add test for multiple page pdf from libre office

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

@@ -0,0 +1,58 @@
+namespace UglyToad.Pdf.Tests.Integration
+{
+    using System;
+    using System.IO;
+    using Content;
+    using Xunit;
+
+    public class TwoPageTextOnlyLibreOfficeTests
+    {
+        private static string GetFilename()
+        {
+            var documentFolder = Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Integration", "Documents"));
+
+            return Path.Combine(documentFolder, "Two Page Text Only - from libre office.pdf");
+        }
+
+        [Fact]
+        public void HasCorrectNumberOfPages()
+        {
+            var file = GetFilename();
+
+            using (var document = PdfDocument.Open(File.ReadAllBytes(file)))
+            {
+                Assert.Equal(2, document.NumberOfPages);
+            }
+        }
+
+        [Fact]
+        public void HasCorrectPageSize()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                var page = document.GetPage(1);
+
+                Assert.Equal(PageSize.A4, page.Size);
+
+                page = document.GetPage(2);
+
+                Assert.Equal(PageSize.A4, page.Size);
+            }
+        }
+
+        [Fact]
+        public void PagesStartWithCorrectText()
+        {
+            using (var document = PdfDocument.Open(GetFilename()))
+            {
+                var page = document.GetPage(1);
+
+                Assert.StartsWith("Apache License", page.Text);
+
+                page = document.GetPage(2);
+
+                Assert.StartsWith("2. Grant of Copyright", page.Text);
+            }
+        }
+    }
+}

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

@@ -16,6 +16,7 @@
     <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 open office.pdf" />
+    <None Remove="Integration\Documents\Two Page Text Only - from libre office.pdf" />
   </ItemGroup>
 
   <ItemGroup>
@@ -43,6 +44,9 @@
     <Content Include="Integration\Documents\Single Page Simple - from open office.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>
+    <Content Include="Integration\Documents\Two Page Text Only - from libre office.pdf">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </Content>
   </ItemGroup>
 
   <ItemGroup>

src/UglyToad.Pdf/Content/Page.cs

@@ -2,6 +2,7 @@
 {
     using System;
     using System.Collections.Generic;
+    using System.Linq;
 
     public class Page
     {
@@ -18,6 +19,8 @@
 
         public IReadOnlyList<Letter> Letters => Content?.Letters ?? new Letter[0];
 
+        public string Text { get; }
+
         /// <summary>
         /// Gets the width of the page in points.
         /// </summary>
@@ -44,11 +47,22 @@
             MediaBox = mediaBox;
             CropBox = cropBox;
             Content = content;
+            Text = GetText(content);
 
             Width = mediaBox.Bounds.Width;
             Height = mediaBox.Bounds.Height;
 
             Size = mediaBox.Bounds.GetPageSize();
         }
+
+        private static string GetText(PageContent content)
+        {
+            if (content?.Letters == null)
+            {
+                return string.Empty;
+            }
+
+            return string.Join(string.Empty, content.Letters.Select(x => x.Value));
+        }
     }
 }

src/UglyToad.Pdf/Content/Pages.cs

@@ -91,6 +91,7 @@
                 bool found = pageNumber == soughtPageNumber;
 
                 locatedPages[pageNumber] = currentPageDictionary;
+                pageNumbersObserved.Add(pageNumber);
 
                 return found;
             }
@@ -117,6 +118,7 @@
                 if (thisPageMatches)
                 {
                     childFound = true;
+                    break;
                 }
             }