source: branches/PushGP/HeuristicLab.PushGP/HeuristicLab.Problem.ProgramSynthesis.BenchmarkSuite/BenchmarkSuiteDataDescriptor.cs @ 14744

namespace HeuristicLab.BenchmarkSuite {
  using System;
  using System.Collections.Generic;
  using System.IO;
  using System.IO.Compression;
  using System.Linq;
  using System.Reflection;
  using System.Text.RegularExpressions;

  using HeuristicLab.BenchmarkSuite.Problems;
  using HeuristicLab.Common;
  using HeuristicLab.Core;
  using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;

  using Microsoft.VisualBasic.FileIO;


  [StorableClass]
  public abstract class BenchmarkSuiteDataDescriptor : NamedItem, IBenchmarkSuiteDataDescriptor {
    private const string InstanceArchiveName = "HeuristicLab.BenchmarkSuite.Data.BenchmarkExamples.zip";
    private const string ResourcePath = @".*\.Data\.";
    private const string ExampleSeparator = ",";

    private const char ArrayValueSeparator = ' ';
    private static readonly char[] ArraySymbolTrim = { '[', ']' };

    protected BenchmarkSuiteDataDescriptor() { }

    [StorableConstructor]
    protected BenchmarkSuiteDataDescriptor(bool deserializing) : base(deserializing) { }


    [Storable]
    private Example[] examples;
    public Example[] Examples
    {
      get
      {
        return this.examples ?? (this.examples = this.ParseData().ToArray());
      }
    }

    public new abstract string Name { get; }
    protected abstract string FileName { get; }
    public new abstract string Description { get; }
    public abstract int OriginalTrainingCount { get; }
    public abstract int OriginalTestCount { get; }
    public abstract int BestResult { get; }
    public abstract int WorstResult { get; }
    public abstract int InputArgumentCount { get; }
    public abstract int OutputArgumentCount { get; }
    public int TotalArgumentCount { get { return InputArgumentCount + OutputArgumentCount; } }

    public abstract Example ParseExample(string[] input, string[] output);

    private IEnumerable<Example> ParseData() {
      using (var file = this.GetType().Assembly.GetManifestResourceStream(InstanceArchiveName))
      using (var archive = new ZipArchive(file, ZipArchiveMode.Read)) {
        var entry = archive.Entries.SingleOrDefault(x => x.Name == FileName);

        using (var parser = new TextFieldParser(entry.Open())) {
          parser.TextFieldType = FieldType.Delimited;
          parser.SetDelimiters(ExampleSeparator);
          parser.TrimWhiteSpace = false;

          //Processing rows
          while (!parser.EndOfData) {
            var fields = parser.ReadFields();

            if (fields.Length != TotalArgumentCount)
              throw new InvalidDataException("Number of values do not fit");

            var input = fields.Take(InputArgumentCount).ToArray();
            var output = fields.Skip(InputArgumentCount).ToArray();

            yield return ParseExample(input, output);
          }
        }
      }
    }

    protected static string GetResourceName(string fileName) {
      return Assembly
          .GetExecutingAssembly()
          .GetManifestResourceNames()
          .SingleOrDefault(x => Regex.Match(x, ResourcePath + fileName).Success);
    }

    protected static double[] ConvertDoubles(string input) {
      return ConvertMultiple(input, double.Parse);
    }

    protected static double ConvertDouble(string input) {
      return double.Parse(input);
    }

    protected static long[] ConvertIntegers(string input) {
      return ConvertMultiple(input, long.Parse);
    }

    protected static long ConvertInteger(string input) {

      return string.IsNullOrEmpty(input) ? default(long) : long.Parse(input);
    }

    protected static T[] ConvertMultiple<T>(string input, Func<string, T> converter) {
      return input
       .Trim(ArraySymbolTrim)
       .Split(ArrayValueSeparator)
       .Where(s => !string.IsNullOrWhiteSpace(s))
       .Select(converter)
       .ToArray();
    }

    protected static IEnumerable<string> SplitByNewLine(string input) {
      return input.Split(new[] { Environment.NewLine, "\n", "\r\n" }, StringSplitOptions.None);
    }

    private static string[] trueFormats = new[] { "true", "t", "1" };
    private static string[] falseFormats = new[] { "false", "f", "0" };

    protected static bool ConvertBool(string input) {
      var str = input.ToLower();

      if (trueFormats.Contains(str)) return true;
      if (falseFormats.Contains(str)) return false;
      throw new InvalidDataException(string.Format("Unable to parse {0} as boolean", str));
    }

    public override IDeepCloneable Clone(Cloner cloner) {
      return this;
    }
  }
}