add more documents to test font size and add tests to check our text positions against other providers

src/UglyToad.Pdf.Tests/Graphics/Operations/TextState/SetFontAndSizeTests.cs

@@ -0,0 +1,70 @@
+namespace UglyToad.Pdf.Tests.Graphics.Operations.TextState
+{
+    using System;
+    using Pdf.Cos;
+    using Pdf.Graphics.Operations.TextState;
+    using Xunit;
+
+    public class SetFontAndSizeTests
+    {
+        private static readonly CosName Font1Name = CosName.Create("Font1");
+
+        [Fact]
+        public void HasCorrectSymbol()
+        {
+            var symbol = SetFontAndSize.Symbol;
+
+            Assert.Equal("Tf", symbol);
+        }
+
+        [Fact]
+        public void SetsValues()
+        {
+            var setFontAndSize = new SetFontAndSize(Font1Name, 12.75m);
+
+            Assert.Equal("Font1", setFontAndSize.Font.Name);
+            Assert.Equal(12.75m, setFontAndSize.Size);
+        }
+
+        [Fact]
+        public void HasCorrectOperator()
+        {
+            var setFontAndSize = new SetFontAndSize(Font1Name, 12);
+
+            Assert.Equal("Tf", setFontAndSize.Operator);
+        }
+
+        [Fact]
+        public void NameNullThrows()
+        {
+            // ReSharper disable once ObjectCreationAsStatement
+            Action action = () => new SetFontAndSize(null, 6);
+
+            Assert.Throws<ArgumentNullException>(action);
+        }
+
+        [Fact]
+        public void StringRepresentationIsCorrect()
+        {
+            var setFontAndSize = new SetFontAndSize(Font1Name, 12.76m);
+
+            Assert.Equal("/Font1 12.76 Tf", setFontAndSize.ToString());
+        }
+
+        [Fact]
+        public void RunSetsFontAndFontSize()
+        {
+            var setFontAndSize = new SetFontAndSize(Font1Name, 69.42m);
+
+            var context = new TestOperationContext();
+            var store = new TestResourceStore();
+
+            setFontAndSize.Run(context, store);
+
+            var state = context.GetCurrentState();
+
+            Assert.Equal(69.42m, state.FontState.FontSize);
+            Assert.Equal(Font1Name, state.FontState.FontName);
+        }
+    }
+}

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

@@ -162,11 +162,7 @@
         {
             var documentFolder = Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Integration", "Documents"));
 
-            var files = Directory.GetFiles(documentFolder);
+            return Path.Combine(documentFolder, "Single Page Simple - from google drive.pdf");
-
-            var file = files[n];
-
-            return file;
         }
     }
 }

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

@@ -92,6 +92,95 @@ namespace UglyToad.Pdf.Tests.Integration
             }
         }
 
+        [Fact]
+        public void LettersHavePdfBoxPositions()
+        {
+            var file = GetFilename();
+
+            var pdfBoxData = GetPdfBoxPositionData();
+            var index = 0;
+
+            using (var document = PdfDocument.Open(File.ReadAllBytes(file)))
+            {
+                var page = document.GetPage(1);
+
+                foreach (var letter in page.Letters)
+                {
+                    // Something a bit weird with how we or PdfBox handle hidden characters and spaces.
+                    if (IgnoredHiddenCharacters.Contains(letter.Value) || string.IsNullOrWhiteSpace(letter.Value))
+                    {
+                        continue;
+                    }
+
+                    var datum = pdfBoxData[index];
+
+                    while (IgnoredHiddenCharacters.Contains(datum.Text))
+                    {
+                        index++;
+                        datum = pdfBoxData[index];
+                    }
+
+                    Assert.Equal(datum.Text, letter.Value);
+                    Assert.Equal(datum.X, letter.Location.X, 2);
+
+                    var transformed = page.Height - letter.Location.Y;
+                    Assert.Equal(datum.Y, transformed, 2);
+
+                    Assert.Equal(datum.Width, letter.Width, 2);
+
+                    Assert.Equal(datum.FontName, letter.FontName);
+
+                    // I think we have font size wrong for now, or right, but differently correct...
+
+                    index++;
+                }
+            }
+        }
+
+        [Fact]
+        public void LettersHaveOtherProviderPositions()
+        {
+            var file = GetFilename();
+
+            var pdfBoxData = GetOtherPositionData1();
+            var index = 0;
+
+            using (var document = PdfDocument.Open(File.ReadAllBytes(file)))
+            {
+                var page = document.GetPage(1);
+
+                foreach (var letter in page.Letters)
+                {
+                    // Something a bit weird with how we or this provider handle hidden characters and spaces.
+                    if (IgnoredHiddenCharacters.Contains(letter.Value) || string.IsNullOrWhiteSpace(letter.Value))
+                    {
+                        continue;
+                    }
+
+                    var datum = pdfBoxData[index];
+
+                    while (IgnoredHiddenCharacters.Contains(datum.Text) || datum.Text == " ")
+                    {
+                        index++;
+                        datum = pdfBoxData[index];
+                    }
+
+                    Assert.Equal(datum.Text, letter.Value);
+                    Assert.Equal(datum.X, letter.Location.X, 2);
+
+                    var transformed = page.Height - letter.Location.Y;
+                    Assert.Equal(datum.Y, transformed, 2);
+
+                    // Until we get width from glyphs we're a bit out.
+                    Assert.True(Math.Abs(datum.Width - letter.Width) < 0.03m);
+
+                    index++;
+                }
+            }
+        }
+
+
+
         private static IReadOnlyList<AssertablePositionData> GetPdfBoxPositionData()
         {
             // X    Y   Width   Letter  FontSize    Font

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

@@ -10,6 +10,8 @@
   <ItemGroup>
     <None Remove="Fonts\TrueType\google-simple-doc.ttf" />
     <None Remove="Fonts\TrueType\Roboto-Regular.ttf" />
+    <None Remove="Integration\Documents\Font Size Test - from libre office.pdf" />
+    <None Remove="Integration\Documents\Font Size Text - from google chrome print pdf.pdf" />
     <None Remove="Integration\Documents\Single Page Simple - from google drive.pdf" />
   </ItemGroup>
 
@@ -20,6 +22,12 @@
     <EmbeddedResource Include="Fonts\TrueType\Roboto-Regular.ttf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </EmbeddedResource>
+    <Content Include="Integration\Documents\Font Size Test - from libre office.pdf">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </Content>
+    <Content Include="Integration\Documents\Font Size Text - from google chrome print pdf.pdf">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </Content>
     <Content Include="Integration\Documents\Single Page Simple - from google drive.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>

src/UglyToad.Pdf/Content/CropBox.cs

@@ -0,0 +1,20 @@
+namespace UglyToad.Pdf.Content
+{
+    using System;
+    using Geometry;
+    using Util.JetBrains.Annotations;
+
+    /// <summary>
+    /// Defines the visible region, contents expanding beyond the crop box should be clipped.
+    /// </summary>
+    public class CropBox
+    {
+        [NotNull]
+        public PdfRectangle Bounds { get; }
+
+        public CropBox(PdfRectangle bounds)
+        {
+            Bounds = bounds ?? throw new ArgumentNullException(nameof(bounds));
+        }
+    }
+}

src/UglyToad.Pdf/Content/Letter.cs

@@ -4,23 +4,46 @@
 
     public class Letter
     {
+        /// <summary>
+        /// The text for this letter or unicode character.
+        /// </summary>
         public string Value { get; }
 
         public PdfPoint Location { get; }
 
+        /// <summary>
+        /// The width of the letter.
+        /// </summary>
         public decimal Width { get; }
 
-        public decimal FontSize { get; }
+        /// <summary>
+        /// Size defined by the Tj operator prior to our possibly incorrect transformation.
+        /// </summary>
+        internal decimal FontSize { get; }
 
+        /// <summary>
+        /// The name of the font.
+        /// </summary>
         public string FontName { get; }
 
-        public Letter(string value, PdfPoint location, decimal width, decimal fontSize, string fontName)
+        /// <summary>
+        /// The size of the font in points.
+        /// </summary>
+        public decimal PointSize { get; }
+
+        internal Letter(string value, PdfPoint location, decimal width, decimal fontSize, string fontName, decimal pointSize)
         {
             Value = value;
             Location = location;
             Width = width;
             FontSize = fontSize;
             FontName = fontName;
+            PointSize = pointSize;
+        }
+
+        public override string ToString()
+        {
+            return $"{Location} {Width} {Value} {FontName} {PointSize}";
         }
     }
 }

src/UglyToad.Pdf/Content/Page.cs

@@ -12,6 +12,8 @@
 
         internal MediaBox MediaBox { get; }
 
+        internal CropBox CropBox { get; }
+
         internal PageContent Content { get; }
 
         public IReadOnlyList<Letter> Letters => Content?.Letters ?? new Letter[0];
@@ -26,7 +28,7 @@
         /// </summary>
         public decimal Height { get; }
 
-        internal Page(int number, MediaBox mediaBox, PageContent content)
+        internal Page(int number, MediaBox mediaBox, CropBox cropBox, PageContent content)
         {
             if (number <= 0)
             {
@@ -35,6 +37,7 @@
 
             Number = number;
             MediaBox = mediaBox;
+            CropBox = cropBox;
             Content = content;
 
             Width = mediaBox.Bounds.Width;

src/UglyToad.Pdf/Content/PageFactory.cs

@@ -40,6 +40,75 @@
                 throw new InvalidOperationException($"Page {number} had its type was specified as {type} rather than 'Page'.");
             }
 
+            MediaBox mediaBox = GetMediaBox(number, dictionary, pageTreeMembers, isLenientParsing);
+            CropBox cropBox = GetCropBox(dictionary, pageTreeMembers, mediaBox);
+
+            if (dictionary.GetItemOrDefault(CosName.RESOURCES) is PdfDictionary resource)
+            {
+                resourceStore.LoadResourceDictionary(resource, reader, isLenientParsing);
+            }
+
+            UserSpaceUnit userSpaceUnit = GetUserSpaceUnits(dictionary);
+
+            PageContent content = default(PageContent);
+
+            var contentObject = dictionary.GetItemOrDefault(CosName.CONTENTS) as CosObject;
+            if (contentObject != null)
+            {
+                var contentStream = pdfObjectParser.Parse(contentObject.ToIndirectReference(), reader, false) as PdfRawStream;
+
+                if (contentStream == null)
+                {
+                    throw new InvalidOperationException("Failed to parse the content for the page: " + number);
+                }
+
+                var contents = contentStream.Decode(filterProvider);
+
+                var operations = pageContentParser.Parse(new ByteArrayInputBytes(contents));
+
+                var context = new ContentStreamProcessor(mediaBox.Bounds, resourceStore, userSpaceUnit);
+
+                content = context.Process(operations);
+            }
+
+            var page = new Page(number, mediaBox, cropBox, content);
+
+            return page;
+        }
+
+        private static UserSpaceUnit GetUserSpaceUnits(PdfDictionary dictionary)
+        {
+            var spaceUnits = UserSpaceUnit.Default;
+            if (dictionary.TryGetValue(CosName.USER_UNIT, out var userUnitCosBase) && userUnitCosBase is ICosNumber userUnitNumber)
+            {
+                spaceUnits = new UserSpaceUnit(userUnitNumber.AsInt());
+            }
+
+            return spaceUnits;
+        }
+
+        private static CropBox GetCropBox(PdfDictionary dictionary, PageTreeMembers pageTreeMembers, MediaBox mediaBox)
+        {
+            CropBox cropBox;
+            if (dictionary.TryGetItemOfType(CosName.CROP_BOX, out COSArray cropBoxArray))
+            {
+                var x1 = cropBoxArray.getInt(0);
+                var y1 = cropBoxArray.getInt(1);
+                var x2 = cropBoxArray.getInt(2);
+                var y2 = cropBoxArray.getInt(3);
+
+                cropBox = new CropBox(new PdfRectangle(x1, y1, x2, y2));
+            }
+            else
+            {
+                cropBox = pageTreeMembers.GetCropBox() ?? new CropBox(mediaBox.Bounds);
+            }
+
+            return cropBox;
+        }
+
+        private static MediaBox GetMediaBox(int number, PdfDictionary dictionary, PageTreeMembers pageTreeMembers, bool isLenientParsing)
+        {
             MediaBox mediaBox;
             if (dictionary.TryGetItemOfType(CosName.MEDIA_BOX, out COSArray mediaboxArray))
             {
@@ -67,35 +136,7 @@
                 }
             }
 
-            if (dictionary.GetItemOrDefault(CosName.RESOURCES) is PdfDictionary resource)
+            return mediaBox;
-            {
-                resourceStore.LoadResourceDictionary(resource, reader, isLenientParsing);
-            }
-
-            PageContent content = default(PageContent);
-
-            var contentObject = dictionary.GetItemOrDefault(CosName.CONTENTS) as CosObject;
-            if (contentObject != null)
-            {
-                var contentStream = pdfObjectParser.Parse(contentObject.ToIndirectReference(), reader, false) as PdfRawStream;
-
-                if (contentStream == null)
-                {
-                    throw new InvalidOperationException("Failed to parse the content for the page: " + number);
-                }
-
-                var contents = contentStream.Decode(filterProvider);
-
-                var operations = pageContentParser.Parse(new ByteArrayInputBytes(contents));
-
-                var context = new ContentStreamProcessor(mediaBox.Bounds, resourceStore);
-
-                content = context.Process(operations);
-            }
-
-            var page = new Page(number, mediaBox, content);
-
-            return page;
         }
     }
 }

src/UglyToad.Pdf/Content/PageTreeMembers.cs

@@ -12,5 +12,10 @@
             // TODO: tree inheritance
             throw new NotImplementedException("Track inherited members");
         }
+
+        public CropBox GetCropBox()
+        {
+            return null;
+        }
     }
 }

src/UglyToad.Pdf/Core/TransformationMatrix.cs

@@ -18,9 +18,15 @@
 
         private readonly decimal[] value;
 
+        /// <summary>
+        /// The scale for the X dimension.
+        /// </summary>
         public decimal A => value[0];
         public decimal B => value[1];
         public decimal C => value[3];
+        /// <summary>
+        /// The scale for the Y dimension.
+        /// </summary>
         public decimal D => value[4];
         public decimal E => value[6];
         public decimal F => value[7];

src/UglyToad.Pdf/Cos/CosName.cs

@@ -518,6 +518,7 @@ namespace UglyToad.Pdf.Cos
         public static readonly CosName UNIX = new CosName("Unix");
         public static readonly CosName URI = new CosName("URI");
         public static readonly CosName URL = new CosName("URL");
+        public static readonly CosName USER_UNIT = new CosName("UserUnit");
         // V
         public static readonly CosName V = new CosName("V");
         public static readonly CosName VERISIGN_PPKVS = new CosName("VeriSign.PPKVS");

src/UglyToad.Pdf/Geometry/UserSpaceUnit.cs

@@ -0,0 +1,36 @@
+namespace UglyToad.Pdf.Geometry
+{
+    using System;
+
+    /// <summary>
+    /// By default user space units correspond to 1/72nd of an inch (a typographic point).
+    /// The UserUnit entry in a page dictionary can define the space units as a different multiple of 1/72 (1 point).
+    /// </summary>
+    public struct UserSpaceUnit
+    {
+        public static readonly UserSpaceUnit Default = new UserSpaceUnit(1);
+
+        /// <summary>
+        /// The number of points (1/72nd of an inch) corresponding to a single unit in user space.
+        /// </summary>
+        public int PointMultiples { get; }
+
+        /// <summary>
+        /// Create a new unit specification for a page.
+        /// </summary>
+        public UserSpaceUnit(int pointMultiples)
+        {
+            if (pointMultiples <= 0)
+            {
+                throw new ArgumentOutOfRangeException("Cannot have a zero or negative value of point multiples: " + pointMultiples);
+            }
+
+            PointMultiples = pointMultiples;
+        }
+
+        public override string ToString()
+        {
+            return PointMultiples.ToString();
+        }
+    }
+}

src/UglyToad.Pdf/Graphics/ContentStreamProcessor.cs

@@ -13,6 +13,7 @@
     internal class ContentStreamProcessor : IOperationContext
     {
         private readonly IResourceStore resourceStore;
+        private readonly UserSpaceUnit userSpaceUnit;
 
         private Stack<CurrentGraphicsState> graphicsStack = new Stack<CurrentGraphicsState>();
 
@@ -22,9 +23,10 @@
         
         public List<Letter> Letters = new List<Letter>();
 
-        public ContentStreamProcessor(PdfRectangle cropBox, IResourceStore resourceStore)
+        public ContentStreamProcessor(PdfRectangle cropBox, IResourceStore resourceStore, UserSpaceUnit userSpaceUnit)
         {
             this.resourceStore = resourceStore;
+            this.userSpaceUnit = userSpaceUnit;
             graphicsStack.Push(new CurrentGraphicsState());
         }
 
@@ -77,10 +79,18 @@
         {
             var font = resourceStore.GetFont(GetCurrentState().FontState.FontName);
 
-            var fontSize = GetCurrentState().FontState.FontSize;
+            var currentState = GetCurrentState();
-            var horizontalScaling = GetCurrentState().FontState.HorizontalScaling;
-            var characterSpacing = GetCurrentState().FontState.CharacterSpacing;
 
+            var fontSize = currentState.FontState.FontSize;
+            var horizontalScaling = currentState.FontState.HorizontalScaling;
+            var characterSpacing = currentState.FontState.CharacterSpacing;
+
+            var transformationMatrix = currentState.CurrentTransformationMatrix;
+
+            // TODO: this does not seem correct, produces the correct result for now but we need to revisit.
+            // see: https://stackoverflow.com/questions/48010235/pdf-specification-get-font-size-in-points
+            var pointSize = decimal.Round(fontSize * transformationMatrix.A, 2);
+            
             while (bytes.MoveNext())
             {
                 var code = font.ReadCharacterCode(bytes, out int codeLength);
@@ -102,7 +112,9 @@
 
                 var displacement = font.GetDisplacement(code);
 
-                ShowGlyph(renderingMatrix, font, code, unicode, displacement, fontSize);
+                var width = (displacement.X * fontSize) * transformationMatrix.A;
+
+                ShowGlyph(renderingMatrix, font, code, unicode, width, fontSize, pointSize);
 
                 decimal tx, ty;
                 if (font.IsVertical)
@@ -122,11 +134,12 @@
             }
         }
 
-        private void ShowGlyph(TransformationMatrix renderingMatrix, IFont font, int characterCode, string unicode, PdfVector displacement, decimal fontSize)
+        private void ShowGlyph(TransformationMatrix renderingMatrix, IFont font, int characterCode, string unicode, decimal width, decimal fontSize,
+            decimal pointSize)
         {
             var location = new PdfPoint(renderingMatrix.E, renderingMatrix.F);
-
+            
-            var letter = new Letter(unicode, location, displacement.X, fontSize, font.Name.Name);
+            var letter = new Letter(unicode, location, width, fontSize, font.Name.Name, pointSize);
 
             Letters.Add(letter);
         }

src/UglyToad.Pdf/Graphics/Operations/TextState/SetFontAndSize.cs

@@ -1,7 +1,9 @@
 namespace UglyToad.Pdf.Graphics.Operations.TextState
 {
+    using System;
     using Content;
     using Cos;
+    using Util.JetBrains.Annotations;
 
     internal class SetFontAndSize : IGraphicsStateOperation
     {
@@ -9,13 +11,21 @@
 
         public string Operator => Symbol;
 
+        /// <summary>
+        /// The name of the font as defined in the resource dictionary.
+        /// </summary>
+        [NotNull]
         public CosName Font { get; }
 
+        /// <summary>
+        /// The font program defines glyphs for a standard size. This standard size is set so that each line of text will occupy 1 unit in user space.
+        /// The size is the scale factor used to scale glyphs from the standard size to the display size rather than the font size in points.
+        /// </summary>
         public decimal Size { get; }
 
         public SetFontAndSize(CosName font, decimal size)
         {
-            Font = font;
+            Font = font ?? throw new ArgumentNullException(nameof(font));
             Size = size;
         }