From 0dfd67ec89b005759fea0d986e47fd7173caaa49 Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Mon, 10 Oct 2016 17:04:29 +0200
Subject: [PATCH] The endianness of binary taxonomy files is now correctly
 checked

---
 src/obidms_taxonomy.c | 73 +++++++++++++++++++++++++++----------------
 src/obidms_taxonomy.h |  4 +++
 2 files changed, 50 insertions(+), 27 deletions(-)

diff --git a/src/obidms_taxonomy.c b/src/obidms_taxonomy.c
index 925dab2..91249e4 100644
--- a/src/obidms_taxonomy.c
+++ b/src/obidms_taxonomy.c
@@ -18,6 +18,7 @@
 
 #include "obidms_taxonomy.h"
 #include "obidms.h"
+#include "obilittlebigman.h"	// TODO the function from this checking the endianness does not seem to work properly
 #include "obidebug.h"
 #include "obierrno.h"
 #include "utils.h"
@@ -25,7 +26,23 @@
 
 #define DEBUG_LEVEL 0	// TODO has to be defined somewhere else (cython compil flag?)
 
-// TODO : the malloc aren't checked but won't exist for long because mapping instead
+// TODO : the malloc aren't checked but shouldn't exist for long because mapping instead
+// error checking and file closing in general aren't done properly yet
+// The endianness eventually shouldn't need checking too, as the machine will write the taxonomy with its endianness.
+
+
+int32_t is_big_endian()
+{
+	int32_t i=1;
+
+	return (int32_t)((char*)&i)[0];
+}
+
+
+int32_t swap_int32_t(int32_t i)
+{
+	return SWAPINT32(i);
+}
 
 
 int compareRankLabel(const void *label1, const void *label2)
@@ -61,22 +78,23 @@ void* read_ecorecord(FILE* f, int32_t* record_size)
 	}
 
 	read = fread(record_size,
-	      		 1,
 	      		 sizeof(int32_t),
+				 1,
 	             f);
 
 	if (feof(f))
 		return NULL;
 
-	if (read != sizeof(int32_t))
+	if (read != 1)
 	{
  		obi_set_errno(OBI_TAXONOMY_ERROR);
  		obidebug(1, "\nError reading a taxonomy file: error reading record size");
  		return NULL;
 	}
 
-//	if (is_big_endian())						// TODO
-//		*recordSize=swap_int32_t(*recordSize);
+//	if (!(obi_is_little_endian()))						// TODO
+	if (is_big_endian())
+		*record_size=swap_int32_t(*record_size);
 
 	if (buffer_size < *record_size)
 	{
@@ -93,11 +111,11 @@ void* read_ecorecord(FILE* f, int32_t* record_size)
 	}
 
 	read = fread(buffer,
-	             1,
 				 *record_size,
+				 1,
 				 f);
 
-	if (read != *record_size)
+	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);
@@ -118,13 +136,14 @@ ecotx_t* readnext_ecotaxon(FILE* f, ecotx_t* taxon)
 	if (!raw)
 		return NULL;
 
-//	if (is_big_endian())		// TODO
-//	{
-//		raw->namelength = swap_int32_t(raw->namelength);
-//		raw->parent     = swap_int32_t(raw->parent);
-//		raw->rank       = swap_int32_t(raw->rank);
-//		raw->taxid      = swap_int32_t(raw->taxid);
-//	}
+//	if (!(obi_is_little_endian()))						// TODO
+	if (is_big_endian())
+	{
+		raw->name_length = swap_int32_t(raw->name_length);
+		raw->parent      = swap_int32_t(raw->parent);
+		raw->rank        = swap_int32_t(raw->rank);
+		raw->taxid       = swap_int32_t(raw->taxid);
+	}
 
 	taxon->parent = (ecotx_t*) ((size_t) raw->parent);
 	taxon->taxid  = raw->taxid;
@@ -146,8 +165,6 @@ FILE* open_ecorecorddb(const char* file_name,
     FILE*		f;
 	int32_t     read;
 
-	fprintf(stderr, "\n%s\n", file_name);
-
 	f = fopen(file_name, "rb");
 
 	if (!f)
@@ -166,19 +183,20 @@ FILE* open_ecorecorddb(const char* file_name,
 	}
 
 	read = fread(count,
-	      		 1,
 	      		 sizeof(int32_t),
+				 1,
 	      		 f);
 
-	if (read != sizeof(int32_t))
+	if (read != 1)
 	{
  		obi_set_errno(OBI_TAXONOMY_ERROR);
  		obidebug(1, "\nError reading taxonomy record size");
  		return NULL;
 	}
 
-//	if (!obi_is_little_endian())		// TODO
-//		*count = swap_int32_t(*count);
+//	if (!(obi_is_little_endian()))						// TODO
+	if (is_big_endian())
+		*count = swap_int32_t(*count);
 
 	return f;
 }
@@ -302,13 +320,14 @@ econame_t* readnext_econame(FILE* f, econame_t* name, OBIDMS_taxonomy_p taxonomy
 	if (!raw)
 		return NULL;
 
-//	if (is_big_endian())		// TODO
-//	{
-//		raw->is_scientificname 	= swap_int32_t(raw->is_scientificname);
-//		raw->namelength 	    = swap_int32_t(raw->namelength);
-//		raw->classlength        = swap_int32_t(raw->classlength);
-//		raw->taxid  	        = swap_int32_t(raw->taxid);
-//	}
+//	if (!(obi_is_little_endian()))						// TODO
+	if (is_big_endian())
+	{
+		raw->is_scientific_name 	= swap_int32_t(raw->is_scientific_name);
+		raw->name_length 	    = swap_int32_t(raw->name_length);
+		raw->class_length        = swap_int32_t(raw->class_length);
+		raw->taxid  	        = swap_int32_t(raw->taxid);
+	}
 
 	name->is_scientific_name = raw->is_scientific_name;
 
diff --git a/src/obidms_taxonomy.h b/src/obidms_taxonomy.h
index fd8a510..ba63feb 100644
--- a/src/obidms_taxonomy.h
+++ b/src/obidms_taxonomy.h
@@ -17,6 +17,10 @@
 #include "obidms.h"
 
 
+#define SWAPINT32(x)     ((((x) << 24) & 0xFF000000) | (((x) <<  8) & 0xFF0000) | \
+                         (((x) >>  8) & 0xFF00)     | (((x) >> 24) & 0xFF))
+
+
 typedef struct {
 	int32_t  taxid;
 	int32_t  rank;