#include "engine.h"
#include <GLFW/glfw3.h>
#include <glad/glad.h>
#include <iostream>
#include <fstream>
#include <sstream>
#include <limits>
#include "rendering/vertex_buffer.h"
#include "rendering/index_buffer.h"

namespace Donut
{
    Engine::Engine() 
        : m_sag_a(glm::vec3(0.0f, 0.0f, 0.0f), 
          static_cast<float>(8.54e36)) 
    {
        GLFWwindow* window = static_cast<GLFWwindow*>(Application::get().get_window().get_native_window());
        glfwGetFramebufferSize(window, &m_width, &m_height);
        m_compute_height = 420;
        m_compute_width = (m_width * 420) / m_height;
        
        m_camera.set_orbital_mode(true);
        m_camera.set_radius(6.34194e10);
        m_camera.set_min_radius(1e10);
        m_camera.set_max_radius(1e12);
        m_camera.set_orbit_speed(0.01f);
        m_camera.set_zoom_speed(25e9f);
        
        m_objects = 
        {
            { glm::vec4(4e11f, 0.0f, 0.0f, 4e10f), glm::vec4(1,1,0,1), static_cast<float>(1.98892e30) },
            { glm::vec4(0.0f, 0.0f, 4e11f, 4e10f), glm::vec4(1,0,0,1), static_cast<float>(1.98892e30) },
            { glm::vec4(0.0f, 0.0f, 0.0f, m_sag_a.m_rs), glm::vec4(0,0,0,1), static_cast<float>(m_sag_a.m_mass) }
        };

        m_quad_shader    = Shader::create("assets/shaders/quad.glsl");
        m_grid_shader    = Shader::create("assets/shaders/grid.glsl");
        m_compute_shader = Shader::create_compute("Geodesic", load_compute_shader("assets/shaders/geodesic.glsl"));
        
        m_camera_ubo = UniformBuffer::create(128, 1);
        m_disk_ubo = UniformBuffer::create(sizeof(float) * 4, 2);
        
        uint32_t obj_ubo_size = sizeof(int) + 3 * sizeof(float)
            + 16 * (sizeof(glm::vec4) + sizeof(glm::vec4))
            + 16 * sizeof(float);
        m_objects_ubo = UniformBuffer::create(obj_ubo_size, 3);

        float quad_vertices[] = 
        {
            -1.0f,  1.0f,  0.0f, 1.0f,
            -1.0f, -1.0f,  0.0f, 0.0f,
             1.0f, -1.0f,  1.0f, 0.0f,
            
            -1.0f,  1.0f,  0.0f, 1.0f,
             1.0f, -1.0f,  1.0f, 0.0f,
             1.0f,  1.0f,  1.0f, 1.0f
        };

        m_quad_vao = VertexArray::create();
        
        auto vertex_buffer = VertexBuffer::create(quad_vertices, static_cast<uint32_t>(sizeof(quad_vertices)));
        VertexBufferLayout layout;
        layout.push<float>(2);
        layout.push<float>(2);
        vertex_buffer->set_layout(layout);
        m_quad_vao->add_vertex_buffer(vertex_buffer);

        m_texture = Texture2D::create(m_width, m_height);
    }

    void Engine::generate_grid(const std::vector<ObjectData>& objects) 
    {
        const int grid_size = 25;
        const float spacing = 1e10f;

        std::vector<glm::vec3> vertices;
        std::vector<uint32_t> indices;

        for (int z = 0; z <= grid_size; ++z) 
        {
            for (int x = 0; x <= grid_size; ++x) 
            {
                float world_x = (x - grid_size / 2) * spacing;
                float world_z = (z - grid_size / 2) * spacing;
                float y = 0.0f;

                for (const auto& obj : objects) 
                {
                    glm::vec3 obj_pos = glm::vec3(obj.m_pos_radius);
                    double mass = obj.m_mass;
                    double radius = obj.m_pos_radius.w;

                    double r_s = 2.0 * G * mass / (c * c);
                    double dx = world_x - obj_pos.x;
                    double dz = world_z - obj_pos.z;
                    double dist = sqrt(dx * dx + dz * dz);

                    if (dist > r_s) 
                    {
                        double delta_y = 2.0 * sqrt(r_s * (dist - r_s));
                        y += static_cast<float>(delta_y) - 3e10f;
                    } 
                    else 
                    {
                        y += 2.0f * static_cast<float>(sqrt(r_s * r_s)) - 3e10f;
                    }
                }

                vertices.emplace_back(world_x, y, world_z);
            }
        }

        for (int z = 0; z < grid_size; ++z) 
        {
            for (int x = 0; x < grid_size; ++x) 
            {
                int i = z * (grid_size + 1) + x;
                indices.push_back(i);
                indices.push_back(i + 1);
                indices.push_back(i);
                indices.push_back(i + grid_size + 1);
            }
        }

        if (!m_grid_vao)
        {
            m_grid_vao = Ref<VertexArray>(VertexArray::create());
            
            auto vertex_buffer = Ref<VertexBuffer>(VertexBuffer::create(vertices.data(), static_cast<uint32_t>(vertices.size() * sizeof(glm::vec3))));
            VertexBufferLayout layout;
            layout.push<float>(3);
            vertex_buffer->set_layout(layout);
            m_grid_vao->add_vertex_buffer(vertex_buffer);

            auto index_buffer = Ref<IndexBuffer>(IndexBuffer::create(indices.data(), static_cast<uint32_t>(indices.size())));
            m_grid_vao->set_index_buffer(index_buffer);
        }
        else
        {
            auto vertex_buffer = m_grid_vao->get_vertex_buffers()[0];
            vertex_buffer->set_data(vertices.data(), static_cast<uint32_t>(vertices.size() * sizeof(glm::vec3)));
        }

        m_grid_index_count = static_cast<int>(indices.size());
    }

    void Engine::draw_grid(const glm::mat4& view_proj) 
    {
        if (!m_grid_shader || !m_grid_vao) 
            return;
        
        m_grid_shader->bind();
        m_grid_shader->set_mat4("viewProj", view_proj);
        
        RenderCommand::disable_depth_test();
        RenderCommand::enable_blending();
        
        m_grid_vao->bind();
        glDrawElements(GL_LINES, m_grid_index_count, GL_UNSIGNED_INT, nullptr);
        
        RenderCommand::enable_depth_test();
    }

    void Engine::update_window_dimensions() 
    {
        GLFWwindow* window = static_cast<GLFWwindow*>(Application::get().get_window().get_native_window());
        int old_width = m_width;
        int old_height = m_height;
        glfwGetFramebufferSize(window, &m_width, &m_height);
        m_compute_height = 420;
        m_compute_width = (m_width * 420) / m_height;
        
        if (old_width != m_width || old_height != m_height)
            m_texture = Texture2D::create(m_width, m_height);
    }

    void Engine::draw_full_screen_quad() 
    {
        RenderCommand::set_viewport(0, 0, m_width, m_height);
        
        m_quad_shader->bind();
        std::cout << "Texture ID: " << m_texture->get_renderer_id() << ", Size: " << m_texture->get_width() << "x" << m_texture->get_height() << std::endl;
        m_texture->bind(0);
        m_quad_shader->set_int("screenTexture", 0);
        std::cout << "Set screenTexture uniform to slot 0\n";

        RenderCommand::disable_depth_test();
        RenderCommand::draw_indexed(m_quad_vao, 6);
        RenderCommand::enable_depth_test();
    }

    void Engine::dispatch_compute(const Camera& cam) 
    {
        int cw = m_compute_width;
        int ch = m_compute_height;

        if (m_texture->get_width() != cw || m_texture->get_height() != ch) 
            m_texture = Texture2D::create(cw, ch);

        m_compute_shader->bind();
        upload_camera_ubo(cam);
        upload_disk_ubo();
        upload_objects_ubo(m_objects);

        m_texture->bind_as_image(0, false);

        uint32_t groups_x = (uint32_t)std::ceil(cw / 16.0f);
        uint32_t groups_y = (uint32_t)std::ceil(ch / 16.0f);
        m_compute_shader->dispatch(groups_x, groups_y, 1);
        m_compute_shader->memory_barrier(IMAGE_ACCESS_BARRIER_BIT);
        
        glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
        glMemoryBarrier(GL_TEXTURE_FETCH_BARRIER_BIT);
    }

    void Engine::upload_camera_ubo(const Camera& cam) 
    {
        struct UBOData 
        {
            glm::vec3 pos; float _pad0;
            glm::vec3 right; float _pad1;
            glm::vec3 up; float _pad2;
            glm::vec3 forward; float _pad3;
            float tan_half_fov;
            float aspect;
            bool moving;
            int _pad4;
        } data;

        glm::vec3 fwd = glm::normalize(cam.get_target() - cam.get_orbital_position());
        glm::vec3 up = glm::vec3(0, 1, 0);
        glm::vec3 right = glm::normalize(glm::cross(fwd, up));
        up = glm::cross(right, fwd);

        data.pos = cam.get_orbital_position();
        data.right = right;
        data.up = up;
        data.forward = fwd;
        data.tan_half_fov = static_cast<float>(tan(glm::radians(60.0f * 0.5f)));
        data.aspect = static_cast<float>(m_width) / static_cast<float>(m_height);
        data.moving = cam.is_dragging() || cam.is_panning();

        m_camera_ubo->set_data(&data, sizeof(UBOData));
    }

    void Engine::upload_objects_ubo(const std::vector<ObjectData>& objs) 
    {
        struct UBOData 
        {
            int num_objects;
            float _pad0, _pad1, _pad2;
            glm::vec4 pos_radius[16];
            glm::vec4 color[16];
            float mass[16];
        } data;

        size_t count = std::min(objs.size(), size_t(16));
        data.num_objects = static_cast<int>(count);

        for (size_t i = 0; i < count; ++i) 
        {
            data.pos_radius[i] = objs[i].m_pos_radius;
            data.color[i] = objs[i].m_color;
            data.mass[i] = objs[i].m_mass;
        }

        m_objects_ubo->set_data(&data, sizeof(data));
    }

    void Engine::upload_disk_ubo() 
    {
        float r1 = static_cast<float>(m_sag_a.m_rs * 2.2);
        float r2 = static_cast<float>(m_sag_a.m_rs * 5.2);
        float num = 2.0f;
        float thickness = 1e9f;
        float disk_data[4] = { r1, r2, num, thickness };

        m_disk_ubo->set_data(disk_data, sizeof(disk_data));
    }

    void Engine::update_physics(float delta_time) 
    {
        for (auto& obj : m_objects) 
        {
            for (auto& obj2 : m_objects) 
            {
                if (&obj == &obj2) continue;
                float dx = obj2.m_pos_radius.x - obj.m_pos_radius.x;
                float dy = obj2.m_pos_radius.y - obj.m_pos_radius.y;
                float dz = obj2.m_pos_radius.z - obj.m_pos_radius.z;
                float distance = sqrt(dx * dx + dy * dy + dz * dz);

                if (distance > 0) 
                {
                    std::vector<double> direction = {dx / distance, dy / distance, dz / distance};
                    double Gforce = (G * obj.m_mass * obj2.m_mass) / (distance * distance);
                    double acc1 = Gforce / obj.m_mass;
                    std::vector<double> acc = {direction[0] * acc1, direction[1] * acc1, direction[2] * acc1};

                    if (m_gravity) 
                    {
                        obj.m_velocity.x += static_cast<float>(acc[0]);
                        obj.m_velocity.y += static_cast<float>(acc[1]);
                        obj.m_velocity.z += static_cast<float>(acc[2]);

                        obj.m_pos_radius.x += static_cast<float>(obj.m_velocity.x);
                        obj.m_pos_radius.y += static_cast<float>(obj.m_velocity.y);
                        obj.m_pos_radius.z += static_cast<float>(obj.m_velocity.z);
                    }
                }
            }
        }
    }

    void Engine::render_scene()
    {
        glm::mat4 view = glm::lookAt(m_camera.get_orbital_position(), m_camera.get_target(), glm::vec3(0,1,0));
        glm::mat4 proj = glm::perspective(glm::radians(60.0f), float(m_width)/m_height, 1e9f, 1e14f);
        glm::mat4 view_proj = proj * view;
        
        generate_grid(m_objects);
        draw_grid(view_proj);
        
        RenderCommand::set_viewport(0, 0, m_width, m_height);
        dispatch_compute(m_camera);
        draw_full_screen_quad();
    }

    std::string Engine::load_compute_shader(const std::string& path)
    {
        std::ifstream in(path);
        if(!in.is_open()) 
        {
            std::cerr << "Failed to open compute shader: " << path << "\n";
            return "";
        }

        std::stringstream ss;
        ss << in.rdbuf();
        std::string result = ss.str();
        if (result.empty()) 
            std::cerr << "Warning: Compute shader file is empty: " << path << "\n";
        else 
            std::cout << "Successfully loaded compute shader: " << path << " (size: " << result.size() << " bytes)\n";
        return result;
    }
}