From f64bb169092c986b60bb48ed21d1842940ca84c5 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Fri, 1 Jun 2007 15:08:15 +0000
Subject: [PATCH] git-svn-id:
 https://www.grenoble.prabi.fr/svn/LECASofts/ecoPCR/branches/refactoring@22
 60f365c0-8329-0410-b2a4-ec073aeeaa1d

---
 src/ecofind.c | 185 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 185 insertions(+)
 create mode 100644 src/ecofind.c

diff --git a/src/ecofind.c b/src/ecofind.c
new file mode 100644
index 0000000..ffc443a
--- /dev/null
+++ b/src/ecofind.c
@@ -0,0 +1,185 @@
+#include "libecoPCR/ecoPCR.h"
+#include <regex.h>
+#include <string.h>
+#include <stdlib.h>
+#include <getopt.h>
+#include <stdio.h>
+#define VERSION "0.1"
+
+
+/* ----------------------------------------------- */
+/* structures				                       */
+/* ----------------------------------------------- */
+typedef struct  {
+	char		*sci_name;	// scientific name
+	int32_t		tax_id;		// tax id use ecoPCR prefilter
+	int32_t		tax_rank;	// taxonomic rank  
+} tax_info ;
+
+
+
+/* ----------------------------------------------- */
+/* functions				                       */
+/* ----------------------------------------------- */
+
+/**
+ * display the result in sdtout
+ **/
+static void printresult(tax_info **taxonlist,int *j){
+	int i;
+	printf("taxonomy rank \t|\t taxonomy id \t|\t  scientific name\n\n");	
+	for(i=0;i<*j;i++){
+		printf("%d \t|\t %d \t|\t %s\n",
+			taxonlist[i]->tax_id,
+			taxonlist[i]->tax_rank,
+			taxonlist[i]->sci_name);	
+	}
+}
+
+/**
+ * printout usage and exit                     
+ **/
+#define PP fprintf(stderr, 
+
+static void ExitUsage(stat)
+        int stat;
+{
+        PP      "usage: ecofind [-d datafile] [-t taxon] [-h]\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      "-t : [T]axonomic 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			numpattern 	= 0;	/* stores number of pattern */
+	int32_t			nummatch 	= 0;
+	int32_t 		k,j 		= 0;
+	int32_t     	errflag 	= 0;
+	char			*prefix		= NULL;
+	char			*taxon 		= NULL;
+	char 			pattern[PATTERN_NUMBER][PATTERN_LENGHT];					/* stores the pattern 		*/
+	tax_info		**taxonlist = ECOMALLOC(sizeof(tax_info) * RESULT_LENGTH,	/* stores the result		*/
+										"error in **taxonlist allocation");		
+	ecotaxonomy_t 	*taxonomy;													/* stores all the taxonomy 	*/	
+	
+	int				re_error;
+	int				re_match;
+	regex_t			re_preg;
+
+	while ((carg = getopt(argc, argv, "hd:t:")) != -1) {
+     switch (carg) {                        
+        case 'd':               	/* path to the database     */
+           prefix = optarg;
+           break;
+
+        case 't':               	/* pattern list			    */
+           strncpy(pattern[numpattern],optarg,PATTERN_LENGHT);
+           numpattern++;
+           break;
+
+        case 'h':               	/* display help	           	*/      
+           PrintHelp();
+           exit(0);
+           break;
+           
+        case '?':               	/* bad option   	        */
+            errflag++;
+     }
+	}
+
+	/**
+	 * check data integrity
+	 * and exit if needed
+	 **/		
+	if (!prefix || !pattern)
+    	errflag++;
+	
+	if (taxonlist == NULL)
+		errflag++;
+		
+	if (errflag)
+		ExitUsage(errflag);
+		
+	/**
+	 * load taxonomy using libecoPCR functions
+	 **/
+	printf("-- opening %s database --\n",prefix);
+	
+	taxonomy = read_taxonomy(prefix);
+	
+	printf("-- %d taxons --\n", taxonomy->taxons->count);
+	
+	/**
+	 * parse taxonomy
+	 **/
+	for (k=0;k<numpattern;k++)
+	{
+		j=0;
+		nummatch=0;
+		printf("\n-- searching for %s pattern --\n",pattern[k]); 
+		do {
+			taxon = taxonomy->taxons->taxon[j].name;
+			re_error = regcomp (&re_preg, pattern[k], REG_NOSUB | REG_EXTENDED);
+	   		if (re_error == 0)
+	   		{
+	   			re_match = regexec (&re_preg, taxon, 0, NULL, 0);
+	   			regfree(&re_preg);
+	   			if (re_match == 0)
+	   			{	
+	   				taxonlist[nummatch] = ECOMALLOC(sizeof(taxonlist),
+	   											"error in taxon j allocation");
+	   				taxonlist[nummatch]->sci_name = taxon;
+	   				taxonlist[nummatch]->tax_id = taxonomy->taxons->taxon[j].taxid;
+	   				taxonlist[nummatch]->tax_rank = taxonomy->taxons->taxon[j].rank;
+	   				nummatch++;
+	   			}
+	   		}
+			j++;
+		} while(j < taxonomy->taxons->count);
+		
+		printf("-- %d records found --\n\n",nummatch);
+	
+		if (nummatch > 0)
+			printresult(taxonlist, &nummatch);
+	}		
+			
+	return 0;
+}
+
+