move catalog parsing to its own factory. parse document information if present and expose publically. add test for itext generated document

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

@@ -8,7 +8,7 @@ namespace UglyToad.Pdf.Tests.Integration
     using Content;
     using Xunit;
 
-    public class SinglePageSimpleTests
+    public class SinglePageSimpleGoogleChromeTests
     {
         private static readonly HashSet<string> IgnoredHiddenCharacters = new HashSet<string>
         {

src/UglyToad.Pdf.Tests/Integration/SinglePageSimpleIText1.cs

@@ -0,0 +1,39 @@
+namespace UglyToad.Pdf.Tests.Integration
+{
+    using System;
+    using System.IO;
+    using Content;
+    using Xunit;
+
+    public class SinglePageSimpleIText1Tests
+    {
+        private static string GetFilename()
+        {
+            var documentFolder = Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Integration", "Documents"));
+
+            return Path.Combine(documentFolder, "Single Page Simple - from itext 1_1.pdf");
+        }
+
+        [Fact]
+        public void HasCorrectNumberOfPages()
+        {
+            var file = GetFilename();
+
+            using (var document = PdfDocument.Open(File.ReadAllBytes(file)))
+            {
+                Assert.Equal(1, document.NumberOfPages);
+            }
+        }
+
+        [Fact]
+        public void HasCorrectPageSize()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                var page = document.GetPage(1);
+
+                Assert.Equal(PageSize.A4, page.Size);
+            }
+        }
+    }
+}

src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj

@@ -14,6 +14,7 @@
     <None Remove="Integration\Documents\Font Size Test - from libre office.pdf" />
     <None Remove="Integration\Documents\Single Page Non Latin - from acrobat distiller.pdf" />
     <None Remove="Integration\Documents\Single Page Simple - from google drive.pdf" />
+    <None Remove="Integration\Documents\Single Page Simple - from itext 1_1.pdf" />
     <None Remove="Integration\Documents\Single Page Simple - from open office.pdf" />
   </ItemGroup>
 
@@ -36,6 +37,9 @@
     <Content Include="Integration\Documents\Single Page Simple - from google drive.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>
+    <Content Include="Integration\Documents\Single Page Simple - from itext 1_1.pdf">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </Content>
     <Content Include="Integration\Documents\Single Page Simple - from open office.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>

src/UglyToad.Pdf/Content/Catalog.cs

@@ -8,9 +8,13 @@
     {
         private readonly PdfDictionary catalogDictionary;
 
-        internal Catalog(PdfDictionary catalogDictionary)
+        public PdfDictionary PagesDictionary { get; }
+
+        internal Catalog(PdfDictionary catalogDictionary, PdfDictionary pagesDictionary)
         {
             this.catalogDictionary = catalogDictionary ?? throw new ArgumentNullException(nameof(catalogDictionary));
+
+            PagesDictionary = pagesDictionary ?? throw new ArgumentNullException(nameof(pagesDictionary));
         }
 
         public CosBase Get(CosName name)

src/UglyToad.Pdf/Content/DocumentInformation.cs

@@ -0,0 +1,88 @@
+namespace UglyToad.Pdf.Content
+{
+    using System.Text;
+    using Util.JetBrains.Annotations;
+
+    /// <summary>
+    /// Metadata for the PDF document.
+    /// </summary>
+    public class DocumentInformation
+    {
+        internal static DocumentInformation Default { get; }
+            = new DocumentInformation(null, null, null, null, null, null);
+
+        private readonly string representation;
+
+        /// <summary>
+        /// The title of this document if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Title { get; }
+
+        /// <summary>
+        /// The name of the person who created this document if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Author { get; }
+
+        /// <summary>
+        /// The subject of this document if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Subject { get; }
+
+        /// <summary>
+        /// Any keywords associated with this document if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Keywords { get; }
+
+        /// <summary>
+        /// The name of the application which created the original document before it was converted to PDF. if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Creator { get; }
+
+        /// <summary>
+        /// The name of the application used to convert the original document to PDF if applicable.
+        /// </summary>
+        [CanBeNull]
+        public string Producer { get; }
+
+        internal DocumentInformation(string title, string author, string subject, string keywords, string creator, string producer)
+        {
+            Title = title;
+            Author = author;
+            Subject = subject;
+            Keywords = keywords;
+            Creator = creator;
+            Producer = producer;
+
+            var builder = new StringBuilder();
+
+            AppendPart("Title", title, builder);
+            AppendPart("Author", author, builder);
+            AppendPart("Subject", subject, builder);
+            AppendPart("Keywords", keywords, builder);
+            AppendPart("Creator", creator, builder);
+            AppendPart("Producer", producer, builder);
+
+            representation = builder.ToString() ?? string.Empty;
+        }
+
+        public override string ToString()
+        {
+            return representation;
+        }
+
+        private static void AppendPart(string name, string value, StringBuilder builder)
+        {
+            if (value == null)
+            {
+                return;
+            }
+
+            builder.Append(name).Append(": ").Append(value).Append(";");
+        }
+    }
+}

src/UglyToad.Pdf/Content/Pages.cs

@@ -31,25 +31,9 @@
                 throw new ArgumentNullException(nameof(catalog));
             }
 
-            var pages = catalog.Get(CosName.PAGES) as CosObject;
+            rootPageDictionary = catalog.PagesDictionary;
 
-            if (pages == null)
-            {
-                throw new InvalidOperationException("No pages were present in the catalog for this PDF document");
-            }
-
-            var pagesObject = pdfObjectParser.Parse(pages.ToIndirectReference(), reader, isLenientParsing);
-
-            if (!(pagesObject is PdfDictionary catalogPageDictionary))
-            {
-                throw new InvalidOperationException("Could not find the root pages object: " + pages);
-            }
-
-            var count = catalogPageDictionary.GetIntOrDefault(CosName.COUNT);
-
-            rootPageDictionary = catalogPageDictionary;
-
-            Count = count;
+            Count = rootPageDictionary.GetIntOrDefault(CosName.COUNT);
 
             this.log = log;
             this.catalog = catalog;

src/UglyToad.Pdf/Exceptions/PdfDocumentFormatException.cs

@@ -0,0 +1,34 @@
+namespace UglyToad.Pdf.Exceptions
+{
+    using System;
+    using System.Runtime.Serialization;
+
+    [Serializable]
+    public class PdfDocumentFormatException : Exception
+    {
+        //
+        // For guidelines regarding the creation of new exception types, see
+        //    http://msdn.microsoft.com/library/default.asp?url=/library/en-us/cpgenref/html/cpconerrorraisinghandlingguidelines.asp
+        // and
+        //    http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dncscol/html/csharp07192001.asp
+        //
+
+        public PdfDocumentFormatException()
+        {
+        }
+
+        public PdfDocumentFormatException(string message) : base(message)
+        {
+        }
+
+        public PdfDocumentFormatException(string message, Exception inner) : base(message, inner)
+        {
+        }
+
+        protected PdfDocumentFormatException(
+            SerializationInfo info,
+            StreamingContext context) : base(info, context)
+        {
+        }
+    }
+}

src/UglyToad.Pdf/Parser/CatalogFactory.cs

@@ -0,0 +1,46 @@
+namespace UglyToad.Pdf.Parser
+{
+    using System;
+    using Content;
+    using ContentStream;
+    using Cos;
+    using Exceptions;
+    using IO;
+
+    internal class CatalogFactory
+    {
+        private readonly IPdfObjectParser pdfObjectParser;
+
+        public CatalogFactory(IPdfObjectParser pdfObjectParser)
+        {
+            this.pdfObjectParser = pdfObjectParser;
+        }
+
+        public Catalog Create(PdfDictionary dictionary, IRandomAccessRead reader, bool isLenientParsing)
+        {
+            if (dictionary == null)
+            {
+                throw new ArgumentNullException(nameof(dictionary));
+            }
+
+            if (dictionary.TryGetName(CosName.TYPE, out var type) && !type.Equals(CosName.CATALOG))
+            {
+                throw new PdfDocumentFormatException($"The type of the catalog dictionary was not Catalog: {dictionary}.");
+            }
+
+            if (!dictionary.TryGetItemOfType(CosName.PAGES, out CosObject value))
+            {
+                throw new PdfDocumentFormatException($"No pages entry was found in the catalog dictionary: {dictionary}.");
+            }
+
+            var pages = pdfObjectParser.Parse(value.ToIndirectReference(), reader, isLenientParsing);
+
+            if (!(pages is PdfDictionary pagesDictionary))
+            {
+                throw new PdfDocumentFormatException($"The pages entry in the catalog {value.ToIndirectReference()} did not link to a dictionary: {pages}.");
+            }
+
+            return new Catalog(dictionary, pagesDictionary);
+        }
+    }
+}

src/UglyToad.Pdf/Parser/DocumentInformationFactory.cs

@@ -0,0 +1,47 @@
+namespace UglyToad.Pdf.Parser
+{
+    using Content;
+    using ContentStream;
+    using Cos;
+    using IO;
+
+    internal class DocumentInformationFactory
+    {
+        public DocumentInformation Create(IPdfObjectParser pdfObjectParser,
+            PdfDictionary rootDictionary, IRandomAccessRead reader, 
+            bool isLenientParsing)
+        {
+            if (!rootDictionary.TryGetItemOfType(CosName.INFO, out CosObject infoBase))
+            {
+                return DocumentInformation.Default;
+            }
+
+            var infoParsed = pdfObjectParser.Parse(infoBase.ToIndirectReference(), reader, isLenientParsing);
+
+            if (!(infoParsed is PdfDictionary infoDictionary))
+            {
+                return DocumentInformation.Default;
+            }
+
+            var title = GetEntryOrDefault(infoDictionary, CosName.TITLE);
+            var author = GetEntryOrDefault(infoDictionary, CosName.AUTHOR);
+            var subject = GetEntryOrDefault(infoDictionary, CosName.SUBJECT);
+            var keywords = GetEntryOrDefault(infoDictionary, CosName.KEYWORDS);
+            var creator = GetEntryOrDefault(infoDictionary, CosName.CREATOR);
+            var producer = GetEntryOrDefault(infoDictionary, CosName.PRODUCER);
+
+            return new DocumentInformation(title, author, subject,
+                keywords, creator, producer);
+        }
+
+        private static string GetEntryOrDefault(PdfDictionary infoDictionary, CosName key)
+        {
+            if (infoDictionary.TryGetItemOfType(key, out CosString str))
+            {
+                return str.GetAscii();
+            }
+
+            return null;
+        }
+    }
+}

src/UglyToad.Pdf/Parser/PageFactory.cs

@@ -1,14 +1,13 @@
-namespace UglyToad.Pdf.Content
+namespace UglyToad.Pdf.Parser
 {
     using System;
+    using Content;
     using ContentStream;
     using Cos;
     using Filters;
     using Geometry;
     using Graphics;
     using IO;
-    using Parser;
-    using Util;
 
     internal class PageFactory : IPageFactory
     {

src/UglyToad.Pdf/Parser/PdfDocumentFactory.cs

@@ -78,6 +78,8 @@
             var resourceContainer = new ResourceContainer(pdfObjectParser, fontFactory);
 
             var pageFactory = new PageFactory(resourceContainer, pdfObjectParser, filterProvider, new PageContentParser(new ReflectionGraphicsStateOperationFactory()));
+            var informationFactory = new DocumentInformationFactory();
+            var catalogFactory = new CatalogFactory(pdfObjectParser);
 
             var root = ParseTrailer(reader, crossReferenceTable, dynamicParser, bruteForceSearcher, pool,
                 isLenientParsing);
@@ -93,9 +95,13 @@
                 rootDictionary.Set(CosName.TYPE, CosName.CATALOG);
             }
 
+            var information = informationFactory.Create(pdfObjectParser, crossReferenceTable.Dictionary, reader, isLenientParsing);
+
+            var catalog = catalogFactory.Create(rootDictionary, reader, isLenientParsing);
+
             var caching = new ParsingCachingProviders(pool, bruteForceSearcher, resourceContainer);
 
-            return new PdfDocument(log, reader, version, crossReferenceTable, isLenientParsing, caching, pageFactory, pdfObjectParser, new Catalog(rootDictionary));
+            return new PdfDocument(log, reader, version, crossReferenceTable, isLenientParsing, caching, pageFactory, pdfObjectParser, catalog, information);
         }
 
         private static CosBase ParseTrailer(IRandomAccessRead reader, CrossReferenceTable crossReferenceTable,

src/UglyToad.Pdf/PdfDocument.cs

@@ -7,7 +7,6 @@
     using Logging;
     using Parser;
     using Parser.Parts;
-    using Util;
     using Util.JetBrains.Annotations;
 
     public class PdfDocument : IDisposable
@@ -30,6 +29,9 @@
         [NotNull]
         internal Pages Pages { get; }
 
+        [NotNull]
+        public DocumentInformation Information { get; }
+
         /// <summary>
         /// Get the number of pages in this document.
         /// </summary>
@@ -40,7 +42,8 @@
             ParsingCachingProviders cachingProviders,
             IPageFactory pageFactory,
             IPdfObjectParser pdfObjectParser,
-            Catalog catalog)
+            Catalog catalog,
+            DocumentInformation information)
         {
             this.log = log;
             this.reader = reader ?? throw new ArgumentNullException(nameof(reader));
@@ -48,6 +51,7 @@
             this.crossReferenceTable = crossReferenceTable ?? throw new ArgumentNullException(nameof(crossReferenceTable));
             this.isLenientParsing = isLenientParsing;
             this.cachingProviders = cachingProviders ?? throw new ArgumentNullException(nameof(cachingProviders));
+            Information = information ?? throw new ArgumentNullException(nameof(information));
             Catalog = catalog ?? throw new ArgumentNullException(nameof(catalog));
             Pages = new Pages(log, Catalog, pdfObjectParser, pageFactory, reader, isLenientParsing);
         }