test all predefined cmaps and fix bugs

src/UglyToad.Pdf.Tests/Fonts/Parser/CMapParserTests.cs

@@ -1,5 +1,9 @@
 namespace UglyToad.Pdf.Tests.Fonts.Parser
 {
+    using System.Collections.Generic;
+    using System.Diagnostics;
+    using System.IO;
+    using IO;
     using Pdf.Fonts.Parser;
     using Xunit;
 
@@ -84,5 +88,42 @@ end";
             Assert.Equal("a", cmap.BaseFontCharacterMap[68]);
             Assert.Equal("x", cmap.BaseFontCharacterMap[91]);
         }
+
+        [Theory]
+        [MemberData(nameof(PredefinedCMaps))]
+        public void CanParseAllPredefinedCMaps(string resourceName)
+        {
+            Debug.WriteLine("Parsing: " + resourceName);
+            
+            var input = new ByteArrayInputBytes(ReadResourceBytes(resourceName));
+
+            var cmap = cMapParser.Parse(input, false);
+
+            Assert.NotNull(cmap);
+        }
+
+        private static byte[] ReadResourceBytes(string name)
+        {
+            using (var resource = typeof(CMapParser).Assembly.GetManifestResourceStream(name))
+            using (var memoryStream = new MemoryStream())
+            {
+                resource.CopyTo(memoryStream);
+
+                return memoryStream.ToArray();
+            }
+        }
+
+        public static IEnumerable<object[]> PredefinedCMaps()
+        {
+            var resources = typeof(CMapParser).Assembly.GetManifestResourceNames();
+
+            foreach (var resource in resources)
+            {
+                if (resource.Contains(".CMap."))
+                {
+                    yield return new object[] {resource};
+                }
+            }
+        }
     }
 }

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

@@ -97,6 +97,45 @@
             throw new InvalidOperationException("The Character Identifier System Information was never set.");
         }
 
+        public void UseCMap(CMap other)
+        {
+            CodespaceRanges = Combine(CodespaceRanges, other.CodespaceRanges);
+            CidCharacterMappings = Combine(CidCharacterMappings, other.CidCharacterMappings);
+            CidRanges = Combine(CidRanges, other.CidRanges);
+
+            if (other.BaseFontCharacterMap != null)
+            {
+                foreach (var keyValuePair in other.BaseFontCharacterMap)
+                {
+                    BaseFontCharacterMap[keyValuePair.Key] = keyValuePair.Value;
+                }
+            }
+        }
+
+        private static IReadOnlyList<T> Combine<T>(IReadOnlyList<T> a, IReadOnlyList<T> b)
+        {
+            if (a == null && b == null)
+            {
+                return new T[0];
+            }
+
+            if (a == null)
+            {
+                return b;
+            }
+
+            if (b == null)
+            {
+                return a;
+            }
+
+            var result = new List<T>(a);
+
+            result.AddRange(b);
+
+            return result;
+        }
+
         private int GetCodeFromArray(IReadOnlyList<byte> data, int length)
         {
             int code = 0;

src/UglyToad.Pdf/Fonts/Parser/CMapParser.cs

@@ -34,7 +34,19 @@
                     switch (operatorToken.Data)
                     {
                         case "usecmap":
-                            throw new NotImplementedException("External CMap files not yet supported, please submit a pull request!");
+                            {
+                                if (previousToken is NameToken name)
+                                {
+                                    var external = ParseExternal(name.Data.Name);
+
+                                    builder.UseCMap(external);
+                                }
+                                else
+                                {
+                                    throw new InvalidOperationException("Unexpected token preceding external cmap call: " + previousToken);
+                                }
+                                break;
+                            }
                         case "begincodespacerange":
                             {
                                 if (previousToken is NumericToken numeric)