/********************************************************
 *  ██████╗  ██████╗████████╗██╗
 * ██╔════╝ ██╔════╝╚══██╔══╝██║
 * ██║  ███╗██║        ██║   ██║
 * ██║   ██║██║        ██║   ██║
 * ╚██████╔╝╚██████╗   ██║   ███████╗
 *  ╚═════╝  ╚═════╝   ╚═╝   ╚══════╝
 * Geophysical Computational Tools & Library (GCTL)
 *
 * Copyright (c) 2022  Yi Zhang (yizhang-geo@zju.edu.cn)
 *
 * GCTL is distributed under a dual licensing scheme. You can redistribute 
 * it and/or modify it under the terms of the GNU Lesser General Public 
 * License as published by the Free Software Foundation, either version 2 
 * of the License, or (at your option) any later version. You should have 
 * received a copy of the GNU Lesser General Public License along with this 
 * program. If not, see <http://www.gnu.org/licenses/>.
 * 
 * If the terms and conditions of the LGPL v.2. would prevent you from using 
 * the GCTL, please consider the option to obtain a commercial license for a 
 * fee. These licenses are offered by the GCTL's original author. As a rule, 
 * licenses are provided "as-is", unlimited in time for a one time fee. Please 
 * send corresponding requests to: yizhang-geo@zju.edu.cn. Please do not forget 
 * to include some description of your company and the realm of its activities. 
 * Also add information on how to contact you by electronic and paper mail.
 ******************************************************/

// linear conjugate gradient solver library
//#include "lcg/solver.h"
// GCTL library
#include "gctl/core.h"
#include "gctl/poly.h"
#include "gctl/io.h"
#include "gctl/utility.h"
#include "gctl/math/boxsort2d.h"
#include "gctl/optimization.h"

#if defined _WINDOWS || __WIN32__
#include "io.h"
// Test for file existence
#define F_OK 0
#endif

using namespace gctl;

struct data_point
{
	point2dc loc;
	std::vector<double> vals;
};

class LBSI : public gctl::lcg_solver
{
public:
	LBSI(){}
	virtual ~LBSI(){}
	virtual void LCG_Ax(const array<double> &a, array<double> &b);
	virtual void LCG_Mx(const array<double> &a, array<double> &b){}

	void Routine(std::string in_name, std::string tar_name, 
		std::string out_name, std::string col_str, gctl::text_descriptor &desc, 
		unsigned int kernel_size, unsigned int box_size, 
		double epsilon);
	void ReadConstrainNodes(std::string filename, gctl::text_descriptor &desc);
	void WriteTargetNodes(std::string filename, const gctl::text_descriptor &desc);
	void InitTargetNodes(std::string para, gctl::text_descriptor &desc);
	void CalKernel();
	void CalKernel(const data_point &tar_node);
	void UpdateTarVec(size_t idx, bool if_global = true);
	void set_kernel_size(unsigned int k){MatSize = k;}
public:
	array<data_point> ConsNodes;
	array<data_point> TargNodes;
	std::vector<data_point*> LocalNodes;
	boxes2d<data_point> PntBoxes;

	double Xmin, Xmax, Ymin, Ymax; // 对输入数据进行归一化处理

	size_t MatSize, ValSize, MaxCol;
	matrix<double> Kernel;
	array<double> Wgts;
	array<double> MidPdt;
	array<double> B;

	//array<double> GK;
	//array<double> DK;
	//array<double> ADK;

	gctl::_1i_vector col_index;
};

void LBSI::Routine(std::string in_name, std::string tar_name, 
	std::string out_name,  std::string col_str, gctl::text_descriptor &desc, 
	unsigned int kernel_size, unsigned int box_size, 
	double epsilon)
{
	gctl::parse_string_to_vector(col_str, ',', col_index);
	if (col_index.size() < 3)
	{
		throw gctl::runtime_error("Invalid column index. From LBSI::Routine(...)");
	}

	MaxCol = 0;
	for (size_t i = 0; i < col_index.size(); i++)
	{
		if (MaxCol < col_index[i]) MaxCol = col_index[i];
	}

	ReadConstrainNodes(in_name, desc);
	InitTargetNodes(tar_name, desc);

	unsigned int k_size = kernel_size;
	if (k_size <= 1)
	{
		throw gctl::runtime_error("Invalid local size. From LBSI::Routine(...)");
	}

	// Throw errors only
	set_lcg_message(LCG_THROW);

	if (k_size >= ConsNodes.size())
	{
		GCTL_ShowWhatError("The local size is equal to or bigger than the input node's size. Reduced to the global algorithm.", GCTL_WARNING_ERROR, 0, 0, 0);

		k_size = ConsNodes.size();
		set_kernel_size(k_size);
		Kernel.resize(MatSize, MatSize);
		Wgts.resize(MatSize);
		MidPdt.resize(MatSize);
		B.resize(MatSize);
		//GK.resize(MatSize);
		//DK.resize(MatSize);
		//ADK.resize(MatSize);

		lcg_para my_para = default_lcg_para();
		//my_para.max_iterations = 1000;
		my_para.epsilon = epsilon;
		set_lcg_para(my_para);

		CalKernel();

		for (size_t s = 0; s < ValSize; s++)
		{
			UpdateTarVec(s, true);

			Wgts.assign(0.0);
			// run the inversion process in factory mode
			//lcg(_AxProduct, nullptr, Wgts.get(), B.get(), MatSize, &my_para, this, GK.get(), DK.get(), ADK.get());
			lcg(Wgts, B);

			double dist, sum;
			for (int i = 0; i < TargNodes.size(); ++i)
			{
				sum = 0.0;
				for (int j = 0; j < MatSize; ++j)
				{
					dist = sqrt((TargNodes[i].loc.x - ConsNodes[j].loc.x)*(TargNodes[i].loc.x - ConsNodes[j].loc.x) 
						+ (TargNodes[i].loc.y - ConsNodes[j].loc.y)*(TargNodes[i].loc.y - ConsNodes[j].loc.y));

					if (dist >= GCTL_ZERO)
					{
						sum += dist*dist*(log(dist)-1.0)*Wgts[j];
					}	
				}
				TargNodes[i].vals[s] = sum;
			}	
		}
		
		WriteTargetNodes(out_name, desc);
		return;
	}

	set_kernel_size(k_size);
	LocalNodes.resize(MatSize);
	Kernel.resize(MatSize, MatSize);
	Wgts.resize(MatSize);
	MidPdt.resize(MatSize);
	B.resize(MatSize);
	//GK.resize(MatSize);
	//DK.resize(MatSize);
	//ADK.resize(MatSize);

	lcg_para my_para = default_lcg_para();
	//my_para.max_iterations = 1000;
	my_para.epsilon = epsilon;
	set_lcg_para(my_para);

	gctl::array<double> xs(ConsNodes.size());
	gctl::array<double> ys(ConsNodes.size());
	for (int i = 0; i < ConsNodes.size(); ++i)
	{
		xs[i] = ConsNodes[i].loc.x;
		ys[i] = ConsNodes[i].loc.y;
	}

	PntBoxes.init(xs, ys, ConsNodes, box_size, box_size);

	double dist, sum;
	progress_bar bar(TargNodes.size());
	for (int i = 0; i < TargNodes.size(); ++i)
	{
		bar.progressed(i);

		Kernel.assign_all(0.0);
		CalKernel(TargNodes[i]);

		for (size_t s = 0; s < ValSize; s++)
		{
			UpdateTarVec(s, false);

			Wgts.assign(0.0);
			// run the inversion process in factory mode
			//lcg(_AxProduct, nullptr, Wgts.get(), B.get(), MatSize, &my_para, this, GK.get(), DK.get(), ADK.get());
			lcg(Wgts, B);

			sum = 0.0;
			for (int j = 0; j < MatSize; ++j)
			{
				dist = sqrt((TargNodes[i].loc.x - LocalNodes[j]->loc.x)*(TargNodes[i].loc.x - LocalNodes[j]->loc.x) 
					+ (TargNodes[i].loc.y - LocalNodes[j]->loc.y)*(TargNodes[i].loc.y - LocalNodes[j]->loc.y));

				if (dist >= GCTL_ZERO)
				{
					sum += dist*dist*(log(dist)-1.0)*Wgts[j];
				}	
			}
			TargNodes[i].vals[s] = sum;
		}
	}

	WriteTargetNodes(out_name, desc);
	return;
}

void LBSI::ReadConstrainNodes(std::string filename, gctl::text_descriptor &desc)
{
	desc.file_name_ = filename;
	gctl::_2d_vector table_data;
	gctl::read_text2vector2d(desc, table_data);

	if (table_data.size() <= 1)
	{
		throw gctl::runtime_error("Not enough constraint points. From LBSI::ReadConstrainNodes(...)");
	}

	if (table_data[0].size() - 1 < MaxCol)
	{
		throw gctl::runtime_error("Invalid constraint point format. From LBSI::ReadConstrainNodes(...)");
	}

	Xmin = Ymin = 1e+30;
	Xmax = Ymax = -1e+30;

	ValSize = col_index.size() - 2;
	ConsNodes.resize(table_data.size());
	for (size_t i = 0; i < table_data.size(); ++i)
	{
		ConsNodes[i].loc.x = table_data[i][col_index[0]];
		ConsNodes[i].loc.y = table_data[i][col_index[1]];
		Xmin = std::min(Xmin, ConsNodes[i].loc.x);
		Xmax = std::max(Xmax, ConsNodes[i].loc.x);
		Ymin = std::min(Ymin, ConsNodes[i].loc.y);
		Ymax = std::max(Ymax, ConsNodes[i].loc.y);

		for (size_t j = 0; j < ValSize; j++)
		{
			ConsNodes[i].vals.push_back(table_data[i][col_index[2+j]]);	
		}
	}

	for (size_t i = 0; i < ConsNodes.size(); i++)
	{
		ConsNodes[i].loc.x = (ConsNodes[i].loc.x - Xmin)/(Xmax - Xmin);
		ConsNodes[i].loc.y = (ConsNodes[i].loc.y - Ymin)/(Ymax - Ymin);
	}

	destroy_vector(table_data);
	return;
}

void LBSI::WriteTargetNodes(std::string filename, const gctl::text_descriptor &desc)
{
	std::ofstream outfile;
	gctl::open_outfile(outfile, filename, ".txt");

	for (size_t i = 0; i < desc.head_num_; i++)
	{
		outfile << desc.head_strs_[i] << "\n";
	}

	for (size_t i = 0; i < TargNodes.size(); i++)
	{
		TargNodes[i].loc.x = TargNodes[i].loc.x*(Xmax - Xmin) + Xmin;
		TargNodes[i].loc.y = TargNodes[i].loc.y*(Ymax - Ymin) + Ymin;

		outfile << TargNodes[i].loc.x << " " << TargNodes[i].loc.y << " " << std::setprecision(12);
		for (size_t j = 0; j < ValSize; j++)
		{
			outfile << TargNodes[i].vals[j] << " ";
		}
		outfile << std::endl;
	}
	
	outfile.close();
	return;
}

void LBSI::InitTargetNodes(std::string para, gctl::text_descriptor &desc)
{
	// try to use the para as a file name
	if (access(para.c_str(), F_OK) != -1)
	{
		desc.file_name_ = para;
		std::vector<point2dc> tmp_vec;
		read_text2vector(desc, tmp_vec);

		TargNodes.resize(tmp_vec.size());
		for (size_t i = 0; i < tmp_vec.size(); ++i)
		{
			TargNodes[i].loc.x = tmp_vec[i].x;
			TargNodes[i].loc.y = tmp_vec[i].y;

			TargNodes[i].loc.x = (TargNodes[i].loc.x - Xmin)/(Xmax - Xmin);
			TargNodes[i].loc.y = (TargNodes[i].loc.y - Ymin)/(Ymax - Ymin);
		}

		for (size_t i = 0; i < TargNodes.size(); i++)
		{
			TargNodes[i].vals.resize(ValSize);
		}

		destroy_vector(tmp_vec);
		return;
	}

	double dx, dy, xmin, xmax, ymin, ymax;
	gctl::parse_string_to_value(para, '/', true, xmin, dx, xmax, ymin, dy, ymax);

	size_t M = floor((ymax - ymin)/dy) + 1;
	size_t N = floor((xmax - xmin)/dx) + 1;

	TargNodes.resize(M*N);
	for (size_t j = 0; j < M; j++)
	{
		for (size_t i = 0; i < N; i++)
		{
			TargNodes[i + j*N].loc.x = xmin + dx*i;
			TargNodes[i + j*N].loc.y = ymin + dy*j;

			TargNodes[i + j*N].loc.x = (TargNodes[i + j*N].loc.x - Xmin)/(Xmax - Xmin);
			TargNodes[i + j*N].loc.y = (TargNodes[i + j*N].loc.y - Ymin)/(Ymax - Ymin);
		}
	}

	for (size_t i = 0; i < TargNodes.size(); i++)
	{
		TargNodes[i].vals.resize(ValSize);
	}
	return;
}

void LBSI::CalKernel()
{
	// 计算出所有成对的格林函数值
	double dist;
	for (int j = 0; j < MatSize-1; ++j)
	{
		for (int k = j+1; k < MatSize; ++k)
		{
			dist = sqrt((ConsNodes[j].loc.x - ConsNodes[k].loc.x)*(ConsNodes[j].loc.x - ConsNodes[k].loc.x) 
				+ (ConsNodes[j].loc.y - ConsNodes[k].loc.y)*(ConsNodes[j].loc.y - ConsNodes[k].loc.y));

			if (dist >= GCTL_ZERO)
			{
				Kernel[j][k] = Kernel[k][j] = dist*dist*(log(dist)-1.0);
			}
		}
	}
	return;
}

void LBSI::CalKernel(const data_point &tar_node)
{
	// 找出距离tar_node最近的一组控制点
	PntBoxes.get_by_number(tar_node.loc.x, tar_node.loc.y, MatSize, LocalNodes);
	
	// 计算出所有成对的格林函数值
	double dist;
	for (int j = 0; j < MatSize-1; ++j)
	{
		for (int k = j+1; k < MatSize; ++k)
		{
			dist = sqrt((LocalNodes[j]->loc.x - LocalNodes[k]->loc.x)*(LocalNodes[j]->loc.x - LocalNodes[k]->loc.x) 
				+ (LocalNodes[j]->loc.y - LocalNodes[k]->loc.y)*(LocalNodes[j]->loc.y - LocalNodes[k]->loc.y));

			if (dist >= GCTL_ZERO)
			{
				Kernel[j][k] = Kernel[k][j] = dist*dist*(log(dist)-1.0);
			}
		}
	}
	return;
}

void LBSI::UpdateTarVec(size_t idx, bool if_global)
{
	if (if_global)
	{
		for (int i = 0; i < MatSize; ++i)
		{
			B[i] = 0;
			for (int j = 0; j < MatSize; ++j)
			{
				B[i] += Kernel[j][i] * ConsNodes[j].vals[idx];
			}
		}
	}
	else
	{
		for (int i = 0; i < MatSize; ++i)
		{
			B[i] = 0;
			for (int j = 0; j < MatSize; ++j)
			{
				B[i] += Kernel[j][i] * LocalNodes[j]->vals[idx];
			}
		}
	}
	return;
}

void LBSI::LCG_Ax(const array<double> &a, array<double> &b)
{
	for (int i = 0; i < MatSize; ++i)
	{
		MidPdt[i] = 0;
		for (int j = 0; j < MatSize; ++j)
		{
			MidPdt[i] += a[j] * Kernel[i][j];
		}
	}

	for (int i = 0; i < MatSize; ++i)
	{
		b[i] = 0;
		for (int j = 0; j < MatSize; ++j)
		{
			b[i] += MidPdt[j] * Kernel[j][i];
		}
	}
	return;
}