#ifndef MAIN_CA_H
#define MAIN_CA_H

#include <err.h>
#include <stdio.h>
#include <string.h>

#include "darknet.h"

#define MAKE_NETWORK_CMD 1
#define WORKSPACE_NETWORK_CMD 2
#define MAKE_CONV_CMD 3
#define MAKE_MAX_CMD 4
#define MAKE_DROP_CMD 5
#define MAKE_CONNECTED_CMD 6
#define MAKE_SOFTMAX_CMD 7
#define MAKE_COST_CMD 8
#define FORWARD_CMD 9
#define BACKWARD_CMD 10
#define BACKWARD_ADD_CMD 11
#define UPDATE_CMD 12
#define NET_TRUTH_CMD 13
#define CALC_LOSS_CMD 14
#define TRANS_WEI_CMD 15
#define OUTPUT_RETURN_CMD 16
#define SAVE_WEI_CMD 17

#define FORWARD_BACK_CMD 18
#define BACKWARD_BACK_CMD 19
#define BACKWARD_BACK_ADD_CMD 20

#define BUF_T(MAX_BUF_LEN) \
	struct { \
	   uint32_t len; \
	   uint8_t buf[MAX_BUF_LEN]; \
	} __attribute__ ((packed))

#define DECLARE_BUF_T(MAX_BUF_LEN) \
	typedef struct { \
	   uint32_t len; \
	   uint8_t buf[MAX_BUF_LEN]; \
	} __attribute__ ((packed))

#define TA_DARKNETP_UUID \
	{ 0x7fc5c039, 0x0542, 0x4ee1, \
		{ 0x80, 0xaf, 0xb4, 0xea, 0xb2, 0xf1, 0x99, 0x8d} }

typedef struct tz_msg_header {
	/** First 4 bytes should always be id: either cmd_id or resp_id */
	uint32_t id;
	uint32_t content_id;
	uint32_t len;
	uint32_t status;
} __attribute__ ((packed)) tz_msg_header_t;

typedef struct {
        int passint[17];
        float passfloat[15];
} __attribute__ ((packed)) make_network_cmd_t;

typedef struct {
        int cond;
        BUF_T(4096) workspace;
} __attribute__ ((packed)) workspace_network_cmd_t;

typedef struct {
        int passint[14];
        float passflo;
        BUF_T(4096) acti;
} __attribute__ ((packed)) make_conv_cmd_t;

typedef struct {
        int passint[7];
} __attribute__ ((packed)) make_max_cmd_t;

typedef struct {
        int passint[5];
        float  passfloat[1];
        BUF_T(4096) net_prev_output;
        BUF_T(4096) net_prev_delta;
} __attribute__ ((packed)) make_drop_cmd_t;

typedef struct {
        int passarg[5];
        BUF_T(4096) actv;
} __attribute__ ((packed)) make_connected_cmd_t;

typedef struct {
        int passint[8];
        float passflo;
} __attribute__ ((packed)) make_softmax_cmd_t;

typedef struct {
        int passint[2];
        float passflo[4];
        BUF_T(4096) passcost;
} __attribute__ ((packed)) make_cost_cmd_t;

typedef struct {
        BUF_T(524288) vec;
        int passint[3];
        char type;
} __attribute__ ((packed)) trans_wei_cmd_t;

typedef struct {
        int passint[2];
        char type;
} __attribute__ ((packed)) save_wei_cmd_t;

typedef struct {
        BUF_T(4096) weights_back; //output sizeof(float)*length
} __attribute__ ((packed)) save_wei_resp_t;

typedef struct {
        BUF_T(524288) net_input;
        int net_train;
} __attribute__ ((packed)) forward_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back;
} __attribute__ ((packed)) forward_back_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back;
} __attribute__ ((packed)) forward_back_resp_t;

typedef struct {
        BUF_T(4096) net_input;
        int net_train;
} __attribute__ ((packed)) backward_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back; //output
        BUF_T(4096) net_delta_back; //output
} __attribute__ ((packed)) backward_add_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back; //output
        BUF_T(4096) net_delta_back; //output
} __attribute__ ((packed)) backward_add_resp_t;

typedef struct {
        BUF_T(4096) net_input;
        BUF_T(4096) net_delta;
} __attribute__ ((packed)) backward_back_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back; //output
} __attribute__ ((packed)) backward_back_add_cmd_t;

typedef struct {
        BUF_T(4096) net_input_back; //output
} __attribute__ ((packed)) backward_back_add_resp_t;

typedef struct {
        int passint[3];
        float passflo[6];
} __attribute__ ((packed)) update_cmd_t;

typedef struct {
        BUF_T(4096) net_truth;
} __attribute__ ((packed)) net_truth_cmd_t;

typedef struct {
        int passint[2];
} __attribute__ ((packed)) calc_loss_cmd_t;

typedef struct {
        BUF_T(4096) net_output_back; //output
} __attribute__ ((packed)) output_return_cmd_t;

typedef struct {
        BUF_T(4096) net_output_back; //output
} __attribute__ ((packed)) output_return_resp_t;

typedef union {
        make_network_cmd_t         make_network_cmd;
        workspace_network_cmd_t    workspace_network_cmd;
        make_conv_cmd_t            make_conv_cmd;
        make_max_cmd_t             make_max_cmd;
        make_drop_cmd_t            make_drop_cmd;
        make_connected_cmd_t       make_connected_cmd;
        make_softmax_cmd_t         make_softmax_cmd;
        make_cost_cmd_t            make_cost_cmd;
        trans_wei_cmd_t            trans_wei_cmd;
        save_wei_cmd_t             save_wei_cmd;
        save_wei_resp_t            save_wei_resp;
        forward_cmd_t              forward_cmd;
        forward_back_cmd_t         forward_back_cmd;
        forward_back_resp_t        forward_back_resp;
        backward_cmd_t             backward_cmd;
        backward_add_cmd_t         backward_add_cmd;
        backward_add_resp_t        backward_add_resp;
        backward_back_cmd_t        backward_back_cmd;
        backward_back_add_cmd_t    backward_back_add_cmd;
        backward_back_add_resp_t   backward_back_add_resp;
        update_cmd_t               update_cmd;
        net_truth_cmd_t            net_truth_cmd;
        calc_loss_cmd_t            calc_loss_cmd;
        output_return_cmd_t        output_return_cmd;
        output_return_resp_t       output_return_resp;

} __attribute__ ((packed)) payload_t;

/**
 * TCI message
 */
typedef struct {
	tz_msg_header_t header;
	payload_t payload;
} __attribute__ ((packed)) tci_message_t;

extern float *net_input_back;
extern float *net_delta_back;
extern float *net_output_back;
extern char state;

void debug_plot(char *filename, int num, float *tobeplot, int length);

void make_network_CA(int n, float learning_rate, float momentum, float decay, int time_steps, int notruth, int batch, int subdivisions, int random, int adam, float B1, float B2, float eps, int h, int w, int c, int inputs, int max_crop, int min_crop, float max_ratio, float min_ratio, int center, float clip, float angle, float aspect, float saturation, float exposure, float hue, int burn_in, float power, int max_batches);

void update_net_agrv_CA_allocateSM(int workspace_size, float *workspace);

void update_net_agrv_CA(int cond, int workspace_size, float *workspace);

void make_convolutional_layer_CA(int batch, int h, int w, int c, int n, int groups, int size, int stride, int padding, ACTIVATION activation, int batch_normalize, int binary, int xnor, int adam, int flipped, float dot);

void make_maxpool_layer_CA(int batch, int h, int w, int c, int size, int stride, int padding);

void make_dropout_layer_CA(int batch, int inputs, float probability, int w, int h, int c, float *net_prev_output, float *net_prev_delta);

void make_connected_layer_CA(int batch, int inputs, int outputs, ACTIVATION activation, int batch_normalize, int adam);

void make_softmax_layer_CA(int batch, int inputs, int groups, float temperature, int w, int h, int c, int spatial, int noloss);

void make_cost_layer_CA(int batch, int inputs, COST_TYPE cost_type, float scale, float ratio, float noobject_scale, float thresh);

void forward_network_CA(float *net_input, int net_inputs, int net_batch, int net_train);

void forward_network_back_CA(float *l_output, int net_inputs, int net_batch);

void backward_network_CA_addidion(float *l_output, float *l_delta, int net_inputs, int net_batch);

void backward_network_CA(float *net_input, int l_inputs, int batch, int net_train);

void backward_network_back_CA_addidion(float *l_output, float *l_delta, int net_inputs, int net_batch);

void backward_network_back_CA(float *net_input, int l_inputs, int net_batch, float *net_delta);

void update_network_CA(update_args a);

void net_truth_CA(float *net_truth, int net_truths, int net_batch);

void calc_network_loss_CA(int n, int batch);

void net_output_return_CA(int net_outputs, int net_batch);

void transfer_weights_CA(float *vec, int length, int layer_i, char type, int additional);

void save_weights_CA(float *vec, int length, int layer_i, char type);

void summary_array(char *print_name, float *arr, int n);

#endif