Refactor a bit

Author: Genbox

Src/FastData.Benchmarks/Benchmarks/AnalyzerBenchmarks.cs

@@ -18,7 +18,7 @@ public AnalyzerBenchmarks()
         _data = Enumerable.Range(1, 100).Select(_ => TestHelper.GenerateRandomString(rng, 50)).ToArray();
         StringProperties props = DataAnalyzer.GetStringProperties(_data);
 
-        _analyzer = new GPerfAnalyzer(_data.Length, props, new GPerfAnalyzerConfig(), new Simulator(new SimulatorConfig()), NullLogger<GPerfAnalyzer>.Instance);
+        _analyzer = new GPerfAnalyzer(_data.Length, props, new GPerfAnalyzerConfig(), new Simulator(), NullLogger<GPerfAnalyzer>.Instance);
     }
 
     [Benchmark]

….Benchmarks/Benchmarks/HashBenchmarks.cs …marks/Benchmarks/StringHashBenchmarks.csSrc/FastData.Benchmarks/Benchmarks/HashBenchmarks.cs renamed to Src/FastData.Benchmarks/Benchmarks/StringHashBenchmarks.cs Src/FastData.Benchmarks/Benchmarks/HashBenchmarks.cs renamed to Src/FastData.Benchmarks/Benchmarks/StringHashBenchmarks.cs

@@ -8,7 +8,7 @@
 namespace Genbox.FastData.Benchmarks.Benchmarks;
 
 [Orderer(SummaryOrderPolicy.FastestToSlowest)]
-public class HashBenchmarks
+public class StringHashBenchmarks
 {
     private string[] _array = null!;
 

Src/FastData.Testbed/Tests/AnalysisTest.cs

@@ -1,5 +1,6 @@
 using System.Diagnostics;
 using System.Runtime.CompilerServices;
+using Genbox.FastData.Internal.Abstracts;
 using Genbox.FastData.Internal.Analysis;
 using Genbox.FastData.Internal.Analysis.Analyzers;
 using Genbox.FastData.Internal.Analysis.Properties;
@@ -66,12 +67,9 @@ private static void RunBruteForce(string[] data, [CallerArgumentExpression(nameo
 
         StringProperties props = DataAnalyzer.GetStringProperties(data);
         using SerilogLoggerFactory loggerFactory = new SerilogLoggerFactory(_logConf);
-        BruteForceAnalyzer analyzer = new BruteForceAnalyzer(props, new BruteForceAnalyzerConfig(), new Simulator(new SimulatorConfig()), loggerFactory.CreateLogger<BruteForceAnalyzer>());
-        analyzer.GetCandidates(data, candidate =>
-        {
-            PrintCandidate(candidate);
-            return false;
-        });
+        BruteForceAnalyzer analyzer = new BruteForceAnalyzer(props, new BruteForceAnalyzerConfig(), new Simulator(), loggerFactory.CreateLogger<BruteForceAnalyzer>());
+        PrintCandidate(GetCandidates(analyzer).OrderByDescending(x => x.Fitness).FirstOrDefault());
+
     }
 
     private static void RunGeneticAnalysis(string[] data, [CallerArgumentExpression(nameof(data))]string? source = null)
@@ -80,12 +78,9 @@ private static void RunGeneticAnalysis(string[] data, [CallerArgumentExpression(
 
         StringProperties props = DataAnalyzer.GetStringProperties(data);
         using SerilogLoggerFactory loggerFactory = new SerilogLoggerFactory(_logConf);
-        GeneticAnalyzer analyzer = new GeneticAnalyzer(props, new GeneticAnalyzerConfig(), new Simulator(new SimulatorConfig()), loggerFactory.CreateLogger<GeneticAnalyzer>());
-        analyzer.GetCandidates(data, candidate =>
-        {
-            PrintCandidate(candidate);
-            return false;
-        });
+        GeneticAnalyzer analyzer = new GeneticAnalyzer(props, new GeneticAnalyzerConfig(), new Simulator(), loggerFactory.CreateLogger<GeneticAnalyzer>());
+        PrintCandidate(GetCandidates(analyzer).OrderByDescending(x => x.Fitness).FirstOrDefault());
+
     }
 
     private static void RunGPerfAnalysis(string[] data, [CallerArgumentExpression(nameof(data))]string? source = null)
@@ -94,17 +89,13 @@ private static void RunGPerfAnalysis(string[] data, [CallerArgumentExpression(na
 
         StringProperties props = DataAnalyzer.GetStringProperties(data);
         using SerilogLoggerFactory loggerFactory = new SerilogLoggerFactory(_logConf);
-        GPerfAnalyzer analyzer = new GPerfAnalyzer(data.Length, props, new GPerfAnalyzerConfig(), new Simulator(new SimulatorConfig()), loggerFactory.CreateLogger<GPerfAnalyzer>());
-        analyzer.GetCandidates(data, candidate =>
-        {
-            PrintCandidate(candidate);
-            return false;
-        });
+        GPerfAnalyzer analyzer = new GPerfAnalyzer(data.Length, props, new GPerfAnalyzerConfig(), new Simulator(), loggerFactory.CreateLogger<GPerfAnalyzer>());
+        PrintCandidate(GetCandidates(analyzer).OrderByDescending(x => x.Fitness).FirstOrDefault());
     }
 
-    private static void Print(string[] data, string? source)
+    private static void Print(string[] data, string? source, [CallerMemberName]string? member = null)
     {
-        Console.WriteLine("###############");
+        Console.WriteLine($"############### {member} ###############");
         Console.WriteLine(source + ": " + string.Join(", ", data.Take(5)));
     }
 
@@ -155,8 +146,11 @@ private static string[] RunFunc(string[] str, double factor, Func<string, double
         return res;
     }
 
-    private static void PrintCandidate(Candidate candidate)
+    private static void PrintCandidate(Candidate? candidate)
     {
+        if (candidate == null)
+            return;
+
         Console.WriteLine();
         Console.WriteLine("#### Candidate ####");
         Console.WriteLine($"{nameof(candidate.Fitness)}: {candidate.Fitness}");
@@ -175,4 +169,15 @@ private static void PrintCandidate(Candidate candidate)
         // Console.WriteLine(candidate.StringHash.GetExpression().ToReadableString());
         Console.WriteLine(ExpressionHelper.Print(candidate.StringHash.GetExpression()));
     }
+
+    private static IEnumerable<Candidate> GetCandidates(IStringHashAnalyzer analyzer)
+    {
+        List<Candidate> candidates = new List<Candidate>();
+        analyzer.GetCandidates(Data, candidate =>
+        {
+            candidates.Add(candidate);
+            return true;
+        });
+        return candidates;
+    }
 }

Src/FastData.Testbed/Tests/GPerfTest.cs

@@ -42,7 +42,7 @@ public static void ProduceOutputs(string path)
             ReadOnlySpan<string> data = File.ReadAllLines(file).AsSpan();
             StringProperties props = DataAnalyzer.GetStringProperties(data);
 
-            GPerfAnalyzer analyzer = new GPerfAnalyzer(data.Length, props, new GPerfAnalyzerConfig(), new Simulator(new SimulatorConfig()), factory.CreateLogger<GPerfAnalyzer>());
+            GPerfAnalyzer analyzer = new GPerfAnalyzer(data.Length, props, new GPerfAnalyzerConfig(), new Simulator(), factory.CreateLogger<GPerfAnalyzer>());
 
             // Candidate hashFunc = analyzer.GetCandidates(data).First(); //TODO: use
 

Src/FastData/FastDataGenerator.cs

@@ -175,9 +175,9 @@ private static ReadOnlySpan<T> Cast<T>(this ReadOnlySpan<object> data) where T :
         return newArr;
     }
 
-    private static IStringHash GetBestHash(string[] data, StringProperties props, SimulatorConfig simConf, ILoggerFactory factory)
+    private static IStringHash GetBestHash(string[] data, StringProperties props, ILoggerFactory factory)
     {
-        Simulator sim = new Simulator(simConf);
+        Simulator sim = new Simulator();
 
         //Run each of the analyzers
         List<Candidate> candidates = new List<Candidate>();

Src/FastData/Generators/StringHash/GPerfStringHash.cs

@@ -109,7 +109,7 @@ private UnaryExpression GetPosition(Expression asso, Expression bytes, int pos)
 
     public override string ToString() =>
         $"""
-         Asso = {string.Join(", ", AssociationValues)}
+         Asso = {string.Join(", ", AssociationValues.Where(x => x != AssociationValues[0]))}
          Alpha = {string.Join(", ", AlphaIncrements)}
          {nameof(Positions)} = {string.Join(", ", Positions)}
          {nameof(MinLen)} = {MinLen}

Src/FastData/Internal/Analysis/Analyzers/Simulator.cs

@@ -3,13 +3,13 @@
 
 namespace Genbox.FastData.Internal.Analysis.Analyzers;
 
-internal sealed class Simulator(SimulatorConfig config)
+internal sealed class Simulator(int capacityFactor = 1)
 {
     private NoEqualityEmulator? _set;
 
     internal Candidate Run(ReadOnlySpan<string> data, IStringHash stringHash, Func<double>? extraFitness = null)
     {
-        uint _capacity = (uint)(data.Length * config.CapacityFactor);
+        uint _capacity = (uint)(data.Length * capacityFactor);
         _set ??= new NoEqualityEmulator(_capacity);
         _set.SetHash(stringHash.GetHashFunction());