Handle TrueType case in CidFontFactory where the font is CFF, implement missing members in PdfCidCompactFontFormatFont and fix #554

src/UglyToad.PdfPig.Tests/Integration/GithubIssuesTests.cs

@@ -7,6 +7,26 @@
 
     public class GithubIssuesTests
     {
+        [Fact]
+        public void Issue554()
+        {
+            var path = IntegrationHelpers.GetSpecificTestDocumentPath("2022.pdf");
+
+            using (var document = PdfDocument.Open(path, new ParsingOptions() { UseLenientParsing = true }))
+            {
+                for (int p = 1; p <= document.NumberOfPages; p++)
+                {
+                    var page = document.GetPage(p);
+                    Assert.NotNull(page.Letters);
+
+                    if (p < document.NumberOfPages)
+                    {
+                        Assert.NotEmpty(page.Letters);
+                    }
+                }
+            }
+        }
+
         [Fact]
         public void Issue822()
         {

src/UglyToad.PdfPig/PdfFonts/CidFonts/PdfCidCompactFontFormatFont.cs

@@ -28,7 +28,8 @@
 
             FontDetails WithWeightValues(bool isBold, int weight) => new FontDetails(null, isBold, weight, font.ItalicAngle != 0);
 
-            return (font.Weight?.ToLowerInvariant()) switch {
+            return (font.Weight?.ToLowerInvariant()) switch
+            {
                 "light"    => WithWeightValues(false, 300),
                 "semibold" => WithWeightValues(true, 600),
                 "bold"     => WithWeightValues(true, FontDetails.BoldWeight),
@@ -61,17 +62,34 @@
 
         public bool TryGetBoundingBox(int characterIdentifier, Func<int, int?> characterCodeToGlyphId, out PdfRectangle boundingBox)
         {
-            throw new NotImplementedException();
+            var font = GetFont();
+
+            int? glyphId = characterCodeToGlyphId.Invoke(characterIdentifier);
+
+            string name = glyphId.HasValue
+                ? GetCharacterName(glyphId.Value)
+                : GetCharacterName(characterIdentifier);
+
+            PdfRectangle? tempBbox = font.GetCharacterBoundingBox(name);
+            if (tempBbox.HasValue)
+            {
+                boundingBox = tempBbox.Value;
+                return true;
+            }
+
+            boundingBox = default;
+            return false;
         }
 
         public bool TryGetBoundingAdvancedWidth(int characterIdentifier, Func<int, int?> characterCodeToGlyphId, out double width)
         {
-            throw new NotImplementedException();
+            return TryGetBoundingAdvancedWidth(characterIdentifier, out width);
         }
 
         public bool TryGetBoundingAdvancedWidth(int characterIdentifier, out double width)
         {
-            throw new NotImplementedException();
+            width = double.NaN;
+            return false;
         }
 
         public int GetFontMatrixMultiplier()
@@ -136,7 +154,25 @@
 
         public bool TryGetPath(int characterCode, Func<int, int?> characterCodeToGlyphId, out IReadOnlyList<PdfSubpath> path)
         {
-            throw new NotImplementedException();
+            path = null;
+
+            int? glyphId = characterCodeToGlyphId.Invoke(characterCode);
+
+            string characterName = glyphId.HasValue
+                ? GetCharacterName(glyphId.Value)
+                : GetCharacterName(characterCode);
+
+            if (string.Equals(characterName, GlyphList.NotDefined, StringComparison.OrdinalIgnoreCase))
+            {
+                return false;
+            }
+
+            if (GetFont().TryGetPath(characterName, out path))
+            {
+                return true;
+            }
+
+            return false;
         }
     }
 }

src/UglyToad.PdfPig/PdfFonts/Parser/Parts/CidFontFactory.cs

@@ -109,6 +109,13 @@
             {
                 case DescriptorFontFile.FontFileType.TrueType:
                     {
+                        if (IsTrueTypeCff(fontFile.Span))
+                        {
+                            logger.Warn("The CID TrueType font has the signature of a CFF font. Using CID CFF instead.");
+                            var font = CompactFontFormatParser.Parse(new CompactFontFormatData(fontFile));
+                            return new PdfCidCompactFontFormatFont(font);
+                        }
+                        
                         var input = new TrueTypeDataBytes(new MemoryInputBytes(fontFile));
                         var ttf = TrueTypeFontParser.Parse(input);
                         return new PdfCidTrueTypeFont(ttf);
@@ -328,5 +335,25 @@
 
             return string.Empty;
         }
+
+        private static bool IsTrueTypeCff(ReadOnlySpan<byte> data)
+        {
+            if (data.Length < 4)
+            {
+                return false;
+            }
+            
+            // See https://docs.fileformat.com/font/cff/
+            // https://adobe-type-tools.github.io/font-tech-notes/pdfs/5176.CFF.pdf
+            byte major = data[0]; // Major version
+            byte minor = data[1]; // Minor version
+            byte hdrSize = data[2]; // Header size 
+            byte offSize = data[3]; // Absolute offset
+
+            return major == 0x01 &&
+                   minor == 0x00 &&
+                   hdrSize >= 0x04 &&
+                   offSize >= 0x01 && offSize <= 0x04;
+        }
     }
 }