Address #672 to ignore errors while reading the descriptor file in CidFontFactory

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

@@ -0,0 +1,74 @@
+using System.Linq;
+using UglyToad.PdfPig.DocumentLayoutAnalysis.PageSegmenter;
+using UglyToad.PdfPig.DocumentLayoutAnalysis.WordExtractor;
+using Xunit;
+
+namespace UglyToad.PdfPig.Tests.Integration
+{
+    public class OpenTypeFontTests
+    {
+        [Fact]
+        public void Issue672()
+        {
+            // NB: The issue is actually not fully fixed: the change are just allowing
+            // to parse the document and get the text without error
+            // but the embedded font data is not properly parsed.
+            // It seems the font bytes are incorrectly parsed using the TrueTypeFontParser
+            // and are actually parsable with CompactFontFormatParser, but with some errors though.
+
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("Why.does.this.not.work")))
+            {
+                var page = document.GetPage(1);
+
+                var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters);
+
+                var lines = DocstrumBoundingBoxes.Instance.GetBlocks(words).SelectMany(b => b.TextLines).ToArray();
+
+                Assert.Equal(3, lines.Length);
+
+                Assert.Equal("THIS TEST SEEMS TO BREAK THE PARSER....", lines[0].Text);
+                Assert.Equal("This is just some test text.", lines[1].Text);
+                Assert.Equal("SO DOES THIS", lines[2].Text);
+            }
+        }
+
+        [Fact]
+        public void Issue672ok()
+        {
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("Test.Doc")))
+            {
+                var page = document.GetPage(1);
+
+                var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters);
+
+                var lines = DocstrumBoundingBoxes.Instance.GetBlocks(words).SelectMany(b => b.TextLines).ToArray();
+
+                Assert.Equal(4, lines.Length);
+
+                Assert.Equal("This is just a bunch of boring text...", lines[0].Text);
+                Assert.Equal("THIS IS SOME SEMPLICITA PRO FONT", lines[1].Text);
+                Assert.Equal("Hopefully font that are not embedded on the server.", lines[2].Text);
+                Assert.Equal("And a bit of Verdana for good measure.", lines[3].Text);
+            }
+        }
+
+        [Fact]
+        public void So74165171()
+        {
+            // https://stackoverflow.com/questions/74165171/embedded-opentype-cff-font-in-a-pdf-shows-strange-behaviour-in-some-viewers
+
+            // Adding this test case as the OpenType font is correctly parsed using TrueTypeFontParser
+            // It seems there are further issues with the extracted test (also the case in Acrobat Reader).
+            // Out of scope for the moment
+
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("test-2_so_74165171")))
+            {
+                var page = document.GetPage(1);
+
+                var words = NearestNeighbourWordExtractor.Instance.GetWords(page.Letters).ToArray();
+
+                Assert.Equal(2, words.Length);
+            }
+        }
+    }
+}

src/UglyToad.PdfPig/Parser/PdfDocumentFactory.cs

@@ -158,7 +158,11 @@
 
             pdfScanner.UpdateEncryptionHandler(encryptionHandler);
 
-            var cidFontFactory = new CidFontFactory(pdfScanner, filterProvider);
+            var cidFontFactory = new CidFontFactory(
+                parsingOptions.Logger,
+                pdfScanner,
+                filterProvider);
+
             var encodingReader = new EncodingReader(pdfScanner);
 
             var type0Handler = new Type0FontHandler(

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

@@ -13,21 +13,24 @@
     using PdfPig.Parser.Parts;
     using Tokenization.Scanner;
     using Tokens;
+    using UglyToad.PdfPig.Logging;
     using Util;
 
     internal class CidFontFactory
     {
         private readonly ILookupFilterProvider filterProvider;
         private readonly IPdfTokenScanner pdfScanner;
+        private readonly ILog logger;
 
-        public CidFontFactory(IPdfTokenScanner pdfScanner, ILookupFilterProvider filterProvider)
+        public CidFontFactory(ILog log, IPdfTokenScanner pdfScanner, ILookupFilterProvider filterProvider)
         {
+            this.logger = log;
             this.pdfScanner = pdfScanner;
             this.filterProvider = filterProvider;
         }
 
         public ICidFont Generate(DictionaryToken dictionary)
-        { 
+        {
             var type = dictionary.GetNameOrDefault(NameToken.Type);
             if (!NameToken.Font.Equals(type))
             {
@@ -50,7 +53,15 @@
                 descriptor = FontDescriptorFactory.Generate(descriptorDictionary, pdfScanner);
             }
 
-            var fontProgram = ReadDescriptorFile(descriptor);
+            ICidFontProgram fontProgram = null;
+            try
+            {
+                fontProgram = ReadDescriptorFile(descriptor);
+            }
+            catch (Exception ex)
+            {
+                logger.Error($"Invalid descriptor in CID font named '{descriptor?.FontName}': {ex.Message}.");
+            }
 
             var baseFont = dictionary.GetNameOrDefault(NameToken.BaseFont);
 
@@ -74,25 +85,7 @@
 
         private bool TryGetFontDescriptor(DictionaryToken dictionary, out DictionaryToken descriptorDictionary)
         {
-            descriptorDictionary = null;
-
-            if (!dictionary.TryGet(NameToken.FontDescriptor, out var baseValue))
-            {
-                return false;
-            }
-
-            try
-            {
-                var descriptor = DirectObjectFinder.Get<DictionaryToken>(baseValue, pdfScanner);
-
-                descriptorDictionary = descriptor;
-            }
-            catch
-            {
-                return false;
-            }
-
-            return true;
+            return dictionary.TryGet(NameToken.FontDescriptor, pdfScanner, out descriptorDictionary);
         }
 
         private ICidFontProgram ReadDescriptorFile(FontDescriptor descriptor)
@@ -267,14 +260,9 @@
                 throw new InvalidFontFormatException($"No CID System Info was found in the CID Font dictionary: {dictionary}");
             }
 
-            if (cidEntry is DictionaryToken cidDictionary)
+            if (!(cidEntry is DictionaryToken cidDictionary))
             {
-
-            }
-            else
-            {
-                cidDictionary =
-                    DirectObjectFinder.Get<DictionaryToken>(cidEntry, pdfScanner);
+                cidDictionary = DirectObjectFinder.Get<DictionaryToken>(cidEntry, pdfScanner);
             }
 
             var registry = SafeKeyAccess(cidDictionary, NameToken.Registry);