using System.Linq;
using System.Threading;
using HeuristicLab.Data;
using HeuristicLab.Encodings.IntegerVectorEncoding;
using HeuristicLab.Problems.GeneralizedQuadraticAssignment;
using HeuristicLab.Random;
using Microsoft.VisualStudio.TestTools.UnitTesting;

namespace UnitTests {
  [TestClass]
  public class ApproximateLocalSearchTest {

    [TestMethod]
    public void ApproximateLocalSearchApplyTest() {
      CollectionAssert.AreEqual(new [] { 2, 0, 1, 1, 2, 3, 0, 3, 0, 0 }, assignment.ToArray());

      var evaluation = instance.Evaluate(assignment);
      Assert.AreEqual(3985258, evaluation.FlowCosts);
      Assert.AreEqual(30, evaluation.InstallationCosts);
      Assert.AreEqual(0, evaluation.ExcessDemand);

      var quality = instance.ToSingleObjective(evaluation);
      Assert.AreEqual(15489822.781533258, quality, 1e-9);

      var evaluatedSolutions = 0;
      ApproximateLocalSearch.Apply(random, assignment, ref quality,
        ref evaluation, 10, 0.5, 100, instance,
        out evaluatedSolutions);
      Assert.AreEqual(61, evaluatedSolutions);
      CollectionAssert.AreEqual(new[] { 2, 0, 0, 0, 2, 1, 0, 3, 0, 0 }, assignment.ToArray());
      Assert.AreEqual(10167912.633734789, quality, 1e-9);
    }

    private const int Equipments = 10, Locations = 5;
    private static GQAPInstance instance;
    private static IntegerVector assignment;
    private static MersenneTwister random;
    
    [ClassInitialize()]
    public static void MyClassInitialize(TestContext testContext) {
      random = new MersenneTwister(42);
      var symmetricDistances = new DoubleMatrix(Locations, Locations);
      var symmetricWeights = new DoubleMatrix(Equipments, Equipments);
      for (int i = 0; i < Equipments - 1; i++) {
        for (int j = i + 1; j < Equipments; j++) {
          symmetricWeights[i, j] = random.Next(Equipments * 100);
          symmetricWeights[j, i] = symmetricWeights[i, j];
        }
      }
      for (int i = 0; i < Locations - 1; i++) {
        for (int j = i + 1; j < Locations; j++) {
          symmetricDistances[i, j] = random.Next(Locations * 100);
          symmetricDistances[j, i] = symmetricDistances[i, j];
        }
      }
      var installationCosts = new DoubleMatrix(Equipments, Locations);
      for (int i = 0; i < Equipments; i++) {
        for (int j = 0; j < Locations; j++) {
          installationCosts[i, j] = random.Next(0, 10);
        }
      }
      var demands = new DoubleArray(Equipments);
      for (int i = 0; i < Equipments; i++) {
        demands[i] = random.Next(1, 10);
      }
      var capacities = new DoubleArray(Locations);
      for (int j = 0; j < Locations; j++) {
        capacities[j] = random.Next(1, 10) * ((double)Equipments / (double)Locations) * 2;
      }

      var transportationCosts = random.NextDouble() * 10;
      var overbookedCapacityPenalty = 1000 * random.NextDouble() + 100;

      instance = new GQAPInstance() {
        Capacities = capacities,
        Demands = demands,
        InstallationCosts = installationCosts,
        PenaltyLevel = overbookedCapacityPenalty,
        TransportationCosts = transportationCosts,
        Weights = symmetricWeights,
        Distances = symmetricDistances
      };

      assignment = GreedyRandomizedSolutionCreator.CreateSolution(random, instance, 100, false, CancellationToken.None);
    }
  }
}