changes to cmap to correctly map for cid fonts. tests for codespace range

src/UglyToad.Pdf.Tests/Fonts/Cmap/CodespaceRangeTests.cs

@@ -0,0 +1,110 @@
+namespace UglyToad.Pdf.Tests.Fonts.Cmap
+{
+    using System;
+    using System.Linq;
+    using Pdf.Fonts.Cmap;
+    using Pdf.Tokenization.Tokens;
+    using Xunit;
+
+    public class CodespaceRangeTests
+    {
+        [Theory]
+        [InlineData("00", "80", 0, 128, 1)]
+        [InlineData("8140", "9ffc", 33088, 40956, 2)]
+        public void CreatesCorrectly(string startHex, string endHex, int startInt, int endInt, int codeLength)
+        {
+            var range = new CodespaceRange(GetHexBytes(startHex.ToCharArray()),
+                GetHexBytes(endHex.ToCharArray()));
+
+            Assert.Equal(startInt, range.StartInt);
+            Assert.Equal(endInt, range.EndInt);
+
+            Assert.Equal(codeLength, range.CodeLength);
+        }
+
+        [Fact]
+        public void MatchesCodeIsNullThrows()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0' );
+
+            var codespace = new CodespaceRange(start, end);
+
+            Action action = () => codespace.Matches(null);
+
+            Assert.Throws<ArgumentNullException>(action);
+        }
+
+        [Fact]
+        public void IsFullMatchCodeIsNullThrows()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0');
+
+            var codespace = new CodespaceRange(start, end);
+
+            Action action = () => codespace.IsFullMatch(null, 2);
+
+            Assert.Throws<ArgumentNullException>(action);
+        }
+
+        [Fact]
+        public void MatchesCodeWrongLengthFalse()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0');
+
+            var codespace = new CodespaceRange(start, end);
+
+            var matches = codespace.Matches(GetHexBytes('6', '9', '0', '1'));
+
+            Assert.False(matches);
+        }
+
+        [Fact]
+        public void MatchesCodeLowerThanStartFalse()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0');
+
+            var codespace = new CodespaceRange(start, end);
+
+            var matches = codespace.Matches(GetHexBytes('0', '1'));
+
+            Assert.False(matches);
+        }
+
+        [Fact]
+        public void MatchesCodeHigherThanEndFalse()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0');
+
+            var codespace = new CodespaceRange(start, end);
+
+            var matches = codespace.Matches(GetHexBytes('9', '6'));
+
+            Assert.False(matches);
+        }
+
+        [Fact]
+        public void MatchesCodeInRangeTrue()
+        {
+            var start = GetHexBytes('0', 'A');
+            var end = GetHexBytes('8', '0');
+
+            var codespace = new CodespaceRange(start, end);
+
+            var matches = codespace.Matches(GetHexBytes('5', 'A'));
+
+            Assert.True(matches);
+        }
+
+        private static byte[] GetHexBytes(params char[] characters)
+        {
+            var token = new HexToken(characters);
+
+            return token.Bytes.ToArray();
+        }
+    }
+}

src/UglyToad.Pdf/Fonts/CharacterIdentifierSystemInfo.cs

@@ -3,7 +3,7 @@
     using CidFonts;
 
     /// <summary>
-    /// Specifies the character collection associated with the <see cref="CharacterIdentifierFont"/> (CIDFont).
+    /// Specifies the character collection associated with the <see cref="ICidFont"/> (CIDFont).
     /// </summary>
     public struct CharacterIdentifierSystemInfo
     {

src/UglyToad.Pdf/Fonts/CidFonts/ICidFont.cs

@@ -1,6 +1,5 @@
 namespace UglyToad.Pdf.Fonts.CidFonts
 {
-    using Cmap;
     using Cos;
 
     /// <summary>
@@ -36,5 +35,9 @@
         CidFontType CidFontType { get; }
 
         FontDescriptor Descriptor { get; }
+
+        decimal GetWidthFromFont(int characterCode);
+
+        decimal GetWidthFromDictionary(int cid);
     }
 }

src/UglyToad.Pdf/Fonts/CidFonts/Type0CidFont.cs

@@ -15,5 +15,15 @@
         public CharacterIdentifierSystemInfo SystemInfo { get; }
         public CidFontType CidFontType => CidFontType.Type0;
         public FontDescriptor Descriptor { get; }
+
+        public decimal GetWidthFromFont(int characterCode)
+        {
+            throw new System.NotImplementedException();
+        }
+
+        public decimal GetWidthFromDictionary(int cid)
+        {
+            throw new System.NotImplementedException();
+        }
     }
 }

src/UglyToad.Pdf/Fonts/CidFonts/Type2CidFont.cs

@@ -35,5 +35,20 @@
             this.verticalWritingMetrics = verticalWritingMetrics;
             this.widths = widths;
         }
+
+        public decimal GetWidthFromFont(int characterCode)
+        {
+            throw new System.NotImplementedException();
+        }
+
+        public decimal GetWidthFromDictionary(int cid)
+        {
+            if (widths.TryGetValue(cid, out var width))
+            {
+                return width;
+            }
+
+            return Descriptor.MissingWidth;
+        }
     }
 }

src/UglyToad.Pdf/Fonts/Cmap/CMapChangeCase.cs

@@ -6,21 +6,36 @@
     using IO;
     using Util.JetBrains.Annotations;
 
+    /// <summary>
+    /// The CMap (character code map) maps character codes to character identifiers (CIDs).
+    /// The set of characters which a CMap refers to is the "character set" (charset).
+    /// </summary>
     internal class CMap
     {
         public CharacterIdentifierSystemInfo Info { get; }
 
+        /// <summary>
+        /// Defines the type of the internal organization of the CMap file.
+        /// </summary>
         public int Type { get; }
 
-        public int WMode { get; }
-
+        /// <summary>
+        /// Defines the name of the CMap file.
+        /// </summary>
         public string Name { get; }
 
+        /// <summary>
+        /// The version number of the CIDFont file.
+        /// </summary>
+        [CanBeNull]
         public string Version { get; }
 
         [NotNull]
         public IReadOnlyDictionary<int, string> BaseFontCharacterMap { get; }
 
+        /// <summary>
+        /// Describes the set of valid input character codes.
+        /// </summary>
         [NotNull]
         public IReadOnlyList<CodespaceRange> CodespaceRanges { get; }
 
@@ -30,6 +45,9 @@
         [NotNull]
         public IReadOnlyList<CidCharacterMapping> CidCharacterMappings { get; }
 
+        /// <summary>
+        /// Controls whether the font associated with the CMap writes horizontally or vertically.
+        /// </summary>
         public WritingMode WritingMode { get; }
 
         public bool HasCidMappings => CidCharacterMappings.Count > 0 || CidRanges.Count > 0;
@@ -54,15 +72,6 @@
             minCodeLength = CodespaceRanges.Min(x => x.CodeLength);
         }
         
-        private string cmapName = null;
-        private string cmapVersion = null;
-        private int cmapType = -1;
-
-        private string registry = null;
-        private string ordering = null;
-        private int supplement = 0;
-
-        
         // CID mappings
         private readonly Dictionary<int, int> codeToCid = new Dictionary<int, int>();
         private readonly List<CidRange> codeToCidRanges = new List<CidRange>();
@@ -111,7 +120,7 @@
         
         public override string ToString()
         {
-            return cmapName;
+            return Name;
         }
 
         public int ReadCode(IInputBytes bytes)
@@ -130,7 +139,7 @@
                 int byteCount = i + 1;
                 foreach (CodespaceRange range in CodespaceRanges)
                 {
-                    if (range.isFullMatch(result, byteCount))
+                    if (range.IsFullMatch(result, byteCount))
                     {
                         return ByteArrayToInt(result, byteCount);
                     }

src/UglyToad.Pdf/Fonts/Cmap/CodespaceRange.cs

@@ -1,5 +1,6 @@
 namespace UglyToad.Pdf.Fonts.Cmap
 {
+    using System;
     using System.Collections.Generic;
 
     /// <summary>
@@ -7,14 +8,23 @@
     /// </summary>
     public class CodespaceRange
     {
+        /// <summary>
+        /// The lower-bound of this range.
+        /// </summary>
         public IReadOnlyList<byte> Start { get; }
 
+        /// <summary>
+        /// The upper-bound of this range.
+        /// </summary>
         public IReadOnlyList<byte> End { get; }
 
         public int StartInt { get; }
 
         public int EndInt { get; }
 
+        /// <summary>
+        /// The number of bytes for numbers in this range.
+        /// </summary>
         public int CodeLength { get; }
 
         /// <summary>
@@ -29,19 +39,29 @@
             CodeLength = start.Count;
         }
 
-        /**
-         * Returns true if the given code bytes match this codespace range.
-         */
-        public bool matches(byte[] code)
+        /// <summary>
+        /// Returns <see langword="true"/> if the given code bytes match this codespace range.
+        /// </summary>
+        public bool Matches(byte[] code)
         {
-            return isFullMatch(code, code.Length);
+            if (code == null)
+            {
+                throw new ArgumentNullException(nameof(code));
             }
 
-        /**
-         * Returns true if the given code bytes match this codespace range.
-         */
-        public bool isFullMatch(byte[] code, int codeLen)
+            return IsFullMatch(code, code.Length);
+        }
+
+        /// <summary>
+        /// Returns true if the given code bytes match this codespace range.
+        /// </summary>
+        public bool IsFullMatch(byte[] code, int codeLen)
         {
+            if (code == null)
+            {
+                throw new ArgumentNullException(nameof(code));
+            }
+
             // code must be the same length as the bounding codes
             if (codeLen == CodeLength)
             {

src/UglyToad.Pdf/Fonts/Composite/Type0Font.cs

@@ -41,7 +41,7 @@
         {
             var current = bytes.CurrentOffset;
 
-            var code = ToUnicode.ReadCode(bytes);
+            var code = CMap.ReadCode(bytes);
 
             codeLength = bytes.CurrentOffset - current;
 
@@ -70,7 +70,19 @@
 
         public PdfVector GetDisplacement(int characterCode)
         {
-            return new PdfVector(0.333m, 0);
+            // This width is in units scaled up by 1000
+            var width = GetWidth(characterCode);
+
+            return new PdfVector(width / 1000, 0);
+        }
+
+        public decimal GetWidth(int characterCode)
+        {
+            var cid = CMap.ConvertToCid(characterCode);
+
+            var fromFont = CidFont.GetWidthFromDictionary(cid);
+            
+            return fromFont;
         }
     }
 }

src/UglyToad.Pdf/Fonts/IFont.cs

@@ -15,5 +15,7 @@
         bool TryGetUnicode(int characterCode, out string value);
 
         PdfVector GetDisplacement(int characterCode);
+
+        decimal GetWidth(int characterCode);
     }
 }

src/UglyToad.Pdf/Graphics/ContentStreamProcessor.cs

@@ -87,6 +87,8 @@
 
                 font.TryGetUnicode(code, out var unicode);
 
+                var width = font.GetWidth(code);
+
                 var wordSpacing = 0m;
                 if (code == ' ' && codeLength == 1)
                 {
@@ -97,7 +99,7 @@
 
                 if (font.IsVertical)
                 {
-                    throw new NotImplementedException("Vertical fonts are currently unsupported, please submit a pull request.");
+                    throw new NotImplementedException("Vertical fonts are currently unsupported, please submit a pull request or issue with an example file.");
                 }
 
                 var displacement = font.GetDisplacement(code);