Merge with enze, Makefile cleanup and correct PyInit

Author: Roman Valls Guimera

Makefile

@@ -10,8 +10,9 @@ tests: python
 valgrind: python
 	valgrind --tool=memcheck --suppressions=facs/utils/valgrind-python.supp nosetests -P -v -s
 
-python:
-	rm -rf build/ ${PROG}.so && python setup.py build_ext && python setup.py install
+python: clean
+	python setup.py build_ext && python setup.py install
 
 clean:
+	rm -rf build dist $(PROG).egg-info
 	cd facs; make clean

facs/big_query.c

@@ -75,9 +75,8 @@ int bq_main(int argc, char** argv)
   return query(source, ref, tole_rate, sampling_rate, list, target_path);
 }
 
-int query(char* query, char* bloom_filter, double tole_rate, double sampling_rate, char* list, char* target_path)
+int query(char* query, char* bloom_filter, float tole_rate, float sampling_rate, char* list, char* target_path)
 {
-  
   gzFile zip;
   int type = 0, normal = 0;
   BIGCAST offset = 0;
@@ -95,7 +94,7 @@ int query(char* query, char* bloom_filter, double tole_rate, double sampling_rat
   if (tole_rate==0)
       tole_rate = mco_suggestion (bl_2->k_mer); 
 
-  if ((get_size (query)<2*ONEG) && !strstr(query,".gz") & !strstr(query, ".tar"))
+  if ((get_size (query)<2*ONEG) && !strstr(query,".gz") && !strstr(query, ".tar"))
       normal = 1; 
   else
       {
@@ -107,7 +106,7 @@ int query(char* query, char* bloom_filter, double tole_rate, double sampling_rat
       normal = 0;
       }
 
-  if (strstr(query, ".fastq") || strstr(query, ".fq"))
+  if (strstr(query, ".fastq")!=NULL || strstr(query, ".fq")!=NULL)
       type = 2;
   else
       type = 1;
@@ -171,9 +170,12 @@ int query(char* query, char* bloom_filter, double tole_rate, double sampling_rat
 
     }				//end while
   if (normal == 0)
-  free (position);
+	  free (position);
+
   evaluate (detail, File_head->filename, File_head);
-  //if(zip != NULL) gzclose(zip);
+  
+  if (normal == 0)
+  	gzclose(zip);
   bloom_destroy (bl_2);
   statistic_save (detail, query, target_path);
 

facs/big_query.h

@@ -11,5 +11,5 @@ extern void clean_list (Queue* head, Queue *tail);
 extern BIGCAST CHUNKer(gzFile zip,BIGCAST offset,int chunk,char *data,int type);
 extern BIGCAST CHUNKgz(gzFile zip, BIGCAST offset,int chunk,char *position,char *extra,int type);
 extern int bq_main (int argc, char **argv);
-extern int query (char* query, char* bloom_filter, double tole_rate, double sampling_rate, char* list, char* target_path);
+extern int query (char* query, char* bloom_filter, float tole_rate, float sampling_rate, char* list, char* target_path);
 #endif

facs/check.h

@@ -4,6 +4,6 @@
 #include "bloom.h"
 extern void evaluate (char *detail, char *filename, F_set *File_head);
 extern void statistic_save (char *detail, char *filename, char* prefix);
-void fasta_process (bloom * bl, Queue * info, Queue * tail, F_set *File_head, float sampling_rate, float tole_rate);
-
+extern void fasta_process (bloom * bl, Queue * info, Queue * tail, F_set *File_head, float sampling_rate, float tole_rate);
+extern void fastq_process (bloom * bl, Queue * info, Queue *tail, F_set * File_head, float sampling_rate, float tole_rate);
 #endif

facs/facs

@@ -28,7 +28,7 @@ PyMODINIT_FUNC init_facs(void)
 {
     PyObject *m = Py_InitModule3("_facs", module_methods, module_docstring);
     if (m == NULL)
-        return -1;
+        return;
 }
 
 

facs/facs.c

@@ -120,7 +120,6 @@ make_list (char *file_user, char *list_user)
 	  if (!strstr (dir_info->d_name, ".bloom"))
 	    continue;
 
-	  printf ("file_path->%s\n", file_path);
 	  F_set *fset = NEW (F_set);
 	  fset->filename = file_path;
 	  fset->number = &number;

facs/file_dir.c

@@ -38,19 +38,19 @@ fastq_process (bloom * bl, Queue * info, Queue *tail, F_set * File_head,
     next = info->next->location;
   } else {
     next = strchr (p, '\0');
+    if ((next-1)=='\n')
+    next-=1;
   }
-
+   
   while (p != next)
     {
-      //printf ("p->%0.50s\n",p);
       temp = jump (p, 2, sampling_rate);	//generate random number and judge if need to scan this read
 
       if (p != temp)
 	{
 	  p = temp;
 	  continue;
 	}
-
 #pragma omp atomic
       File_head->reads_num++;
 
@@ -59,7 +59,6 @@ fastq_process (bloom * bl, Queue * info, Queue *tail, F_set * File_head,
 #pragma omp atomic
 	File_head->reads_contam++;
       }
-
       p = strchr (p, '\n') + 1;
       p = strchr (p, '\n') + 1;
       p = strchr (p, '\n') + 1;
@@ -84,7 +83,11 @@ fasta_process (bloom * bl, Queue * info, Queue * tail, F_set * File_head,
   else if (info->next != tail)
     next = info->next->location;
   else
-    next = strchr (info->location, '\0');
+    {
+      next = strchr (info->location, '\0');
+      if ((next-1)=='\n')
+      next -= 1;
+    }
 
   char *p = info->location;
 

facs/simple_check_1_ge.c

@@ -67,8 +67,6 @@ int remove_reads(char *source, char *ref, char *list, char *prefix, float tole_r
   /*-------------------------------------*/
   int type = 1;
   char *position;
-  //char *clean;
-  //char *contam;
   char *clean2;
   char *contam2;
   /*-------------------------------------*/
@@ -83,6 +81,7 @@ int remove_reads(char *source, char *ref, char *list, char *prefix, float tole_r
   /*-------------------------------------*/
   position = mmaping (source);
   type = get_parainfo (position, head);
+  //printf ("position->%0.10s\n",head->next->location);
   clean = (char *) malloc (strlen (position) * sizeof (char));
   contam = (char *) malloc (strlen (position) * sizeof (char));
   clean2 = clean;
@@ -96,6 +95,7 @@ int remove_reads(char *source, char *ref, char *list, char *prefix, float tole_r
 
       if (tole_rate==0)
       	tole_rate = mco_suggestion (bl_2->k_mer);
+      //head = head->next;
 #pragma omp parallel
       {
 #pragma omp single nowait

facs/simple_remove.c

@@ -1,3 +1,2 @@
 [egg_info]
-tag_build = dev
-tag_svn_revision = true
+tag_svn_revision = false