Merge pull request #9 from stefannikolei/sn/fix/issue96

Port GenericTests from Rust lib

Author: JimBobSquarePants

src/PolygonClipper/FloatExtensions.cs

@@ -0,0 +1,75 @@
+// Copyright (c) Six Labors.
+// Licensed under the Six Labors Split License.
+
+using System;
+using System.Runtime.CompilerServices;
+
+namespace PolygonClipper;
+
+/// <summary>
+/// Provides extension methods for floating-point numbers.
+/// </summary>
+internal static class FloatExtensions
+{
+    /// <summary>
+    /// Returns the next representable double value in the direction of y.
+    /// </summary>
+    /// <remarks><see href="https://docs.rs/float_next_after/latest/src/float_next_after/lib.rs.html"/></remarks>
+    /// <param name="x">The starting floating-point number.</param>
+    /// <param name="y">The target floating-point number.</param>
+    /// <returns>The next representable value of x towards y.</returns>
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
+    public static double NextAfter(this double x, double y)
+    {
+        // Special cases
+        if (double.IsNaN(x) || double.IsNaN(y))
+        {
+            return double.NaN;
+        }
+
+        if (x == y)
+        {
+            return y;
+        }
+
+        if (double.IsPositiveInfinity(x))
+        {
+            return double.PositiveInfinity;
+        }
+
+        if (double.IsNegativeInfinity(x))
+        {
+            return double.NegativeInfinity;
+        }
+
+        // Handle stepping from zero
+        if (x == 0D)
+        {
+            return Math.CopySign(double.Epsilon, y); // Smallest positive subnormal double
+        }
+
+        // Convert double to raw bits
+        long bits = BitConverter.DoubleToInt64Bits(x);
+
+        // Adjust bits to get the next representable value
+        if ((y > x) == (x > 0D)) // Moving in the same sign direction
+        {
+            bits++;
+        }
+        else
+        {
+            bits--;
+        }
+
+        // Convert bits back to double
+        double next = BitConverter.Int64BitsToDouble(bits);
+
+        // Ensure correct handling of signed zeros
+        if (next == 0D)
+        {
+            return Math.CopySign(next, x);
+        }
+
+        return next;
+    }
+}