Preserve line break pre for OSX/Windows

onizet · Jul 23, 2024 · 0a27948 · 0a27948
1 parent 042efa1
commit 0a27948
Show file tree

Hide file tree

Showing 5 changed files with 53 additions and 7 deletions.
diff --git a/examples/Demo/Program.cs b/examples/Demo/Program.cs
@@ -17,7 +17,15 @@ static async Task Main(string[] args)
             const string filename = "test.docx";
             string html = ResourceHelper.GetString("Resources.CompleteRunTest.html");
             if (File.Exists(filename)) File.Delete(filename);
+const string preformattedText = @"
+              ^__^
+              (oo)\_______
+              (__)\       )\/\
+                  ||----w |
+                  ||     ||";
 
+            html = @$"<pre role='img' aria-label='ASCII COW'>
+{preformattedText}</pre>";
             using (MemoryStream generatedDocument = new MemoryStream())
             {
                 // Uncomment and comment the second using() to open an existing template document

diff --git a/src/Html2OpenXml/Expressions/ParsingContext.cs b/src/Html2OpenXml/Expressions/ParsingContext.cs
@@ -34,7 +34,7 @@ sealed class ParsingContext(HtmlConverter converter, MainDocumentPart mainPart)
     private Dictionary<string, object?> propertyBag = [];
 
     /// <summary>Whether the text content should preserve the line breaks.</summary>
-    public bool PreverseLinebreaks { get; set; }
+    public bool PreserveLinebreaks { get; set; }
 
     /// <summary>Whether the text content should collapse the whitespaces.</summary>
     public bool CollapseWhitespaces { get; set; } = true;

diff --git a/src/Html2OpenXml/Expressions/PreElementExpression.cs b/src/Html2OpenXml/Expressions/PreElementExpression.cs
@@ -26,7 +26,7 @@ public override IEnumerable<OpenXmlElement> Interpret(ParsingContext context)
     {
         ComposeStyles(context);
         var childContext = context.CreateChild(this);
-        childContext.PreverseLinebreaks = true;
+        childContext.PreserveLinebreaks = true;
         childContext.CollapseWhitespaces = false;
         var childElements = Interpret(childContext, node.ChildNodes);
 

diff --git a/src/Html2OpenXml/Expressions/TextExpression.cs b/src/Html2OpenXml/Expressions/TextExpression.cs
@@ -31,7 +31,7 @@ public override IEnumerable<OpenXmlElement> Interpret (ParsingContext context)
         string text = node.TextContent.Normalize();
         if (text.Trim().Length == 0) return [];
 
-        if (!context.PreverseLinebreaks)
+        if (!context.PreserveLinebreaks)
             text = text.CollapseLineBreaks();
         if (context.CollapseWhitespaces && text[0].IsWhiteSpaceCharacter() &&
             node.PreviousSibling is IHtmlImageElement)
@@ -41,9 +41,36 @@ public override IEnumerable<OpenXmlElement> Interpret (ParsingContext context)
         else if (context.CollapseWhitespaces)
             text = text.CollapseAndStrip();
 
-        Run run = new(
-            new Text(text)
-        );
+        if (!context.PreserveLinebreaks)
+            return [new Run(new Text(text))];
+
+        var run = new Run();
+        char[] chars = text.ToCharArray();
+        int shift = 0, c = 0;
+        bool wasCR = false; // avoid adding 2 breaks for \r\n
+        for ( ; c < chars.Length ; c++)
+        {
+            if (!chars[c].IsLineBreak())
+            {
+                wasCR = false;
+                continue;
+            }
+
+            if (wasCR) continue;
+            wasCR = chars[c] == Symbols.CarriageReturn;
+
+            if (c > 1)
+            {
+                run.Append(new Text(new string(chars, shift, c - shift)) 
+                    { Space = SpaceProcessingModeValues.Preserve });
+                run.Append(new Break());
+            }
+            shift = c + 1;
+        }
+
+        if (c > shift)
+            run.Append(new Text(new string(chars, shift, c - shift)) 
+                { Space = SpaceProcessingModeValues.Preserve });
 
         return [run];
     }

diff --git a/test/HtmlToOpenXml.Tests/TableTests.cs b/test/HtmlToOpenXml.Tests/TableTests.cs
@@ -387,11 +387,22 @@ public void ParsePreAsTable()
             var cell = cells.First();
             Assert.Multiple(() =>
             {
-                Assert.That(cell.InnerText, Is.EqualTo(preformattedText));
                 Assert.That(cell.TableCellProperties?.TableCellBorders.ChildElements.Count(), Is.EqualTo(4));
                 Assert.That(cell.TableCellProperties?.TableCellBorders.ChildElements, Has.All.InstanceOf<BorderType>());
                 Assert.That(cell.TableCellProperties?.TableCellBorders.Elements<BorderType>().All(b => b.Val.Value == BorderValues.Single), Is.True);
             });
+
+            var run = cell.GetFirstChild<Paragraph>()?.GetFirstChild<Run>();
+            Assert.Multiple(() =>
+            {
+                var odds = run.ChildElements.Where((item, index) => index % 2 != 0);
+                Assert.That(odds, Has.All.TypeOf<Break>());
+                Assert.That(run.ChildElements.ElementAt(0).InnerText, Is.EqualTo("              ^__^"));
+                Assert.That(run.ChildElements.ElementAt(2).InnerText, Is.EqualTo("              (oo)\\_______"));
+                Assert.That(run.ChildElements.ElementAt(4).InnerText, Is.EqualTo("              (__)\\       )\\/\\"));
+                Assert.That(run.ChildElements.ElementAt(6).InnerText, Is.EqualTo("                  ||----w |"));
+                Assert.That(run.ChildElements.ElementAt(8).InnerText, Is.EqualTo("                  ||     ||"));
+            });
         }
 
         [Test]