fix bugs with reading documents from microsoft print to pdf

src/UglyToad.Pdf.Tests/Integration/LocalTests.cs

@@ -10,6 +10,7 @@
         [Fact]
         public void Tests()
         {
+            
         }
     }
 }

src/UglyToad.Pdf.Tests/Integration/OldGutnishTests.cs

@@ -0,0 +1,41 @@
+namespace UglyToad.Pdf.Tests.Integration
+{
+    using System;
+    using System.IO;
+    using Xunit;
+
+    public class OldGutnishTests
+    {
+        private static string GetFilename()
+        {
+            var documentFolder = Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Integration", "Documents"));
+
+            return Path.Combine(documentFolder, "Old Gutnish Internet Explorer.pdf");
+        }
+
+        [Fact]
+        public void HasCorrectNumberOfPages()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                Assert.Equal(3, document.NumberOfPages);
+            }
+        }
+
+        [Fact]
+        public void HasCorrectContentAfterReadingPreviousPage()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                var page = document.GetPage(1);
+
+                Assert.Contains("Þissi þieluar hafþi ann sun sum hit hafþi. En hafþa cuna hit huita stierna", page.Text);
+
+                page = document.GetPage(2);
+
+                Assert.Contains("Greipur sem annar hét; og Gunnfjón sá þriðji", page.Text);
+            }
+        }
+
+    }
+}

src/UglyToad.Pdf/Fonts/Parser/Handlers/Type0FontHandler.cs

@@ -38,16 +38,20 @@
 
             if (TryGetFirstDescendant(dictionary, out var descendantObject))
             {
-                var parsed = DirectObjectFinder.Find<PdfDictionary>(descendantObject, pdfObjectParser, reader, isLenientParsing);
+                PdfDictionary descendantFontDictionary;
 
-                if (parsed is PdfDictionary descendantFontDictionary)
+                if (descendantObject is CosObject obj)
                 {
-                    cidFont = ParseDescendant(descendantFontDictionary, reader, isLenientParsing);
+                    var parsed = DirectObjectFinder.Find<PdfDictionary>(obj, pdfObjectParser, reader, isLenientParsing);
+
+                    descendantFontDictionary = parsed;
                 }
                 else
                 {
-                    throw new InvalidFontFormatException("Expected to find a Descendant Font dictionary, instead it was: " + parsed);
+                    descendantFontDictionary = (PdfDictionary) descendantObject;
                 }
+
+                cidFont = ParseDescendant(descendantFontDictionary, reader, isLenientParsing);
             }
             else
             {
@@ -76,7 +80,7 @@
             return font;
         }
 
-        private static bool TryGetFirstDescendant(PdfDictionary dictionary, out CosObject descendant)
+        private static bool TryGetFirstDescendant(PdfDictionary dictionary, out CosBase descendant)
         {
             descendant = null;
 
@@ -91,9 +95,21 @@
                 return true;
             }
 
-            if (value is COSArray array && array.Count > 0 && array.get(0) is CosObject objArr)
+            if (value is COSArray array && array.Count > 0)
             {
+                if (array.get(0) is CosObject objArr)
+                {
                 descendant = objArr;
+                }
+                else if (array.get(0) is PdfDictionary dict)
+                {
+                    descendant = dict;
+                }
+                else
+                {
+                    return false;
+                }
+
                 return true;
             }
 

src/UglyToad.Pdf/Fonts/Parser/Parts/CidFontFactory.cs

@@ -75,6 +75,12 @@
 
             if (!dictionary.TryGetValue(CosName.FONT_DESC, out var baseValue) || !(baseValue is CosObject obj))
             {
+                if (baseValue is PdfDictionary baseDictionary)
+                {
+                    descriptorDictionary = baseDictionary;
+                    return true;
+                }
+
                 return false;
             }
 
@@ -239,11 +245,33 @@
                 throw new InvalidFontFormatException($"No CID System Info was found in the CID Font dictionary: {dictionary}");
             }
 
-            var registry = (CosString) cidDictionary.GetItemOrDefault(CosName.REGISTRY);
-            var ordering = (CosString)cidDictionary.GetItemOrDefault(CosName.ORDERING);
+            var registry = SafeKeyAccess(cidDictionary, CosName.REGISTRY, reader, isLenientParsing);
+            var ordering = SafeKeyAccess(cidDictionary, CosName.ORDERING, reader, isLenientParsing);
             var supplement = cidDictionary.GetIntOrDefault(CosName.SUPPLEMENT, 0);
 
             return new CharacterIdentifierSystemInfo(registry.GetAscii(), ordering.GetAscii(), supplement);
         }
+
+        private CosString SafeKeyAccess(PdfDictionary dictionary, CosName keyName, IRandomAccessRead reader, bool isLenientParsing)
+        {
+            var item = dictionary.GetItemOrDefault(keyName);
+
+            if (item == null)
+            {
+                return new CosString(string.Empty);
+            }
+
+            if (item is CosString str)
+            {
+                return str;
+            }
+
+            if (item is CosObject obj)
+            {
+                return DirectObjectFinder.Find<CosString>(obj, pdfObjectParser, reader, isLenientParsing);
+            }
+
+            return new CosString(string.Empty);
+        }
     }
 }