From c2645566291904dc4f9e2b074f0a4dee23445bf0 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Wed, 13 Jun 2007 09:56:09 +0000
Subject: [PATCH] git-svn-id:
 https://www.grenoble.prabi.fr/svn/LECASofts/ecoPCR/branches/refactoring@61
 60f365c0-8329-0410-b2a4-ec073aeeaa1d

---
 src/ecogrep.c | 80 ++++++++++++++++++++++++++++++++-------------------
 1 file changed, 51 insertions(+), 29 deletions(-)

diff --git a/src/ecogrep.c b/src/ecogrep.c
index 357f560..e4d3f57 100644
--- a/src/ecogrep.c
+++ b/src/ecogrep.c
@@ -15,6 +15,15 @@ void freememory(char **tab, int32_t num){
 	}
 }
 
+/**
+ * Check if a pattern match a string using mamberall libapat function
+ * @param	line		array containing sequence information
+ * @param	pattern		array containing patterns to test on the sequence
+ * @param	numpattern	number of pattern in pattern array
+ * @param	error_max	error rate allowed by the user
+ * 
+ * @return	int			1 if a pattern match, else 0
+ **/
 int	ispatternmatching(char **line, char **pattern, int32_t numpattern, int32_t error_max){
 	int 			i;
 	SeqPtr 	    	apatseq	= NULL;
@@ -38,6 +47,11 @@ int	ispatternmatching(char **line, char **pattern, int32_t numpattern, int32_t e
 	return 0;
 }
 
+/**
+ *returns the result on standard output
+ * @param 	line	array containing sequence information
+ * @param	i		length of line 
+ **/
 void printline(char **line, int32_t i){
 	int32_t k=0;
 	for (k=0; k < i; k++)
@@ -99,36 +113,31 @@ static void ExitUsage(stat)
 
 int main(int argc, char **argv){
 	int32_t     	carg 			= 0;
-	int32_t			r				= 0;				// number of restricted taxid
-	int32_t			g				= 0;				// number of ignored taxid
-	int32_t			v				= 0;			
-	int32_t			p				= 0;				// number of pattern
+	int32_t			r				= 0;			// number of restricted taxid
+	int32_t			g				= 0;			// number of ignored taxid
+	int32_t			v				= 0;			// stores if -v mode is active
+	int32_t			p				= 0;			// number of pattern
 	int32_t			i				= 0;
 	int32_t			errflag			= 0;
-	int32_t       	error_max		= 0;
-	int32_t			matchingresult 	= 0;
+	int32_t       	error_max		= 0;			// stores the error rate allowed by the user
+	int32_t			matchingresult 	= 0;			// stores number of matching result
 	
-	ecotaxonomy_t	*taxonomy;
+	ecotaxonomy_t	*taxonomy;						// stores the taxonomy
 		
 	char			*database			= NULL;		// stores the database path (for taxonomy)
 	int32_t		  	*restricted_taxid 	= NULL;		// stores the restricted taxid
 	int32_t       	*ignored_taxid		= NULL;		// stores the ignored taxid
 	char 			**pattern			= NULL;		// stores the regex pattern
 	char 			*line[19]			= {0};		// stores the line 
-	
-	FILE			*file				= NULL;
-	char			*stream				= ECOMALLOC(sizeof(char *)*10000,"error stream buffer allocation"); 
 	int				current_taxid;
-	char 			*buffer; 
 	
-	struct 	stat 	buf;
+	FILE			*file				= NULL;		// stores the data stream, stdin by default
+	char			*stream				= ECOMALLOC(sizeof(char *)*10000,"error stream buffer allocation"); 
+	char 			*buffer;
 	
-	// check stdin 
-	if (fstat(fileno(stdin),&buf) == 0)
-		if ( (buf.st_dev) == 0)
-			file = stdin;
-	
-
+	/**
+	 * Parse commande line options
+	 **/
 	while ((carg = getopt(argc, argv, "f:d:i:r:e:vh")) != -1) {
 
 	    switch (carg) {                              
@@ -178,7 +187,7 @@ int main(int argc, char **argv){
 	}
 	
 	/**
-	 * Get the leftover command line arguments back 
+	 * Get the left-over command line arguments back 
 	 * and check the pattern is not more than 32 character long
 	 */
 	pattern = ECOMALLOC(sizeof *pattern * (argc - optind), "Error in pattern allocation");
@@ -188,15 +197,26 @@ int main(int argc, char **argv){
 		else
 		{
 			printf("# Sorry, ecogrep doesn't handle pattern longer than 32 characters.\
-					\n# Please correct this one : %s\n",argv[optind]);
+					\n# Please check it out : %s\n",argv[optind]);
 			exit(0);
 		}
 	}
 	
+	/**
+	 *  check standard input if no file name given in -f option 
+	 **/
+	 if (file == NULL)
+	 {
+	 	if (isatty(fileno(stdin)))
+	 		errflag++;
+	 	else
+	 		file = stdin;
+	 }
+	 
 	/**
 	 * try to get the database name from environment variable
 	 * if no database name specified in the -d option
-	 */
+	 **/
 	if (database == NULL)
 	{
 		database = getenv("ECOPCRDB");
@@ -205,12 +225,10 @@ int main(int argc, char **argv){
 	}
 	
 	/**
-	 * check everything required is provided
+	 * check at leat one processing is asked
+	 * either patterns or taxid filters
 	 */			
-	if ( !file 		||
-		 !database 	||
-		 (!p && restricted_taxid == NULL && ignored_taxid == NULL)
-	   )
+	if ( !p && restricted_taxid == NULL && ignored_taxid == NULL )
 		errflag++;
 			
 	if (errflag)
@@ -224,7 +242,7 @@ int main(int argc, char **argv){
 	/**
 	* Parse the stream
 	*/ 
-	while( fgets(stream, 10000, file) ){
+	while( fgets(stream, 10000, file) != NULL ){
 	 		
 	 	if (stream[0]!= '#')
 	 	{
@@ -232,6 +250,7 @@ int main(int argc, char **argv){
 					buffer != NULL;
 					i++, buffer = strtok(NULL,"|"))
 			{
+				printf("¤¤ %s\n",buffer);
 				line[i] = strdup(buffer);
 			}
 			
@@ -254,7 +273,7 @@ int main(int argc, char **argv){
 			       continue;
 			}
 			 	
-			else // v mode, invert ignore and restrict taxid
+			else // v mode, invert ignore and restrict options
 			{
 				if ( (r > 0) && (eco_is_taxid_included(	taxonomy, 
 		                            					restricted_taxid, 
@@ -280,12 +299,15 @@ int main(int argc, char **argv){
 	 
 	printf("# %d matching result\n",matchingresult);
 	
+	/**
+	 * clean, close and free before leaving
+	 **/	
 	if ( file != stdin )
     	fclose(file);
-	
 	freememory(line,i);
 	freememory(pattern,p);
 	ECOFREE(pattern,"Error in free pattern");
 	ECOFREE(stream,"Error in free stream");
+	
 	return 0;
 }
\ No newline at end of file