/*
  This file is part of CDO. CDO is a collection of Operators to manipulate and analyse Climate model Data.

  Author: Uwe Schulzweida

*/

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#ifdef HAVE_LIBFFTW3
#include <fftw3.h>
#include <mutex>
static std::mutex fftwMutex;
#endif

#include <cdi.h>

#include "process_int.h"
#include "param_conversion.h"
#include "cdo_fft.h"
#include "cdo_options.h"
#include "cdo_omp.h"
#include "field_functions.h"

namespace
{
struct FourierMemory
{
  Varray<double> real;
  Varray<double> imag;
  Varray<double> work_r;
  Varray<double> work_i;
#ifdef HAVE_LIBFFTW3
  fftw_complex *in_fft{};
  fftw_complex *out_fft{};
  fftw_plan plan;
#endif
};
}  // namespace

static void
fourier_fftw(int sign, int varID, int levelID, int nts, size_t gridsize, double missval, FieldVector3D &vars,
             std::vector<FourierMemory> &fourierMemory)
{
  (void) sign;
#ifdef HAVE_LIBFFTW3
  auto norm = 1.0 / std::sqrt(nts);

#ifdef _OPENMP
#pragma omp parallel for default(shared) schedule(static)
#endif
  for (size_t i = 0; i < gridsize; ++i)
  {
    auto ompthID = cdo_omp_get_thread_num();
    auto &fm = fourierMemory[ompthID];

    auto hasMissvals = false;
    for (int tsID = 0; tsID < nts; ++tsID)
    {
      auto real = vars[tsID][varID][levelID].vec_d[2 * i];
      auto imag = vars[tsID][varID][levelID].vec_d[2 * i + 1];
      fm.in_fft[tsID][0] = real;
      fm.in_fft[tsID][1] = imag;
      if (fp_is_equal(real, missval) || fp_is_equal(imag, missval)) hasMissvals = true;
    }

    if (hasMissvals)
    {
      for (int tsID = 0; tsID < nts; ++tsID)
      {
        vars[tsID][varID][levelID].vec_d[2 * i] = missval;
        vars[tsID][varID][levelID].vec_d[2 * i + 1] = missval;
      }
    }
    else
    {
      fftw_execute(fm.plan);

      for (int tsID = 0; tsID < nts; ++tsID)
      {
        vars[tsID][varID][levelID].vec_d[2 * i] = fm.out_fft[tsID][0] * norm;
        vars[tsID][varID][levelID].vec_d[2 * i + 1] = fm.out_fft[tsID][1] * norm;
      }
    }
  }
#endif
}

static void
fourier_intrinsic(int sign, int varID, int levelID, int nts, size_t gridsize, double missval, FieldVector3D &vars,
                  std::vector<FourierMemory> &fourierMemory)
{
  auto isPower2 = ((nts & (nts - 1)) == 0);

#ifdef _OPENMP
#pragma omp parallel for default(shared) schedule(static)
#endif
  for (size_t i = 0; i < gridsize; ++i)
  {
    auto ompthID = cdo_omp_get_thread_num();
    auto &fm = fourierMemory[ompthID];

    auto hasMissvals = false;
    for (int tsID = 0; tsID < nts; ++tsID)
    {
      auto real = vars[tsID][varID][levelID].vec_d[2 * i];
      auto imag = vars[tsID][varID][levelID].vec_d[2 * i + 1];
      fm.real[tsID] = real;
      fm.imag[tsID] = imag;
      if (fp_is_equal(real, missval) || fp_is_equal(imag, missval)) hasMissvals = true;
    }

    if (hasMissvals)
    {
      for (int tsID = 0; tsID < nts; ++tsID)
      {
        vars[tsID][varID][levelID].vec_d[2 * i] = missval;
        vars[tsID][varID][levelID].vec_d[2 * i + 1] = missval;
      }
    }
    else
    {
      if (isPower2)  // nts is a power of 2
        cdo::fft(fm.real.data(), fm.imag.data(), nts, sign);
      else
        cdo::ft_r(fm.real.data(), fm.imag.data(), nts, sign, fm.work_r.data(), fm.work_i.data());

      for (int tsID = 0; tsID < nts; ++tsID)
      {
        vars[tsID][varID][levelID].vec_d[2 * i] = fm.real[tsID];
        vars[tsID][varID][levelID].vec_d[2 * i + 1] = fm.imag[tsID];
      }
    }
  }
}

class Fourier : public Process
{
public:
  using Process::Process;
  inline static CdoModule module = {
    .name = "Fourier",
    .operators = { { "fourier", FourierHelp } },
    .aliases = {},
    .mode = EXPOSED,     // Module mode: 0:intern 1:extern
    .number = CDI_COMP,  // Allowed number type
    .constraints = { 1, 1, NoRestriction },
  };
  inline static auto registration = RegisterEntry<Fourier>();

  size_t nalloc = 0;

  CdoStreamID streamID1;
  CdoStreamID streamID2;

  int taxisID1{ CDI_UNDEFID };
  int taxisID2{ CDI_UNDEFID };

  int vlistID1{ CDI_UNDEFID };

  bool use_fftw = false;
  int sign{};

  VarList varList;

public:
  void
  init() override
  {
    if (Options::Use_FFTW)
    {
#ifdef HAVE_LIBFFTW3
      if (Options::cdoVerbose) cdo_print("Using fftw3 lib");
      use_fftw = true;
#else
      if (Options::cdoVerbose) cdo_print("LIBFFTW3 support not compiled in!");
#endif
    }

    if (Options::cdoVerbose && !use_fftw) cdo_print("Using intrinsic FFT function!");

    operator_input_arg("the sign of the exponent (-1 for normal or 1 for reverse transformation)!");
    sign = parameter_to_int(cdo_operator_argv(0));

    streamID1 = cdo_open_read(0);

    vlistID1 = cdo_stream_inq_vlist(streamID1);
    auto vlistID2 = vlistDuplicate(vlistID1);

    taxisID1 = vlistInqTaxis(vlistID1);
    taxisID2 = taxisDuplicate(taxisID1);
    vlistDefTaxis(vlistID2, taxisID2);

    streamID2 = cdo_open_write(1);
    cdo_def_vlist(streamID2, vlistID2);

    varList = VarList(vlistID1);
  }

  void
  run() override
  {
    FieldVector3D varDataList;
    std::vector<CdiDateTime> vDateTimes;

    auto numVars = varList.numVars();
    int tsID = 0;
    while (true)
    {
      auto numFields = cdo_stream_inq_timestep(streamID1, tsID);
      if (numFields == 0) break;

      if ((size_t) tsID >= nalloc)
      {
        constexpr size_t NALLOC_INC = 1024;
        nalloc += NALLOC_INC;
        vDateTimes.resize(nalloc);
        varDataList.resize(nalloc);
      }

      vDateTimes[tsID] = taxisInqVdatetime(taxisID1);

      field2D_init(varDataList[tsID], varList);

      for (int fieldID = 0; fieldID < numFields; ++fieldID)
      {
        auto [varID, levelID] = cdo_inq_field(streamID1);
        auto gridsize = varList.vars[varID].gridsize;
        varDataList[tsID][varID][levelID].resize(2 * gridsize);
        cdo_read_field(streamID1, varDataList[tsID][varID][levelID]);
      }

      tsID++;
    }

    int nts = tsID;

    std::vector<FourierMemory> fourierMemory(Threading::ompNumMaxThreads);

    if (use_fftw)
    {
#ifdef HAVE_LIBFFTW3
      for (auto &fm : fourierMemory)
      {
        fm.in_fft = fftw_alloc_complex(nts);
        fm.out_fft = fftw_alloc_complex(nts);
        std::scoped_lock lock(fftwMutex);
        fm.plan = fftw_plan_dft_1d(nts, fm.in_fft, fm.out_fft, sign, FFTW_ESTIMATE);
      }
      if (Options::cdoVerbose) fftw_print_plan(fourierMemory[0].plan);
#endif
    }
    else
    {
      auto isPowerOf2 = (nts > 0 && (nts & (nts - 1)) == 0);
      for (auto &fm : fourierMemory)
      {
        fm.real.resize(nts);
        fm.imag.resize(nts);
        if (!isPowerOf2) fm.work_r.resize(nts);
        if (!isPowerOf2) fm.work_i.resize(nts);
      }
    }

    for (int varID = 0; varID < numVars; ++varID)
    {
      auto const &var = varList.vars[varID];
      for (int levelID = 0; levelID < var.nlevels; ++levelID)
      {
        if (use_fftw)
          fourier_fftw(sign, varID, levelID, nts, var.gridsize, var.missval, varDataList, fourierMemory);
        else
          fourier_intrinsic(sign, varID, levelID, nts, var.gridsize, var.missval, varDataList, fourierMemory);
      }
    }

#ifdef HAVE_LIBFFTW3
    if (use_fftw)
    {
      for (auto &fm : fourierMemory)
      {
        fftw_free(fm.in_fft);
        fftw_free(fm.out_fft);
        std::scoped_lock lock(fftwMutex);
        fftw_destroy_plan(fm.plan);
      }
      fftw_cleanup();
    }
#endif

    for (tsID = 0; tsID < nts; ++tsID)
    {
      taxisDefVdatetime(taxisID2, vDateTimes[tsID]);
      cdo_def_timestep(streamID2, tsID);

      for (int varID = 0; varID < numVars; ++varID)
      {
        auto numLevels = varList.vars[varID].nlevels;
        for (int levelID = 0; levelID < numLevels; ++levelID)
        {
          if (!varDataList[tsID][varID][levelID].empty())
          {
            cdo_def_field(streamID2, varID, levelID);
            cdo_write_field(streamID2, varDataList[tsID][varID][levelID]);
          }
        }
      }
    }
  }

  void
  close() override
  {
    cdo_stream_close(streamID2);
    cdo_stream_close(streamID1);
  }
};