#include "load_influenza_faa.h"
#include "check_error.h"
#include "check_h5_error.h"
#include "hdf5_hl.h"
#include <string.h>
#include <stdlib.h>

#define SEQUENCE_DATA_FIELD_NUM 4

void
load_influenza_faa (hid_t file_id)
{
  typedef struct
  {
    int gi;
    char gb[9];
    char description[196];
    char protein_type[7];
  } sequence_data;
  
  size_t dst_size = sizeof (sequence_data);
  size_t dst_offset[SEQUENCE_DATA_FIELD_NUM] =
    { HOFFSET (sequence_data, gi),
      HOFFSET (sequence_data, gb),
      HOFFSET (sequence_data, description),
      HOFFSET (sequence_data, protein_type)
    };

  sequence_data dst_buf[1];

  size_t dst_sizes[SEQUENCE_DATA_FIELD_NUM] = {
    sizeof (dst_buf[0].gi),
    sizeof (dst_buf[0].gb),
    sizeof (dst_buf[0].description),
    sizeof (dst_buf[0].protein_type)
  };

  hid_t field_type[SEQUENCE_DATA_FIELD_NUM];

  field_type[0] = H5T_NATIVE_INT;

  hid_t gb_type = H5Tcopy (H5T_C_S1);
  H5Tset_size (gb_type, 9);
  field_type[1] = gb_type;

  hid_t description_type = H5Tcopy (H5T_C_S1);
  H5Tset_size (description_type, 196);
  field_type[2] = description_type;

  hid_t protein_type_type = H5Tcopy (H5T_C_S1);
  H5Tset_size (protein_type_type, 7);
  field_type[3] = protein_type_type;

  const char *field_names[SEQUENCE_DATA_FIELD_NUM] = 
    { "GI",
      "GB",
      "Description",
      "Protein Type" };

  hsize_t chunk_size = 10;
  int *fill_data = NULL;
  int compress = 0;

  sequence_data p_data;
  FILE *dat = fopen ("/home/don/exp004/genomes/INFLUENZA/influenza.faa",
		     "r");
  if (dat == NULL)
    check_error (__FILE__, __LINE__);
  char *line = NULL;
  size_t len = 0;
  int current_line = 0;

  while (getline (&line, &len, dat) != -1)
    {
      current_line++;      

      // Header line.
      if (line[0] == '>')
	{
	  char *running = strdup (line);
	  char *token = NULL;
	  
	  // Eat the ">gi".
	  strsep (&running, "|");
	  
	  // GI value.
	  token = strsep (&running, "|");
	  p_data.gi = atoi (token);

	  // Eat the "gb"
	  strsep (&running, "|");

	  // GB value.
	  strncpy (p_data.gb, strsep(&running, "|"), sizeof (p_data.gb));

	  // Description value.
	  strncpy (p_data.description, strsep (&running, "|"),
		   sizeof (p_data.description));

	  strncpy (p_data.protein_type, "", sizeof (p_data.protein_type));

	  if (current_line == 1)
	    {
	      herr_t status = H5TBmake_table ("influenza.faa", file_id,
					      "influenza.faa",
					      SEQUENCE_DATA_FIELD_NUM, 1,
					      dst_size, field_names,
					      dst_offset, field_type,
					      chunk_size, fill_data, compress,
					      &p_data);
	      if (status < 0)
		check_h5_error (status, __FILE__, __LINE__);
	    }
	  else
	    {
	      herr_t status =
		H5TBappend_records (file_id, "influenza.faa", 1, dst_size,
				    dst_offset, dst_sizes, &p_data);
	      if (status < 0)
		check_h5_error (status, __FILE__, __LINE__);
	    }

	  if (running)
	    free (running);

	}
      
    }

  if (line)
    free (line);

  fclose (dat);

  H5Tclose (gb_type);
  H5Tclose (description_type);
  H5Tclose (protein_type_type);

  return;
}