Fix DOCX font measurement and table layout to match Word

.github/workflows/ci.yml

@@ -61,6 +61,38 @@ jobs:
           name: test-results
           path: '**/test-results.trx'
 
+  docx-font-tests:
+    runs-on: windows-latest
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Setup .NET
+        uses: actions/setup-dotnet@v4
+        with:
+          dotnet-version: '9.0.x'
+
+      - name: Verify required DOCX fonts
+        shell: pwsh
+        run: |
+          $fontDirectory = Join-Path $env:WINDIR 'Fonts'
+          foreach ($font in @('cambria.ttc', 'cambriab.ttf', 'cambriaz.ttf', 'seguisym.ttf', 'seguiemj.ttf')) {
+            if (-not (Test-Path -LiteralPath (Join-Path $fontDirectory $font))) {
+              throw "Required DOCX test font is unavailable: $font"
+            }
+          }
+
+      - name: Test DOCX rendering with system fonts
+        run: dotnet test tests/MiniPdf.Tests/MiniPdf.Tests.csproj --configuration Release --filter "FullyQualifiedName~Docx|FullyQualifiedName~SystemFontTheoryAttributeTests" --verbosity normal --logger "trx;LogFileName=docx-font-results.trx"
+
+      - name: Upload DOCX test results
+        uses: actions/upload-artifact@v4
+        if: always()
+        with:
+          name: docx-font-test-results
+          path: '**/docx-font-results.trx'
+
   ai-security-scan:
     if: github.event_name == 'pull_request'
     runs-on: ubuntu-latest

CONTRIBUTING.md

@@ -37,6 +37,12 @@ dotnet build --configuration Release
 dotnet test tests/MiniPdf.Tests --configuration Release
 ```
 
+DOCX tests that measure installed Cambria or Segoe fonts report an explicit skip
+when their required fonts are unavailable. CI runs the portable suite on Ubuntu
+and a separate DOCX suite on Windows, where it verifies the required fonts
+before testing. Install these fonts through a licensed system/Office installation
+to run the font-dependent checks locally; do not add font binaries to the repository.
+
 ### Rust
 
 Install the stable Rust toolchain, then run:

FEATURE_PARITY.md

@@ -149,6 +149,24 @@ Word compatibility modes up to 14 align implicit left indents with first-cell
 content; modern implicit indents continue to align the table border.
 Inline images in table cells honor the paragraph's left, center, or right
 alignment in both body and header/footer tables.
+Cambria table cells use the same preferred-font advances for wrapping and
+row-height calculation, including header/footer tables. Auto-spaced body
+paragraphs do not carry the next text-line ascent into a following table.
+Large first-paragraph runs advance subsequent content from their rendered
+baseline rather than the smaller inherited paragraph-mark baseline.
+Cell line height distinguishes explicit paragraph-mark sizing from inherited
+defaults. Collapsed table borders contribute half their stroke widths to natural
+row heights and text insets; center/bottom alignment uses the remaining content
+area. Cambria/Calibri paragraph flow tracks font ascents across text, empty marks,
+and inline-image hosts. Auto-spaced image-only paragraphs add proportional font
+leading without scaling image height. Hard section starts collapse paragraph
+space-before against the closing section's space-after.
+Image-only paragraphs reclaim the next text ascent before body tables, while
+border-only empty marks reset text-ascent tracking. Courier transitions use
+Courier font metrics. Symbol cells account for the Segoe fallback's ascent and
+line height when the PDF writer selects that font.
+Mixed-format Cambria paragraphs measure each run's bold/italic font advances and
+spaces consistently for wrapping, run positioning, alignment, and run shading.
 
 ## XLSX Rendering Gaps
 

src/MiniPdf/DocxReader.cs

@@ -130,7 +130,7 @@
         var xpath = dataBinding.Attribute(W + "xpath")?.Value;
         var prefixMappings = dataBinding.Attribute(W + "prefixMappings")?.Value ?? "";
         if (string.IsNullOrEmpty(storeItemID) || string.IsNullOrEmpty(xpath)) return null;
         if (!ctx.XmlStores.TryGetValue(storeItemID, out var doc) || doc.Root == null) return null;
 
         var nsMgr = new System.Xml.XmlNamespaceManager(new System.Xml.NameTable());
         foreach (System.Text.RegularExpressions.Match m in System.Text.RegularExpressions.Regex.Matches(
@@ -871,7 +871,7 @@
         if (string.IsNullOrWhiteSpace(style))
             return values;
 
         foreach (var declaration in style.Split(';', StringSplitOptions.RemoveEmptyEntries))
         {
             var separator = declaration.IndexOf(':');
             if (separator <= 0)
@@ -912,6 +912,7 @@
         float spacingAfter = -1;
         bool spacingAfterExplicit = false;
         bool spacingBeforeExplicit = false;
+        float paragraphMarkFontSize = 0;
         float lineSpacing = 0;
         bool lineSpacingAbsolute = false;
         bool lineSpacingExact = false;
@@ -1183,7 +1184,10 @@
             {
                 var sz = rPr.Element(W + "sz")?.Attribute(W + "val")?.Value;
                 if (float.TryParse(sz, out var s))
+                {
                     fontSize = s / 2f; // half-points to points
+                    paragraphMarkFontSize = fontSize;
+                }
                 // Paragraph-level character spacing
                 var spacingEl2 = rPr.Element(W + "spacing");
                 if (spacingEl2 != null && int.TryParse(spacingEl2.Attribute(W + "val")?.Value, out var pcs))
@@ -1509,7 +1513,8 @@
             ListSuff: listSuff,
             ListTabStop: listTabStop,
             OutlineLevel: ReadOutlineLevel(pPr),
-            StyleIndentFirstLine: styleIndentFirstLineForListLabel);
+            StyleIndentFirstLine: styleIndentFirstLineForListLabel,
+            ParagraphMarkFontSize: paragraphMarkFontSize);
     }
 
     private static void EmitPageFieldPlaceholder(string instruction, List<DocxRun> runs, bool bold, bool italic, float fontSize, PdfColor? color)
@@ -5286,7 +5291,8 @@
     // not position the automatic-numbering suffix.
     float? ListTabStop = null,
     int OutlineLevel = -1,
-    float StyleIndentFirstLine = 0
+    float StyleIndentFirstLine = 0,
+    float ParagraphMarkFontSize = 0
 ) : DocxElement;
 
 /// <summary>Represents a single border edge.  Width=0 is a sentinel for an