obitools3/src/obidms_taxonomy.c

/********************************************************************
 * OBIDMS taxonomy functions                                        *
 ********************************************************************/

/**
 * @file obidms_taxonomy.c
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 * @date March 2nd 2016
 * @brief Functions for reading binary taxonomy files.
 */


#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <stdbool.h>
#include <fcntl.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <unistd.h>

#include "obidms_taxonomy.h"
#include "obidms.h"
#include "obidebug.h"
#include "obierrno.h"
#include "utils.h"


#define DEBUG_LEVEL 0	// TODO has to be defined somewhere else (cython compil flag?)


int cmp_rank_labels(const void* label1, const void* label2)
{
	return strcmp((const char*)label1,*(const char**)label2);
}


static int cmp_taxids(const void* ptaxid, const void* ptaxon)
{
  ecotx_t* current_taxon = (ecotx_t*) ptaxon;
  int32_t  taxid = (int32_t) ((size_t) ptaxid);
  return taxid - current_taxon->taxid;
}


static int cmp_str(const void* s1, const void* s2)
{
    return strcmp(*((char**)s1), *((char**)s2));
}


static int cmp_names(const void* n1, const void* n2)
{
	econame_t name1 = *((econame_t*)n1);
	econame_t name2 = *((econame_t*)n2);

    return strcmp(name1.name, name2.name);
}


char* get_taxonomy_path(OBIDMS_p dms, const char* tax_name)
{
	char*   all_tax_dir_path;
	char*   tax_path;

	all_tax_dir_path = obi_dms_get_full_path(dms, TAXONOMY_DIR_NAME);
	if (all_tax_dir_path == NULL)
		return NULL;

	tax_path = (char*) malloc((strlen(all_tax_dir_path) + strlen(tax_name) + 2)*sizeof(char));
	if (tax_path == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for taxonomy path");
		free(all_tax_dir_path);
		return NULL;
	}

	if (sprintf(tax_path, "%s/%s", all_tax_dir_path, tax_name) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError building taxonomy path");
		free(all_tax_dir_path);
		return NULL;
	}

	free(all_tax_dir_path);

	return tax_path;
}


int32_t rank_index(const char* label, ecorankidx_t* ranks)
{
	char **rep;

	rep = bsearch(label, ranks->label, ranks->count, sizeof(char*), cmp_rank_labels);

	if (rep)
		return rep-ranks->label;

	return -1;
}


void* read_ecorecord(FILE* f, int32_t* record_size)
{
	static void* buffer = NULL;
	int32_t      buffer_size = 0;
	int32_t      read;

	if (!record_size)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError reading a taxonomy file: record_size can not be NULL");
 		return NULL;
	}

	read = fread(record_size,
	      		 sizeof(int32_t),
				 1,
	             f);

	if (feof(f))
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError reading a taxonomy file: reached end of file");
 		return NULL;
	}

	if (read != 1)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError reading a taxonomy file: error reading record size");
 		return NULL;
	}

	if (buffer_size < *record_size)
	{
		if (buffer)
			buffer = realloc(buffer, *record_size);
		else
			buffer = malloc(*record_size);
		if (buffer == NULL)
		{
	 		obi_set_errno(OBI_MALLOC_ERROR);
	 		obidebug(1, "\nError reading a taxonomy file: error allocating memory");
	 		return NULL;
		}
	}

	read = fread(buffer,
				 *record_size,
				 1,
				 f);

	if (read != 1)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError reading a taxonomy file: error reading a record %d, %d", read, *record_size);
 		free(buffer);
 		return NULL;
	}

	return buffer;
};


ecotx_t* readnext_ecotaxon(FILE* f, ecotx_t* taxon)
{
	ecotxformat_t* raw;
	int32_t        record_length;

	raw = read_ecorecord(f, &record_length);
	if (!raw)
		return NULL;

	taxon->parent = (ecotx_t*) ((size_t) raw->parent);
	taxon->taxid  = raw->taxid;
	taxon->rank   = raw->rank;
	taxon->farest = -1;

	taxon->name   = malloc((raw->name_length+1) * sizeof(char));
	if (taxon->name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError reading a taxonomy file: error allocating memory");
 		return NULL;
	}

	strncpy(taxon->name, raw->name, raw->name_length);
	taxon->name[raw->name_length] = 0;					// TODO note: this line is probably missing in ROBITaxonomy and source of a bug

	return taxon;
}


FILE* open_ecorecorddb(const char* file_name,
                       int32_t*    count,
                       int32_t     abort_on_open_error)
{
    FILE*		f;
	int32_t     read;

	f = fopen(file_name, "rb");

	if (!f)
	{
		if (abort_on_open_error)
		{
	 		obi_set_errno(OBI_TAXONOMY_ERROR);
	 		obidebug(1, "\nCouldn't open a taxonomy file");
	 		fclose(f);
	 		return NULL;
		}
	 	else
	 	{
	 		*count = 0;
	 		fclose(f);
	 		return NULL;
	 	}
	}

	read = fread(count,
	      		 sizeof(int32_t),
				 1,
	      		 f);

	if (read != 1)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError reading taxonomy record size");
 		fclose(f);
 		return NULL;
	}

	return f;
}


ecorankidx_t* read_rankidx(const char* ranks_file_name)
{
	int32_t      			count;
	FILE*        			ranks_file;
	ecorankidx_t*           ranks_index;
	int32_t      			i;
	int32_t      			rank_length;
	char*					buffer;

	ranks_file = open_ecorecorddb(ranks_file_name, &count, 0);
	if (ranks_file==NULL)
		return NULL;

	ranks_index = (ecorankidx_t*) malloc(sizeof(ecorankidx_t) + sizeof(char*) * (count-1));
	if (ranks_index == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for taxonomy rank structure");
 		fclose(ranks_file);
 		return NULL;
	}

	ranks_index->count = count;

	for (i=0; i < count; i++)
	{
		buffer = read_ecorecord(ranks_file, &rank_length);
		if (buffer == NULL)
		{
	 		obi_set_errno(OBI_TAXONOMY_ERROR);
	 		obidebug(1, "\nError reading a value in a taxonomy file");
	 		fclose(ranks_file);
	 		free(ranks_index);
	 		return NULL;
		}
		ranks_index->label[i] = (char*) malloc(rank_length+1);
		if (ranks_index->label[i] == NULL)
		{
	 		obi_set_errno(OBI_MALLOC_ERROR);
	 		obidebug(1, "\nError allocating memory for taxonomy rank label");
	 		fclose(ranks_file);
	 		free(ranks_index);
	 		free(buffer);
	 		return NULL;
		}
		strncpy(ranks_index->label[i], buffer, rank_length);
		(ranks_index->label[i])[rank_length] = 0;
	}

	fclose(ranks_file);

	return ranks_index;
}


ecotxidx_t* read_taxonomyidx(const char* taxa_file_name, const char* local_taxa_file_name)
{
	int32_t      	  count_taxa;
	int32_t      	  count_local_taxa;
	FILE*			  f_taxa;
	FILE* 			  f_local_taxa;
	ecotxidx_t*		  taxa_index;
	struct ecotxnode* t;
	int32_t      	  i;
	int32_t      	  j;

	f_taxa = open_ecorecorddb(taxa_file_name, &count_taxa, 1);
	if (f_taxa == NULL)
	{
 		obidebug(1, "\nError reading taxonomy taxa file");
 		return NULL;
	}

	f_local_taxa = open_ecorecorddb(local_taxa_file_name, &count_local_taxa, 0);

	taxa_index = (ecotxidx_t*) malloc(sizeof(ecotxidx_t) + sizeof(ecotx_t) * (count_taxa + count_local_taxa - 1));
	if (taxa_index == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for taxonomy structure");
 		fclose(f_taxa);
 		fclose(f_local_taxa);
 		return NULL;
	}

	taxa_index->count = count_taxa + count_local_taxa;
	taxa_index->ncbi_count = count_taxa;
	taxa_index->local_count = count_local_taxa;
	taxa_index->buffer_size = taxa_index->count;

	taxa_index->max_taxid = 0;
	printf("Reading %d taxa...\n", count_taxa);
	for (i=0; i<count_taxa; i++)
	{
		readnext_ecotaxon(f_taxa, &(taxa_index->taxon[i]));
		taxa_index->taxon[i].parent = taxa_index->taxon + (size_t) taxa_index->taxon[i].parent;
		taxa_index->taxon[i].parent->farest = 0;
		if (taxa_index->taxon[i].taxid > taxa_index->max_taxid)
			taxa_index->max_taxid = taxa_index->taxon[i].taxid;
	}

	if (count_local_taxa > 0)
		printf("Reading %d local taxa...\n", count_local_taxa);
	else
		printf("No local taxa\n");

	count_taxa = taxa_index->count;

	for (; i < count_taxa; i++){
		readnext_ecotaxon(f_local_taxa, &(taxa_index->taxon[i]));
		taxa_index->taxon[i].idx = i;
		taxa_index->taxon[i].parent = taxa_index->taxon + (size_t) taxa_index->taxon[i].parent;
		taxa_index->taxon[i].parent->farest=0;
		if (taxa_index->taxon[i].taxid > taxa_index->max_taxid)
			taxa_index->max_taxid = taxa_index->taxon[i].taxid;
	}

	for (i=0; i < count_taxa; i++)
	{
		t = taxa_index->taxon+i;
		if (t->farest == -1)
		{
			t->farest=0;
            while (t->parent != t)
            {
            	j = t->farest + 1;
            	if (j > t->parent->farest)
            	{
            		t->parent->farest = j;
            		t=t->parent;
            	}
            	else
            		t = taxa_index->taxon;
            }
		}
	}

	fclose(f_taxa);
	if (f_local_taxa != NULL)
		fclose(f_local_taxa);

	return taxa_index;
}


econame_t* readnext_econame(FILE* f, econame_t* name, OBIDMS_taxonomy_p taxonomy)
{
	econameformat_t* raw;
	int32_t          record_length;

	raw = read_ecorecord(f, &record_length);
	if (raw == NULL)
 		return NULL;

	name->is_scientific_name = raw->is_scientific_name;

	name->name = malloc((raw->name_length + 1) * sizeof(char));
	if (name->name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for a taxon name");
 		free(raw);
 		return NULL;
	}
	strncpy(name->name, raw->names, raw->name_length);
	name->name[raw->name_length] = 0;

	name->class_name = malloc((raw->class_length+1) * sizeof(char));
	if (name->class_name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for a taxon class name");
 		free(name->name);
 		free(raw);
 		return NULL;
	}
	strncpy(name->class_name,(raw->names + raw->name_length), raw->class_length);
	name->class_name[raw->class_length] = 0;

	name->taxon = taxonomy->taxa->taxon + raw->taxid;

	return name;
}


econameidx_t* read_nameidx(const char *file_name, OBIDMS_taxonomy_p taxonomy)
{
	int32_t      		count;
	FILE*				f;
	econameidx_t*		index_names;
	int32_t      		i;

	f = open_ecorecorddb(file_name, &count, 0);
	if (f == NULL)
	{
 		obidebug(1, "\nError reading taxonomy name file");
 		return NULL;
	}

	index_names = (econameidx_t*) malloc(sizeof(econameidx_t) + sizeof(econame_t) * (count-1));
	if (index_names == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError reading taxonomy name file");
 		return NULL;
	}

	index_names->count = count;

	for (i=0; i < count; i++)
	{
		readnext_econame(f, (index_names->names)+i, taxonomy);
		if ((index_names->names)+i == NULL)
		{
	 		obi_set_errno(OBI_TAXONOMY_ERROR);
	 		obidebug(1, "\nError reading taxonomy name file");
	 		free(index_names);
	 		return NULL;
		}
	}

	fclose(f);

	return index_names;
}






// Functions to write taxonomy structure to binary files

int write_rankidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name)		// TODO prefix in taxonomy struct?
{
	int     i;
	char* 	file_name;
	int 	file_descriptor;
	off_t 	file_size;
	char*   taxonomy_path;
	int32_t	length;

	// Compute file size
	file_size = sizeof(int32_t);
	for (i=0; i < (tax->ranks)->count; i++)
	{
		file_size = file_size + sizeof(int32_t);	// To store label size
		file_size = file_size + strlen(((tax->ranks)->label)[i]);	// To store label
	}

	// Build the taxonomy directory path
	taxonomy_path = get_taxonomy_path(dms, taxonomy_name);

	file_name =	(char*) malloc((strlen(taxonomy_path) + strlen(taxonomy_name) + 5)*sizeof(char));
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating the memory for a binary taxonomy file name");
		return -1;
	}

	// Build the file path
	if (sprintf(file_name, "%s/%s.rdx", taxonomy_path, taxonomy_name) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError building a binary taxonomy file name");
		return -1;
	}

	free(taxonomy_path);

	// Create file
	file_descriptor = open(file_name, O_RDWR | O_CREAT | O_EXCL, 0777);
	if (file_descriptor < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError creating a binary taxonomy file %s", file_name);
		free(file_name);
		return -1;
	}

	free(file_name);

	// Truncate the file to the right size
	if (ftruncate(file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError truncating a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write rank count
	if (write(file_descriptor, &((tax->ranks)->count), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError writing in a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write ranks
	for (i=0; i < (tax->ranks)->count; i++)
	{
		length = strlen(((tax->ranks)->label)[i]);

		// Write rank size
		if (write(file_descriptor, &length, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write rank label
		if (write(file_descriptor, ((tax->ranks)->label)[i], length) < ((ssize_t) length))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
	}

	// Close file
	if (close(file_descriptor) < 0)
	{
		obi_set_errno(OBIDMS_UNKNOWN_ERROR);
		obidebug(1, "\nError closing a DMS information file");
		return -1;
	}

	return 0;
}


int write_taxonomyidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name)		// TODO prefix in taxonomy struct? keep argument but if NULL, use the one in struct?
{
	int     i;
	char* 	file_name;
	int 	file_descriptor;
	off_t 	file_size;
	char*   taxonomy_path;
	int32_t	name_length;
	int32_t record_size;

	// Compute file size
	file_size = sizeof(int32_t);	// To store record count
	for (i=0; i < (tax->taxa)->ncbi_count; i++)
	{
		file_size = file_size + sizeof(int32_t) * 5;				// To store record size, taxid, rank index, parent index, and name length
		file_size = file_size + strlen(tax->taxa->taxon[i].name);	// To store name
	}

	// Build the taxonomy directory path
	taxonomy_path = get_taxonomy_path(dms, taxonomy_name);

	file_name =	(char*) malloc((strlen(taxonomy_path) + strlen(taxonomy_name) + 5)*sizeof(char));
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating the memory for a binary taxonomy file name");
		return -1;
	}

	// Build the file path
	if (sprintf(file_name, "%s/%s.tdx", taxonomy_path, taxonomy_name) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError building a binary taxonomy file name");
		return -1;
	}

	free(taxonomy_path);

	// Create file
	file_descriptor = open(file_name, O_RDWR | O_CREAT | O_EXCL, 0777);
	if (file_descriptor < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError creating a binary taxonomy file");
		free(file_name);
		return -1;
	}

	free(file_name);

	// Truncate the file to the right size
	if (ftruncate(file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError truncating a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write record count
	if (write(file_descriptor, &(tax->taxa->ncbi_count), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError writing in a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write records
	for (i=0; i < (tax->taxa)->ncbi_count; i++)
	{
		name_length = strlen(tax->taxa->taxon[i].name);
		record_size = 4*sizeof(int32_t) + name_length;

		// Write record size
		if (write(file_descriptor, &record_size, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write taxid
		if (write(file_descriptor, &(tax->taxa->taxon[i].taxid), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write rank index
		if (write(file_descriptor, &(tax->taxa->taxon[i].rank), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write parent index
		if (write(file_descriptor, &((tax->taxa->taxon[i].parent)->idx), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name length
		if (write(file_descriptor, &name_length, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name
		if (write(file_descriptor, tax->taxa->taxon[i].name, name_length) < ((ssize_t) name_length))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
	}

	// Close file
	if (close(file_descriptor) < 0)
	{
		obi_set_errno(OBIDMS_UNKNOWN_ERROR);
		obidebug(1, "\nError closing a DMS information file");
		return -1;
	}

	return 0;
}


int write_local_taxonomy_idx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name)		// TODO prefix in taxonomy struct? keep argument but if NULL, use the one in struct?
{
	int     i;
	char* 	file_name;
	int 	file_descriptor;
	off_t 	file_size;
	char*   taxonomy_path;
	int32_t	name_length;
	int32_t record_size;

	// Compute file size
	file_size = sizeof(int32_t);	// To store record count
	for (i=(tax->taxa)->ncbi_count; i < (tax->taxa)->count; i++)
	{
		file_size = file_size + sizeof(int32_t) * 5;				// To store record size, taxid, rank index, parent index, and name length
		file_size = file_size + strlen(tax->taxa->taxon[i].name);	// To store name
	}

	// Build the taxonomy directory path
	taxonomy_path = get_taxonomy_path(dms, taxonomy_name);

	file_name =	(char*) malloc((strlen(taxonomy_path) + strlen(taxonomy_name) + 5)*sizeof(char));
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating the memory for a binary taxonomy file name");
		return -1;
	}

	// Build the file path
	if (sprintf(file_name, "%s/%s.ldx", taxonomy_path, taxonomy_name) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError building a binary taxonomy file name");
		return -1;
	}

	free(taxonomy_path);

	// Create file
	file_descriptor = open(file_name, O_RDWR | O_CREAT, 0777);
	if (file_descriptor < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError creating a binary taxonomy file");
		free(file_name);
		return -1;
	}

	free(file_name);

	// Truncate the file to the right size
	if (ftruncate(file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError truncating a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write record count
	if (write(file_descriptor, &((tax->taxa)->local_count), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError writing in a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write records
	for (i=(tax->taxa)->ncbi_count; i < (tax->taxa)->count; i++)
	{
		name_length = strlen(tax->taxa->taxon[i].name);
		record_size = 4*sizeof(int32_t) + name_length;

		// Write record size
		if (write(file_descriptor, &record_size, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write taxid
		if (write(file_descriptor, &(tax->taxa->taxon[i].taxid), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write rank index
		if (write(file_descriptor, &(tax->taxa->taxon[i].rank), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write parent index
		if (write(file_descriptor, &((tax->taxa->taxon[i].parent)->idx), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name length
		if (write(file_descriptor, &name_length, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name
		if (write(file_descriptor, tax->taxa->taxon[i].name, name_length) < ((ssize_t) name_length))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
	}

	// Close file
	if (close(file_descriptor) < 0)
	{
		obi_set_errno(OBIDMS_UNKNOWN_ERROR);
		obidebug(1, "\nError closing a DMS information file");
		return -1;
	}

	return 0;
}


int write_nameidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name)		// TODO prefix in taxonomy struct? keep argument but if NULL, use the one in struct?
{
	int     i;
	char* 	file_name;
	int 	file_descriptor;
	off_t 	file_size;
	char*   taxonomy_path;
	int32_t	name_length;
	int32_t	class_length;
	int32_t record_size;

	// Compute file size
	file_size = sizeof(int32_t);	// To store record count
	for (i=0; i < (tax->names)->count; i++)
	{
		file_size = file_size + sizeof(int32_t) * 5;						// To store record size, taxid, rank index, parent index, and name length
		file_size = file_size + strlen(tax->names->names[i].name);			// To store name
		file_size = file_size + strlen(tax->names->names[i].class_name);	// To store name
	}

	// Build the taxonomy directory path
	taxonomy_path = get_taxonomy_path(dms, taxonomy_name);

	file_name =	(char*) malloc((strlen(taxonomy_path) + strlen(taxonomy_name) + 5)*sizeof(char));
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating the memory for a binary taxonomy file name");
		return -1;
	}

	// Build the file path
	if (sprintf(file_name, "%s/%s.ndx", taxonomy_path, taxonomy_name) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError building a binary taxonomy file name");
		return -1;
	}

	free(taxonomy_path);

	// Create file
	file_descriptor = open(file_name, O_RDWR | O_CREAT | O_EXCL, 0777);
	if (file_descriptor < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError creating a binary taxonomy file");
		free(file_name);
		return -1;
	}

	free(file_name);

	// Truncate the file to the right size
	if (ftruncate(file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError truncating a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write record count
	if (write(file_descriptor, &(tax->names->count), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError writing in a binary taxonomy file");
		close(file_descriptor);
		return -1;
	}

	// Write records
	for (i=0; i < tax->names->count; i++)
	{
		name_length = strlen(tax->names->names[i].name);
		class_length = strlen(tax->names->names[i].class_name);
		record_size = 4*sizeof(int32_t) + name_length + class_length;

		// Write record size
		if (write(file_descriptor, &record_size, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write if the name is a scientific name
		if (write(file_descriptor, &(tax->names->names[i].is_scientific_name), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name length
		if (write(file_descriptor, &name_length, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write class length
		if (write(file_descriptor, &class_length, sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write taxid index
		if (write(file_descriptor, &(tax->names->names[i].taxon->idx), sizeof(int32_t)) < ((ssize_t) sizeof(int32_t)))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write name
		if (write(file_descriptor, tax->names->names[i].name, name_length) < ((ssize_t) name_length))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
		// Write class
		if (write(file_descriptor, tax->names->names[i].class_name, class_length) < ((ssize_t) class_length))
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError writing in a binary taxonomy file");
			close(file_descriptor);
			return -1;
		}
	}

	// Close file
	if (close(file_descriptor) < 0)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError closing a DMS information file");
		return -1;
	}

	return 0;
}



int obi_write_taxonomy(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* tax_name)
{
	char* taxonomy_path;

	// Build the taxonomy directory path
	taxonomy_path = get_taxonomy_path(dms, tax_name);
	if (taxonomy_path == NULL)
		return -1;

	// Try to create the directory
	if (mkdir(taxonomy_path, 00777) < 0)
	{
		if (errno == EEXIST)
			obidebug(1, "\nA taxonomy already exists with this name.");
		obidebug(1, "\nProblem creating a new taxonomy directory");
		free(taxonomy_path);
		return -1;
	}

	free(taxonomy_path);

    if (write_rankidx(dms, tax, tax_name) < 0)
         return -1;
     if (write_taxonomyidx(dms, tax, tax_name) < 0)
         return -1;
     if (write_nameidx(dms, tax, tax_name) < 0)
         return -1;
     // Check if there are local taxa (if so last taxon is local)
     if ((tax->taxa)->local_count > 0)
    	 if (write_local_taxonomy_idx(dms, tax, tax_name) < 0)
    		 return -1;

	return 0;
}


OBIDMS_taxonomy_p obi_read_taxdump(const char* taxdump)
{
	OBIDMS_taxonomy_p tax;
	struct dirent*    dp;
	DIR*              tax_dir;
	FILE*             file;
	bool 			  nodes_found=false;
	bool			  names_found=false;
	char 			  line[2048];			// TODO large enough?
	char*			  elt;
	char*			  file_name;
	int				  buffer_size;
	int			      i, j;
	int				  n;
	char**			  rank_names;
	int*			  parent_taxids;
	int				  taxid;
	bool			  already_in;
	ecotx_t*		  t;

	// Initialize taxonomy structure
	tax = (OBIDMS_taxonomy_p) malloc(sizeof(OBIDMS_taxonomy_t));
	if (tax == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating the memory for a taxonomy structure");
		return NULL;
	}
	tax->ranks = NULL;
	tax->taxa  = NULL;
	tax->names = NULL;

	tax->dms = NULL;
	(tax->tax_name)[0] = '\0';

	// TODO check if taxdump path is for a gz file to unzip or a directory

	tax_dir = opendir(taxdump);
	if (tax_dir == NULL)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nProblem opening a taxdump directory");
		free(tax);
		return NULL;
	}

	// Go through taxonomy files
	while ((dp = readdir(tax_dir)) != NULL)
	{
		if (strcmp(dp->d_name, "nodes.dmp") == 0)
		{
			nodes_found = true;
			buffer_size = 10000;

			// Initializing the taxa structure
			tax->taxa = (ecotxidx_t*) malloc(sizeof(ecotxidx_t) + sizeof(ecotx_t) * buffer_size);
			if (tax->taxa == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a taxonomy structure");
				free(tax);
				closedir(tax_dir);
				return NULL;
			}

			// Initialize rank names and parent taxids arrays
			parent_taxids = malloc(buffer_size * sizeof(int));
			if (file_name == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a file name");
				obi_close_taxonomy(tax);
				closedir(tax_dir);
				return NULL;
			}

			rank_names = malloc(buffer_size * sizeof(char*));
			if (file_name == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a file name");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				closedir(tax_dir);
				return NULL;
			}

			// Allocating the memory for the file name
			file_name =	(char*) malloc((strlen(taxdump) + 10)*sizeof(char));
			if (file_name == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a file name");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				free(rank_names);
				closedir(tax_dir);
				return NULL;
			}

			// Build the file path
			if (sprintf(file_name, "%s/nodes.dmp", taxdump) < 0)
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nError building a taxonomy file name");
				obi_close_taxonomy(tax);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				free(file_name);
				return NULL;
			}

			file = fopen(file_name, "r");
			if (file == NULL)
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nProblem opening a taxonomy file");
				obi_close_taxonomy(tax);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				free(file_name);
				return NULL;
			}

			free(file_name);

			(tax->taxa)->max_taxid = 0;
			n = 0;
			while (fgets(line, sizeof(line), file))
			{
				// Enlarge structures if needed
				if (n == buffer_size)
				{
					buffer_size = buffer_size * 2;

					tax->taxa = (ecotxidx_t*) realloc(tax->taxa, sizeof(ecotxidx_t) + sizeof(ecotx_t) * buffer_size);
					if (tax->taxa == NULL)
					{
						obi_set_errno(OBI_MALLOC_ERROR);
						obidebug(1, "\nError reallocating memory for a taxonomy structure");
						obi_close_taxonomy(tax);
						fclose(file);
						closedir(tax_dir);
						free(parent_taxids);
						free(rank_names);
						return NULL;
					}

					parent_taxids = (int*) realloc(parent_taxids, sizeof(int) * buffer_size);
					if (parent_taxids == NULL)
					{
						obi_set_errno(OBI_MALLOC_ERROR);
						obidebug(1, "\nError reallocating memory for a taxonomy structure");
						obi_close_taxonomy(tax);
						fclose(file);
						closedir(tax_dir);
						free(parent_taxids);
						free(rank_names);
						return NULL;
					}

					rank_names = (char**) realloc(rank_names, sizeof(char*) * buffer_size);
					if (rank_names == NULL)
					{
						obi_set_errno(OBI_MALLOC_ERROR);
						obidebug(1, "\nError reallocating memory for a taxonomy structure");
						obi_close_taxonomy(tax);
						fclose(file);
						closedir(tax_dir);
						free(parent_taxids);
						free(rank_names);
						return NULL;
					}
				}

				// Check for terminal '\n' character (line complete)
				if (line[strlen(line) - 1] != '\n')
				{
					obi_set_errno(OBI_TAXONOMY_ERROR);
					obidebug(1, "\nError: line buffer size not large enough for line in taxonomy file");
					obi_close_taxonomy(tax);
					fclose(file);
					closedir(tax_dir);
					free(parent_taxids);
					free(rank_names);
					return NULL;
				}

				(tax->taxa)->taxon[n].idx = n;

				// Parse 3 first elements separated by '|'

				elt = strtok(line, "|");

				// Remove the last character (tab character)
				elt[strlen(elt)-1] = '\0';

				// First element: taxid
				(tax->taxa)->taxon[n].taxid = atoi(elt);

				// Update max taxid
				if ((tax->taxa)->taxon[n].taxid > (tax->taxa)->max_taxid)
					(tax->taxa)->max_taxid = (tax->taxa)->taxon[n].taxid;

				// Initialize farest taxid value
				(tax->taxa)->taxon[n].farest = -1;

				i = 1;
				while (i < 3)
				{
					elt = strtok(NULL, "|");

					// Remove the first and the last characters (tab characters)
					elt = elt+1;
					elt[strlen(elt)-1] = '\0';

					if (i == 1)
						parent_taxids[n] = atoi(elt);
					else if (i == 2)
					{
						rank_names[n] = (char*) malloc((strlen(elt)+1) * sizeof(char));
						if (rank_names[n] == NULL)
						{
							obi_set_errno(OBI_MALLOC_ERROR);
							obidebug(1, "\nError allocating memory for taxon rank name");
							obi_close_taxonomy(tax);
							fclose(file);
							closedir(tax_dir);
							free(parent_taxids);
							free(rank_names);
							return NULL;
						}
						strcpy(rank_names[n], elt);
					}
					i++;
				}
				n++;
			}

			// Check that fgets stopped because it reached EOF
			if (!feof(file))
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nError: file reading was stopped before end of file");
				obi_close_taxonomy(tax);
				fclose(file);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}

			// Store count
			(tax->taxa)->count = n;
			(tax->taxa)->ncbi_count = n;
			(tax->taxa)->local_count = 0;

			// Truncate the structure memory to the right size
			tax->taxa = (ecotxidx_t*) realloc(tax->taxa, sizeof(ecotxidx_t) + sizeof(ecotx_t) * (tax->taxa)->count);
			if (tax->taxa == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError reallocating memory for taxonomy structure");
				obi_close_taxonomy(tax);
				fclose(file);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}

			fclose(file);
		}
	}
	closedir(tax_dir);


	// Go through directory again for next file
	tax_dir = opendir(taxdump);
	if (tax_dir == NULL)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nProblem opening a taxdump directory");
		obi_close_taxonomy(tax);
		free(parent_taxids);
		free(rank_names);
		return NULL;
	}

	// Go through taxonomy files
	while ((dp = readdir(tax_dir)) != NULL)
	{
		if (strcmp(dp->d_name, "names.dmp") == 0)
		{
			names_found = true;
			buffer_size = 10000;

			// Initializing the names structure
			tax->names = (econameidx_t*) malloc(sizeof(econameidx_t) + sizeof(econame_t) * buffer_size);
			if (tax->names == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a taxonomy structure");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				free(rank_names);
				closedir(tax_dir);
				return NULL;
			}

			// Allocating the memory for the file name
			file_name =	(char*) malloc((strlen(taxdump) + 10)*sizeof(char));
			if (file_name == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating the memory for a file name");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				free(rank_names);
				closedir(tax_dir);
				return NULL;
			}

			// Build the file path
			if (sprintf(file_name, "%s/names.dmp", taxdump) < 0)
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nError building a taxonomy file name");
				obi_close_taxonomy(tax);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				free(file_name);
				return NULL;
			}

			file = fopen(file_name, "r");
			if (file == NULL)
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nProblem opening a taxonomy file");
				obi_close_taxonomy(tax);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				free(file_name);
				return NULL;
			}

			free(file_name);

			n = 0;
			j = 0;
			while (fgets(line, sizeof(line), file))
			{
				// Enlarge structures if needed
				if (n == buffer_size)
				{
					buffer_size = buffer_size * 2;
					tax->names = (econameidx_t*) realloc(tax->names, sizeof(econameidx_t) + sizeof(econame_t) * buffer_size);
					if (tax->names == NULL)
					{
						obi_set_errno(OBI_MALLOC_ERROR);
						obidebug(1, "\nError reallocating memory for a taxonomy structure");
						obi_close_taxonomy(tax);
						fclose(file);
						closedir(tax_dir);
						free(parent_taxids);
						free(rank_names);
						return NULL;
					}
				}

				// Check for terminal '\n' character (line complete)
				if (line[strlen(line) - 1] != '\n')
				{
					obi_set_errno(OBI_TAXONOMY_ERROR);
					obidebug(1, "\nError: line buffer size not large enough for line in taxonomy file");
					obi_close_taxonomy(tax);
					fclose(file);
					closedir(tax_dir);
					free(parent_taxids);
					free(rank_names);
					return NULL;
				}

				// Parse 4 first elements separated by '|'

				elt = strtok(line, "|");

				// Remove the last character (tab character)
				elt[strlen(elt)-1] = '\0';

				// First element: taxid
				taxid = atoi(elt);
				// Find taxid in taxa structure and store pointer in names structure
				i = j;
				while ((i < (tax->taxa)->count) && ((tax->taxa)->taxon[i].taxid != taxid))
					i++;
				if (i == (tax->taxa)->count)
				{
					obi_set_errno(OBI_TAXONOMY_ERROR);
					obidebug(1, "\nError: could not find taxon associated to name when reading taxdump");
					obi_close_taxonomy(tax);
					fclose(file);
					closedir(tax_dir);
					free(parent_taxids);
					free(rank_names);
					return NULL;
				}
				j = i;	// Because there are several names by taxon but they are in the same order
				(tax->names)->names[n].taxon = ((tax->taxa)->taxon)+i;

				i = 1;
				while (i < 4)
				{
					elt = strtok(NULL, "|");

					// Remove the first and the last characters (tab characters)
					elt = elt+1;
					elt[strlen(elt)-1] = '\0';

					if (i == 1)		// Name
					{
						(tax->names)->names[n].name = (char*) malloc((strlen(elt) + 1) * sizeof(char));
						if ((tax->names)->names[n].name == NULL)
						{
							obi_set_errno(OBI_MALLOC_ERROR);
							obidebug(1, "\nError allocating memory for a taxon name");
							obi_close_taxonomy(tax);
							fclose(file);
							closedir(tax_dir);
							free(parent_taxids);
							free(rank_names);
							return NULL;
						}
						strcpy((tax->names)->names[n].name, elt);
					}
					else if (i == 3)	// Class name
					{
						(tax->names)->names[n].class_name = (char*) malloc((strlen(elt) + 1) * sizeof(char));
						if ((tax->names)->names[n].class_name == NULL)
						{
							obi_set_errno(OBI_MALLOC_ERROR);
							obidebug(1, "\nError allocating memory for a taxon class name");
							obi_close_taxonomy(tax);
							fclose(file);
							closedir(tax_dir);
							free(parent_taxids);
							free(rank_names);
							return NULL;
						}
						strcpy((tax->names)->names[n].class_name, elt);
						if (strcmp(elt, "scientific name") == 0)
						{
							(tax->names)->names[n].is_scientific_name = 1;
						}
						else
							(tax->names)->names[n].is_scientific_name = 0;
					}
					i++;
				}
				n++;
			}

			// Check that fgets stopped because it reached EOF
			if (!feof(file))
			{
				obi_set_errno(OBI_TAXONOMY_ERROR);
				obidebug(1, "\nError: file reading was stopped before end of file");
				obi_close_taxonomy(tax);
				fclose(file);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}

			// Store count
			(tax->names)->count = n;

			// Truncate the structure memory to the right size
			tax->names = (econameidx_t*) realloc(tax->names, sizeof(econameidx_t) + sizeof(econame_t) * (tax->names)->count);
			if (tax->names == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError reallocating memory for a a taxonomy structure");
				obi_close_taxonomy(tax);
				fclose(file);
				closedir(tax_dir);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}

			fclose(file);
		}
	}
	closedir(tax_dir);

	if (!nodes_found)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nProblem reading taxdump: nodes.dmp file not found");
		obi_close_taxonomy(tax);
		free(parent_taxids);
		free(rank_names);
		return NULL;
	}
	if (!names_found)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nProblem reading taxdump: names.dmp file not found");
		obi_close_taxonomy(tax);
		free(parent_taxids);
		free(rank_names);
		return NULL;
	}

	// Go through data to fill the taxonomy structure

	// Build rank list

	// Initialize rank structure
	buffer_size = 10;
	tax->ranks = (ecorankidx_t*) malloc(sizeof(ecorankidx_t) + sizeof(char*) * buffer_size);
	if (tax->ranks == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for taxon rank array");
		obi_close_taxonomy(tax);
		free(parent_taxids);
		free(rank_names);
		return NULL;
	}
	(tax->ranks)->count = 0;
	for (i=0; i < (tax->taxa)->count; i++)
	{
		already_in = false;
		for (j=0; j < (tax->ranks)->count; j++)
		{
			if (strcmp(rank_names[i], ((tax->ranks)->label)[j]) == 0)
			{
				already_in = true;
				break;
			}
		}
		if (!already_in)
		{
			// Realloc rank structure if needed
			if ((tax->ranks)->count == buffer_size)
			{
				buffer_size = buffer_size + 10;
				tax->ranks = (ecorankidx_t*) realloc(tax->ranks, sizeof(ecorankidx_t) + sizeof(char*) * buffer_size);
				if (tax->ranks == NULL)
				{
					obi_set_errno(OBI_MALLOC_ERROR);
					obidebug(1, "\nError reallocating memory for taxon ranks");
					obi_close_taxonomy(tax);
					free(parent_taxids);
					free(rank_names);
					return NULL;
				}
			}

			// Store new rank
			((tax->ranks)->label)[(tax->ranks)->count] = (char*) malloc((strlen(rank_names[i]) + 1) * sizeof(char));
			if (((tax->ranks)->label)[(tax->ranks)->count] == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError allocating memory for taxon rank names");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}
			strcpy(((tax->ranks)->label)[(tax->ranks)->count], rank_names[i]);
			((tax->ranks)->count)++;
		}
	}

	// Truncate to the number of ranks recorded
	tax->ranks = (ecorankidx_t*) realloc(tax->ranks, sizeof(ecorankidx_t) + sizeof(char*) * (tax->ranks)->count);
	if (tax->ranks == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError reallocating memory for taxon ranks");
		obi_close_taxonomy(tax);
		free(parent_taxids);
		free(rank_names);
		return NULL;
	}

	// Sort in alphabetical order
	qsort((tax->ranks)->label, (tax->ranks)->count, sizeof(char*), cmp_str);

	// Associate the taxa with their rank indices
	for (i=0; i < (tax->taxa)->count; i++)
	{
		for (j=0; j < (tax->ranks)->count; j++)
		{
			if (strcmp(rank_names[i], ((tax->ranks)->label)[j]) == 0)
			{
				((tax->taxa)->taxon)[i].rank = j;
				break;
			}
		}
	}

	// Associate the taxa with their scientific name
	for (i=0; i < (tax->names)->count; i++)
	{
		if ((tax->names)->names[i].is_scientific_name)
		{
			((tax->names)->names[i].taxon)->name = (char*) malloc((strlen((((tax->names)->names)[i]).name) + 1) * sizeof(char));
			if (((tax->names)->names[i].taxon)->name == NULL)
			{
				obi_set_errno(OBI_MALLOC_ERROR);
				obidebug(1, "\nError reallocating memory for taxon ranks");
				obi_close_taxonomy(tax);
				free(parent_taxids);
				free(rank_names);
				return NULL;
			}
			strcpy(((tax->names)->names[i].taxon)->name, (((tax->names)->names)[i]).name);
		}
	}

	// Sort names in alphabetical order
	qsort((tax->names)->names, (tax->names)->count, sizeof(econame_t), cmp_names);

	// Associate the taxa with their parent
	for (i=0; i < (tax->taxa)->count; i++)
	{
		((tax->taxa)->taxon)[i].parent = obi_taxo_get_taxon_with_taxid(tax, parent_taxids[i]);
		if (((tax->taxa)->taxon)[i].parent == NULL)
		{
			obi_set_errno(OBI_TAXONOMY_ERROR);
			obidebug(1, "\nError: taxon parent not found");
			obi_close_taxonomy(tax);
			free(parent_taxids);
			free(rank_names);
			return NULL;
		}
		(((tax->taxa)->taxon)[i].parent)->farest = 0;
	}

	(tax->taxa)->buffer_size = (tax->taxa)->count;

	// Compute longest branches    TODO what is this for???
	for (i=0; i < (tax->taxa)->count; i++)
	{
		t = (((tax->taxa))->taxon)+i;
		if (t->farest == -1)
		{
			t->farest=0;
            while (t->parent != t)
            {
            	j = t->farest + 1;
            	if (j > t->parent->farest)
            	{
            		t->parent->farest = j;
            		t=t->parent;
            	}
            	else
            		t = (tax->taxa)->taxon;
            }
		}
	}

	// Freeing
	free(parent_taxids);
	for (i=0; i < (tax->taxa)->count; i++)
		free(rank_names[i]);
	free(rank_names);

	return tax;
}


int obi_taxonomy_add_local_taxon(OBIDMS_taxonomy_p tax, const char* name, const char* rank_name, int32_t parent_taxid, int32_t min_taxid)
{
	int32_t    taxid;
	ecotx_t*   taxon;
	econame_t* name_struct;
	int       i;

	// Enlarge the structure memory for a new taxon
	tax->taxa = (ecotxidx_t*) realloc(tax->taxa, sizeof(ecotxidx_t) + sizeof(ecotx_t) * (((tax->taxa)->count) + 1));
	if (tax->taxa == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError reallocating memory for a taxonomy structure to add a new taxon");
		return -1;
	}

	// Compute new taxid that must be equal or greater than 1E7 and greater than the maximum taxid existing in the taxonomy
	if (min_taxid < MIN_LOCAL_TAXID)
		min_taxid = MIN_LOCAL_TAXID;
	if (min_taxid > (tax->taxa)->max_taxid)
		taxid = min_taxid;
	else
		taxid = ((tax->taxa)->max_taxid) + 1;

	// Fill the ecotx_t node structure
	taxon = ((tax->taxa)->taxon)+((tax->taxa)->count);
	taxon->taxid = taxid;
	taxon->idx = (tax->taxa)->count;
	taxon->local = true;
	taxon->name = (char*) malloc((strlen(name) + 1) * sizeof(char));
	if (taxon->name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for a taxon name to add a new taxon");
		return -1;
	}
	strcpy(taxon->name, name);
	taxon->rank = -1;
	for (i=0; i < (tax->ranks)->count; i++)
	{
		if (strcmp(rank_name, ((tax->ranks)->label)[i]) == 0)
		{
			taxon->rank = i;
			break;
		}
	}
	if (taxon->rank == -1)	// TODO Discuss possibility of creating rank if doesn't exist
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError: taxon rank not found when adding a new taxon");
		return -1;
	}
	taxon->parent = obi_taxo_get_taxon_with_taxid(tax, parent_taxid);
	if (taxon->parent == NULL)
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError: taxon parent not found when adding a new taxon");
		return -1;
	}
	taxon->farest = 0;	// TODO not sure

	// Update taxonomy counts etc
	(tax->taxa)->max_taxid = taxid;
	((tax->taxa)->count)++;
	((tax->taxa)->local_count)++;
	(tax->taxa)->buffer_size = (tax->taxa)->count;

	// Add new name in names structure		// TODO discuss because in OBITools1 the new names were not written in .ndx
	// Allocate memory for new name
	tax->names = (econameidx_t*) realloc(tax->names, sizeof(econameidx_t) + sizeof(econame_t) * ((tax->names)->count + 1));
	if (tax->names == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError reallocating memory for a taxonomy structure to add a new taxon");
		return -1;
	}

	// Add new name
	name_struct = (tax->names)->names + ((tax->names)->count);
	name_struct->name = (char*) malloc((strlen(name) + 1) * sizeof(char));
	if (name_struct->name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for a taxon name to add a new taxon");
		return -1;
	}
	strcpy(name_struct->name, name);
	name_struct->class_name = (char*) malloc((strlen("scientific name") + 1) * sizeof(char));
	if (name_struct->class_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for a taxon class name to add a new taxon");
		return -1;
	}
	strcpy(name_struct->class_name, "scientific name");
	name_struct->is_scientific_name = true;
	name_struct->taxon = ((tax->taxa)->taxon) + ((tax->taxa)->count) - 1;

	// Sort names in alphabetical order
	qsort((tax->names)->names, (tax->names)->count, sizeof(econame_t), cmp_names);

	// Update name count
	((tax->names)->count)++;

	return taxid;
}


/////// PUBLIC /////////


OBIDMS_taxonomy_p obi_read_taxonomy(OBIDMS_p dms, const char* taxonomy_name, bool read_alternative_names)
{
	OBIDMS_taxonomy_p  tax;
	char*			   taxonomy_path;
	char*  			   ranks_file_name;
	char*              taxa_file_name;
	char*			   local_taxa_file_name;
	char*			   alter_names_file_name;
	int                buffer_size;

	tax = (OBIDMS_taxonomy_p) malloc(sizeof(OBIDMS_taxonomy_t));
	if (tax == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for a taxonomy structure");
 		return NULL;
	}

	tax->ranks = NULL;
	tax->taxa  = NULL;
	tax->names = NULL;

	tax->dms = dms;

	strcpy(tax->tax_name, taxonomy_name);

	buffer_size = 2048;

	taxonomy_path = get_taxonomy_path(dms, taxonomy_name);
	if (taxonomy_path == NULL)
		return NULL;

	// Read ranks
	ranks_file_name = (char*) malloc(buffer_size*sizeof(char));
	if (ranks_file_name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for ranks file name");
		free(taxonomy_path);
		free(tax);
		return NULL;
	}
	if (snprintf(ranks_file_name, buffer_size, "%s/%s.rdx", taxonomy_path, taxonomy_name) < 0)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError building ranks file name");
		free(taxonomy_path);
		free(ranks_file_name);
		free(tax);
		return NULL;
	}
	tax->ranks = read_rankidx(ranks_file_name);
	if (tax->ranks == NULL)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError building ranks file name");
		free(taxonomy_path);
		free(ranks_file_name);
		free(tax);
		return NULL;
	}
	free(ranks_file_name);

	// Read taxa
	taxa_file_name = (char*) malloc(buffer_size*sizeof(char));
	if (taxa_file_name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for taxa file name");
		free(taxonomy_path);
		obi_close_taxonomy(tax);
		return NULL;
	}
	if (snprintf(taxa_file_name, buffer_size, "%s/%s.tdx", taxonomy_path, taxonomy_name) < 0)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError building taxa file name");
		free(taxonomy_path);
		free(taxa_file_name);
		obi_close_taxonomy(tax);
		return NULL;
	}
	local_taxa_file_name = (char*) malloc(buffer_size*sizeof(char));
	if (local_taxa_file_name == NULL)
	{
 		obi_set_errno(OBI_MALLOC_ERROR);
 		obidebug(1, "\nError allocating memory for taxa file name");
		free(taxonomy_path);
		free(taxa_file_name);
		obi_close_taxonomy(tax);
		return NULL;
	}
	if (snprintf(local_taxa_file_name, buffer_size, "%s/%s.ldx", taxonomy_path, taxonomy_name) < 0)
	{
 		obi_set_errno(OBI_TAXONOMY_ERROR);
 		obidebug(1, "\nError building local taxa file name");
		free(taxonomy_path);
		free(taxa_file_name);
		free(local_taxa_file_name);
		obi_close_taxonomy(tax);
		return NULL;
	}
	tax->taxa = read_taxonomyidx(taxa_file_name, local_taxa_file_name);
	if (tax->taxa == NULL)
	{
		free(taxonomy_path);
		free(taxa_file_name);
		free(local_taxa_file_name);
		obi_close_taxonomy(tax);
		return NULL;
	}
	free(taxa_file_name);
	free(local_taxa_file_name);

	// Read alternative names
	if (read_alternative_names)
	{
		alter_names_file_name = (char*) malloc(buffer_size*sizeof(char));
		if (alter_names_file_name == NULL)
		{
	 		obi_set_errno(OBI_MALLOC_ERROR);
	 		obidebug(1, "\nError allocating memory for alternative names file name");
			free(taxonomy_path);
			obi_close_taxonomy(tax);
			return NULL;
		}
		if (snprintf(alter_names_file_name, buffer_size, "%s/%s.ndx", taxonomy_path, taxonomy_name) < 0)
		{
	 		obi_set_errno(OBI_TAXONOMY_ERROR);
	 		obidebug(1, "\nError building alternative names file name");
			free(taxonomy_path);
			free(alter_names_file_name);
			obi_close_taxonomy(tax);
			return NULL;
		}
        tax->names = read_nameidx(alter_names_file_name, tax);
    	if (tax->names == NULL)
    	{
			free(taxonomy_path);
			free(alter_names_file_name);
			obi_close_taxonomy(tax);
			return NULL;
    	}
        free(alter_names_file_name);
	}

	free(taxonomy_path);

	return tax;
}


int obi_close_taxonomy(OBIDMS_taxonomy_p taxonomy)
{
	int i;

	// Update local informations (local taxa and preferred names) if there are any
	if ((taxonomy->taxa)->local_count > 0)
	{
		if (taxonomy->dms == NULL)
		{
	 		obi_set_errno(OBI_TAXONOMY_ERROR);
	 		obidebug(1, "\nError closing a taxonomy with local files but no DMS associated (probably read directly from taxdump)");		// TODO discuss
		}
		if (write_local_taxonomy_idx(taxonomy->dms, taxonomy, taxonomy->tax_name) < 0)
			return -1;
	}

	if (taxonomy)
	{
		if (taxonomy->ranks)
		{
			for (i=0; i < (taxonomy->ranks)->count; i++)
			{
				if ((taxonomy->ranks)->label[i])
					free((taxonomy->ranks)->label[i]);
			}
			free(taxonomy->ranks);
		}

		if (taxonomy->names)
		{
			for (i=0; i < (taxonomy->names)->count; i++)
			{
				if (((taxonomy->names)->names[i]).name)
					free(((taxonomy->names)->names[i]).name);
				if (((taxonomy->names)->names[i]).class_name)
					free(((taxonomy->names)->names[i]).class_name);
			}
			free(taxonomy->names);
		}

		if (taxonomy->taxa)
		{
			for (i=0; i < (taxonomy->taxa)->count; i++)
			{
				if (((taxonomy->taxa)->taxon[i]).name)
					free(((taxonomy->taxa)->taxon[i]).name);
			}
			free(taxonomy->taxa);
		}

		free(taxonomy);
	}

	return 0;
}


//////////////////////////////////////////////////////////////////////////


ecotx_t* obi_taxo_get_parent_at_rank(ecotx_t* taxon, int32_t rankidx)
{
	ecotx_t* current_taxon;
	ecotx_t* next_taxon;

	current_taxon = taxon;
	next_taxon    = current_taxon->parent;

	while ((current_taxon != next_taxon) &&  // root node
		   (current_taxon->rank != rankidx))
	{
		current_taxon = next_taxon;
		next_taxon    = current_taxon->parent;
	}

	if (current_taxon->rank == rankidx)
		return current_taxon;
	else
		return NULL;
}


ecotx_t* obi_taxo_get_taxon_with_taxid(OBIDMS_taxonomy_p taxonomy, int32_t taxid)
{
	ecotx_t    *current_taxon;
	int32_t     count;

	count = (taxonomy->taxa)->count;

	current_taxon = (ecotx_t*) bsearch((const void *) ((size_t) taxid),
                                       (const void *) taxonomy->taxa->taxon,
                                       count,
                                       sizeof(ecotx_t),
                                       cmp_taxids);
	return current_taxon;
}


bool obi_taxo_is_taxon_under_taxid(ecotx_t* taxon, int32_t other_taxid)
{
	ecotx_t* next_parent;

	next_parent = taxon->parent;

	while ((other_taxid != next_parent->taxid) && (strcmp(next_parent->name, "root")))
		next_parent = next_parent->parent;

	if (other_taxid == next_parent->taxid)
		return 1;
	else
		return 0;
}


ecotx_t* obi_taxo_get_species(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
{
	static OBIDMS_taxonomy_p tax = NULL;
	static int32_t		  rankindex = -1;

	if (taxonomy && (tax != taxonomy))
	{
		rankindex = rank_index("species", taxonomy->ranks);
		tax = taxonomy;
	}

	if (!tax || (rankindex < 0))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError trying to get the species associated with a taxon: No taxonomy defined");
		return NULL;
	}

	return obi_taxo_get_parent_at_rank(taxon, rankindex);
}


ecotx_t* obi_taxo_get_genus(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
{
	static OBIDMS_taxonomy_p tax = NULL;
	static int32_t		  rankindex = -1;

	if (taxonomy && (tax != taxonomy))
	{
		rankindex = rank_index("genus", taxonomy->ranks);
		tax = taxonomy;
	}

	if (!tax || (rankindex < 0))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError trying to get the genus associated with a taxon: No taxonomy defined");
		return NULL;
	}

	return obi_taxo_get_parent_at_rank(taxon, rankindex);
}


ecotx_t* obi_taxo_get_family(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
{
	static OBIDMS_taxonomy_p tax = NULL;
	static int32_t		  rankindex = -1;

	if (taxonomy && (tax != taxonomy))
	{
		rankindex = rank_index("family", taxonomy->ranks);
		tax = taxonomy;
	}

	if (!tax || (rankindex < 0))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError trying to get the family associated with a taxon: No taxonomy defined");
		return NULL;
	}

	return obi_taxo_get_parent_at_rank(taxon, rankindex);
}


ecotx_t* obi_taxo_get_kingdom(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
{
	static OBIDMS_taxonomy_p tax = NULL;
	static int32_t		  rankindex = -1;

	if (taxonomy && (tax != taxonomy))
	{
		rankindex = rank_index("kingdom", taxonomy->ranks);
		tax = taxonomy;
	}

	if (!tax || (rankindex < 0))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError trying to get the kingdom associated with a taxon: No taxonomy defined");
		return NULL;
	}

	return obi_taxo_get_parent_at_rank(taxon, rankindex);
}


ecotx_t* obi_taxo_get_superkingdom(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
{
	static OBIDMS_taxonomy_p tax = NULL;
	static int32_t		  rankindex = -1;

	if (taxonomy && (tax != taxonomy))
	{
		rankindex = rank_index("superkingdom", taxonomy->ranks);
		tax = taxonomy;
	}

	if (!tax || (rankindex < 0))
	{
		obi_set_errno(OBI_TAXONOMY_ERROR);
		obidebug(1, "\nError trying to get the superkingdom associated with a taxon: No taxonomy defined");
		return NULL;
	}

	return obi_taxo_get_parent_at_rank(taxon, rankindex);
}