ParadoxSaveParser/ParadoxSaveParser.Lib/Parser.cs

global using System;
global using System.IO;
global using System.Text;
global using System.Collections.Generic;

namespace ParadoxSaveParser.Lib;

public class Parser
{
    protected Stream _saveFile;
    private List<Token> _tokens = new(4_194_304);
    private int _tokenIndex;

    public Parser(Stream savefile)
    {
        _saveFile = savefile;
    }

    protected enum TokenType : byte
    {
        Invalid,
        String,
        Equals,
        BracketOpen,
        BracketClose,
    }

    protected struct Token
    {
        public TokenType type;
        public short column;
        public int line;
        public string? value;

        public override string ToString()
        {
            string s;
            switch (type)
            {
                case TokenType.Invalid:
                    s = "INVALID_TOKEN";
                    break;
                case TokenType.String:
                    s = value ?? "NULL";
                    break;
                case TokenType.Equals:
                    s = "=";
                    break;
                case TokenType.BracketOpen:
                    s = "{";
                    break;
                case TokenType.BracketClose:
                    s = "}";
                    break;
                default:
                    throw new ArgumentOutOfRangeException(type.ToString());
            }
            
            return $"{line}:{column} '{s}'";
        }
    }

    protected void Lex()
    {
        _tokens.Clear();
        
        string expectedHeader = "EU4txt";
        byte[] headBytes = new byte[expectedHeader.Length];
        _saveFile.ReadExactly(headBytes);
        string headStr = Encoding.UTF8.GetString(headBytes);
        if (headStr != expectedHeader)
            throw new Exception($"Invalid gamestate header: '{headStr}'");
        
        StringBuilder str = new();
        int line = 2;
        int column = 0;
        bool isQuoteOpen = false;
        bool isStrInQuotes = false;

        void CompleteStringToken()
        {
            if (isQuoteOpen)
                return;
            // strings in quotes can be empty
            if (!isStrInQuotes && (str.Length <= 0 || str[0] == '#')) 
                return;
            _tokens.Add(new Token
            {
                type = TokenType.String,
                column = (short)(column - str.Length),
                line = line,
                value = str.ToString()
            });
            str.Clear();
            isStrInQuotes = false;
        }

        while (_saveFile.CanRead)
        {
            int c = _saveFile.ReadByte();
            column++;
            switch (c)
            {
                case -1:
                    CompleteStringToken();
                    return;
                case '\"':
                    isQuoteOpen = !isQuoteOpen;
                    isStrInQuotes = true;
                    break;
                case ' ':
                case '\t':
                case '\r':
                    CompleteStringToken();
                    break;
                case '\n':
                    CompleteStringToken();
                    line++;
                    column = 0;
                    break;
                case '=':
                    CompleteStringToken();
                    _tokens.Add(new Token
                    {
                        type = TokenType.Equals,
                        line = line, column = (short)column
                    });
                    break;
                case '{':
                    CompleteStringToken();
                    _tokens.Add(new Token
                    {
                        type = TokenType.BracketOpen,
                        line = line, column = (short)column
                    });
                    break;
                case '}':
                    CompleteStringToken();
                    _tokens.Add(new Token
                    {
                        type = TokenType.BracketClose,
                        line = line, column = (short)column
                    });
                    break;
                default:
                    str.Append((char)c);
                    break;
            }
        }
    }

    protected class UnexpectedTokenException : Exception
    {
        public UnexpectedTokenException(Token token, int tokenIndex) :
            base($"Unexpected token at index {tokenIndex}: {token}")
        {}
    }
    

    private object? ParseValue()
    {
        Token tok = _tokens[_tokenIndex++];
        switch (tok.type)
        {
            case TokenType.String:
                return tok.value!;
            case TokenType.BracketOpen:
                return ParseListOrDict();
            case TokenType.BracketClose:
                return null;
            default:
                throw new UnexpectedTokenException(tok, _tokenIndex - 1);
        }
    }

    private object ParseListOrDict()
    {
        Token first = _tokens[_tokenIndex];
        Token second = _tokens[_tokenIndex + 1];
        if (first.type == TokenType.String && second.type == TokenType.Equals) 
            return ParseDict();
        
        return ParseList();
    }

    private List<object> ParseList()
    {
        List<object> list = new();
        while(true)
        {
            object? value = ParseValue();
            if (value == null)
                break;
            list.Add(value);
        }
        return list;
    }

    
    private Dictionary<string, List<object>> ParseDict()
    {
        Dictionary<string, List<object>> dict = new();
        // root is a dict without closing bracket, so this method must check _tokenIndex < _tokens.Count
        while (_tokenIndex < _tokens.Count)
        {
            Token tok = _tokens[_tokenIndex++];
            // end of dictionary 
            if (tok.type == TokenType.BracketClose)
                break;
            
            // Saves may contain some blocks without key.
            // Such blocks are skipped because idk where to put them.
            // Example: `technology_group=tech_cannorian{ }
            //          { } { } { }`
            if (tok.type == TokenType.BracketOpen)
            {
                int bracketBalance = 1;
                while (bracketBalance != 0)
                {
                    tok = _tokens[_tokenIndex++];
                    if (tok.type == TokenType.BracketOpen)
                        bracketBalance++;
                    else if (tok.type == TokenType.BracketClose)
                        bracketBalance--;
                }
                
                continue;
            }
            
            if(tok.type != TokenType.String)
                throw new UnexpectedTokenException(tok, _tokenIndex - 1);
            
            string key = tok.value!;
            
            tok = _tokens[_tokenIndex++];
            if (tok.type == TokenType.BracketOpen)
            {
                // Saves may contain key-value definition without `=`.
                // Example: `map_area_data{` instead of `map_area_data = {`
                _tokenIndex--;
            }
            else if(tok.type != TokenType.Equals)
                throw new UnexpectedTokenException(tok, _tokenIndex - 1);
            
            
            object? value = ParseValue();
            if (value == null)
                throw new UnexpectedTokenException(_tokens[_tokenIndex - 1], _tokenIndex - 1);
            if(!dict.TryGetValue(key, out List<object>? list))
            {
                list = new List<object>();
                dict.Add(key, list);
            }
            list.Add(value);
        }

        return dict;
    }
    
    public Dictionary<string, List<object>> Parse()
    {
        Lex();
        if (_tokens.Count == 0)
            throw new Exception("Save file is empty");

        _tokenIndex = 0;
        var root = ParseDict();
        return root;
    }
}