StringBuilderPool reduced memory allocation in 3000 times

ParadoxSaveParser.Lib/BufferedEnumerator.cs

@@ -37,35 +37,73 @@ namespace ParadoxSaveParser.Lib;
 /// 2 3 | 4 | 5
 /// 3 4 | 5 |
 /// </code>
-public class BufferedEnumerator<T> : IEnumerator<LinkedListNode<T>>
+public class BufferedEnumerator<T> : IEnumerator<BufferedEnumerator<T>.Node>
 {
-    private readonly int _bufferSize;
+    public class Node
-    private LinkedListNode<T>? _currentNode;
+    {
-    private int _currentNodeIndex = -1;
+#nullable disable
+        public Node Previous;
+        public Node Next;
+        public T Value;
+#nullable enable
+    }
+    
     private readonly IEnumerator<T> _enumerator;
-    private readonly LinkedList<T> _llist = new();
+    private readonly Node[] _ringBuffer;
+    private Node? _currentNode;
+    private int _currentBufferIndex = -1;
+    private int _lastValueIndex = -1;
 
     public BufferedEnumerator(IEnumerator<T> enumerator, int bufferSize)
     {
         _enumerator = enumerator;
-        _bufferSize = bufferSize;
+        _ringBuffer = new Node[bufferSize];
     }
 
+    private void InitBuffer()
+    {
+        _ringBuffer[0] = new Node
+        {
+            Value = default!
+        };
+        for (int i = 1; i < _ringBuffer.Length; i++)
+        {
+            _ringBuffer[i] = new Node
+            {
+                Previous = _ringBuffer[i - 1],
+                Value = default!,
+            };
+            _ringBuffer[i - 1].Next = _ringBuffer[i];
+        }
+        _ringBuffer[^1].Next = _ringBuffer[0];
+        _ringBuffer[0].Previous = _ringBuffer[^1];
+    }
+    
     public bool MoveNext()
     {
-        if (_currentNodeIndex >= _bufferSize / 2)
+        if (_currentBufferIndex == -1)
-            _llist.RemoveFirst();
-        
-        while (_llist.Count < _bufferSize && _enumerator.MoveNext())
         {
-            _llist.AddLast(_enumerator.Current);
+            InitBuffer();
+            
+            int beforeMidpoint = _ringBuffer.Length / 2 - 1;
+            for (int i = 0; i <= beforeMidpoint && _enumerator.MoveNext(); i++)
+            {
+                _ringBuffer[i].Value = _enumerator.Current;
+            }
         }
-        if (_llist.Count == 0)
+        
+        _currentBufferIndex = (_currentBufferIndex + 1) % _ringBuffer.Length;
+        if (_enumerator.MoveNext())
+        {
+            int midpoint = (_currentBufferIndex + _ringBuffer.Length / 2) % _ringBuffer.Length;
+            _ringBuffer[midpoint].Value = _enumerator.Current;
+            _lastValueIndex = midpoint;
+        }
+        if(_currentBufferIndex == (_lastValueIndex + 1) % _ringBuffer.Length)
             return false;
-
+        
-        _currentNodeIndex++;
+        _currentNode = _ringBuffer[_currentBufferIndex];
-        _currentNode = _currentNode is null ? _llist.First : _currentNode.Next;
+        return true;
-        return _currentNode is not null;
     }
 
     public void Reset()
@@ -73,7 +111,7 @@ public class BufferedEnumerator<T> : IEnumerator<LinkedListNode<T>>
         throw new NotImplementedException();
     }
 
-    public LinkedListNode<T> Current => _currentNode!;
+    public Node Current => _currentNode!;
 
     object IEnumerator.Current => Current;
 

ParadoxSaveParser.Lib/ParadoxSaveParser.Lib.csproj

@@ -4,5 +4,8 @@
         <ImplicitUsings>disable</ImplicitUsings>
         <Nullable>enable</Nullable>
     </PropertyGroup>
+    <ItemGroup>
+      <PackageReference Include="Microsoft.Extensions.ObjectPool" Version="9.0.3" />
+    </ItemGroup>
 
 </Project>

ParadoxSaveParser.Lib/SaveParserEU4.cs

@@ -2,6 +2,7 @@
 global using System.Collections.Generic;
 global using System.IO;
 global using System.Text;
+using Microsoft.Extensions.ObjectPool;
 
 namespace ParadoxSaveParser.Lib;
 
@@ -10,9 +11,13 @@ namespace ParadoxSaveParser.Lib;
 /// </summary>
 public class SaveParserEU4
 {
-    protected Stream _saveFile;
+    protected readonly Stream _saveFile;
-    private ISearchExpression? _searchExprCurrent;
     private readonly BufferedEnumerator<Token> _tokens;
+    private readonly ObjectPool<StringBuilder> _stringBuilderPool;
+    private ISearchExpression? _searchExprCurrent;
+
+    public int SBPoolGetCount = 0;
+    public int SBPoolReturnCount = 0;
 
     /// <param name="savefile">
     ///     Uncompressed stream of <c>gamestate</c> file which can be extracted from save archive
@@ -23,9 +28,16 @@ public class SaveParserEU4
     /// </param>
     public SaveParserEU4(Stream savefile, ISearchExpression? query)
     {
-        _tokens = new BufferedEnumerator<Token>(LexTextSave(), 5);
         _saveFile = savefile;
         _searchExprCurrent = query;
+        const int tokenBufSize = 5;
+        _tokens = new BufferedEnumerator<Token>(LexTextSave(), tokenBufSize);
+        _stringBuilderPool = new DefaultObjectPool<StringBuilder>(
+            new StringBuilderPooledObjectPolicy
+            {
+                InitialCapacity = tokenBufSize * 13,
+                MaximumRetainedCapacity = tokenBufSize * 13,
+            });
     }
 
     protected IEnumerator<Token> LexTextSave()
@@ -37,7 +49,8 @@ public class SaveParserEU4
         if (headStr != expectedHeader)
             throw new Exception($"Invalid gamestate header. Expected '{expectedHeader}', got '{headStr}'.");
 
-        StringBuilder str = new();
+        StringBuilder strb = _stringBuilderPool.Get();
+        SBPoolGetCount++;
         int line = 2;
         int column = 0;
         bool isQuoteOpen = false;
@@ -46,7 +59,8 @@ public class SaveParserEU4
         {
             type = TokenType.Invalid,
             column = -1,
-            line = -1
+            line = -1,
+            value = null,
         };
 
         bool TryCompleteStringToken()
@@ -55,17 +69,18 @@ public class SaveParserEU4
                 return false;
 
             // strings in quotes may be empty
-            if (!isStrInQuotes && (str.Length <= 0 || str[0] == '#'))
+            if (!isStrInQuotes && (strb.Length <= 0 || strb[0] == '#'))
                 return false;
 
             strToken = new Token
             {
                 type = TokenType.StringOrNumber,
-                column = (short)(column - str.Length),
+                column = (short)(column - strb.Length),
                 line = line,
-                value = str.ToString()
+                value = strb,
             };
-            str.Clear();
+            strb = _stringBuilderPool.Get();
+            SBPoolGetCount++;
             isStrInQuotes = false;
             return true;
         }
@@ -79,6 +94,8 @@ public class SaveParserEU4
                 case -1:
                     if (TryCompleteStringToken())
                         yield return strToken;
+                    _stringBuilderPool.Return(strb);
+                    SBPoolReturnCount++;
                     yield break;
                 case '\"':
                     isQuoteOpen = !isQuoteOpen;
@@ -127,10 +144,13 @@ public class SaveParserEU4
                     // Skip control characters, which are invisible and causing frontend bugs.
                     // I dont know why there are so many of them in strings.
                     if (c >= 0x20)
-                        str.Append((char)c);
+                        strb.Append((char)c);
                     break;
             }
         }
+
+        _stringBuilderPool.Return(strb);
+        SBPoolReturnCount++;
     }
 
 
@@ -141,15 +161,16 @@ public class SaveParserEU4
         switch (tok.type)
         {
             case TokenType.StringOrNumber:
-                if (string.IsNullOrEmpty(tok.value))
+                string tokStr = tok.value!.ToString();
-                    return string.Empty;
+                _stringBuilderPool.Return(tok.value);
-                if (tok.value[0] != '-' && !char.IsDigit(tok.value[0]))
+                SBPoolReturnCount++;
-                    return tok.value;
+                if (tokStr[0] != '-' && !char.IsDigit(tokStr[0]))
-                if (tok.value.Contains('.') && double.TryParse(tok.value, out double d))
+                    return tokStr;
+                if (tokStr.Contains('.') && double.TryParse(tokStr, out double d))
                     return d;
-                if (long.TryParse(tok.value, out long l))
+                if (long.TryParse(tokStr, out long l))
                     return l;
-                return tok.value;
+                return tokStr;
             case TokenType.BracketOpen:
                 object obj = ParseListOrDict();
                 return obj;
@@ -166,13 +187,22 @@ public class SaveParserEU4
     private bool SkipValue()
     {
         var tok = _tokens.Current.Value;
-        if (tok.type == TokenType.BracketOpen)
+        switch (tok.type)
         {
-            SkipObject();
+            case TokenType.BracketOpen:
-            return true;
+                SkipObject();
+                return true;
+            case TokenType.StringOrNumber:
+                _stringBuilderPool.Return(tok.value!);
+                SBPoolReturnCount++;
+                return true;
+            case TokenType.Equals:
+                return true;
+            case TokenType.BracketClose:
+                return false;
+            default:
+                throw new UnexpectedTokenException(tok);
         }
-
-        return tok.type != TokenType.BracketClose;
     }
 
     // skips all tokens inside curly braces block
@@ -185,6 +215,11 @@ public class SaveParserEU4
                 bracketBalance++;
             else if (tok.type == TokenType.BracketClose)
                 bracketBalance--;
+            else if (tok.type == TokenType.StringOrNumber)
+            {
+                _stringBuilderPool.Return(tok.value!);
+                SBPoolReturnCount++;
+            }
         }
     }
 
@@ -242,7 +277,7 @@ public class SaveParserEU4
             if (tok.type != TokenType.StringOrNumber)
                 throw new UnexpectedTokenException(tok);
 
-            string key = tok.value!;
+            var keySB = tok.value!;
 
             // next token should be `=` or `{`
             if (!_tokens.MoveNext())
@@ -263,9 +298,11 @@ public class SaveParserEU4
 
             ISearchExpression? searchExprNext = null;
             if (_searchExprCurrent != null
-                && !_searchExprCurrent.DoesMatch(new SearchArgs(key, localIndex), out searchExprNext))
+                && !_searchExprCurrent.DoesMatch(new SearchArgs(localIndex, keySB), out searchExprNext))
             {
                 SkipValue();
+                _stringBuilderPool.Return(keySB);
+                SBPoolReturnCount++;
                 continue;
             }
 
@@ -276,10 +313,13 @@ public class SaveParserEU4
                 throw new UnexpectedTokenException(_tokens.Current.Value);
             _searchExprCurrent = searExpressionPrevious;
 
-            if (!dict.TryGetValue(key, out var list))
+            string keyStr = keySB.ToString();
+            _stringBuilderPool.Return(keySB);
+            SBPoolReturnCount++;
+            if (!dict.TryGetValue(keyStr, out var list))
             {
                 list = new List<object>();
-                dict.Add(key, list);
+                dict.Add(keyStr, list);
             }
 
             list.Add(value);
@@ -308,7 +348,7 @@ public class SaveParserEU4
         public required TokenType type;
         public required short column;
         public required int line;
-        public string? value;
+        public StringBuilder? value;
 
         public override string ToString()
         {
@@ -319,7 +359,9 @@ public class SaveParserEU4
                     s = "INVALID_TOKEN";
                     break;
                 case TokenType.StringOrNumber:
-                    s = value ?? "NULL";
+                    if (value == null || value.Length == 0)
+                        s = "NULL";
+                    else s = value.ToString();
                     break;
                 case TokenType.Equals:
                     s = "=";

ParadoxSaveParser.Lib/SearchExpression.cs

@@ -1,6 +1,25 @@
 namespace ParadoxSaveParser.Lib;
 
-public record SearchArgs(string key, int localIndex);
+public record SearchArgs
+{
+    public readonly string KeyStr;
+    public readonly StringBuilder? KeySB;
+    public readonly int LocalIndex;
+
+    public SearchArgs(int localIndex, string keyStr)
+    {
+        KeyStr = keyStr;
+        KeySB = null;
+        LocalIndex = localIndex;
+    }
+    
+    public SearchArgs(int localIndex, StringBuilder keySb)
+    {
+        KeyStr = string.Empty;
+        KeySB = keySb;
+        LocalIndex = localIndex;
+    }
+}
 
 public interface ISearchExpression
 {
@@ -109,7 +128,7 @@ public static class SearchExpressionCompiler
     {
         public bool DoesMatch(SearchArgs args, out ISearchExpression? nextSearchExpression)
         {
-            if (args.localIndex == index)
+            if (args.LocalIndex == index)
             {
                 nextSearchExpression = next;
                 return true;
@@ -124,7 +143,7 @@ public static class SearchExpressionCompiler
     {
         public bool DoesMatch(SearchArgs args, out ISearchExpression? nextSearchExpression)
         {
-            if (args.key == key)
+            if ((args.KeySB != null && args.KeySB.Equals(key)) || args.KeyStr == key)
             {
                 nextSearchExpression = next;
                 return true;

ParadoxSaveParser.WebAPI/Program.cs

@@ -12,6 +12,7 @@ global using Directory = DTLib.Filesystem.Directory;
 global using File = DTLib.Filesystem.File;
 global using Path = DTLib.Filesystem.Path;
 using System.Collections.Concurrent;
+using System.Diagnostics;
 using System.IO;
 using System.Text.Encodings.Web;
 using DTLib.Dtsod;
@@ -54,6 +55,21 @@ public static partial class Program
 
         try
         {
+            Stopwatch stopwatch = new();
+            using var save = File.OpenRead("data/gamestate");
+            stopwatch.Start();
+            var parser = new SaveParserEU4(save, SearchExpressionCompiler.Compile("saved_event_target"));
+            var result = parser.Parse();
+            stopwatch.Stop();
+            using (var resultFile = File.OpenWrite("data/parsed.json"))
+            {
+                JsonSerializer.Serialize(resultFile, result, _saveSerializerOptions);
+            }
+            Console.WriteLine($"get: {parser.SBPoolGetCount} return: {parser.SBPoolReturnCount} " +
+                              $"delta: {parser.SBPoolGetCount - parser.SBPoolReturnCount}");
+            Console.WriteLine(stopwatch.Elapsed);
+            return;
+            
             // config
             if (!File.Exists(_configPath))
             {