source: branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/Grammar.cs @ 15832

using System.Collections.Generic;
using System.Diagnostics;
using System.Linq;
using HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration.GrammarEnumeration;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.Problems.DataAnalysis.Symbolic;

namespace HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration {
  public class Grammar {

    public Symbol StartSymbol { get; }

    public Hasher<int> Hasher { get; }

    #region Symbols
    public VariableSymbol Var;

    public NonterminalSymbol Expr;
    public NonterminalSymbol Term;
    public NonterminalSymbol Factor;
    public NonterminalSymbol LogFactor;
    public NonterminalSymbol ExpFactor;
    public NonterminalSymbol SinFactor;
    public NonterminalSymbol CosFactor;

    public NonterminalSymbol SimpleExpr;
    public NonterminalSymbol SimpleTerm;

    public NonterminalSymbol InvExpr;
    public NonterminalSymbol InvTerm;

    public TerminalSymbol Addition;
    public TerminalSymbol Multiplication;
    public TerminalSymbol Log;
    public TerminalSymbol Exp;
    public TerminalSymbol Sin;
    public TerminalSymbol Cos;
    public TerminalSymbol Inv;

    // For infix notation
    public TerminalSymbol OpeningBracket;
    public TerminalSymbol ClosingBracket;

    #endregion

    #region HL Symbols for Parsing ExpressionTrees

    private ISymbol constSy;
    private ISymbol varSy;

    private ISymbol addSy;
    private ISymbol mulSy;
    private ISymbol logSy;
    private ISymbol expSy;
    private ISymbol divSy;
    private ISymbol sinSy;
    private ISymbol cosSy;

    private ISymbol rootSy;
    private ISymbol startSy;

    private InfixExpressionFormatter infixExpressionFormatter;
    #endregion

    public Grammar(string[] variables) {
      #region Define Symbols
      Var = new VariableSymbol("var", variables);

      Expr = new NonterminalSymbol("Expr");
      Term = new NonterminalSymbol("Term");
      Factor = new NonterminalSymbol("Factor");
      LogFactor = new NonterminalSymbol("LogFactor");
      ExpFactor = new NonterminalSymbol("ExpFactor");
      SinFactor = new NonterminalSymbol("SinFactor");
      CosFactor = new NonterminalSymbol("CosFactor");

      SimpleExpr = new NonterminalSymbol("SimpleExpr");
      SimpleTerm = new NonterminalSymbol("SimpleTerm");

      InvExpr = new NonterminalSymbol("InvExpr");
      InvTerm = new NonterminalSymbol("InvTerm");

      Addition = new TerminalSymbol("+");
      Multiplication = new TerminalSymbol("*");
      Log = new TerminalSymbol("log");
      Exp = new TerminalSymbol("exp");
      Sin = new TerminalSymbol("sin");
      Cos = new TerminalSymbol("cos");
      Inv = new TerminalSymbol("inv");

      OpeningBracket = new TerminalSymbol("(");
      ClosingBracket = new TerminalSymbol(")");
      #endregion

      #region Production rules
      StartSymbol = Expr;

      Expr.AddProduction(Term, Expr, Addition);
      Expr.AddProduction(Term);

      Term.AddProduction(Factor, Term, Multiplication);
      Term.AddProduction(Factor);
      Term.AddProduction(InvExpr, Inv);

      Factor.AddProduction(Var);
      Factor.AddProduction(LogFactor);
      Factor.AddProduction(ExpFactor);
      Factor.AddProduction(SinFactor);
      Factor.AddProduction(CosFactor);

      LogFactor.AddProduction(SimpleExpr, Log);
      ExpFactor.AddProduction(SimpleTerm, Exp);
      SinFactor.AddProduction(SimpleExpr, Sin);
      CosFactor.AddProduction(SimpleExpr, Cos);

      SimpleExpr.AddProduction(SimpleTerm, SimpleExpr, Addition);
      SimpleExpr.AddProduction(SimpleTerm);

      SimpleTerm.AddProduction(Var, SimpleTerm, Multiplication);
      SimpleTerm.AddProduction(Var);

      InvExpr.AddProduction(InvTerm, InvExpr, Addition);
      InvExpr.AddProduction(InvTerm);

      InvTerm.AddProduction(Factor, InvTerm, Multiplication);
      InvTerm.AddProduction(Factor);
      #endregion

      #region Parsing to SymbolicExpressionTree
      var symbolicExpressionGrammar = new TypeCoherentExpressionGrammar();
      symbolicExpressionGrammar.ConfigureAsDefaultRegressionGrammar();

      constSy = symbolicExpressionGrammar.Symbols.OfType<Constant>().First();
      varSy = symbolicExpressionGrammar.Symbols.OfType<Variable>().First();
      addSy = symbolicExpressionGrammar.Symbols.OfType<Addition>().First();
      mulSy = symbolicExpressionGrammar.Symbols.OfType<Multiplication>().First();
      logSy = symbolicExpressionGrammar.Symbols.OfType<Logarithm>().First();
      expSy = symbolicExpressionGrammar.Symbols.OfType<Exponential>().First();
      divSy = symbolicExpressionGrammar.Symbols.OfType<Division>().First();
      sinSy = symbolicExpressionGrammar.Symbols.OfType<Sine>().First();
      cosSy = symbolicExpressionGrammar.Symbols.OfType<Cosine>().First();

      rootSy = symbolicExpressionGrammar.Symbols.OfType<ProgramRootSymbol>().First();
      startSy = symbolicExpressionGrammar.Symbols.OfType<StartSymbol>().First();

      infixExpressionFormatter = new InfixExpressionFormatter();
      #endregion

      Hasher = new IntHasher(this);
    }

    #region Parse to SymbolicExpressionTree

    public string ToInfixString(SymbolString sentence) {
      Debug.Assert(sentence.Any(), "Trying to evaluate empty sentence!");
      Debug.Assert(sentence.All(s => s is TerminalSymbol), "Trying to evaluate symbol sequence with nonterminalsymbols!");

      return infixExpressionFormatter.Format(ParseSymbolicExpressionTree(sentence));
    }

    public SymbolicExpressionTree ParseSymbolicExpressionTree(SymbolString sentence) {
      Debug.Assert(sentence.Any(), "Trying to evaluate empty sentence!");
      Debug.Assert(sentence.All(s => s is TerminalSymbol), "Trying to evaluate symbol sequence with nonterminalsymbols!");

      var rootNode = rootSy.CreateTreeNode();
      var startNode = startSy.CreateTreeNode();
      rootNode.AddSubtree(startNode);

      Stack<TerminalSymbol> parseStack = new Stack<TerminalSymbol>(sentence.OfType<TerminalSymbol>());
      startNode.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      return new SymbolicExpressionTree(rootNode);
    }

    public ISymbolicExpressionTreeNode ParseSymbolicExpressionTree(Stack<TerminalSymbol> parseStack) {
      TerminalSymbol currentSymbol = parseStack.Pop();

      ISymbolicExpressionTreeNode parsedSubTree = null;

      if (currentSymbol == Addition) {
        parsedSubTree = addSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack)); // left part
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack)); // right part

      } else if (currentSymbol == Multiplication) {
        parsedSubTree = mulSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack)); // left part
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack)); // right part

      } else if (currentSymbol == Log) {
        parsedSubTree = logSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      } else if (currentSymbol == Exp) {
        parsedSubTree = expSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      } else if (currentSymbol == Sin) {
        parsedSubTree = sinSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      } else if (currentSymbol == Cos) {
        parsedSubTree = cosSy.CreateTreeNode();
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      } else if (currentSymbol == Inv) {
        parsedSubTree = divSy.CreateTreeNode();
        ConstantTreeNode dividend = (ConstantTreeNode)constSy.CreateTreeNode();
        dividend.Value = 1.0;
        parsedSubTree.AddSubtree(dividend);
        parsedSubTree.AddSubtree(ParseSymbolicExpressionTree(parseStack));

      } else if (currentSymbol is VariableTerminalSymbol) {
        VariableTreeNode varNode = (VariableTreeNode)varSy.CreateTreeNode();
        varNode.Weight = 1.0;
        varNode.VariableName = currentSymbol.StringRepresentation;
        parsedSubTree = varNode;
      }

      Debug.Assert(parsedSubTree != null);
      return parsedSubTree;
    }
    #endregion
  }
}