#ifndef RENDERER_H
#define RENDERER_H

#include <vector>
#include <cmath> // trig functions
#include <cstdlib> // calloc

// TODO: decide on extension library
//#include <GL/glew.h>
#include <GL/gl3w.h>

#include <SDL.h>
#include <glm/glm.hpp>
#include <glm/gtc/matrix_transform.hpp>

#include "hexlib.h"
#include "hexgame.h"


const char * VERTEX_SHADER_CODE =
	"#version 330 core\n"
	"in vec3 vertexPosition_modelspace;\n"
	"in vec3 vertexColor;\n"
	"out vec3 fragmentColor;\n"
	"uniform mat4 MVP;\n"
	"void main()\n"
	"{\n"
	"	gl_Position = MVP * vec4(vertexPosition_modelspace, 1);\n"
	"	fragmentColor = vertexColor;\n"
	"}";

const char * FRAGMENT_SHADER_CODE =
	"#version 330 core\n"
	"in vec3 fragmentColor;\n"
	"out vec3 color;\n"
	"void main()\n"
	"{\n"
	"	color = fragmentColor;\n"
	"}";

const char * LINE_FRAGMENT_SHADER_CODE =
	"#version 330 core\n"
	"out vec3 color;\n"
	"void main()\n"
	"{\n"
	"	color = vec3(0,0,0);\n"
	"}";

const char * DEBUG_FRAGMENT_SHADER_CODE =
	"#version 330 core\n"
	"out vec3 color;\n"
	"void main()\n"
	"{\n"
	"	color = vec3(1,0,0);\n"
	"}";

typedef struct clear_col
{
	real32 R;
	real32 G;
	real32 B;
	real32 A;
} clear_col;
clear_col g_clear_col { 75.f / 255.f, 135.f / 255.f, 135.f / 255.f, 1.f };

typedef struct gl_matrix_info
{
	glm::mat4 projection;
	glm::mat4 view;
	glm::mat4 model;
	glm::mat4 MVP;
} gl_matrix_info;

typedef struct gl_buffer
{
	GLuint buffer_id;
	size_t buffer_len; // NOTE: number of elements in buffer
	GLfloat* buffer;
} gl_buffer;

typedef struct gl_render_group
{
	// NOTE: For now the renderFrame function will assume these are the same size
	gl_buffer vertex_buffer;
	gl_buffer color_buffer;
	GLuint program_id;
	GLuint matrix_id; // NOTE: reference to a vertex shader's MVP matrix uniform
	GLuint vertex_array_id;
} gl_render_group;

gl_matrix_info g_scene_matrices;
gl_render_group g_filled_hex_render_group;
gl_render_group g_hex_line_render_group;
gl_render_group g_debug_render_group;

// Interface

bool initRenderer(SDL_Handles &handles, v2i vpDims);
bool createScene(std::vector<hex_info>* hexes);
void renderFrame(const std::vector<hex_info> *hexes);
void renderDebug(hex_info *start_hex, hex_info * current_hex);
void freeBuffers();

// forward declarations

bool initShaderProgram(gl_render_group &rg, const char * vertex_code, const char * frag_code, const char * mat_id);
// enable debug output https://www.khronos.org/opengl/wiki/OpenGL_Error
void openglDebugCallback(GLenum source, GLenum type, GLuint id, GLenum severity,
						GLsizei length, const GLchar* message, const void* userParam);
void fillTriangleBufferFromHex(GLfloat buf[], int idx,  const hex_info &hex);
void fillColorBuffer(GLfloat buf[], int len, std::vector<hex_info> *hexes);
void fillHexLineBuffer(GLfloat buf[], int len, std::vector<hex_info>* hexes);
bool checkGLBufferSize(GLenum buf_type, int expected_size, int line_num);
bool initGLBufferObject(gl_buffer* buf_obj, int len, GLenum usage, GLfloat data[]);
void drawRenderGroup(gl_render_group* rg, GLenum draw_mode, bool update_vertex_data);

bool
initRenderer(SDL_Handles &handles, v2i vpDims)
{
	SDL_GL_SetAttribute(SDL_GL_CONTEXT_FLAGS, SDL_GL_CONTEXT_FORWARD_COMPATIBLE_FLAG);
	SDL_GL_SetAttribute(SDL_GL_CONTEXT_PROFILE_MASK, SDL_GL_CONTEXT_PROFILE_CORE);
	SDL_GL_SetAttribute(SDL_GL_DOUBLEBUFFER, 1);
	SDL_GL_SetAttribute(SDL_GL_DEPTH_SIZE, 24);
	SDL_GL_SetAttribute(SDL_GL_STENCIL_SIZE, 8);
	SDL_GL_SetAttribute(SDL_GL_CONTEXT_MAJOR_VERSION, 3);
	SDL_GL_SetAttribute(SDL_GL_CONTEXT_MINOR_VERSION, 3);
	SDL_GL_SetSwapInterval(1);	// vsync
	SDL_GetCurrentDisplayMode(0, &handles.currentDisplayMode);
	handles.window = 
		SDL_CreateWindow("hexgame", SDL_WINDOWPOS_CENTERED, SDL_WINDOWPOS_CENTERED,
							vpDims.x, vpDims.y, SDL_WINDOW_OPENGL|SDL_WINDOW_RESIZABLE);
	
	if (handles.window == NULL)
	{
		// TODO: log error SDL_GetError()
		return false;
	}

	handles.glContext = SDL_GL_CreateContext(handles.window);
	
	// TODO: decide on extension library
	gl3wInit();
#if 0
	// Initialize GLEW
	// glewExperimental is only needed in GLEW <= 1.13.0
	// we can require version 2.0.0+
	//glewExperimental = true; // Needed for core profile
	GLenum err = glewInit();
	if (err != GLEW_OK) {
		LOG(ERROR) << "Failed to initilize GLEW" << glewGetErrorString(err) << "\n";
		return false;
	}
#endif

	LOG(INFO) << "opengl vendor: " << glGetString(GL_VENDOR) << "\n";
	LOG(INFO)<< "opengl renderer: " << glGetString(GL_RENDERER) << "\n";
	LOG(INFO) << "opengl version: " << glGetString(GL_VERSION) << "\n";
	
	// TODO: blending, these options break the http://www.opengl-tutorial.org tutorials atm
	// Setup render state: alpha-blending enabled, polygon fill
#if 0
	glEnable(GL_BLEND);
	glBlendEquation(GL_FUNC_ADD);
	glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
	glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
#endif

	// TODO: glDebugMessageCallback is only availabe  from >v4.3
	//  check and warn if context doesn't support this function here
	glEnable (GL_DEBUG_OUTPUT);
	glDebugMessageCallback((GLDEBUGPROC) openglDebugCallback, 0);
	// hide VRAM debug messages
	glDebugMessageControl(GL_DONT_CARE, 33361, GL_DONT_CARE, 0, 0, GL_FALSE);
	
	// TODO: check for errors when compiling shaders
	initShaderProgram(g_filled_hex_render_group, VERTEX_SHADER_CODE, FRAGMENT_SHADER_CODE, "MVP");
	initShaderProgram(g_hex_line_render_group, VERTEX_SHADER_CODE, LINE_FRAGMENT_SHADER_CODE, "MVP");
	initShaderProgram(g_debug_render_group, VERTEX_SHADER_CODE, DEBUG_FRAGMENT_SHADER_CODE, "MVP");

	return true;
}

// NOTE: mat_id should match the matrix variable name in the shader source
bool
initShaderProgram(gl_render_group &rg, const char * vertex_code, const char * frag_code, const char * mat_id)
{
	glGenVertexArrays(1, &rg.vertex_array_id);
	glBindVertexArray(rg.vertex_array_id);
	GLuint vertex_shader_id = glCreateShader(GL_VERTEX_SHADER);
	GLuint fragment_shader_id = glCreateShader(GL_FRAGMENT_SHADER);

	glShaderSource(vertex_shader_id, 1, &vertex_code, NULL);
	glShaderSource(fragment_shader_id, 1, &frag_code, NULL);
	glCompileShader(vertex_shader_id);
	glCompileShader(fragment_shader_id);

	rg.program_id = glCreateProgram();
	glAttachShader(rg.program_id, vertex_shader_id);
	glAttachShader(rg.program_id, fragment_shader_id);
	glLinkProgram(rg.program_id);

	rg.matrix_id = glGetUniformLocation(rg.program_id, mat_id);

	glDetachShader(rg.program_id, vertex_shader_id);
	glDetachShader(rg.program_id, fragment_shader_id);
	glDeleteShader(vertex_shader_id);
	glDeleteShader(fragment_shader_id);

	// TODO: check for errors when compiling shaders
	return true;
}

bool
createScene(std::vector<hex_info>* hexes)
{
	// Model, View, Projection Matrices
	
	// left, right, bottom, top, zNear, zFar
	g_scene_matrices.projection = glm::ortho(0.f, 1280.0f, 0.f, 720.0f, 0.1f, 100.0f);

	g_scene_matrices.view = glm::lookAt(
		glm::vec3(0.0f, 0.0f, 2.0f),	// camera position
		glm::vec3(0.0f, 0.0f, 0.0f),	// look at position
		glm::vec3(0,1,0)				// "up" vector
	);

	g_scene_matrices.model = glm::mat4(1.0f);
	g_scene_matrices.MVP = g_scene_matrices.projection * g_scene_matrices.view * g_scene_matrices.model; 

	// Vertex Data
	
	// TODO: index duplicate vertices
	// http://www.opengl-tutorial.org/intermediate-tutorials/tutorial-9-vbo-indexing/
	int hex_count = (int) hexes->size();
	// 6 triangles * 3 vertices per triangle * 3 floats per vertex = 54
	int vbuf_len = hex_count * 6 * 3 * 3;
	
	// TODO: surely there's a way to use indexed drawing for line vertices eg) line_loop
	// 	and still use one buffer for multiple shapes/paths gldrawarraysintanced?, gldrawelements?
	// 	https://gamedev.stackexchange.com/questions/104310/opengl-4-5-primitive-restart-vs-base-index
	int line_vertices_per_hex = 6 * 2; // 12 vertices since we're using line segments atm
	//gl_buffer *line_vertex_buffer = &g_hex_line_render_group.vertex_buffer;
	int line_buf_len = hexes->size() * line_vertices_per_hex * 3; // 3 floats per vertex

	// temporary buffers
	GLfloat* vbuf = (GLfloat*) std::calloc(vbuf_len, sizeof(GLfloat));
	GLfloat* cbuf = (GLfloat*) std::calloc(vbuf_len, sizeof(GLfloat));
	GLfloat* line_buf = (GLfloat*) std::calloc(vbuf_len, sizeof(GLfloat));

	for (int i = 0; i < hex_count; i++)
		fillTriangleBufferFromHex(vbuf, 54 * i, (*hexes)[i]);
	

	if (!initGLBufferObject(&g_filled_hex_render_group.vertex_buffer, vbuf_len, GL_DYNAMIC_DRAW, vbuf))
		return false;

	// color data for hex vertices
	
	fillColorBuffer(cbuf, vbuf_len, hexes);

	if (!initGLBufferObject(&g_filled_hex_render_group.color_buffer, vbuf_len, GL_DYNAMIC_DRAW, cbuf))
		return false;

	// hex line vertex data
	
	fillHexLineBuffer(line_buf, line_buf_len, hexes);
	if (!initGLBufferObject(&g_hex_line_render_group.vertex_buffer, line_buf_len, GL_STATIC_DRAW, line_buf))
		return false;

	// free temporary buffers
	std::free(vbuf);
	std::free(cbuf);
	std::free(line_buf);

	// debug draw vertexes
	
	// 4 vertices, 3 floats per vertex
	int len = 4 * 3;
	if (!initGLBufferObject(&g_debug_render_group.vertex_buffer, len, GL_DYNAMIC_DRAW, 0))
		return false;

	return true;
}

void
fillTriangleBufferFromHex(GLfloat buf[], int idx,  const hex_info &hex)
{
	// triangles
	for (int i = 0; i < 6; i++)
	{
		if (i == 5)	// re-use the first point for the last triangle
		{
			// vertex 0
			buf[idx + 0] = (GLfloat) hex.XPos;
			buf[idx + 1] = (GLfloat) hex.YPos;
			buf[idx + 2] = (GLfloat) 0.f;

			// vertex 1
			buf[idx + 3] = (GLfloat) hex.vertices[i].x;
			buf[idx + 4] = (GLfloat) hex.vertices[i].y;
			buf[idx + 5] = (GLfloat) 0.f;

			// vertex 2
			buf[idx + 6] = (GLfloat) hex.vertices[0].x;
			buf[idx + 7] = (GLfloat) hex.vertices[0].y;
			buf[idx + 8] = (GLfloat) 0.f;
		}
		else
		{
			// vertex 0
			buf[idx + 0] = (GLfloat) hex.XPos;
			buf[idx + 1] = (GLfloat) hex.YPos;
			buf[idx + 2] = (GLfloat) 0.f;

			// vertex 1
			buf[idx + 3] = (GLfloat) hex.vertices[i].x;
			buf[idx + 4] = (GLfloat) hex.vertices[i].y;
			buf[idx + 5] = (GLfloat) 0.f;

			// vertex 2
			buf[idx + 6] = (GLfloat) hex.vertices[i + 1].x;
			buf[idx + 7] = (GLfloat) hex.vertices[i + 1].y;
			buf[idx + 8] = (GLfloat) 0.f;
		}
		
		// we've added 9 GLfloats per loop
		idx += 9;
	}
}

void
fillColorBuffer(GLfloat buf[], int len, std::vector<hex_info>* hexes)
{
	int buf_idx;
	int buf_len_per_hex = 54; // NOTE: 3 * 3 * 6
	for (int i = 0; i < (int) hexes->size(); i++)
	{
		buf_idx = i * buf_len_per_hex;
		hex_info hxi = (*hexes)[i];
		GLfloat color = (GLfloat) hxi.current_color / (GLfloat) UINT32_MAX;
		
		for (int j = 0; j < buf_len_per_hex; j++)
			buf[buf_idx + j] = color;
	}
}

void
fillHexLineBuffer(GLfloat buf[], int len, std::vector<hex_info>* hexes)
{
	Point p1, p2;
	int idx = 0;
	for (int i = 0; i < (int) hexes->size(); i++)
	{
		hex_info hxi = (*hexes)[i];
		for (int j = 0; j < 6; j ++)
		{
			if (j == 5) // wrap
			{
				p1 = hxi.vertices[j];
				p2 = hxi.vertices[0];
			}
			else
			{
				p1 = hxi.vertices[j];
				p2 = hxi.vertices[j + 1];
			}
			
			buf[idx + 0] = p1.x;
			buf[idx + 1] = p1.y;
			buf[idx + 2] = 0.f;
			buf[idx + 3] = p2.x;
			buf[idx + 4] = p2.y;
			buf[idx + 5] = 0.f;
			idx += 6;
		}
	}
}

// NOTE: this will only work after a call to glBindBuffer
bool
checkGLBufferSize(GLenum buf_type, int expected_size, int line_num)
{
	GLint gl_size;
	glGetBufferParameteriv(buf_type, GL_BUFFER_SIZE ,&gl_size);

	if (expected_size != gl_size)
	{
		LOG(ERROR) << "line: " << line_num << "gl buffer size mismatch\n";
		return false;
	}

	return true;
}

bool
initGLBufferObject(gl_buffer* buf_obj, int len, GLenum usage, GLfloat data[])
{
	buf_obj->buffer_len = (size_t) len;
	buf_obj->buffer = (GLfloat*) std::calloc(len, sizeof(GLfloat));

	if (data)
	{
		for (int i = 0; i < len; i++)
			buf_obj->buffer[i] = data[i];
	}

	glGenBuffers(1, &buf_obj->buffer_id);
	glBindBuffer(GL_ARRAY_BUFFER, buf_obj->buffer_id);
	glBufferData(GL_ARRAY_BUFFER, len * sizeof(GLfloat), buf_obj->buffer, usage);

	if (!checkGLBufferSize(GL_ARRAY_BUFFER, len * sizeof(GLfloat), __LINE__))
		return false;
	
	return true;
}

void
openglDebugCallback(GLenum source, GLenum type, GLuint id, GLenum severity,
					GLsizei length, const GLchar* message, const void* userParam)
{
	LOG((type == GL_DEBUG_TYPE_ERROR) ? ERROR : DEBUG)
		<< (type == GL_DEBUG_TYPE_ERROR ? "** GL ERROR **" : "")
		<< ", type: " << type
		<< ", severity: " << severity
		<< ", message: " << message << "\n";
}

void
drawRenderGroup(gl_render_group* rg, GLenum draw_mode, bool update_vertex_data = false)
{
	glUseProgram(rg->program_id);
	// Send our transformation to the currently bound shader, in the "g_MVP" uniform
	glUniformMatrix4fv(rg->matrix_id, 1, GL_FALSE, &g_scene_matrices.MVP[0][0]);
	
	// 1st attribute buffer : vertices
	glEnableVertexAttribArray(0);
	glBindBuffer(GL_ARRAY_BUFFER, rg->vertex_buffer.buffer_id);
	glVertexAttribPointer(0, 3, GL_FLOAT, GL_FALSE, 0, (void*) 0);
	
	if (update_vertex_data)
	{
		glBufferSubData(GL_ARRAY_BUFFER, 0, rg->vertex_buffer.buffer_len * sizeof(GLfloat),
			rg->vertex_buffer.buffer);
	}

	// 2nd attribute buffer : colors
	if (rg->color_buffer.buffer)
	{
		glEnableVertexAttribArray(1);
		glBindBuffer(GL_ARRAY_BUFFER, rg->color_buffer.buffer_id);
		glVertexAttribPointer(1, 3, GL_FLOAT, GL_FALSE, 0, (void*) 0);
		
		// TODO: maybe add an option to not send color data every frame?
		glBufferSubData(GL_ARRAY_BUFFER, 0, rg->color_buffer.buffer_len * sizeof(GLfloat),
				rg->color_buffer.buffer);
	}
	// draw
	glDrawArrays(draw_mode, 0, rg->vertex_buffer.buffer_len / 3);
	
	// cleanup
	glDisableVertexAttribArray(0);
	if (rg->color_buffer.buffer)
		glDisableVertexAttribArray(1);
}

void
renderFrame(std::vector<hex_info> *hexes)
{
	glClearColor(g_clear_col.R, g_clear_col.G, g_clear_col.B, g_clear_col.A);
	glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT);

	// filled hexes

	// get new colors every frame
	gl_render_group* rg = &g_filled_hex_render_group;
	fillColorBuffer(rg->color_buffer.buffer, rg->color_buffer.buffer_len, hexes);
	drawRenderGroup(rg, GL_TRIANGLES);

	// hex lines

	drawRenderGroup(&g_hex_line_render_group, GL_LINES);
}

void
renderDebug(hex_info *start_hex, hex_info *current_hex)
{
	if (!start_hex || !current_hex)
		return;

	// fill vertex buffer
	Point p1(start_hex->XPos, start_hex->YPos);
	Point p2(current_hex->XPos, current_hex->YPos);
	real64 angle = std::atan2(p2.y - p1.y, p2.x - p1.x);
	real64 len = std::hypot(p2.y - p1.y, p2.x - p1.x);
	real64 coneAngle = 24.5f * M_PI / 180; // M_PI is non-standard and may not be portable
	
	// TODO: add matrix math library for rotation transform?
	// |x|   | cos(a + b) -sin(a + b)  0 |
	// |y| * | sin(a + b)  cos(a + b)  0 |
	// |0|   | 0           0           1 |
	//
	real64 x1 = len * std::cos(angle);
	real64 y1 = len * std::sin(angle);
	// top of cone
	real64 topX = x1 * std::cos(coneAngle) - y1 * std::sin(coneAngle) + p1.x;
	real64 topY = x1 * std::sin(coneAngle) + y1 * std::cos(coneAngle) + p1.y;
	// bottom of cone
	real64 botX = x1 * std::cos(coneAngle) + y1 * std::sin(coneAngle) + p1.x;
	real64 botY = x1 * std::sin(-1 * coneAngle) + y1 * std::cos(coneAngle) + p1.y;
	// TODO: indexed line drawing
	real64 buf[4 * 3] = {
		p1.x, p1.y, 0,
		botX, botY, 0,
		p2.x, p2.y, 0,
		topX, topY, 0,
	};

	// copy vertexes to render group
	gl_render_group* rg = &g_debug_render_group;
	for (int i = 0; i < 12; i++)
		rg->vertex_buffer.buffer[i] = buf[i];

	drawRenderGroup(rg, GL_LINE_LOOP, true);
}

void
freeBuffers()
{
	std::vector<gl_render_group> groups = {
		g_filled_hex_render_group,
		g_hex_line_render_group,
		g_debug_render_group
	};
	
	for (gl_render_group group : groups)
	{
		if (group.vertex_buffer.buffer)
		{
			std::free(group.vertex_buffer.buffer);
			group.vertex_buffer.buffer = 0;
		}
	
		if (group.color_buffer.buffer)
		{
			std::free(group.color_buffer.buffer);
			group.color_buffer.buffer = 0;
		}
	}
		 
}

#endif // RENDERER_H