Handle alternate Unicode name representation cXXX and fix #943 (#944)

UglyToad · Nov 24, 2024 · 2080424 · 2080424
1 parent bcc8cce
commit 2080424
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 0 deletions.
diff --git a/src/UglyToad.PdfPig.Fonts/GlyphList.cs b/src/UglyToad.PdfPig.Fonts/GlyphList.cs
@@ -152,6 +152,13 @@ public string NameToUnicode(string name)
 
                 unicode = char.ConvertFromUtf32(codePoint);
             }
+            else if (name.StartsWith("c", StringComparison.OrdinalIgnoreCase) && name.Length >= 3 && name.Length <= 4)
+            {
+                // name representation cXXX
+                var codePoint = int.Parse(name.AsSpanOrSubstring(1), NumberStyles.Integer, CultureInfo.InvariantCulture);
+                System.Diagnostics.Debug.Assert(codePoint > 0);
+                unicode = char.ConvertFromUtf32(codePoint);
+            }
             else
             {
                 return null;

diff --git a/src/UglyToad.PdfPig.Tests/Integration/GithubIssuesTests.cs b/src/UglyToad.PdfPig.Tests/Integration/GithubIssuesTests.cs
@@ -1,9 +1,32 @@
 namespace UglyToad.PdfPig.Tests.Integration
 {
     using Content;
+    using DocumentLayoutAnalysis.PageSegmenter;
+    using DocumentLayoutAnalysis.WordExtractor;
 
     public class GithubIssuesTests
     {
+        [Fact]
+        public void Issue943()
+        {
+            var path = IntegrationHelpers.GetDocumentPath("MOZILLA-10225-0.pdf");
+
+            using (var document = PdfDocument.Open(path))
+            {
+                var page = document.GetPage(1);
+                Assert.NotNull(page);
+
+                var letters = page.Letters;
+                Assert.NotNull(letters);
+
+                var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters);
+                var blocks = DocstrumBoundingBoxes.Instance.GetBlocks(words);
+
+                Assert.Equal("Rocket and Spacecraft Propulsion", blocks[0].TextLines[0].Text);
+                Assert.Equal("Principles, Practice and New Developments (Second Edition)", blocks[0].TextLines[1].Text);
+            }
+        }
+
         [Fact]
         public void Issue736()
         {