#ifndef DETECTOREFFICIENCYVARIATION_H_
#define DETECTOREFFICIENCYVARIATION_H_

#include <cxxtest/TestSuite.h>
#include "MantidTestHelpers/WorkspaceCreationHelper.h"

#include "MantidAlgorithms/DetectorEfficiencyVariation.h"
#include "MantidKernel/UnitFactory.h"
#include "MantidAPI/AnalysisDataService.h"
#include "MantidAPI/WorkspaceFactory.h"
#include "MantidDataObjects/Workspace2D.h"
#include "MantidDataHandling/LoadInstrument.h"
#include <boost/shared_ptr.hpp>
#include <boost/lexical_cast.hpp>
#include <Poco/Path.h>
#include <cmath>
#include <sstream>
#include <fstream>
#include <ios>
#include <string>

using namespace Mantid::Kernel;
using namespace Mantid::Geometry;
using namespace Mantid::API;
using namespace Mantid::Algorithms;
using namespace Mantid::DataObjects;

class DetectorEfficiencyVariationTest : public CxxTest::TestSuite {
public:
  static DetectorEfficiencyVariationTest *createSuite() {
    return new DetectorEfficiencyVariationTest();
  }
  static void destroySuite(DetectorEfficiencyVariationTest *suite) {
    delete suite;
  }

  bool runInit(DetectorEfficiencyVariation &alg) // this is run by both tests so
                                                 // I thought I'd take it out
                                                 // and split things up
  {
    TS_ASSERT_THROWS_NOTHING(alg.initialize());
    bool good = alg.isInitialized();
    // Set the properties
    alg.setPropertyValue("WhiteBeamBase", m_WB1Name);
    alg.setPropertyValue("WhiteBeamCompare", m_WB2Name);
    alg.setPropertyValue("OutputWorkspace", "DetEfficVariTestWSO");
    return good;
  }

  void testWorkspace() {
    DetectorEfficiencyVariation alg;
    // the spectra were setup in the constructor and passed to our algorithm
    // through this function
    TS_ASSERT_THROWS_NOTHING(TS_ASSERT(runInit(alg)))

    // these are realistic values that I just made up
    double variation = 1.1;
    alg.setProperty("Variation", variation);

    // we are using the defaults on StartSpectrum, EndSpectrum,
    // RangeLower and RangeUpper which is to use the whole spectrum

    alg.setRethrows(true);

    alg.execute();
    TS_ASSERT(alg.isExecuted());

    // Get back the saved workspace
    Workspace_sptr output;
    TS_ASSERT_THROWS_NOTHING(output = AnalysisDataService::Instance().retrieve(
                                 "DetEfficVariTestWSO"));
    MatrixWorkspace_sptr outputMat =
        boost::dynamic_pointer_cast<MatrixWorkspace>(output);
    TS_ASSERT(outputMat);
    TS_ASSERT_EQUALS(outputMat->YUnit(), "");

    const int numFailed = alg.getProperty("NumberOfFailures");
    TS_ASSERT_EQUALS(numFailed, 65);

    int firstGoodSpec = (Nhist / 2) - int((variation - 1) / m_ramp) + 1;
    int lastGoodSpec = (Nhist / 2) + int((variation - 1) / m_ramp) - 1;
    for (int lHist = 0; lHist < firstGoodSpec; lHist++) {
      TS_ASSERT_EQUALS(static_cast<double>(outputMat->readY(lHist).front()),
                       static_cast<double>(BadVal))
    }
    for (int lHist = firstGoodSpec; lHist <= lastGoodSpec; lHist++) {
      TS_ASSERT_EQUALS(static_cast<double>(outputMat->readY(lHist).front()),
                       static_cast<double>(GoodVal))
    }
    for (int lHist = lastGoodSpec + 1; lHist < Nhist; lHist++) {
      TS_ASSERT_EQUALS(static_cast<double>(outputMat->readY(lHist).front()),
                       static_cast<double>(BadVal))
    }
  }

  DetectorEfficiencyVariationTest()
      : m_WB1Name("DetEfficVariTestWSI1"), m_WB2Name("DetEfficVariTestWSI2"),
        m_ramp(0.01) {
    using namespace Mantid;
    // Set up a small workspace for testing
    Workspace_sptr spaceA =
        WorkspaceFactory::Instance().create("Workspace2D", Nhist, NXs, NXs - 1);
    Workspace_sptr spaceB =
        WorkspaceFactory::Instance().create("Workspace2D", Nhist, NXs, NXs - 1);
    Workspace2D_sptr inputA = boost::dynamic_pointer_cast<Workspace2D>(spaceA);
    Workspace2D_sptr inputB = boost::dynamic_pointer_cast<Workspace2D>(spaceB);
    boost::shared_ptr<MantidVec> x(new MantidVec(NXs));
    for (int i = 0; i < NXs; ++i) {
      (*x)[i] = i * 1000;
    }
    // random numbers that will be copied into the workspace spectra
    const short ySize = NXs - 1;
    double yArray[ySize] = {
        0.2, 4,  50,  14,    0.001, 0,        0,    0,     1,  0, 1e-3, 15,
        4,   0,  9,   0.001, 2e-10, 1,        0,    8,     0,  7, 1e-4, 1,
        7,   11, 101, 6,     53,    0.345324, 3444, 13958, 0.8}; // NXs = 34 so
                                                                 // we need that
                                                                 // many numbers

    // the error values aren't used and aren't tested so we'll use some basic
    // data
    boost::shared_ptr<MantidVec> errors(new MantidVec(ySize, 1));
    boost::shared_ptr<MantidVec> forInputA, forInputB;

    for (int j = 0; j < Nhist; ++j) {
      inputA->setX(j, x);
      // both workspaces must have the same x bins
      inputB->setX(j, x);
      forInputA.reset(new MantidVec);
      forInputB.reset(new MantidVec);
      // the spectravalues will be multiples of the random numbers above
      for (int l = 0; l < ySize; ++l) {
        forInputA->push_back(yArray[l]);
        // there is going to be a small difference between the workspaces that
        // will vary with histogram number
        forInputB->push_back(forInputA->back() *
                             (1 + m_ramp * (j - (Nhist / 2))));
      }
      // insert a particularly large value to pick up later
      m_LargeValue = 3.1;
      if (j == Nhist - 1)
        for (int l = 0; l < ySize; ++l)
          (*forInputB)[l] = (*forInputA)[l] * m_LargeValue;

      inputA->setData(j, forInputA, errors);
      inputB->setData(j, forInputB, errors);
      // Just set the spectrum number to match the index, spectra numbers and
      // detector maps must be indentical for both
      inputA->getSpectrum(j)->setSpectrumNo(j + 1);
      inputB->getSpectrum(j)->setSpectrumNo(j + 1);
    }

    // Register the input workspaces to the ADS where they can be accessed by
    // the algorithm
    AnalysisDataService::Instance().add(m_WB1Name, inputA);
    AnalysisDataService::Instance().add(m_WB2Name, inputB);

    // Load the instrument data
    Mantid::DataHandling::LoadInstrument loader;
    loader.initialize();
    // Path to test input file assumes Test directory checked out from SVN
    std::string inputFile = "INES_Definition.xml";
    loader.setPropertyValue("Filename", inputFile);
    loader.setPropertyValue("Workspace", m_WB1Name);
    loader.execute();
    // both workspaces should use the same instrument information
    loader.setPropertyValue("Workspace", m_WB2Name);
    loader.execute();

    inputA->getAxis(0)->unit() = UnitFactory::Instance().create("TOF");
    inputB->getAxis(0)->unit() = UnitFactory::Instance().create("TOF");
  }

private:
  std::string m_WB1Name, m_WB2Name;
  double m_ramp, m_LargeValue;
  enum {
    Nhist = 84,
    NXs = 34,
    // these values must match the values in DetectorEfficiencyVariation.h
    BadVal = 1,
    GoodVal = 0
  };
};

#endif /*DETECTOREFFICIENCYVARIATION_H_*/