ecopcr/src/ecofind.c

#include "libecoPCR/ecoPCR.h"
#include <regex.h>
#include <string.h>
#include <stdlib.h>
#include <getopt.h>
#include <stdio.h>
#define VERSION "0.1"

/**
 * display the result in sdtout
 **/
 
static void printresult(ecotx_t *taxon,ecotaxonomy_t *taxonomy){
	char* rankname;
	
	rankname= taxonomy->ranks->label[taxon->rank];	
	
	printf("%10d \t| %15s \t|\t %s\n",
			taxon->taxid,
			rankname,
			taxon->name);	
}

/**
 * printout usage and exit                     
 **/
#define PP fprintf(stderr, 

static void ExitUsage(stat)
        int stat;
{
        PP      "usage: ecofind [-d datafile] [-h] <taxon name pattern> ... \n");
        PP      "type \"ecofind -h\" for help\n");
        if (stat)
            exit(stat);
}

#undef  PP

/**
 * printout help
 **/
#define PP fprintf(stdout, 

static void PrintHelp()
{
        PP      "------------------------------------------\n");
        PP      " ecofind Version %s\n", VERSION);
        PP      "------------------------------------------\n");
        PP      "synopsis : searching for scientific name, taxonomic\n");
        PP		"	rank and taxonomy id for given regular\n");
        PP		"	expression patterns\n");
        PP      "usage: ecofind [options]\n");
        PP      "------------------------------------------\n");
        PP      "options:\n");
        PP      "-d : [D]atafile containing the taxonomy\n");
        PP		"	to match the expected format, the database\n");
        PP		"	has to be formated first by the ecoPCRFormat.py\n");
        PP		"	program located in the tools directory.\n");
        PP		"	Write the datafile radical without any extension.\n\n");
        PP		"-h	: [H]elp - print <this> help\n\n");
        PP      "<taxon> name pattern bearing regular expressions\n\n");
        PP      "------------------------------------------\n");
}

/* ----------------------------------------------- */

#define PATTERN_NUMBER 	10
#define PATTERN_LENGHT 	40
#define	RESULT_LENGTH	100

int main(int argc, char **argv) 
{
	int32_t     	carg		= 0;
	int32_t			nummatch 	= 0;
	int32_t 		k,j 		= 0;
	int32_t     	errflag 	= 0;
	int32_t			tax_count	= 0;
	char			*prefix		= NULL;
	ecotx_t			*taxon 		= NULL;
	ecotaxonomy_t 	*taxonomy;	
	
	int				re_error;
	int				re_match;
	regex_t			re_preg;

	while ((carg = getopt(argc, argv, "hd:")) != -1) {
	    switch (carg) {                        
	       case 'd':               	/* path to the database     */
	          prefix = optarg;
	          break;
	
	       case 'h':               	/* display help	           	*/      
	          PrintHelp();
	          exit(0);
	          break;
	           
	       case '?':               	/* bad option   	        */
	          errflag++;
	     }
	}
	
    /**
	 * check the path to the database is given as last argument
	 **/
	if ((argc - optind) < 1)
        errflag++;
		
	if (errflag)
		ExitUsage(errflag);
		
	/**
	 * load taxonomy using libecoPCR functions
	 **/
	printf("# \n#  opening %s database\n",prefix);
	
	taxonomy = read_taxonomy(prefix);
	tax_count = taxonomy->taxons->count;
	
	printf("#  %d taxons\n", tax_count);
	
	/**
	 * parse taxonomy
	 **/
	for (k=optind;k<argc;k++)
	{
		printf("#\n#  searching for '%s' pattern\n",argv[k]); 
		
		re_error = regcomp (&re_preg, argv[k], REG_NOSUB | REG_EXTENDED | REG_ICASE);
		if (re_error)
		{
			fprintf(stderr,"#  misformed pattern '%s'\n",argv[k]);
			exit(1);
		}
		
		nummatch=0;
		
		printf("#  taxonomy id \t|\t taxonomy rank \t|\t  scientific name\n#\n");	
		
		for (j=0,taxon=taxonomy->taxons->taxon;
		     j < tax_count;
		     taxon++,j++)
		{
   			re_match = regexec (&re_preg, taxon->name, 0, NULL, 0);
   			if (re_match == 0)
   			{
   				printresult(taxon,taxonomy);
   				nummatch++;
   			}
		} 
		
		printf("#  %d records found \n",nummatch);
	   	regfree(&re_preg);
	}		
			
	return 0;
}