectedFromWord(expectedParagraphs[i]));
                 // Get the actual result by calling UnicodeMapper.RunToString on the test content runs
                 var actualResult = testContentParagraph.Descendants(W.r)
                     .Select(UnicodeMapper.RunToString)
@@ -242,7 +264,7 @@ public void TreatsXmlSpaceLikeWord()
 
         // Extracts the expected text from Word’s canonicalized output for the whitespace tests.
         // This helper intentionally handles *only* the constructs that Word emits in the saved
-        // version of UM-whitespace-test.docx:
+        // versions of the UM-Whitespace fixtures:
         //   • <w:t>      → literal text
         //   • <w:tab/>   → '\t'
         //   • <w:lastRenderedPageBreak/> (intentionally ignored)

OpenXmlPowerTools/OpenXmlPowerTools.csproj

@@ -43,9 +43,9 @@
 	</ItemGroup>
 	<ItemGroup>
                 <PackageReference Include="DocumentFormat.OpenXml" Version="3.5.1" />
-                <PackageReference Include="SkiaSharp.NativeAssets.Linux.NoDependencies" Version="4.152.0" />
-                <PackageReference Include="SkiaSharp" Version="4.152.0" />
-                <PackageReference Include="SonarAnalyzer.CSharp" Version="10.25.0.139117">
+                <PackageReference Include="SkiaSharp.NativeAssets.Linux.NoDependencies" Version="4.153.1" />
+                <PackageReference Include="SkiaSharp" Version="4.153.1" />
+                <PackageReference Include="SonarAnalyzer.CSharp" Version="10.35.0.4138">
                         <PrivateAssets>all</PrivateAssets>
                         <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
                 </PackageReference>

OpenXmlPowerTools/UnicodeMapper.cs

@@ -155,10 +155,10 @@ private static string NormalizeWhitespace(string text, bool preserve)
         {
             if (string.IsNullOrEmpty(text))
                 return string.Empty;
-            // Trim leading & trailing whitespace when NOT preserving
+            // Word trims XML whitespace at text boundaries, not Unicode spaces such as NBSP.
             ReadOnlySpan<char> span = preserve
                 ? text.AsSpan()
-                : text.AsSpan().Trim();
+                : text.AsSpan().Trim(" \t\r\n");
             if (span.Length == 0)
                 return string.Empty;
             var sb = new System.Text.StringBuilder(span.Length);

OpenXmlPowerToolsExamples/MarkupSimplifierApp/MarkupSimplifierApp.csproj

@@ -7,7 +7,7 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="System.Configuration.ConfigurationManager" Version="10.0.9" />
+    <PackageReference Include="System.Configuration.ConfigurationManager" Version="10.0.14" />
   </ItemGroup>
 
   <ItemGroup>

TestFiles/UM-Whitespace-Unicode-test.xml

@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+  <w:body><w:p><w:r><w:t>NBSP leading xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t> foo</w:t></w:r></w:p><w:p><w:r><w:t>NBSP trailing xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t>foo </w:t></w:r></w:p><w:p><w:r><w:t>NBSP only xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t> </w:t></w:r></w:p><w:p><w:r><w:t>NBSP mixed xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>NBSP split field xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t>{[elseif</w:t></w:r><w:r><w:t> TakingTitle == "Unequal"]}</w:t></w:r></w:p><w:p><w:r><w:t>NBSP leading xml:space default</w:t></w:r></w:p><w:p><w:r><w:t xml:space="default"> foo</w:t></w:r></w:p><w:p><w:r><w:t>NBSP trailing xml:space default</w:t></w:r></w:p><w:p><w:r><w:t xml:space="default">foo </w:t></w:r></w:p><w:p><w:r><w:t>NBSP only xml:space default</w:t></w:r></w:p><w:p><w:r><w:t xml:space="default"> </w:t></w:r></w:p><w:p><w:r><w:t>NBSP mixed xml:space default</w:t></w:r></w:p><w:p><w:r><w:t xml:space="default"> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>NBSP split field xml:space default</w:t></w:r></w:p><w:p><w:r><w:t>{[elseif</w:t></w:r><w:r><w:t xml:space="default"> TakingTitle == "Unequal"]}</w:t></w:r></w:p><w:p><w:r><w:t>NBSP leading xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve"> foo</w:t></w:r></w:p><w:p><w:r><w:t>NBSP trailing xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve">foo </w:t></w:r></w:p><w:p><w:r><w:t>NBSP only xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve"> </w:t></w:r></w:p><w:p><w:r><w:t>NBSP mixed xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve"> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>NBSP split field xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t>{[elseif</w:t></w:r><w:r><w:t xml:space="preserve"> TakingTitle == "Unequal"]}</w:t></w:r></w:p><w:p><w:r><w:t>NBSP trailing keyword split field</w:t></w:r></w:p><w:p><w:r><w:t>{[elseif </w:t></w:r><w:r><w:t>TakingTitle == "Unequal"]}</w:t></w:r></w:p><w:p><w:r><w:t>NBSP standalone separator run</w:t></w:r></w:p><w:p><w:r><w:t>{[elseif</w:t></w:r><w:r><w:t> </w:t></w:r><w:r><w:t>TakingTitle == "Unequal"]}</w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+0085 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10;…foo…&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+1680 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2000 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2001 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2002 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2003 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2004 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2005 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2006 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2007 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2008 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2009 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+200A mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2028 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10;
foo
&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+2029 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10;
foo
&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+202F mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+205F mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10; foo &#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>Unicode U+3000 mixed boundaries</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10;　foo　&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>XML whitespace edges xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t> &#9;&#13;&#10;foo&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>referenced CR xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t>foo&#13;bar</w:t></w:r></w:p><w:p><w:r><w:t>referenced CRLF xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t>foo&#13;&#10;bar</w:t></w:r></w:p><w:p><w:r><w:t>internal tab xml:space absent</w:t></w:r></w:p><w:p><w:r><w:t>foo&#9;bar</w:t></w:r></w:p><w:p><w:r><w:t>XML whitespace edges xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve"> &#9;&#13;&#10;foo&#13;&#10;&#9; </w:t></w:r></w:p><w:p><w:r><w:t>referenced CR xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve">foo&#13;bar</w:t></w:r></w:p><w:p><w:r><w:t>referenced CRLF xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve">foo&#13;&#10;bar</w:t></w:r></w:p><w:p><w:r><w:t>internal tab xml:space preserve</w:t></w:r></w:p><w:p><w:r><w:t xml:space="preserve">foo&#9;bar</w:t></w:r></w:p><w:p/></w:body>
+</w:document>

TestFiles/UM-Whitespace.md

@@ -0,0 +1,56 @@
+# UnicodeMapper whitespace fixtures
+
+`UnicodeMapperTests.TreatsXmlSpaceLikeWord` compares each crafted input paragraph
+with the literal text and tabs in the corresponding paragraph saved by Microsoft
+Word. Each case has a name paragraph followed by a content paragraph. Each document
+ends with one blank paragraph. The tests check paragraph counts and matching names
+so a missing or reordered case cannot silently pass.
+
+The original `UM-Whitespace-test.docx` and `UM-Whitespace-Word-saved.docx` contain
+107 cases from the January 2026 investigation. They are unchanged.
+
+The supplementary `UM-Whitespace-Unicode-test.docx` and
+`UM-Whitespace-Unicode-Word-saved.docx` contain 43 cases. The expected document was
+opened and saved without content edits using Microsoft Word for Mac 16.114 on
+October 21, 2026. These are synthetic documents; no customer document was used.
+
+The supplementary cases cover:
+
+- Leading, trailing, standalone, and mixed-boundary NBSP with `xml:space` absent,
+  `default`, and `preserve`.
+- An NBSP separator at either side of a run boundary and in a standalone run.
+- All other XML-valid characters recognized as whitespace by .NET: U+0085,
+  U+1680, U+2000 through U+200A, U+2028, U+2029, U+202F, U+205F, and U+3000.
+  Each is surrounded by ordinary XML whitespace to distinguish trimming the XML
+  whitespace from trimming the Unicode character.
+- Ordinary spaces, tabs, CR, and CRLF, with preservation absent and present.
+  CR and LF use character references so the XML parser does not normalize them
+  before Word or UnicodeMapper sees them. Literal XML line endings in the original
+  fixture are normalized by XML parsing and do not make that distinction.
+
+Word preserves the tested Unicode characters, trims only boundary space/tab/CR/LF
+when preservation is absent or `default`, converts internal line endings to spaces,
+and retains tabs when preservation is enabled. These observations cover the listed
+inputs and Word version, not every possible whitespace or formatting context.
+
+`UM-Whitespace-Unicode-test.xml` is the exact `word/document.xml` payload in the
+supplementary input DOCX, retained for review and reproduction. To regenerate the
+input from the repository root using Python's standard library:
+
+```python
+from pathlib import Path
+from zipfile import ZipFile
+
+root = Path("TestFiles")
+xml = (root / "UM-Whitespace-Unicode-test.xml").read_bytes()
+with ZipFile(root / "UM-Whitespace-test.docx") as source:
+    with ZipFile(root / "UM-Whitespace-Unicode-test.docx", "w") as target:
+        for entry in source.infolist():
+            data = xml if entry.filename == "word/document.xml" else source.read(entry.filename)
+            target.writestr(entry, data)
+```
+
+Keep the crafted input unchanged. Open it in Word and use Save As to create the
+separate `UM-Whitespace-Unicode-Word-saved.docx`. Do not manufacture expected output
+with UnicodeMapper or another word processor. Review any changes in the Word-saved
+text before updating expectations.