handle messed up numbers in content #355

src/UglyToad.PdfPig.Tests/Tokenization/NumericTokenizerTests.cs

@@ -96,6 +96,18 @@
             Assert.Equal(0m, AssertNumericToken(token).Data);
         }
 
+        [Fact]
+        public void HandleDoubleDashedNumber()
+        {
+            // This is a really weird format but seen in the wild. PDF, shine on, you crazy diamond.
+            var input = StringBytesTestConverter.Convert("--10.25");
+
+            var result = tokenizer.TryTokenize(input.First, input.Bytes, out var token);
+
+            Assert.True(result);
+            Assert.Equal(-10.25m, AssertNumericToken(token).Data);
+        }
+
         [Fact]
         public void HandlesDot()
         {

src/UglyToad.PdfPig.Tests/Tokenization/Scanner/CoreTokenScannerTests.cs

@@ -170,6 +170,32 @@ endobj";
             AssertCorrectToken<NumericToken, decimal>(tokens[3], -91);
         }
 
+        [Fact]
+        public void ScansStringWithWeirdWeirdDoubleSymbolNumerics()
+        {
+            const string content = @"
+                0.00 --21.72 TD
+                /F1 8.00 Tf";
+
+            var tokens = new List<IToken>();
+
+            var scanner = scannerFactory(StringBytesTestConverter.Convert(content, false).Bytes);
+
+            while (scanner.MoveNext())
+            {
+                tokens.Add(scanner.CurrentToken);
+            }
+
+            Assert.Equal(6, tokens.Count);
+
+            AssertCorrectToken<NumericToken, decimal>(tokens[0], 0m);
+            AssertCorrectToken<NumericToken, decimal>(tokens[1], -21.72m);
+            AssertCorrectToken<OperatorToken, string>(tokens[2], "TD");
+            AssertCorrectToken<NameToken, string>(tokens[3], "F1");
+            AssertCorrectToken<NumericToken, decimal>(tokens[4], 8m);
+            AssertCorrectToken<OperatorToken, string>(tokens[5], "Tf");
+
+        }
         private static void AssertCorrectToken<T, TData>(IToken token, TData expected) where T : IDataToken<TData>
         {
             var cast = Assert.IsType<T>(token);

src/UglyToad.PdfPig.Tokenization/NumericTokenizer.cs

@@ -20,28 +20,45 @@
             token = null;
 
             StringBuilder characters;
+            var initialSymbol = currentByte == '-' || currentByte == '+';
 
-            if ((currentByte >= Zero && currentByte <= Nine) || currentByte == '-' || currentByte == '+' || currentByte == '.')
+            if ((currentByte >= Zero && currentByte <= Nine) || currentByte == '.')
             {
                 characters = stringBuilder;
                 characters.Append((char)currentByte);
             }
+            else if (initialSymbol)
+            {
+                characters = stringBuilder;
+                characters.Append((char) currentByte);
+            }
             else
             {
                 return false;
             }
 
+            var previousSymbol = initialSymbol;
+
             while (inputBytes.MoveNext())
             {
                 var b = inputBytes.CurrentByte;
 
-                if ((b >= Zero && b <= Nine) ||
+                if (b == '+' || b == '-')
-                    b == '-' ||
-                    b == '+' ||
-                    b == '.' ||
-                    b == 'E' ||
-                    b == 'e')
                 {
+                    if (previousSymbol)
+                    {
+                        continue;
+                    }
+
+                    characters.Append((char) b);
+                    previousSymbol = true;
+                }
+                else if ((b >= Zero && b <= Nine) ||
+                         b == '.' ||
+                         b == 'E' ||
+                         b == 'e')
+                {
+                    previousSymbol = false;
                     characters.Append((char)b);
                 }
                 else

src/UglyToad.PdfPig/Graphics/ReflectionGraphicsStateOperationFactory.cs

@@ -271,18 +271,8 @@ namespace UglyToad.PdfPig.Graphics
                         throw new InvalidOperationException($"Tried to create a move to next line and show text operation with operand type: {operands[0]?.GetType().Name ?? "null"}");
                     }
                 case MoveToNextLineWithOffset.Symbol:
-                    if (operands.Count < 2)
-                    {
-                        return null;
-                    }
-
                     return new MoveToNextLineWithOffset(OperandToDecimal(operands[0]), OperandToDecimal(operands[1]));
                 case MoveToNextLineWithOffsetSetLeading.Symbol:
-                    if (operands.Count < 2)
-                    {
-                        return null;
-                    }
-
                     return new MoveToNextLineWithOffsetSetLeading(OperandToDecimal(operands[0]), OperandToDecimal(operands[1]));
                 case PaintShading.Symbol:
                     return new PaintShading((NameToken)operands[0]);