#483 make skip missing fonts even more resilient to nonsense files

UglyToad · Dec 11, 2022 · c8874c5 · c8874c5
1 parent 2aed996
commit c8874c5
Show file tree

Hide file tree

Showing 6 changed files with 64 additions and 11 deletions.
diff --git a/src/UglyToad.PdfPig.Fonts/CorruptCompressedDataException.cs b/src/UglyToad.PdfPig.Fonts/CorruptCompressedDataException.cs
@@ -0,0 +1,34 @@
+namespace UglyToad.PdfPig.Fonts
+{
+    using System;
+    using System.Runtime.Serialization;
+
+    /// <summary>
+    /// Thrown when a PDF contains an invalid compressed data stream.
+    /// </summary>
+    [Serializable]
+    public class CorruptCompressedDataException : Exception
+    {
+        /// <inheritdoc />
+        public CorruptCompressedDataException()
+        {
+        }
+
+        /// <inheritdoc />
+        public CorruptCompressedDataException(string message) : base(message)
+        {
+        }
+
+        /// <inheritdoc />
+        public CorruptCompressedDataException(string message, Exception inner) : base(message, inner)
+        {
+        }
+
+        /// <inheritdoc />
+        protected CorruptCompressedDataException(
+            SerializationInfo info,
+            StreamingContext context) : base(info, context)
+        {
+        }
+    }
+}
diff --git a/src/UglyToad.PdfPig/Content/IResourceStore.cs b/src/UglyToad.PdfPig/Content/IResourceStore.cs
@@ -6,7 +6,7 @@
 
     internal interface IResourceStore
     {
-        void LoadResourceDictionary(DictionaryToken resourceDictionary);
+        void LoadResourceDictionary(DictionaryToken resourceDictionary, InternalParsingOptions parsingOptions);
 
         /// <summary>
         /// Remove any named resources and associated state for the last resource dictionary loaded.

diff --git a/src/UglyToad.PdfPig/Content/ResourceStore.cs b/src/UglyToad.PdfPig/Content/ResourceStore.cs
@@ -33,7 +33,7 @@ public ResourceStore(IPdfTokenScanner scanner, IFontFactory fontFactory)
             this.fontFactory = fontFactory;
         }
 
-        public void LoadResourceDictionary(DictionaryToken resourceDictionary)
+        public void LoadResourceDictionary(DictionaryToken resourceDictionary, InternalParsingOptions parsingOptions)
         {
             lastLoadedFont = (null, null);
 
@@ -43,7 +43,7 @@ public void LoadResourceDictionary(DictionaryToken resourceDictionary)
             {
                 var fontDictionary = DirectObjectFinder.Get<DictionaryToken>(fontBase, scanner);
 
-                LoadFontDictionary(fontDictionary);
+                LoadFontDictionary(fontDictionary, parsingOptions);
             }
 
             if (resourceDictionary.TryGet(NameToken.Xobject, out var xobjectBase))
@@ -132,7 +132,7 @@ public void UnloadResourceDictionary()
             currentResourceState.Pop();
         }
 
-        private void LoadFontDictionary(DictionaryToken fontDictionary)
+        private void LoadFontDictionary(DictionaryToken fontDictionary, InternalParsingOptions parsingOptions)
         {
             lastLoadedFont = (null, null);
 
@@ -157,7 +157,18 @@ private void LoadFontDictionary(DictionaryToken fontDictionary)
                         continue;
                     }
 
-                    loadedFonts[reference] = fontFactory.Get(fontObject);
+                    try
+                    { 
+                        loadedFonts[reference] = fontFactory.Get(fontObject);
+                    }
+                    catch
+                    {
+                        if (!parsingOptions.SkipMissingFonts)
+                        {
+                            throw;
+                        }
+                    }
+
                 }
                 else if (pair.Value is DictionaryToken fd)
                 {

diff --git a/src/UglyToad.PdfPig/Filters/FlateFilter.cs b/src/UglyToad.PdfPig/Filters/FlateFilter.cs
@@ -1,5 +1,6 @@
 namespace UglyToad.PdfPig.Filters
 {
+    using Fonts;
     using System;
     using System.Collections.Generic;
     using System.IO;
@@ -79,10 +80,17 @@ private byte[] Decompress(byte[] input)
                 memoryStream.ReadByte();
                 memoryStream.ReadByte();
 
-                using (var deflate = new DeflateStream(memoryStream, CompressionMode.Decompress))
+                try
                 {
-                    deflate.CopyTo(output);
-                    return output.ToArray();
+                    using (var deflate = new DeflateStream(memoryStream, CompressionMode.Decompress))
+                    {
+                        deflate.CopyTo(output);
+                        return output.ToArray();
+                    }
+                }
+                catch (InvalidDataException ex)
+                {
+                    throw new CorruptCompressedDataException("Invalid Flate compressed stream encountered", ex);
                 }
             }
         }

diff --git a/src/UglyToad.PdfPig/Graphics/ContentStreamProcessor.cs b/src/UglyToad.PdfPig/Graphics/ContentStreamProcessor.cs
@@ -479,7 +479,7 @@ private void ProcessFormXObject(StreamToken formStream)
             var hasResources = formStream.StreamDictionary.TryGet<DictionaryToken>(NameToken.Resources, pdfScanner, out var formResources);
             if (hasResources)
             {
-                resourceStore.LoadResourceDictionary(formResources);
+                resourceStore.LoadResourceDictionary(formResources, parsingOptions);
             }
 
             // 1. Save current state.

diff --git a/src/UglyToad.PdfPig/Parser/PageFactory.cs b/src/UglyToad.PdfPig/Parser/PageFactory.cs
@@ -63,13 +63,13 @@ public Page Create(int number, DictionaryToken dictionary, PageTreeMembers pageT
             {
                 var resource = pageTreeMembers.ParentResources.Dequeue();
 
-                resourceStore.LoadResourceDictionary(resource);
+                resourceStore.LoadResourceDictionary(resource, parsingOptions);
                 stackDepth++;
             }
 
             if (dictionary.TryGet(NameToken.Resources, pdfScanner, out DictionaryToken resources))
             {
-                resourceStore.LoadResourceDictionary(resources);
+                resourceStore.LoadResourceDictionary(resources, parsingOptions);
                 stackDepth++;
             }