From bfdca3079f3a62c1ae759479ec85de635dc1be97 Mon Sep 17 00:00:00 2001
From: Eliot Jones <elioty@hotmail.co.uk>
Date: Wed, 3 Jan 2018 19:13:12 +0000
Subject: [PATCH] change the itext document test to reflect its text being form
 content. fix readme typo

---
 README.md                                        |   2 +-
 ...ingle Page Form Content - from itext 1_1.pdf} | Bin
 ...t1.cs => SinglePageFormContentIText1Tests.cs} |  15 +++++++++++++--
 src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj |   4 ++--
 src/UglyToad.Pdf/Parser/PageFactory.cs           |   5 ++++-
 5 files changed, 20 insertions(+), 6 deletions(-)
 rename src/UglyToad.Pdf.Tests/Integration/Documents/{Single Page Simple - from itext 1_1.pdf => Single Page Form Content - from itext 1_1.pdf} (100%)
 rename src/UglyToad.Pdf.Tests/Integration/{SinglePageSimpleIText1.cs => SinglePageFormContentIText1Tests.cs} (68%)

diff --git a/README.md b/README.md
index 55a272f4..124c00ef 100644
--- a/README.md
+++ b/README.md
@@ -39,7 +39,7 @@ The ```Page``` contains the page width and height in points as well as mapping t
 
     PageSize size = Page.Size;
     
-    bool sA4 = size == PageSize.A4;
+    bool isA4 = size == PageSize.A4;
 
 The ```PdfDocument``` will also support opening from byte arrays (as well as streams eventually):
 
diff --git a/src/UglyToad.Pdf.Tests/Integration/Documents/Single Page Simple - from itext 1_1.pdf b/src/UglyToad.Pdf.Tests/Integration/Documents/Single Page Form Content - from itext 1_1.pdf
similarity index 100%
rename from src/UglyToad.Pdf.Tests/Integration/Documents/Single Page Simple - from itext 1_1.pdf
rename to src/UglyToad.Pdf.Tests/Integration/Documents/Single Page Form Content - from itext 1_1.pdf
diff --git a/src/UglyToad.Pdf.Tests/Integration/SinglePageSimpleIText1.cs b/src/UglyToad.Pdf.Tests/Integration/SinglePageFormContentIText1Tests.cs
similarity index 68%
rename from src/UglyToad.Pdf.Tests/Integration/SinglePageSimpleIText1.cs
rename to src/UglyToad.Pdf.Tests/Integration/SinglePageFormContentIText1Tests.cs
index 45a49bd8..35156401 100644
--- a/src/UglyToad.Pdf.Tests/Integration/SinglePageSimpleIText1.cs
+++ b/src/UglyToad.Pdf.Tests/Integration/SinglePageFormContentIText1Tests.cs
@@ -5,13 +5,13 @@
     using Content;
     using Xunit;
 
-    public class SinglePageSimpleIText1Tests
+    public class SinglePageFormContentIText1Tests
     {
         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");
+            return Path.Combine(documentFolder, "Single Page Form Content - from itext 1_1.pdf");
         }
 
         [Fact]
@@ -35,5 +35,16 @@
                 Assert.Equal(PageSize.A4, page.Size);
             }
         }
+
+        [Fact]
+        public void DoesNotExtractText()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                var page = document.GetPage(1);
+
+                Assert.Empty(page.Letters);
+            }
+        }
     }
 }
diff --git a/src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj b/src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj
index 75b7bb07..fad17196 100644
--- a/src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj
+++ b/src/UglyToad.Pdf.Tests/UglyToad.Pdf.Tests.csproj
@@ -12,9 +12,9 @@
     <None Remove="Fonts\TrueType\Roboto-Regular.ttf" />
     <None Remove="Integration\Documents\Font Size Test - from google chrome print pdf.pdf" />
     <None Remove="Integration\Documents\Font Size Test - from libre office.pdf" />
+    <None Remove="Integration\Documents\Single Page Form Content - from itext 1_1.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>
 
@@ -37,7 +37,7 @@
     <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">
+    <Content Include="Integration\Documents\Single Page Form Content - from itext 1_1.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>
     <Content Include="Integration\Documents\Single Page Simple - from open office.pdf">
diff --git a/src/UglyToad.Pdf/Parser/PageFactory.cs b/src/UglyToad.Pdf/Parser/PageFactory.cs
index cf206e2e..4d812f6f 100644
--- a/src/UglyToad.Pdf/Parser/PageFactory.cs
+++ b/src/UglyToad.Pdf/Parser/PageFactory.cs
@@ -8,6 +8,7 @@
     using Geometry;
     using Graphics;
     using IO;
+    using Util;
 
     internal class PageFactory : IPageFactory
     {
@@ -60,7 +61,9 @@
                 }
 
                 var contents = contentStream.Decode(filterProvider);
-                
+
+                var txt = OtherEncodings.BytesAsLatin1String(contents);
+
                 var operations = pageContentParser.Parse(new ByteArrayInputBytes(contents));
 
                 var context = new ContentStreamProcessor(cropBox.Bounds, resourceStore, userSpaceUnit);