#define _GNU_SOURCE
#include "error/check_error.h"
#include "error/check_h5_error.h"
#include "model/blast_scores_data.h"
#include "model/blast_scores_data_init.h"
#include "load_blast_scores.h"
#include <hdf5_hl.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>

/*
 * A simple sanity check can be performed by comparing the HDF5
 * content with the content of the input text file.  Data block output
 * from the first command should be equal to the first 5 lines of the
 * input file while data block output from the second command should
 * be equal to the last 5 lines of the input file.
 *
 *   h5dump --dataset=blast --start "0" --count "5" influenza.h5
 *
 *   h5dump --dataset=blast --start "5749892" --count "5" influenza.h5
 */
void
load_blast_scores (hid_t file_id, const char *file_name)
{
  size_t dst_size;
  size_t dst_offset[BLAST_SCORES_DATA_FIELD_NUM];
  size_t dst_sizes[BLAST_SCORES_DATA_FIELD_NUM];
  hid_t field_type[BLAST_SCORES_DATA_FIELD_NUM];

  blast_scores_data_init (&dst_size, dst_offset, dst_sizes, field_type);

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

  blast_scores_data p_data[10000];
  FILE *dat = fopen (file_name, "r");
  if (dat == NULL)
    check_error (__FILE__, __LINE__);
  char *line = NULL;
  size_t len = 0;
  int i = -1;
  int written = 0;
  int current_line = 0;

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

      char *running = strdup (line);
      char *token = NULL;
      
      token = strsep (&running, ",");
      p_data[i].source_gi = atoi (&token[4]);

      token = strsep (&running, ",");
      p_data[i].source_start = atoi (token);

      token = strsep (&running, ",");
      p_data[i].source_end = atoi (token);

      token = strsep (&running, ",");
      p_data[i].target_gi = atoi (&token[4]);

      token = strsep (&running, ",");
      p_data[i].target_start = atoi (token);

      token = strsep (&running, ",");
      p_data[i].target_end = atoi (token);

      token = strsep (&running, ",");
      p_data[i].score = atoi (token);

      token = strsep (&running, ",");
      p_data[i].bit_score = strtod (token, NULL);

      token = strsep (&running, ",");
      p_data[i].evalue = strtod (token, NULL);

      if (written == 0)
	{

	  printf ("Creating blast table.\n");
	  
	  const char *blast_scores_data_field_names[BLAST_SCORES_DATA_FIELD_NUM] =
	    BLAST_SCORES_DATA_FIELD_NAMES;

	  herr_t status = H5TBmake_table ("blast", file_id,
					  "blast",
					  BLAST_SCORES_DATA_FIELD_NUM, 1,
					  dst_size,
					  blast_scores_data_field_names,
					  dst_offset, field_type,
					  chunk_size, fill_data,
					  compress,
					  &p_data);

	  if (status < 0)
	    check_h5_error (__FILE__, __LINE__);

	  written = 1;
	  i = -1;
	  current_line = 0;

	}

      if ((current_line % 10000 == 0) && (current_line > 0))
	{
	  
	  herr_t status =
	    H5TBappend_records (file_id, "blast", 10000,
				dst_size, dst_offset, dst_sizes,
				&p_data[0]);
	  if(status < 0)
	    check_h5_error (__FILE__, __LINE__);

	  status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
	  if (status < 0)
	    check_h5_error (__FILE__, __LINE__);

	  printf ("Processed %i records.\n", current_line);

	  written += 10000;
	  i = -1;

	}

      if (running)
	free (running);

    } // End for each line of the input file.

  if (i >= 0)
    {
      herr_t status =
	H5TBappend_records (file_id, "blast", i+1,
			    dst_size, dst_offset, dst_sizes, &p_data[0]);
      if (status < 0)
	check_h5_error (__FILE__, __LINE__);
 
      printf ("Wrote %i records.\n", written + i + 1);
    }

  herr_t status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
  if (status < 0)
    check_h5_error (__FILE__, __LINE__);

  if (line)
    free (line);

  fclose (dat);

  return;
}