From 268947fb5ee7807513c5e883a71836ca6fccf646 Mon Sep 17 00:00:00 2001
From: Wesley Moret <wesmoret98@gmail.com>
Date: Sat, 1 May 2021 11:44:56 -0400
Subject: [PATCH] Allow to have a metadata stream in the document information
 entry (#322)

---
 .../Parser/DocumentInformationFactory.cs      | 45 ++++++++++++++-----
 1 file changed, 34 insertions(+), 11 deletions(-)

diff --git a/src/UglyToad.PdfPig/Parser/DocumentInformationFactory.cs b/src/UglyToad.PdfPig/Parser/DocumentInformationFactory.cs
index 6028d80e..14515c63 100644
--- a/src/UglyToad.PdfPig/Parser/DocumentInformationFactory.cs
+++ b/src/UglyToad.PdfPig/Parser/DocumentInformationFactory.cs
@@ -5,6 +5,7 @@
     using Parts;
     using Tokenization.Scanner;
     using Tokens;
+    using UglyToad.PdfPig.Core;
 
     /// <summary>
     /// Parse the dictionary from a PDF file trailer.
@@ -21,19 +22,41 @@
                 return DocumentInformation.Default;
             }
 
-            var infoParsed = DirectObjectFinder.Get<DictionaryToken>(trailer.Info.Value, pdfTokenScanner);
+            var token = DirectObjectFinder.Get<IToken>(trailer.Info.Value, pdfTokenScanner);
+            if (token is DictionaryToken infoParsed)
+            {
+                var title = GetEntryOrDefault(infoParsed, NameToken.Title);
+                var author = GetEntryOrDefault(infoParsed, NameToken.Author);
+                var subject = GetEntryOrDefault(infoParsed, NameToken.Subject);
+                var keywords = GetEntryOrDefault(infoParsed, NameToken.Keywords);
+                var creator = GetEntryOrDefault(infoParsed, NameToken.Creator);
+                var producer = GetEntryOrDefault(infoParsed, NameToken.Producer);
+                var creationDate = GetEntryOrDefault(infoParsed, NameToken.CreationDate);
+                var modifiedDate = GetEntryOrDefault(infoParsed, NameToken.ModDate);
 
-            var title = GetEntryOrDefault(infoParsed, NameToken.Title);
-            var author = GetEntryOrDefault(infoParsed, NameToken.Author);
-            var subject = GetEntryOrDefault(infoParsed, NameToken.Subject);
-            var keywords = GetEntryOrDefault(infoParsed, NameToken.Keywords);
-            var creator = GetEntryOrDefault(infoParsed, NameToken.Creator);
-            var producer = GetEntryOrDefault(infoParsed, NameToken.Producer);
-            var creationDate = GetEntryOrDefault(infoParsed, NameToken.CreationDate);
-            var modifiedDate = GetEntryOrDefault(infoParsed, NameToken.ModDate);
+                return new DocumentInformation(infoParsed, title, author, subject,
+                    keywords, creator, producer, creationDate, modifiedDate);
+            }
+            else if (token is StreamToken streamToken)
+            {
+                var streamDictionary = streamToken.StreamDictionary;
+                if (!streamDictionary.TryGet(NameToken.Type, out NameToken typeNameToken) || typeNameToken != "Metadata")
+                {
+                    throw new PdfDocumentFormatException($"Unknown document metadata type was found");
+                }
 
-            return new DocumentInformation(infoParsed, title, author, subject,
-                keywords, creator, producer, creationDate, modifiedDate);
+                if (!streamDictionary.TryGet(NameToken.Subtype, out NameToken subtypeToken) || subtypeToken != "XML")
+                {
+                    throw new PdfDocumentFormatException($"Unknown document metadata subtype was found");
+                }
+
+                // We are not fully supporting XMP Stream so we left the user fully deserialize the stream
+                return DocumentInformation.Default;
+            }
+            else
+            {
+                throw new PdfDocumentFormatException($"Unknown document information token was found {token.GetType().Name}");
+            }
         }
 
         private static string GetEntryOrDefault(DictionaryToken infoDictionary, NameToken key)