#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2015 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * HeuristicLab is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System;
using System.Collections;
using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Analysis;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.Optimization;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
using HeuristicLab.Problems.DataAnalysis;
using HeuristicLab.Problems.DataAnalysis.Symbolic;
using HeuristicLab.Problems.Instances;


namespace HeuristicLab.Problems.GeneticProgramming.GlucosePrediction {
  [Item("Blood Glucose Forecast", "See MedGEC Workshop at GECCO 2016")]
  [Creatable(CreatableAttribute.Categories.GeneticProgrammingProblems, Priority = 999)]
  [StorableClass]
  public sealed class Problem : SymbolicExpressionTreeProblem, IRegressionProblem, IProblemInstanceConsumer<IRegressionProblemData>, IProblemInstanceExporter<IRegressionProblemData> {

    #region parameter names
    private const string ProblemDataParameterName = "ProblemData";
    #endregion

    #region Parameter Properties
    IParameter IDataAnalysisProblem.ProblemDataParameter { get { return ProblemDataParameter; } }

    public IValueParameter<IRegressionProblemData> ProblemDataParameter {
      get { return (IValueParameter<IRegressionProblemData>)Parameters[ProblemDataParameterName]; }
    }
    #endregion

    #region Properties
    public IRegressionProblemData ProblemData {
      get { return ProblemDataParameter.Value; }
      set { ProblemDataParameter.Value = value; }
    }
    IDataAnalysisProblemData IDataAnalysisProblem.ProblemData { get { return ProblemData; } }
    #endregion

    public event EventHandler ProblemDataChanged;

    public override bool Maximization {
      get { return true; }
    }

    #region item cloning and persistence
    // persistence
    [StorableConstructor]
    private Problem(bool deserializing) : base(deserializing) { }
    [StorableHook(HookType.AfterDeserialization)]
    private void AfterDeserialization() {
      RegisterEventHandlers();
    }

    // cloning 
    private Problem(Problem original, Cloner cloner)
      : base(original, cloner) {
      RegisterEventHandlers();
    }
    public override IDeepCloneable Clone(Cloner cloner) { return new Problem(this, cloner); }
    #endregion

    public Problem()
      : base() {
      Parameters.Add(new ValueParameter<IRegressionProblemData>(ProblemDataParameterName, "The data for the glucose prediction problem", new RegressionProblemData()));

      var g = new SimpleSymbolicExpressionGrammar(); // empty grammar is replaced in UpdateGrammar()
      base.Encoding = new SymbolicExpressionTreeEncoding(g, 100, 17);

      UpdateGrammar();
      RegisterEventHandlers();
    }


    public override double Evaluate(ISymbolicExpressionTree tree, IRandom random) {
      var problemData = ProblemData;
      var target = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, problemData.TrainingIndices);
      var allPredicted = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0), problemData.Dataset, problemData.AllIndices).ToArray();
      var predicted = problemData.TrainingIndices.Select(r => allPredicted[r]);

      // var predicted1 = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(1), problemData.Dataset, rows);
      // var predicted2 = Interpreter.Apply(tree.Root.GetSubtree(0).GetSubtree(0).GetSubtree(2), problemData.Dataset, rows);

      var pred0_rsq = Rsq(predicted, target);
      // var pred1_rsq = Rsq(predicted1, target);
      // var pred2_rsq = Rsq(predicted2, target);
      return pred0_rsq; // + pred1_rsq + pred2_rsq;
    }

    private double Rsq(IEnumerable<double> predicted, IEnumerable<double> target) {
      // only take predictions for which the target is not NaN
      var selectedTuples = target.Zip(predicted, Tuple.Create).Where(t => !double.IsNaN(t.Item1)).ToArray();
      target = selectedTuples.Select(t => t.Item1);
      predicted = selectedTuples.Select(t => t.Item2);

      OnlineCalculatorError errorState;
      var r = OnlinePearsonsRCalculator.Calculate(target, predicted, out errorState);
      if (errorState != OnlineCalculatorError.None) r = 0;
      return r * r;
    }

    public override void Analyze(ISymbolicExpressionTree[] trees, double[] qualities, ResultCollection results,
      IRandom random) {
      base.Analyze(trees, qualities, results, random);

      if (!results.ContainsKey("Solution")) {
        results.Add(new Result("Solution", typeof(IRegressionSolution)));
      }
      if (!results.ContainsKey("ScaledTree")) {
        results.Add(new Result("ScaledTree", typeof(ISymbolicExpressionTree)));
      }
      // if (!results.ContainsKey("Terms")) {
      //   results.Add(new Result("Terms", typeof(DataTable)));
      // }

      var bestTree = trees.First();
      var bestQuality = qualities.First();
      for (int i = 1; i < trees.Length; i++) {
        if (qualities[i] > bestQuality) {
          bestQuality = qualities[i];
          bestTree = trees[i];
        }
      }

      bestTree = (ISymbolicExpressionTree)bestTree.Clone();
      var expressionNode = bestTree.Root.GetSubtree(0).GetSubtree(0);
      // scale

      var problemData = ProblemData;
      var rows = problemData.AllIndices.ToArray();
      var target = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows).ToArray();
      var predicted =
        Interpreter.Apply(expressionNode.GetSubtree(0), problemData.Dataset, rows)
          .ToArray();

      var filteredPredicted = rows.Where(r => !double.IsNaN(target[r])).Select(r => predicted[r]).ToArray();
      var filteredTarget = target.Where(t => !double.IsNaN(t)).ToArray();
      OnlineCalculatorError error;
      double alpha;
      double beta;
      OnlineLinearScalingParameterCalculator.Calculate(filteredPredicted, filteredTarget, out alpha, out beta, out error);

      var prod = new SimpleSymbol("*", "*", 2, 2).CreateTreeNode();
      var sum = new SimpleSymbol("+", "+", 2, 2).CreateTreeNode();
      var constAlpha = (ConstantTreeNode)(new Constant()).CreateTreeNode();
      constAlpha.Value = alpha;
      var constBeta = (ConstantTreeNode)(new Constant()).CreateTreeNode();
      constBeta.Value = beta;

      var originalTree = expressionNode.GetSubtree(0);
      expressionNode.RemoveSubtree(0);
      expressionNode.AddSubtree(sum);
      sum.AddSubtree(prod);
      sum.AddSubtree(constAlpha);
      prod.AddSubtree(originalTree);
      prod.AddSubtree(constBeta);

      var model = new Model(bestTree, problemData.TargetVariable, problemData.AllowedInputVariables.ToArray());
      model.Name = "Scaled Model";
      model.Description = "Scaled Model";
      results["Solution"].Value = model.CreateRegressionSolution(problemData);
      results["ScaledTree"].Value = bestTree;

    }

    #region events
    private void RegisterEventHandlers() {
      ProblemDataParameter.ValueChanged += new EventHandler(ProblemDataParameter_ValueChanged);
      if (ProblemDataParameter.Value != null) ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
    }

    private void ProblemDataParameter_ValueChanged(object sender, EventArgs e) {
      ProblemDataParameter.Value.Changed += new EventHandler(ProblemData_Changed);
      OnProblemDataChanged();
      OnReset();
    }

    private void ProblemData_Changed(object sender, EventArgs e) {
      OnReset();
    }

    private void OnProblemDataChanged() {
      UpdateGrammar();

      var handler = ProblemDataChanged;
      if (handler != null) handler(this, EventArgs.Empty);
    }

    private void UpdateGrammar() {
      // whenever ProblemData is changed we create a new grammar with the necessary symbols
      var g = new Grammar();
      Encoding.Grammar = g;
    }
    #endregion

    #region Import & Export
    public void Load(IRegressionProblemData data) {
      Name = data.Name;
      Description = data.Description;
      ProblemData = data;
    }

    public IRegressionProblemData Export() {
      return ProblemData;
    }
    #endregion
  }
}