refactored SearchExpression

ParadoxSaveParser.Lib/SaveParserEU4.cs

@@ -14,13 +14,20 @@ public class SaveParserEU4
     private const int DefaultBufferSize = 64 * 1024;
     private static ReadOnlySpan<byte> Header => "EU4txt"u8;
 
+    /// <summary>
+    ///     Windows-1252 is the default encoding of the save files
+    /// </summary>
+    public static Encoding DefaultEncoding => ParadoxEncodings.Windows1252;
+
     private readonly Tokenizer _tokens;
+    private readonly SearchExpressionCompilation? _query;
+
+    // step of the query the parser is currently at, compiled by Parse()
     private ISearchExpression? _searchExprCurrent;
 
     /// <summary>
     ///     Encoding of the strings inside the save. Saves of different localizations use
-    ///     different ones, so it can be changed; <see cref="System.Text.Encoding.Latin1" /> maps
-    ///     every byte to one character and never fails, which makes it a safe default.
+    ///     different ones, so it can be changed; <see cref="DefaultEncoding" /> is what EU4 writes.
     /// </summary>
     public Encoding Encoding { get; }
 
@@ -28,21 +35,23 @@ public class SaveParserEU4
     ///     Uncompressed stream of <c>gamestate</c> file which can be extracted from save archive
     /// </param>
     /// <param name="query">
-    ///     Parsing whole save takes 10 seconds on mid pc and takes 1GB of RAM,
+    ///     Parsing whole save may take a few seconds on mid pc and takes 1GB of RAM,
     ///     so you should specify what exactly you want to get from save file
     /// </param>
-    /// <param name="encoding">Encoding of the strings inside the save. Latin1 by default.</param>
+    /// <param name="encoding">Encoding of the strings inside the save. <see cref="DefaultEncoding"/></param>
     /// <param name="bufferSize">Size of the read buffer. Mostly useful for tests.</param>
-    public SaveParserEU4(Stream savefile, ISearchExpression? query,
+    public SaveParserEU4(Stream savefile, SearchExpressionCompilation? query,
         Encoding? encoding = null, int bufferSize = DefaultBufferSize)
     {
-        Encoding = encoding ?? Encoding.Latin1;
-        _searchExprCurrent = query;
+        Encoding = encoding ?? DefaultEncoding;
+        _query = query;
         _tokens = new Tokenizer(savefile, bufferSize);
     }
 
     public Dictionary<string, object> Parse()
     {
+        // compiled here, so the keys of the query are always encoded like the save
+        _searchExprCurrent = _query?.Compile(Encoding);
         _tokens.ReadHeader(Header, Encoding);
         return ParseDict();
     }
@@ -194,7 +203,7 @@ public class SaveParserEU4
             ISearchExpression? searchExprNext = null;
             bool matches = _searchExprCurrent == null
                            || _searchExprCurrent.DoesMatch(
-                               new MatchCandidate(localIndex, _tokens.Text, Encoding), out searchExprNext);
+                               new MatchCandidate(localIndex, _tokens.Text), out searchExprNext);
             string? keyStr = matches ? DecodeString(_tokens.Text) : null;
 
             // next token should be `=` or `{`