Fix #807 (#808)

UglyToad · Mar 23, 2024 · e789691 · e789691
1 parent 69e2b7b
commit e789691
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 11 deletions.
diff --git a/src/UglyToad.PdfPig.Tests/Integration/Documents/Diacritics_export.pdf b/src/UglyToad.PdfPig.Tests/Integration/Documents/Diacritics_export.pdf
diff --git a/src/UglyToad.PdfPig.Tests/Integration/Type1FontSimpleTests.cs b/src/UglyToad.PdfPig.Tests/Integration/Type1FontSimpleTests.cs
@@ -0,0 +1,24 @@
+namespace UglyToad.PdfPig.Tests.Integration
+{
+    using System.Linq;
+
+    public class Type1FontSimpleTests
+    {
+        [Fact]
+        public void Issue807()
+        {
+            var file = IntegrationHelpers.GetDocumentPath("Diacritics_export.pdf");
+
+            using (var document = PdfDocument.Open(file))
+            {
+                var page = document.GetPage(1);
+                var words = page.GetWords().ToArray();
+
+                Assert.Equal(3, words.Length);
+                Assert.Equal("Espinosa", words[0].Text);
+                Assert.Equal("Spínola", words[1].Text);
+                Assert.Equal("Moraña,", words[2].Text);
+            }
+        }
+    }
+}
diff --git a/src/UglyToad.PdfPig/PdfFonts/Simple/Type1FontSimple.cs b/src/UglyToad.PdfPig/PdfFonts/Simple/Type1FontSimple.cs
@@ -90,14 +90,13 @@ public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
 
         public bool TryGetUnicode(int characterCode, [NotNullWhen(true)] out string? value)
         {
-            if (toUnicodeCMap.CanMapToUnicode)
+            value = null;
+            if (toUnicodeCMap.CanMapToUnicode && toUnicodeCMap.TryGet(characterCode, out value))
             {
-                return toUnicodeCMap.TryGet(characterCode, out value);
+                return true;
             }
 
-            value = null;
-
-            if (encoding == null)
+            if (encoding is null)
             {
                 try
                 {
@@ -132,7 +131,7 @@ public bool TryGetUnicode(int characterCode, [NotNullWhen(true)] out string? val
                 return false;
             }
 
-            return true;
+            return value is not null;
         }
 
         public CharacterBoundingBox GetBoundingBox(int characterCode)

diff --git a/src/UglyToad.PdfPig/PdfFonts/Simple/Type3Font.cs b/src/UglyToad.PdfPig/PdfFonts/Simple/Type3Font.cs
@@ -52,16 +52,29 @@ public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
 
         public bool TryGetUnicode(int characterCode, [NotNullWhen(true)] out string? value)
         {
-            if (toUnicodeCMap.CanMapToUnicode)
+            value = null;
+
+            if (toUnicodeCMap.CanMapToUnicode && toUnicodeCMap.TryGet(characterCode, out value))
             {
-                return toUnicodeCMap.TryGet(characterCode, out value);
+                return true;
             }
 
-            var name = encoding.GetName(characterCode);
+            if (encoding is null)
+            {
+                return false;
+            }
 
-            value = GlyphList.AdobeGlyphList.NameToUnicode(name);
+            try
+            {
+                var name = encoding.GetName(characterCode);
+                value = GlyphList.AdobeGlyphList.NameToUnicode(name);
+            }
+            catch
+            {
+                return false;
+            }
 
-            return true;
+            return value is not null;
         }
 
         public CharacterBoundingBox GetBoundingBox(int characterCode)