fix: replace UseAdvancedExtensions with selective pipeline to fix phantom table columns (#68)

csharp-version/src/MarkdownToDocx.Core/Markdown/MarkdigParser.cs

@@ -1,4 +1,5 @@
 using Markdig;
+using Markdig.Extensions.EmphasisExtras;
 using Markdig.Syntax;
 using MarkdownToDocx.Core.Interfaces;
 
@@ -10,9 +11,38 @@ namespace MarkdownToDocx.Core.Markdown;
 /// <param name="pipeline">Markdig pipeline configuration for parsing</param>
 public sealed class MarkdigParser(MarkdownPipeline? pipeline = null) : IMarkdownParser
 {
-    private readonly MarkdownPipeline _pipeline = pipeline ?? new MarkdownPipelineBuilder()
-        .UseAdvancedExtensions()
-        .Build();
+    private readonly MarkdownPipeline _pipeline = pipeline ?? BuildDefaultPipeline();
+
+    /// <summary>
+    /// Builds the default Markdig pipeline with selective extensions.
+    /// Uses explicit extension registration instead of UseAdvancedExtensions() to avoid
+    /// the subscript extension (~), which causes Markdig to misparse table cell boundaries
+    /// when tilde appears inside bold markers (e.g., **~$1.50**).
+    /// </summary>
+    private static MarkdownPipeline BuildDefaultPipeline()
+    {
+        return new MarkdownPipelineBuilder()
+            .UseAlertBlocks()
+            .UseAbbreviations()
+            .UseAutoIdentifiers()
+            .UseCitations()
+            .UseCustomContainers()
+            .UseDefinitionLists()
+            .UseEmphasisExtras(EmphasisExtraOptions.Strikethrough)
+            .UseFigures()
+            .UseFooters()
+            .UseFootnotes()
+            .UseGridTables()
+            .UseMathematics()
+            .UseMediaLinks()
+            .UsePipeTables()
+            .UseListExtras()
+            .UseTaskLists()
+            .UseDiagrams()
+            .UseAutoLinks()
+            .UseGenericAttributes()
+            .Build();
+    }
 
     /// <inheritdoc/>
     public MarkdownDocument Parse(string markdown)

csharp-version/tests/MarkdownToDocx.Tests/Unit/TableBoldTildeRegressionTests.cs

@@ -0,0 +1,128 @@
+using FluentAssertions;
+using Markdig;
+using Markdig.Extensions.Tables;
+using Markdig.Syntax;
+using MarkdownToDocx.Core.Markdown;
+using Xunit;
+
+namespace MarkdownToDocx.Tests.Unit;
+
+/// <summary>
+/// Regression tests for issue #68: Bold markers with tilde inside table cells
+/// cause extra phantom columns due to Markdig subscript extension.
+/// </summary>
+public class TableBoldTildeRegressionTests
+{
+    private static readonly MarkdownPipeline Pipeline = new MarkdownPipelineBuilder()
+        .UseAdvancedExtensions()
+        .Build();
+
+    private static Table ParseTable(string markdown)
+    {
+        var doc = Markdown.Parse(markdown, Pipeline);
+        return doc.Descendants<Table>().FirstOrDefault()
+            ?? throw new InvalidOperationException("No table found in markdown");
+    }
+
+    [Fact]
+    public void Extract_BoldWithTildeInCells_ShouldNotCreatePhantomColumns()
+    {
+        // Arrange — exact reproduction from issue #68
+        const string md = """
+            | Component | Per query | Per 1,000 queries |
+            |-----------|-----------|-------------------|
+            | Embedding | ~$0.00001 | ~$0.01 |
+            | LLM       | ~$0.0015  | ~$1.50 |
+            | **Total** | **~$0.0015** | **~$1.50** |
+            """;
+
+        // Act — use MarkdigParser (which now uses the fixed pipeline)
+        var parser = new MarkdigParser();
+        var doc = parser.Parse(md);
+        var table = doc.Descendants<Table>().First();
+        var data = TableExtractor.Extract(table);
+
+        // Assert
+        data.ColumnCount.Should().Be(3, "bold+tilde should not create phantom columns");
+        data.Rows.Should().HaveCount(4); // 1 header + 3 body rows
+        data.Rows.Should().OnlyContain(r => r.Cells.Count == 3);
+    }
+
+    [Fact]
+    public void Extract_BoldWithTildeInCells_ShouldPreserveBoldFormatting()
+    {
+        const string md = """
+            | A | B |
+            |---|---|
+            | **~$1.50** | normal |
+            """;
+
+        var parser = new MarkdigParser();
+        var doc = parser.Parse(md);
+        var table = doc.Descendants<Table>().First();
+        var data = TableExtractor.Extract(table);
+
+        data.ColumnCount.Should().Be(2);
+        var boldCell = data.Rows[1].Cells[0];
+        boldCell.Runs.Should().Contain(r => r.Bold && r.Text.Contains("$1.50"));
+    }
+
+    [Fact]
+    public void Extract_MultipleRowsWithBoldTilde_ShouldAllHaveCorrectColumnCount()
+    {
+        // From issue #68: "Real-world impact" example
+        const string md = """
+            | Item | Cost | Monthly |
+            |------|------|---------|
+            | **Total without rerank** | **~$0.0015** | **~$1.50** |
+            | **Total with rerank**    | **~$0.0019** | **~$1.90** |
+            """;
+
+        var parser = new MarkdigParser();
+        var doc = parser.Parse(md);
+        var table = doc.Descendants<Table>().First();
+        var data = TableExtractor.Extract(table);
+
+        data.ColumnCount.Should().Be(3);
+        data.Rows.Should().OnlyContain(r => r.Cells.Count == 3);
+    }
+
+    [Fact]
+    public void Extract_StrikethroughInTable_ShouldStillWork()
+    {
+        // Ensure ~~strikethrough~~ is not broken by the fix
+        const string md = """
+            | Feature | Status |
+            |---------|--------|
+            | ~~Removed~~ | Done |
+            """;
+
+        var parser = new MarkdigParser();
+        var doc = parser.Parse(md);
+        var table = doc.Descendants<Table>().First();
+        var data = TableExtractor.Extract(table);
+
+        data.ColumnCount.Should().Be(2);
+        data.Rows[1].Cells[0].Runs.Should().HaveCount(1);
+    }
+
+    [Fact]
+    public void Extract_TildeWithoutBold_ShouldPreserveLiteralTilde()
+    {
+        const string md = """
+            | Value |
+            |-------|
+            | ~$0.01 |
+            """;
+
+        var parser = new MarkdigParser();
+        var doc = parser.Parse(md);
+        var table = doc.Descendants<Table>().First();
+        var data = TableExtractor.Extract(table);
+
+        data.ColumnCount.Should().Be(1);
+        var text = string.Join("", data.Rows[1].Cells[0].Runs.Select(r => r.Text));
+        text.Should().Contain("~");
+        text.Should().Contain("$0.01");
+    }
+}