voxel-raycaster/include/CLCaster.h

#pragma once
#include <GL/glew.h>
#include <iostream>
#include <map>
#include <string.h>
#include <unordered_map>
#include <vector>
#include "Application.h"
#include "Camera.h"
#include "Gui.h"
#include "LightController.h"
#include "Logger.h"
#include "map/Map.h"
#include "Vector4.hpp"

#ifdef linux
#include <CL/cl.h>
#include <CL/opencl.h>
#include <GL/gl.h>
#include <GL/glx.h>
#include <GL/glu.h>
//#include <EGL/egl.h>
//#include <GL/glx.h>

#elif defined _WIN32
#include <CL/cl_gl.h>
#include <CL/cl.h>
#include <CL/opencl.h>

// Note: windows.h must be included before Gl/GL.h
#include <windows.h>
#include <GL/GL.h>

#elif defined TARGET_OS_MAC
#include <OpenGL/OpenGL.h>
#include <OpenCL/opencl.h>
#include <OpenGL/gl.h>

#endif

#undef ERROR

struct device_info {
	cl_uint cl_device_address_bits;
	cl_bool cl_device_available;
	cl_bool cl_device_compiler_available;
	cl_bool cl_device_endian_little;
	cl_bool cl_device_error_correction_support;
	char cl_device_extensions[1024];
	cl_ulong cl_device_global_mem_cache_size;
	cl_uint cl_device_global_mem_cacheline_size;
	cl_ulong cl_device_global_mem_size;
	cl_bool cl_device_image_support;
	size_t cl_device_image2d_max_height;
	size_t cl_device_image2d_max_width;
	size_t cl_device_image3d_max_depth;
	size_t cl_device_image3d_max_height;
	size_t cl_device_image3d_max_width;
	cl_ulong cl_device_local_mem_size;
	cl_uint cl_device_max_clock_frequency;
	cl_uint cl_device_max_compute_units;
	cl_uint cl_device_max_constant_args;
	cl_ulong cl_device_max_constant_buffer_size;
	cl_ulong cl_device_max_mem_alloc_size;
	size_t cl_device_max_parameter_size;
	cl_uint cl_device_max_read_image_args;
	cl_uint cl_device_max_samplers;
	size_t cl_device_max_work_group_size;
	cl_uint cl_device_max_work_item_dimensions;
	size_t cl_device_max_work_item_sizes[3];
	cl_uint cl_device_max_write_image_args;
	cl_uint cl_device_mem_base_addr_align;
	cl_uint cl_device_min_data_type_align_size;
	char cl_device_name[128];
	cl_platform_id cl_device_platform;
	cl_uint cl_device_preferred_vector_width_char;
	cl_uint cl_device_preferred_vector_width_short;
	cl_uint cl_device_preferred_vector_width_int;
	cl_uint cl_device_preferred_vector_width_long;
	cl_uint cl_device_preferred_vector_width_float;
	cl_uint cl_device_preferred_vector_width_double;
	char cl_device_profile[256];
	size_t cl_device_profiling_timer_resolution;
	cl_device_type device_type;
	char cl_device_vendor[128];
	cl_uint cl_device_vendor_id;
	char cl_device_version[128];
	char cl_driver_version[128];
};

struct PackedData;

class CLCaster : private Gui, public VrEventSubscriber {

public:

	/**
	 * CLCaster is the beginning and end to all interaction with the GPU.
	 * 
	 * It queries devices, manages the creation of various data structures as well
	 * as they syncing between the GPU. It Handles computing of the cast as well
	 * as rendering of the computed cast.
	 * 
	 */

	CLCaster();
	virtual ~CLCaster();
	
	// Queries hardware, creates the command queue and context, and compiles kernel
	bool init();

	// Creates a texture to send to the GPU via height and width
	// Creates a viewport vector array via vertical and horizontal fov
	bool create_viewport(int width, int height, float v_fov, float h_fov) ;
	bool release_viewport();

	// Light controllers own the copy of the PackedData array.
	// We receive a pointer to the array and USE_HOST_POINTER to map the memory to the GPU
	bool assign_lights(std::vector<PackedData> *data) ;

    // TODO: Double maps??
	// We take a ptr to the map and create the map, and map_dimensions buffer for the GPU
	bool assign_map(std::shared_ptr<Map> map);
	bool release_map();

	// We take a ptr to the map and create the map, and map_dimensions buffer for the GPU
	bool assign_octree(std::shared_ptr<Map> map);
	bool release_octree();

	// We take a ptr to the camera and create a camera direction and position buffer
	bool assign_camera(std::shared_ptr<Camera> camera);
	bool release_camera();

	// Creates 3 buffers relating to the texture atlas: texture_atlas, atlas_dim, and tile_dim
	// With these on the GPU we can texture any quad with an atlas tile
	bool create_texture_atlas(sf::Texture *t, sf::Vector2i tile_dim);
	
	// Check to make sure that the buffers have been initiated and set them as kernel args
	bool validate();

	// Aquires the GL objects, runs the kernel, releases back the GL objects
	bool compute();

	// Take the viewport sprite and draw it to the screen
	void draw(sf::RenderWindow* window);

	// Load the saved device config from a file
	bool load_config();

	// Save the chosen device config to a file
	void save_config();

    // Set a define
    void set_define(std::string name, std::string value);
	void remove_define(std::string name);

    bool create_settings_buffer();
    bool release_settings_buffer();
    bool add_to_settings_buffer(std::string setting_name, std::string define_accessor_name, int64_t *value);
    bool remove_from_settings_buffer(std::string setting_name);

	// ================================== DEBUG =======================================
	
	// Re compile the kernel and revalidate the args
	bool debug_quick_recompile();

	// Modify the viewport matrix
	void test_edit_viewport(int width, int height, float v_fov, float h_fov);


	// ============= GUI ==============
	virtual void render_gui() override;
	virtual void update_gui() override;

	virtual void event_handler(VrEventPublisher *publisher, std::unique_ptr<vr::Event> event) override;

	// ================================

private:

	/**
	* Device is a storage container for device data we retrieve from OpenCL
	*
	* The data are mainly queries as strings or integer types and stored into
	* respective containers. We store this data into a file and retrieve it later
	* to let users select a preferred compute device and keep track of their choice
	*/
	class device {

	public:

#pragma pack(push, 1)
		struct packed_data {

			cl_device_type device_type;
			cl_uint clock_frequency;
			char opencl_version[64];
			cl_uint compute_units;
			char device_extensions[1024];
			char device_name[256];
			char platform_name[128];
		};
#pragma pack(pop)

		device(cl_device_id device_id, cl_platform_id platform_id);
		device(const device& d);
		void print(std::ostream& stream) const;
		void print_packed_data(std::ostream& stream);

		cl_device_id getDeviceId() const { return device_id; };
		cl_platform_id getPlatformId() const { return platform_id; };

	private:

		packed_data data;

		cl_device_id device_id;
		cl_platform_id platform_id;

		cl_bool is_little_endian = false;
		bool cl_gl_sharing = false;

	};

	// Cycle through the OpenCL devices and store *all* of their data, not super useful
	bool query_hardware();

	// Cycle through the OpenCL devices and store only the minimal amount of data that we need
	bool aquire_hardware();

	// Create a shared cl_gl context with respect to the individual platforms implementation of sharing
	bool create_shared_context();

	// Using the context and the device create a command queue for them
	bool create_command_queue();

	// Buffer operations
	// All of these functions create and store a buffer in a map with the key representing their name
	
	// Create an image buffer from an SF texture. Access Type is the read/write specifier required by OpenCL
	bool create_image_buffer(std::string buffer_name, cl_uint size, sf::Texture* texture, cl_int access_type);

	// Create a buffer with CL_MEM_READ_ONLY and CL_MEM_COPY_HOST_PTR
	bool create_buffer(std::string buffer_name, cl_uint size, void* data);

	// Create a buffer with user defined data flags
	bool create_buffer(std::string buffer_name, cl_uint size, void* data, cl_mem_flags flags);
	
	// Store a cl_mem object in the buffer map <string:name, cl_mem:buffer>
	bool store_buffer(cl_mem buffer, std::string buffer_name);

	// Using CL release the memory object and remove the KVP associated with the buffer name
	bool release_buffer(std::string buffer_name);
	
	// Compile the kernel with either a full src string or by is_path=true and kernel_source = a valid path
	bool compile_kernel(std::string kernel_source, bool is_path, std::string kernel_name);

	// Set the arg index for the specified kernel and buffer
	bool set_kernel_arg(std::string kernel_name, int index, std::string buffer_name);

	// Run the kernel using a 1d work size
	bool run_kernel(std::string kernel_name, const int work_dim_x, const int work_dim_y);

	// Run a test kernel that prints out the kernel args
	void print_kernel_arguments();

	// Cl can return 0 or 1 for success, greater or lower for fail. Return true or false based on that requirement
	static bool cl_assert(int error_code);
	
	// Take an integer error code and return the string of the related CL ENUM
	static std::string cl_err_lookup(int error_code);

	// Setters and getters
	cl_device_id getDeviceID();
	cl_platform_id getPlatformID();
	cl_context getContext();
	cl_kernel getKernel(std::string kernel_name);
	cl_command_queue getCommandQueue();

	// List of queried devices
	std::vector<device> device_list;

	// Our picked device data
	cl_platform_id platform_id;
	cl_device_id device_id;

	// And state
	cl_context context;
	cl_command_queue command_queue;

	// Containers holding the kernels and buffers
	std::map<std::string, cl_kernel> kernel_map;
	std::map<std::string, cl_mem> buffer_map;
	std::unordered_map<std::string, std::pair<sf::Sprite, std::unique_ptr<sf::Texture>>> image_map;

    const unsigned int SETTINGS_BUFFER_SIZE = 64;
    unsigned int settings_buffer_position = 0;
    int64_t* settings_buffer = nullptr;

    // name of setting, position in the settings buffer
    std::map<std::string, unsigned int> settings_buffer_indices;

    // name of define, value
    std::map<std::string, std::string> defines_map;

	// Hardware caster holds and renders its own textures
	sf::Sprite viewport_sprite;
	sf::Texture viewport_texture;

	sf::Uint8 *viewport_image = nullptr;
	sf::Vector4f *viewport_matrix = nullptr;
	sf::Vector2i viewport_resolution;

	std::shared_ptr<Camera> camera;
	std::shared_ptr<Map> map;

	std::vector<PackedData> *lights;
	int light_count = 0;

	int error = 0;

};