/*-------------------------------------------------------------------------------
 * soft_viterbi_k5.c
 * Soft Decision Viterbi K5 Contraint for M17, NXDN, and YSF Fusion
 *
 * Viterbi Code from or Modified from libM17 (see bottom)
 * Wojciech Kaczmarski, SP5WWP
 *
 * LWVMOBILE
 * 2026-01 DSD-FME Florida Man Edition
 *-----------------------------------------------------------------------------*/

#include "dsd.h"

//Ripped from libM17
#define K	5                       //constraint length
#define NUM_STATES (1 << (K - 1)) //number of states

static uint32_t prevMetrics[NUM_STATES];
static uint32_t currMetrics[NUM_STATES];
static uint32_t prevMetricsData[NUM_STATES];
static uint32_t currMetricsData[NUM_STATES];
static uint16_t viterbi_history[600*2];

/**
* @brief Decode unpunctured convolutionally encoded data.
*
* @param out Destination array where decoded data is written.
* @param in Input data.
* @param len Input length in bits.
* @return Number of bit errors corrected.
*/
uint32_t viterbi_decode(uint8_t* out, const uint16_t* in, const uint16_t len)
{

	viterbi_reset();

	size_t pos = 0;
	for(size_t i = 0; i < len; i += 2)
	{
		uint16_t s0 = in[i];
		uint16_t s1 = in[i + 1];

		viterbi_decode_bit(s0, s1, pos);
		pos++;
	}
	uint32_t err = viterbi_chainback(out, pos, len/2);

	//debug
	// fprintf (stderr, "\n vcb: \n");
	// for (size_t i = 0; i < len; i++)
	// 	fprintf (stderr, "%02X", out[i]);

	return err;
}

/**
* @brief Decode punctured convolutionally encoded data.
*
* @param out Destination array where decoded data is written.
* @param in Input data.
* @param punct Puncturing matrix.
* @param in_len Input data length.
* @param p_len Puncturing matrix length (entries).
* @return Number of bit errors corrected.
*/
uint32_t viterbi_decode_punctured(uint8_t* out, const uint16_t* in, const uint8_t* punct, const uint16_t in_len, const uint16_t p_len)
{

	uint16_t umsg[600*2]; //unpunctured message
	uint16_t p=0;		      //puncturer matrix entry
	uint16_t u=0;		      //bits count - unpunctured message
	uint16_t i=0;         //bits read from the input message

	while(i<in_len)
	{
		if(punct[p])
		{
			umsg[u]=in[i];
			i++;
		}
		else
		{
			umsg[u]=0x7FFF;
		}

		u++;
		p++;
		p%=p_len;
	}

	//debug
	// fprintf (stderr, " p: %d, u: %d; p_len: %d; ", p, u, p_len); //

	return viterbi_decode(out, umsg, u) - (u-in_len)*0x7FFF;
}

/**
* @brief Decode one bit and update trellis.
*
* @param s0 Cost of the first symbol.
* @param s1 Cost of the second symbol.
* @param pos Bit position in history.
*/
void viterbi_decode_bit(uint16_t s0, uint16_t s1, const size_t pos)
{
	static const uint16_t COST_TABLE_0[] = {0, 0, 0, 0, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF};
	static const uint16_t COST_TABLE_1[] = {0, 0xFFFF, 0xFFFF, 0, 0, 0xFFFF, 0xFFFF, 0};

	for(uint8_t i = 0; i < NUM_STATES/2; i++)
	{
		uint32_t metric = q_abs_diff(COST_TABLE_0[i], s0)
										+ q_abs_diff(COST_TABLE_1[i], s1);


		uint32_t m0 = prevMetrics[i] + metric;
		uint32_t m1 = prevMetrics[i + NUM_STATES/2] + (0x1FFFE - metric);

		uint32_t m2 = prevMetrics[i] + (0x1FFFE - metric);
		uint32_t m3 = prevMetrics[i + NUM_STATES/2] + metric;

		uint8_t i0 = 2 * i;
		uint8_t i1 = i0 + 1;

		if(m0 >= m1)
		{
				viterbi_history[pos]|=(1<<i0);
				currMetrics[i0] = m1;
		}
		else
		{
				viterbi_history[pos]&=~(1<<i0);
				currMetrics[i0] = m0;
		}

		if(m2 >= m3)
		{
				viterbi_history[pos]|=(1<<i1);
				currMetrics[i1] = m3;
		}
		else
		{
				viterbi_history[pos]&=~(1<<i1);
				currMetrics[i1] = m2;
		}
	}

	//swap
	uint32_t tmp[NUM_STATES];
	for(uint8_t i=0; i<NUM_STATES; i++)
	{
		tmp[i]=currMetrics[i];
	}
	for(uint8_t i=0; i<NUM_STATES; i++)
	{
		currMetrics[i]=prevMetrics[i];
		prevMetrics[i]=tmp[i];
	}
}

/**
* @brief History chainback to obtain final byte array.
*
* @param out Destination byte array for decoded data.
* @param pos Starting position for the chainback.
* @param len Length of the output in bits.
* @return Minimum Viterbi cost at the end of the decode sequence.
*/
uint32_t viterbi_chainback(uint8_t* out, size_t pos, uint16_t len)
{

	uint8_t state = 0;
	size_t bitPos = len+4;

	while(pos > 0)
	{
		bitPos--;
		pos--;
		uint16_t bit = viterbi_history[pos]&((1<<(state>>4)));
		state >>= 1;
		if(bit)
		{
			state |= 0x80;
			out[bitPos/8]|=1<<(7-(bitPos%8));
		}
	}

	uint32_t cost = prevMetrics[0];

	for(size_t i = 0; i < NUM_STATES; i++)
	{
		uint32_t m = prevMetrics[i];
		if(m < cost) cost = m;
	}

	//debug
	// fprintf (stderr, "\n cb: \n");
	// for (size_t i = 0; i < len; i++)
	// 	fprintf (stderr, "%02X", out[i]);

	return cost;
}

/**
 * @brief Reset the decoder state. No args.
 *
 */
void viterbi_reset(void)
{
	memset(viterbi_history, 0, sizeof(viterbi_history));
	memset(currMetrics, 0, sizeof(currMetrics));
	memset(prevMetrics, 0, sizeof(prevMetrics));
	memset(currMetricsData, 0, sizeof(currMetricsData));
	memset(prevMetricsData, 0, sizeof(prevMetricsData));
}

uint16_t q_abs_diff(const uint16_t v1, const uint16_t v2)
{
	if(v2 > v1) return v2 - v1;
	return v1 - v2;
}

//same as symbol levels in m17.h (m17-fme version)
const int8_t symbol_list[4]={-3, -1, +1, +3};

/**
 * @brief Slice payload symbols into soft dibits.
 * Input (RRC filtered baseband sampled at symbol centers)
 * should be already normalized to {-3, -1, +1 +3}.
 * @param out Soft valued dibits (type-4).
 * @param inp Array of 184 floats (1 sample per symbol).
 */
void slice_symbols(uint16_t out[2*SYM_PER_PLD], const float inp[SYM_PER_PLD])
{

  for(uint8_t i=0; i<SYM_PER_PLD; i++)
  {
    //bit 0
    if(inp[i]>=symbol_list[3])
    {
      out[i*2+1]=0xFFFF;
    }
    else if(inp[i]>=symbol_list[2])
    {
      out[i*2+1]=-(float)0xFFFF/(symbol_list[3]-symbol_list[2])*symbol_list[2]+inp[i]*(float)0xFFFF/(symbol_list[3]-symbol_list[2]);
    }
    else if(inp[i]>=symbol_list[1])
    {
      out[i*2+1]=0x0000;
    }
    else if(inp[i]>=symbol_list[0])
    {
      out[i*2+1]=(float)0xFFFF/(symbol_list[1]-symbol_list[0])*symbol_list[1]-inp[i]*(float)0xFFFF/(symbol_list[1]-symbol_list[0]);
    }
    else
    {
      out[i*2+1]=0xFFFF;
    }

    //bit 1
    if(inp[i]>=symbol_list[2])
    {
      out[i*2]=0x0000;
    }
    else if(inp[i]>=symbol_list[1])
    {
      out[i*2]=0x7FFF-inp[i]*(float)0xFFFF/(symbol_list[2]-symbol_list[1]);
    }
    else
    {
      out[i*2]=0xFFFF;
    }
  }
}

/**
 * @brief Slice payload symbols into soft dibits.
 * Input (RRC filtered baseband sampled at symbol centers)
 * should be already normalized to {-3, -1, +1 +3}.
 * @param out Soft valued dibits (type-4).
 * @param inp Array of len floats (1 sample per symbol).
 * @param len legnth value to run this at
 */
void slice_symbols_to_len(uint16_t * out, const float * inp, int len)
{

	for(int i=0; i<len; i++)
	{
		//bit 0
		if(inp[i]>=symbol_list[3])
		{
			out[i*2+1]=0xFFFF;
		}
		else if(inp[i]>=symbol_list[2])
		{
			out[i*2+1]=-(float)0xFFFF/(symbol_list[3]-symbol_list[2])*symbol_list[2]+inp[i]*(float)0xFFFF/(symbol_list[3]-symbol_list[2]);
		}
		else if(inp[i]>=symbol_list[1])
		{
			out[i*2+1]=0x0000;
		}
		else if(inp[i]>=symbol_list[0])
		{
			out[i*2+1]=(float)0xFFFF/(symbol_list[1]-symbol_list[0])*symbol_list[1]-inp[i]*(float)0xFFFF/(symbol_list[1]-symbol_list[0]);
		}
		else
		{
			out[i*2+1]=0xFFFF;
		}

		//bit 1
		if(inp[i]>=symbol_list[2])
		{
			out[i*2]=0x0000;
		}
		else if(inp[i]>=symbol_list[1])
		{
			out[i*2]=0x7FFF-inp[i]*(float)0xFFFF/(symbol_list[2]-symbol_list[1]);
		}
		else
		{
			out[i*2]=0xFFFF;
		}
	}
}

/**
 * @brief Soft logic NOT.
 * 
 * @param a Input A.
 * @return uint16_t Output = not A.
 */
uint16_t soft_bit_NOT(const uint16_t a)
{
  return 0xFFFFU-a;
}

//randomizing pattern for M17
const uint8_t rand_seq[46]=
{
  0xD6, 0xB5, 0xE2, 0x30, 0x82, 0xFF, 0x84, 0x62, 0xBA, 0x4E, 0x96, 0x90, 0xD8, 0x98, 0xDD, 0x5D, 0x0C, 0xC8, 0x52, 0x43, 0x91, 0x1D, 0xF8,
  0x6E, 0x68, 0x2F, 0x35, 0xDA, 0x14, 0xEA, 0xCD, 0x76, 0x19, 0x8D, 0xD5, 0x80, 0xD1, 0x33, 0x87, 0x13, 0x57, 0x18, 0x2D, 0x29, 0x78, 0xC3
};

/**
 * @brief Randomize type-4 soft bits.
 * 
 * @param inp Input 368 soft type-4 bits.
 */
void randomize_soft_bits(uint16_t inp[SYM_PER_PLD*2])
{
  for(uint16_t i=0; i<SYM_PER_PLD*2; i++)
  {
    if((rand_seq[i/8]>>(7-(i%8)))&1) //flip bit if '1'
    {
      inp[i]=soft_bit_NOT(inp[i]);
    }
  }
}

//interleaver pattern for M17
const uint16_t intrl_seq[SYM_PER_PLD*2]=
{
	0, 137, 90, 227, 180, 317, 270, 39, 360, 129, 82, 219, 172, 309, 262, 31,
	352, 121, 74, 211, 164, 301, 254, 23, 344, 113, 66, 203, 156, 293, 246, 15,
	336, 105, 58, 195, 148, 285, 238, 7, 328, 97, 50, 187, 140, 277, 230, 367,
	320, 89, 42, 179, 132, 269, 222, 359, 312, 81, 34, 171, 124, 261, 214, 351,
	304, 73, 26, 163, 116, 253, 206, 343, 296, 65, 18, 155, 108, 245, 198, 335,
	288, 57, 10, 147, 100, 237, 190, 327, 280, 49, 2, 139, 92, 229, 182, 319,
	272, 41, 362, 131, 84, 221, 174, 311, 264, 33, 354, 123, 76, 213, 166, 303,
	256, 25, 346, 115, 68, 205, 158, 295, 248, 17, 338, 107, 60, 197, 150, 287,
	240, 9, 330, 99, 52, 189, 142, 279, 232, 1, 322, 91, 44, 181, 134, 271,
	224, 361, 314, 83, 36, 173, 126, 263, 216, 353, 306, 75, 28, 165, 118, 255,
	208, 345, 298, 67, 20, 157, 110, 247, 200, 337, 290, 59, 12, 149, 102, 239,
	192, 329, 282, 51, 4, 141, 94, 231, 184, 321, 274, 43, 364, 133, 86, 223,
	176, 313, 266, 35, 356, 125, 78, 215, 168, 305, 258, 27, 348, 117, 70, 207,
	160, 297, 250, 19, 340, 109, 62, 199, 152, 289, 242, 11, 332, 101, 54, 191,
	144, 281, 234, 3, 324, 93, 46, 183, 136, 273, 226, 363, 316, 85, 38, 175,
	128, 265, 218, 355, 308, 77, 30, 167, 120, 257, 210, 347, 300, 69, 22, 159,
	112, 249, 202, 339, 292, 61, 14, 151, 104, 241, 194, 331, 284, 53, 6, 143,
	96, 233, 186, 323, 276, 45, 366, 135, 88, 225, 178, 315, 268, 37, 358, 127,
	80, 217, 170, 307, 260, 29, 350, 119, 72, 209, 162, 299, 252, 21, 342, 111,
	64, 201, 154, 291, 244, 13, 334, 103, 56, 193, 146, 283, 236, 5, 326, 95,
	48, 185, 138, 275, 228, 365, 318, 87, 40, 177, 130, 267, 220, 357, 310, 79,
	32, 169, 122, 259, 212, 349, 302, 71, 24, 161, 114, 251, 204, 341, 294, 63,
	16, 153, 106, 243, 196, 333, 286, 55, 8, 145, 98, 235, 188, 325, 278, 47
};

/**
 * @brief Reorder (interleave) 368 soft bits.
 * 
 * @param outp Reordered soft bits.
 * @param inp Input soft bits.
 */
void reorder_soft_bits(uint16_t outp[SYM_PER_PLD*2], const uint16_t inp[SYM_PER_PLD*2])
{
  for(uint16_t i=0; i<SYM_PER_PLD*2; i++)
    outp[i]=inp[intrl_seq[i]];
}

//NXDN all-in-one function to go from punctured bits to valid output, return viterbi error
uint32_t nxdn_soft_decision_viterbi(uint8_t * bits, const uint16_t * interleave, uint8_t * puncture, int d_len, int p_len, int num_bytes, int offset, uint8_t * viterbi_bits, uint8_t * viterbi_bytes)
{
  uint32_t error = 0;
  uint16_t soft_bit[1000];
  uint8_t temp_bits[500];
  memset(temp_bits, 0, sizeof(temp_bits));
  memset(soft_bit, 0, sizeof(soft_bit));

  //deinterleaved bits
  uint8_t deinterleaved_bits[500];
  memset(deinterleaved_bits, 0, sizeof(deinterleaved_bits));

  //deinterleaving the bits
  for (int i = 0; i < d_len; i++)
    deinterleaved_bits[interleave[i]] = bits[i];

  //converting back to dibits
  uint8_t dbuf[500];
  memset(dbuf, 0, sizeof(dbuf));
  for (int i = 0; i < d_len/2; i++)
    dbuf[i] = (deinterleaved_bits[(i*2)+0] << 1) | deinterleaved_bits[(i*2)+1];

  float sbuf[500];
  memset(sbuf, 0.0f, sizeof(sbuf));

  //convert dbuf into a symbol array
  for (int i = 0; i < d_len/2; i++)
  {
    if      (dbuf[i] == 0) sbuf[i] = +1.0f;
    else if (dbuf[i] == 1) sbuf[i] = +3.0f;
    else if (dbuf[i] == 2) sbuf[i] = -1.0f;
    else if (dbuf[i] == 3) sbuf[i] = -3.0f;
    else                   sbuf[i] = +0.0f;
  }

  //slice symbols to soft dibits
  slice_symbols_to_len(soft_bit, sbuf, d_len/2);

  //viterbi
  error = viterbi_decode_punctured(viterbi_bytes, soft_bit, puncture, d_len, p_len);

  //debug
  // fprintf (stderr, " Ve: %1.1f; ", (float)error/(float)0xFFFF);

  //debug
  // for (int i = 0; i < num_bytes+1; i++)
  // 	fprintf(stderr, "%02X", viterbi_bytes[i]);

  //load viterbi_bytes into bits
  unpack_byte_array_into_bit_array(viterbi_bytes, temp_bits, num_bytes+1);

  //rearrange bits to get correct offset
  for (int i = 0; i < num_bytes*8; i++)
    viterbi_bits[i] = temp_bits[i+offset];

  //zero out viterbi_bytes to reload packed appropriately
  for (int i = 0; i < num_bytes; i++)
    viterbi_bytes[i] = 0;

  pack_bit_array_into_byte_array(viterbi_bits, viterbi_bytes, num_bytes);

  return error;
}

//ysf fake puncture
uint8_t ysf_puncture[4] = {1,1,1,1};

//YSF all-in-one function to go from dibits to valid output, return viterbi error
uint32_t ysf_soft_decision_viterbi(uint8_t * dbuf, int d_len, int num_bytes, int offset, uint8_t * viterbi_bits, uint8_t * viterbi_bytes)
{
  uint32_t error = 0;
  uint16_t soft_bit[1000];
  uint8_t temp_bits[500];
  memset(temp_bits, 0, sizeof(temp_bits));
  memset(soft_bit, 0, sizeof(soft_bit));

  float sbuf[500];
  memset(sbuf, 0.0f, sizeof(sbuf));

  //convert dbuf into a symbol array
  for (int i = 0; i < d_len/2; i++)
  {
    if      (dbuf[i] == 0) sbuf[i] = +1.0f;
    else if (dbuf[i] == 1) sbuf[i] = +3.0f;
    else if (dbuf[i] == 2) sbuf[i] = -1.0f;
    else if (dbuf[i] == 3) sbuf[i] = -3.0f;
    else                   sbuf[i] = +0.0f;
  }

  //slice symbols to soft dibits
  slice_symbols_to_len(soft_bit, sbuf, d_len/2);

  //viterbi
  error = viterbi_decode_punctured(viterbi_bytes, soft_bit, ysf_puncture, d_len, 4);

  //debug
  // fprintf (stderr, " Ve: %1.1f; ", (float)error/(float)0xFFFF);

  //debug
  // for (int i = 0; i < num_bytes+1; i++)
  // 	fprintf(stderr, "%02X", viterbi_bytes[i]);

  //load viterbi_bytes into bits
  unpack_byte_array_into_bit_array(viterbi_bytes, temp_bits, num_bytes+1);

  //rearrange bits to get correct offset
  for (int i = 0; i < num_bytes*8; i++)
    viterbi_bits[i] = temp_bits[i+offset];

  //zero out viterbi_bytes to reload packed appropriately
  for (int i = 0; i < num_bytes; i++)
    viterbi_bytes[i] = 0;

  pack_bit_array_into_byte_array(viterbi_bits, viterbi_bytes, num_bytes);

  return error;
}

//original sources for viterbi code and misc math related
//--------------------------------------------------------------------
// M17 C library - decode/viterbi.c
//
// This file contains:
// - the Viterbi decoder
//
// Wojciech Kaczmarski, SP5WWP
// M17 Project, 29 December 2023
//--------------------------------------------------------------------

//--------------------------------------------------------------------
// M17 C library - math/math.c
//
// This file contains:
// - absolute difference value
// - Euclidean norm (L2) calculation for n-dimensional vectors (float)
// - soft-valued arrays to integer conversion (and vice-versa)
// - fixed-valued multiplication and division
//
// Wojciech Kaczmarski, SP5WWP
// M17 Project, 29 December 2023
//--------------------------------------------------------------------