added benchmarks

2026-09-14 23:09:25 +02:00
parent f1aca261ae
commit 3dde5e5acf
11 changed files with 823 additions and 18 deletions
@@ -0,0 +1,80 @@
+using System;
+using System.IO;
+using System.Linq;
+using System.Text.RegularExpressions;
+using BenchmarkDotNet.Running;
+using PCRE;
+using ParadoxSaveParser.Lib;
+
+namespace ParadoxSaveParser.Benchmarks;
+
+public static class Program
+{
+    public static void Main(string[] args)
+    {
+        switch (args.FirstOrDefault())
+        {
+            case "prepare":
+                BenchData.PrepareJson();
+                return;
+            case "verify":
+                Verify();
+                return;
+            default:
+                BenchmarkSwitcher.FromTypes([typeof(ExtractionBenchmarks), typeof(JqBenchmarks), typeof(ParserInputBenchmarks)])
+                    .Run(args);
+                return;
+        }
+    }
+
+    /// <summary>
+    ///     Runs every extractor once and prints its checksum and wall time,
+    ///     so that differences in what each engine actually finds are visible before benchmarking.
+    /// </summary>
+    private static void Verify()
+    {
+        Console.WriteLine($"save:  {BenchData.PdxPath}  {new FileInfo(BenchData.PdxPath).Length / 1024 / 1024} MB");
+        Console.WriteLine($"json:  {BenchData.JsonPath}  {new FileInfo(BenchData.JsonPath).Length / 1024 / 1024} MB");
+
+        var pdx = BenchData.PdxBytes;
+        var text = BenchData.PdxText;
+        var json = File.ReadAllBytes(BenchData.JsonPath);
+        var query = SearchExpressionCompiler.Compile(BenchData.PdxQuery);
+
+        Time("SearchExpression", () => Extractors.SearchExpression(pdx, query));
+        Time("FullParse", () => Extractors.FullParseThenSelect(pdx));
+        Time("Regex flat compiled",
+            () => Extractors.RegexScan(new Regex(Extractors.FlatPattern, RegexOptions.Compiled), text));
+        Time("Regex path aware compiled",
+            () => Extractors.RegexScan(new Regex(Extractors.PathAwarePattern, RegexOptions.Compiled), text));
+        Time("Regex path aware nonbacktracking",
+            () => Extractors.RegexScan(new Regex(Extractors.PathAwarePattern, RegexOptions.NonBacktracking), text));
+        Time("Regex balanced two stage",
+            () => Extractors.RegexTwoStage(new Regex(Extractors.CountriesBlockPattern, RegexOptions.Compiled),
+                new Regex(Extractors.FlatPattern, RegexOptions.Compiled), text));
+        Time("PCRE flat", () => Extractors.PcreScan(new PcreRegex(Extractors.FlatPattern, PcreOptions.Compiled), text));
+        Time("PCRE path aware",
+            () => Extractors.PcreScan(new PcreRegex(Extractors.PathAwarePattern, PcreOptions.Compiled), text));
+        Time("Utf8JsonReader", () => Extractors.Utf8JsonReaderScan(json));
+        Mispairing.Report(pdx, text);
+        Time("JsonDocument", () => Extractors.JsonDocumentScan(json));
+    }
+
+    private static void Time(string name, Func<long> f)
+    {
+        var sw = System.Diagnostics.Stopwatch.StartNew();
+        long checksum;
+        try
+        {
+            checksum = f();
+        }
+        catch (Exception ex)
+        {
+            Console.WriteLine($"{name,-34} FAILED: {ex.Message}");
+            return;
+        }
+
+        sw.Stop();
+        Console.WriteLine($"{name,-34} countries={checksum / 1_000_000,5}  techSum={checksum % 1_000_000,7}  {sw.ElapsedMilliseconds,7} ms");
+    }
+}