#pragma once
#include "gp/function.hpp"
#include "gp/buffer.hpp"
#include "gp/math.hpp"
#include "gp/bitops.hpp"
#include "gp/algorithm/tmp_manip.hpp"


#include <iostream>

namespace gp{
	template<bool lazy, typename color_type>
	class bmp_viewport {
		using src_t = typename gp::either<lazy,
			gp::function<color_type(gp::vec2_g<int32_t>)>,
			gp::buffer<gp::buffer<color_type>>
		>::type;
		src_t source;
		gp::vec2_g<int32_t> resolution;
		color_type get(int32_t x, int32_t y) {
			gp_config::assertion(x>=0, "getting an x below zero");
			gp_config::assertion(y>=0, "getting an y below zero");

			if constexpr (lazy) {
				return source({x, y});
			} else {
				return source[x][y];
			}
		}
	public:
		bmp_viewport(gp::vec2_g<int32_t> res, src_t src)
		: source{src}
		, resolution{res}
		{}

		gp::buffer<char>::associated_iterator write(gp::buffer<char> destination) {
			using sle16 = /*gp::endian_wrapper<*/int16_t/*, gp::endian::little>*/;
			using sbe16 = /*gp::endian_wrapper<*/int16_t/*, gp::endian::big>*/;
			using sle32 = /*gp::endian_wrapper<*/int32_t/*, gp::endian::little>*/;
			using sbe32 = /*gp::endian_wrapper<*/int32_t/*, gp::endian::big>*/;

			auto it = destination.begin();
			*(it++) = 'B';
			*(it++) = 'M';
			auto& filesize = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
			it = it+4;

			*(it++) = 0;
			*(it++) = 0;
			*(it++) = 0;
			*(it++) = 0;
			auto& pixel_array_offset = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
			it = it+4;

			auto dib_start = it;

				auto& dibsize = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;

				auto& width = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				width = resolution.x;
				it = it+4;
				auto& height = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				height = resolution.y;

				auto& plane_cnt = gp::buffer<char>{it.data, (it+2).data}.cast<sle16>()[0];
				it = it+2;
				plane_cnt = 1;
				auto& bit_per_pixel = gp::buffer<char>{it.data, (it+2).data}.cast<sle16>()[0];
				it = it+2;
				bit_per_pixel = sizeof(color_type)*8; // TODO: correct the size

				auto& compression_method = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				compression_method = 0;

				auto& image_size = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;

				auto& h_pixel_per_meter = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				h_pixel_per_meter = 2835;
				auto& v_pixel_per_meter = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				v_pixel_per_meter = 2835;

				auto& colors_in_palette = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				colors_in_palette = 0;
				auto& important_colors = gp::buffer<char>{it.data, (it+4).data}.cast<sle32>()[0];
				it = it+4;
				important_colors = 0;

			auto dib_end = it;
			dibsize = dib_end - dib_start;
			auto pixel_array_start = it;

				for(int32_t line = resolution.y - 1; line >= 0; line--)
				{
					int32_t len = 0;
					for(int32_t row = 0; row < resolution.x; row++)
					{
						// TODO: add more default color modes
						if constexpr (std::is_same<color_type, gp::vec4_g<uint8_t>>::value)
						{
							auto color = get(row, line);
							*(it++) = color.x;
							*(it++) = color.y;
							*(it++) = color.z;
							*(it++) = color.w;
						} else {
							it = it + sizeof(color_type);
						}
						len+=sizeof(color_type);
					}
					for(;len % 4; ++len)
					{
							*(it++) = 0;
					}
				}

			auto pixel_array_end = it;
			pixel_array_offset = pixel_array_start - destination.begin();
			filesize = pixel_array_end - destination.begin();
			image_size = pixel_array_end - pixel_array_start;


				std::cout << width << std::endl;
				std::cout << height << std::endl;
				std::cout << filesize << std::endl;
				std::cout << image_size << std::endl;


			return it;
		}
	};
}