Check and handle circular references when processing XObject forms and fix #671

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

@@ -9,6 +9,10 @@
     public class IntegrationDocumentTests
     {
         private static readonly Lazy<string> DocumentFolder = new Lazy<string>(() => Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Integration", "Documents")));
+        private static readonly HashSet<string> _documentsToIgnore = new HashSet<string>()
+        {
+            "issue_671.pdf"
+        };
 
         [Theory]
         [MemberData(nameof(GetAllDocuments))]
@@ -101,7 +105,7 @@ public static IEnumerable<object[]> GetAllDocuments
                 var files = Directory.GetFiles(DocumentFolder.Value, "*.pdf");
 
                 // Return the shortname so we can see it in the test explorer.
-                return files.Select(x => new object[] { Path.GetFileName(x) });
+                return files.Where(x => !_documentsToIgnore.Any(i => x.EndsWith(i))).Select(x => new object[] { Path.GetFileName(x) });
             }
         }
     }

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

@@ -0,0 +1,37 @@
+namespace UglyToad.PdfPig.Tests.Integration
+{
+    using UglyToad.PdfPig.Core;
+    using Xunit;
+
+    public class XObjectFormTests
+    {
+        [Fact]
+        public void CanReadDocumentWithoutStackOverflowIssue671()
+        {
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("issue_671")))
+            {
+                var page = document.GetPage(1);
+            }
+        }
+
+        [Fact]
+        public void CanReadDocumentThrowsIssue671()
+        {
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("issue_671"), ParsingOptions.LenientParsingOff))
+            {
+                var exception = Assert.Throws<PdfDocumentFormatException>(() => document.GetPage(1));
+                Assert.Contains("is referencing itself which can cause unexpected behaviour", exception.Message);
+            }
+        }
+
+        [Fact]
+        public void CanReadDocumentMOZILLA_3136_0()
+        {
+            // This document does not actually contain circular references
+            using (var document = PdfDocument.Open(IntegrationHelpers.GetDocumentPath("MOZILLA-3136-0"), ParsingOptions.LenientParsingOff))
+            {
+                var page = document.GetPage(1);
+            }
+        }
+    }
+}

src/UglyToad.PdfPig/Graphics/ContentStreamProcessor.cs

@@ -486,15 +486,15 @@ public void ApplyXObject(NameToken xObjectName)
             }
             else if (subType.Equals(NameToken.Form))
             {
-                ProcessFormXObject(xObjectStream);
+                ProcessFormXObject(xObjectStream, xObjectName);
             }
             else
             {
                 throw new InvalidOperationException($"XObject encountered with unexpected SubType {subType}. {xObjectStream.StreamDictionary}.");
             }
         }
 
-        private void ProcessFormXObject(StreamToken formStream)
+        private void ProcessFormXObject(StreamToken formStream, NameToken xObjectName)
         {
             /*
              * When a form XObject is invoked the following should happen:
@@ -603,6 +603,20 @@ private void ProcessFormXObject(StreamToken formStream)
             // 3. We don't respect clipping currently.
 
             // 4. Paint the objects.
+            bool hasCircularReference = HasFormXObjectCircularReference(formStream, xObjectName, operations);
+            if (hasCircularReference)
+            {
+                if (parsingOptions.UseLenientParsing)
+                {
+                    operations = operations.Where(o => o is not InvokeNamedXObject xo || xo.Name != xObjectName).ToArray();
+                    parsingOptions.Logger.Warn($"An XObject form named '{xObjectName}' is referencing itself which can cause unexpected behaviour. The self reference was removed from the operations before further processing.");
+                }
+                else
+                {
+                    throw new PdfDocumentFormatException($"An XObject form named '{xObjectName}' is referencing itself which can cause unexpected behaviour.");
+                }
+            }
+
             ProcessOperations(operations);
 
             // 5. Restore saved state.
@@ -614,6 +628,20 @@ private void ProcessFormXObject(StreamToken formStream)
             }
         }
 
+        /// <summary>
+        /// Check for circular reference in the XObject form.
+        /// </summary>
+        /// <param name="formStream">The original form stream.</param>
+        /// <param name="xObjectName">The form's name.</param>
+        /// <param name="operations">The form operations parsed from original form stream.</param>
+        private bool HasFormXObjectCircularReference(StreamToken formStream, NameToken xObjectName, IReadOnlyList<IGraphicsStateOperation> operations)
+        {
+            return xObjectName != null
+                && operations.OfType<InvokeNamedXObject>()?.Any(o => o.Name == xObjectName) == true // operations contain another form with same name
+                && resourceStore.TryGetXObject(xObjectName, out var result)
+                && result.Data.SequenceEqual(formStream.Data); // The form contained in the operations has identical data to current form
+        }
+
         public void BeginSubpath()
         {
             if (CurrentPath == null)