Fix GPerf hashing

Fix bug where hashing would increment idx twice
Make a version of the hash that does not allocate a new string

Author: Genbox

Src/FastData.Generator.CPlusPlus.Tests/Generated/PerfectHashGPerf_String_100.output

@@ -1,3 +1,5 @@
+// This file is auto-generated. Do not edit manually.
+// Structure: PerfectHashGPerf
 #include <string>
 #include <array>
 #include <cstdint>
@@ -95,4 +97,6 @@ public:
 
 
     static constexpr int item_count = 100;
+    static constexpr int min_length = 1;
+    static constexpr int max_length = 2;
 };

Src/FastData.Generator.CPlusPlus.Tests/Generated/PerfectHashGPerf_String_3.output

@@ -1,3 +1,5 @@
+// This file is auto-generated. Do not edit manually.
+// Structure: PerfectHashGPerf
 #include <string>
 #include <array>
 #include <cstdint>
@@ -58,4 +60,6 @@ public:
 
 
     static constexpr int item_count = 3;
+    static constexpr int min_length = 5;
+    static constexpr int max_length = 5;
 };

Src/FastData.Generator.CPlusPlus.Tests/Generated/PerfectHashGPerf_String_7.output

@@ -1,3 +1,5 @@
+// This file is auto-generated. Do not edit manually.
+// Structure: PerfectHashGPerf
 #include <string>
 #include <array>
 #include <cstdint>
@@ -81,4 +83,6 @@ public:
 
 
     static constexpr int item_count = 7;
+    static constexpr int min_length = 3;
+    static constexpr int max_length = 10;
 };

Src/FastData.Generator.CSharp.Tests/Generated/PerfectHashGPerf_String_100.output

@@ -1,14 +1,11 @@
 // <auto-generated />
 // Structure: PerfectHashGPerf
 #nullable enable
-using Genbox.FastData.Abstracts;
-using Genbox.FastData.Generator.CSharp.Abstracts;
-using Genbox.FastData.Helpers;
-using Genbox.FastData;
+using System;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
-using System.Text;
-using System;
+using Genbox.FastData.Helpers;
+using Genbox.FastData.Generator.CSharp.Abstracts;
 
 
 internal partial class PerfectHashGPerf_String_100 : IFastSet<String>

Src/FastData.Generator.CSharp.Tests/Generated/PerfectHashGPerf_String_3.output

@@ -1,14 +1,11 @@
 // <auto-generated />
 // Structure: PerfectHashGPerf
 #nullable enable
-using Genbox.FastData.Abstracts;
-using Genbox.FastData.Generator.CSharp.Abstracts;
-using Genbox.FastData.Helpers;
-using Genbox.FastData;
+using System;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
-using System.Text;
-using System;
+using Genbox.FastData.Helpers;
+using Genbox.FastData.Generator.CSharp.Abstracts;
 
 
 internal partial class PerfectHashGPerf_String_3 : IFastSet<String>

Src/FastData.Generator.CSharp.Tests/Generated/PerfectHashGPerf_String_7.output

@@ -1,14 +1,11 @@
 // <auto-generated />
 // Structure: PerfectHashGPerf
 #nullable enable
-using Genbox.FastData.Abstracts;
-using Genbox.FastData.Generator.CSharp.Abstracts;
-using Genbox.FastData.Helpers;
-using Genbox.FastData;
+using System;
 using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
-using System.Text;
-using System;
+using Genbox.FastData.Helpers;
+using Genbox.FastData.Generator.CSharp.Abstracts;
 
 
 internal partial class PerfectHashGPerf_String_7 : IFastSet<String>

Src/FastData.Tests/HashSpecTests.cs

@@ -27,6 +27,6 @@ public static IEnumerable<object[]> GetSpecs()
         yield return [new GeneticHashSpec(2, 1, 2, 1, [new StringSegment(0, -1, Alignment.Left)]), 401880771];
 
         yield return [new HeuristicHashSpec([1]), 101];
-        yield return [new HeuristicHashSpec([0, 1]), 104];
+        yield return [new HeuristicHashSpec([0, 1]), 1765];
     }
 }

Src/FastData/HashFunctions/PJWHash.cs

@@ -41,6 +41,7 @@ the hash is a perfect hash.
  */
 
 /// <summary>Finds the least number of positions in a string that hashes to a unique value for all inputs.</summary>
+[SuppressMessage("Performance", "MA0159:Use \'Order\' instead of \'OrderBy\'")]
 internal class HeuristicAnalyzer(object[] data, StringProperties props, HeuristicAnalyzerConfig config, Simulator simulator) : IHashAnalyzer<HeuristicHashSpec>
 {
     public Candidate<HeuristicHashSpec> Run()
@@ -232,7 +233,6 @@ private void MergePositions(int max, HashSet<int> mandatory, ref HashSet<int> cu
 
     private double CalculateFitness(HashSet<int> set) => CalculateFitnessInternal(set).Fitness * -1; //The algorithm here works with less fitness = better. At least for now.
 
-    [SuppressMessage("Performance", "MA0159:Use \'Order\' instead of \'OrderBy\'")]
     private Candidate<HeuristicHashSpec> CalculateFitnessInternal(HashSet<int> set)
     {
         Candidate<HeuristicHashSpec> cand = new Candidate<HeuristicHashSpec>(new HeuristicHashSpec(set.OrderBy(x => x).ToArray()));
@@ -241,7 +241,6 @@ private Candidate<HeuristicHashSpec> CalculateFitnessInternal(HashSet<int> set)
     }
 
     [Conditional("DebugPrint")]
-    [SuppressMessage("Performance", "MA0159:Use \'Order\' instead of \'OrderBy\'")]
     private void Print(string stage, HashSet<int> set)
     {
         Candidate<HeuristicHashSpec> cand = CalculateFitnessInternal(set);

Src/FastData/Internal/Analysis/Analyzers/Heuristics/HeuristicAnalyzer.cs

@@ -10,12 +10,57 @@ internal readonly record struct HeuristicHashSpec(int[] Positions) : IHashSpec
     public HashFunc GetHashFunction()
     {
         int[] localPos = Positions;
-        return x => PJWHash.Hash(x, localPos);
+        return x => Hash(x, localPos);
     }
 
     public EqualFunc GetEqualFunction()
     {
         int[] localPos = Positions;
-        return (a, b) => PJWHash.GetString(a, localPos) == PJWHash.GetString(b, localPos);
+        return (a, b) => Equal(a, b, localPos);
+    }
+
+    private static bool Equal(string a, string b, int[] positions)
+    {
+        foreach (int pos in positions)
+        {
+            if (pos == -1) //This if-case should come first, or else it will overlap with the next
+            {
+                if (a[a.Length - 1] != b[b.Length - 1])
+                    return false;
+            }
+            else if (pos <= a.Length - 1 && pos <= b.Length - 1)
+            {
+                if (a[pos] != b[pos])
+                    return false;
+            }
+        }
+
+        return true;
+    }
+
+    private static uint Hash(string input, int[] positions)
+    {
+        //This hash function is PJW hash
+
+        uint h = 0;
+        foreach (int pos in positions)
+        {
+            char c;
+
+            if (pos == -1)
+                c = input[input.Length - 1];
+            else if (pos <= input.Length - 1)
+                c = input[pos];
+            else
+                continue;
+
+            h = (h << 4) + c;
+
+            uint high = h & 0xf0000000;
+
+            if (high != 0)
+                h = h ^ (high >> 24) ^ high;
+        }
+        return h;
     }
 }