#ifndef HARDSPI_H
#define HARDSPI_H

#include "../Debug.h"
#include "driver/spi_master.h"
#include <cmath>
#include <string.h>

//static spi_transaction_t trans[2];

#define PIN_NUM_MISO	GPIO_NUM_19
#define PIN_NUM_MOSI	GPIO_NUM_23
#define PIN_NUM_CLK		GPIO_NUM_18
#define PIN_NUM_CS		GPIO_NUM_2


template <int PIN_NUM_MISO, int PIN_NUM_MOSI, int PIN_NUM_CLK> class HardSPI {
//class HardSPI {

	static constexpr int MAX_LEN = 4000;

	spi_device_handle_t spi;

	spi_bus_config_t buscfg;

	spi_device_interface_config_t devcfg;

	static constexpr const char* TAG = "hwSPI";

public:

	// can be called before every transmit
//	static void IRAM_ATTR pre_transfer_callback(spi_transaction_t *t) {
//	}

	/** ctor */
	HardSPI() {
		init();
	}

private:

	void init() {

		int ret;

		// BUS specific options
		memset(&buscfg, 0, sizeof(buscfg));
		buscfg.miso_io_num = PIN_NUM_MISO;
		buscfg.mosi_io_num = PIN_NUM_MOSI;
		buscfg.sclk_io_num = PIN_NUM_CLK;
		buscfg.quadwp_io_num = -1;
		buscfg.quadhd_io_num = -1;
		buscfg.max_transfer_sz = MAX_LEN;//320*240*2;
		buscfg.flags = 0;
		buscfg.intr_flags = 0;

		// DEVICE specific options
		memset(&devcfg, 0, sizeof(devcfg));
		devcfg.clock_speed_hz = 10 * 1000 * 1000;
		devcfg.mode = 0;
		devcfg.spics_io_num = -1;//PIN_NUM_CS;		// currently not used
		devcfg.queue_size = 2;
		devcfg.pre_cb = nullptr;
		//devcfg.pre_cb = pre_transfer_callback;	// can be called before every transmit

		// Initialize the SPI bus
		ret = spi_bus_initialize(HSPI_HOST, &buscfg, 1);
		ESP_ERROR_CHECK(ret);

		// Attach the device to the SPI bus
		ret = spi_bus_add_device(HSPI_HOST, &devcfg, &spi);
		ESP_ERROR_CHECK(ret);

	}

public:

	/** blocking transmit the given bytes */
	void write(const uint8_t* data, size_t len) {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));

		while(len > 0) {

			const size_t chunkLen = min(MAX_LEN, (len));
			t.length = 8 * chunkLen;		// in bits
			t.tx_buffer = data;				// MOSI only
			t.flags = 0;

			// blocking transfer
			esp_err_t ret = spi_device_polling_transmit(spi, &t);
			ESP_ERROR_CHECK(ret);

			// update remaining length and data pointer
			len -= chunkLen;
			data += chunkLen;

		}

	}

	void read(uint8_t* dst, size_t len) {

		// sanity check
		{
			const size_t addr = reinterpret_cast<size_t>(dst);
			if ((addr % 4) != 0) {
				ESP_LOGE(TAG, "read-buffer must be 32-bit aligned");
			}
		}

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));

		//debugMod1(TAG, "read %d bytes", len);

		while(len > 0) {

			const size_t chunkLen = min(MAX_LEN, (len));
			//ESP_LOGI(TAG, "- read %d byte chunk", chunkLen);
			t.length = 8 * chunkLen;		// in bits
			t.rx_buffer = dst;				// MISO only
			t.flags = 0;

			// blocking transfer
			esp_err_t ret = spi_device_polling_transmit(spi, &t);
			ESP_ERROR_CHECK(ret);

			// update remaining length and destination-data pointer
			len -= chunkLen;
			dst += chunkLen;

		}

	}

	/** send 4 bytes onto the bus while reading 4 bytes response */
	uint32_t readWriteQuad(uint32_t write) {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));
		t.length = 8*4;
		t.rxlength = 8*4;
		t.tx_data[0] = (write >> 24) & 0xFF;
		t.tx_data[1] = (write >> 16) & 0xFF;
		t.tx_data[2] = (write >>  8) & 0xFF;
		t.tx_data[3] = (write >>  0) & 0xFF;
		t.flags = SPI_TRANS_USE_RXDATA | SPI_TRANS_USE_TXDATA;

		esp_err_t ret = spi_device_polling_transmit(spi, &t);
		ESP_ERROR_CHECK(ret);

		return
			(static_cast<uint32_t>(t.rx_data[0]) << 24) |
			(static_cast<uint32_t>(t.rx_data[1]) << 16) |
			(static_cast<uint32_t>(t.rx_data[2]) <<  8) |
			(static_cast<uint32_t>(t.rx_data[3]) <<  0);

	}


	/** read one byte from the bus */
	uint8_t readByte() {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));
		t.length = 8*1;
		t.flags = SPI_TRANS_USE_RXDATA;

		esp_err_t ret = spi_device_polling_transmit(spi, &t);
		ESP_ERROR_CHECK(ret);

		return t.rx_data[0];

	}



	void writeByte(const uint8_t byte) {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));
		t.length = 8*1;
		t.tx_data[0] = byte;
		t.flags = SPI_TRANS_USE_TXDATA;

		esp_err_t ret = spi_device_polling_transmit(spi, &t);
		ESP_ERROR_CHECK(ret);

	}

	void writeWord(const uint16_t word) {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));
		t.length = 8*2;
		t.tx_data[0] = static_cast<uint8_t>(word>>8);
		t.tx_data[1] = static_cast<uint8_t>(word>>0);
		t.flags = SPI_TRANS_USE_TXDATA;

		esp_err_t ret = spi_device_polling_transmit(spi, &t);
		ESP_ERROR_CHECK(ret);

	}

	void writeQuad(const uint32_t quad) {

		spi_transaction_t t;
		memset(&t, 0, sizeof(spi_transaction_t));
		t.length = 8*4;
		t.tx_data[0] = static_cast<uint8_t>(quad>>24);
		t.tx_data[1] = static_cast<uint8_t>(quad>>16);
		t.tx_data[2] = static_cast<uint8_t>(quad>>8);
		t.tx_data[3] = static_cast<uint8_t>(quad>>0);
		t.flags = SPI_TRANS_USE_TXDATA;

		esp_err_t ret = spi_device_polling_transmit(spi, &t);
		ESP_ERROR_CHECK(ret);

	}



};

#endif // HARDSPI_H