Added the possibility to clone a column, with or without its data

python/obitools3/obidms/obidmscolumn/capidmscolumn.pxd

@@ -35,6 +35,7 @@ cdef extern from "obidmscolumn.h" nogil:
     OBIType_t obi_column_get_type(OBIDMS_column_p column)
     int obi_close_column(OBIDMS_column_p column)
     OBIType_t obi_column_get_data_type_from_name(OBIDMS_p dms, const char* column_name)
+    OBIDMS_column_p obi_clone_column(OBIDMS_p dms, const char* column_name, obiversion_t version_number, bint clone_data)
 
 
 cdef class OBIDMS_column:

python/obitools3/obidms/obidmscolumn/capidmscolumn.pyx

@@ -12,9 +12,10 @@ cdef class OBIDMS_column:
 
     #Should only be initialized through a subclass (or open())
     def __init__(self, dms_name, column_name, 
-                 create, version_number, 
+                 create, clone, clone_data,
-                 type, nb_elements, 
+                 version_number, type,
-                 nb_elements_per_line, elements_names):
+                 nb_elements, nb_elements_per_line, 
+                 elements_names):
         column_name_b = column_name.encode(encoding='UTF-8')
         dms_name_b = dms_name.encode(encoding='UTF-8')
         self.dms_name = dms_name
@@ -31,13 +32,10 @@ cdef class OBIDMS_column:
         else :
             dms = obi_open_dms(dms_name_b)
             self.dms = dms
-            self.pointer = obi_open_column(self.dms, column_name_b, version_number)
+            if clone :
-
+                self.pointer = obi_clone_column(self.dms, column_name_b, version_number, clone_data)
-    def create(self, dms_name, column_name, type, nb_elements, nb_elements_per_line=1, elements_names=None):
+            else :
-        if elements_names == None :  
+                self.pointer = obi_open_column(self.dms, column_name_b, version_number)
-            elements_names = column_name
-        self.dms = obi_dms(dms_name)
-        self.pointer = obi_create_column(self.dms, column_name, type, nb_elements, nb_elements_per_line, elements_names)     
     
     def __iter__(self):
         elements_names = self.get_elements_names()
@@ -78,9 +76,10 @@ cdef class OBIDMS_column:
     
     @staticmethod
     def open(dms_name, column_name, 
-             create=False, version_number=-1, 
+             create=False, clone=False, clone_data=True,
-             type=None, nb_elements=None, 
+             version_number=-1, type=None, 
-             nb_elements_per_line=1, elements_names=None):
+             nb_elements=None, nb_elements_per_line=1, 
+             elements_names=None):
         
         column_name_b = column_name.encode(encoding='UTF-8')
         dms_name_b = dms_name.encode(encoding='UTF-8')
@@ -94,33 +93,38 @@ cdef class OBIDMS_column:
 
         if type == 1 :
             column = OBIDMS_column_int(dms_name, column_name, 
-                                       create, version_number, 
+                                       create, clone, clone_data, 
-                                       type, nb_elements, 
+                                       version_number, type, 
-                                       nb_elements_per_line, elements_names)
+                                       nb_elements, nb_elements_per_line, 
+                                       elements_names)
         
         elif type == 2 :
             column = OBIDMS_column_float(dms_name, column_name, 
-                                       create, version_number, 
+                                       create, clone, clone_data, 
-                                       type, nb_elements, 
+                                       version_number, type, 
-                                       nb_elements_per_line, elements_names)
+                                       nb_elements, nb_elements_per_line, 
+                                       elements_names)
 
         elif type == 3 :
             column = OBIDMS_column_bool(dms_name, column_name, 
-                                       create, version_number, 
+                                       create, clone, clone_data, 
-                                       type, nb_elements, 
+                                       version_number, type, 
-                                       nb_elements_per_line, elements_names)
+                                       nb_elements, nb_elements_per_line, 
+                                       elements_names)
 
-#         elif type == 4 :
+#        elif type == 4 :
-#             column = OBIDMS_column_char(dms_name, column_name, 
+#            column = OBIDMS_column_char(dms_name, column_name, 
-#                                        create, version_number, 
+#                                       create, clone, clone_data, 
-#                                        type, nb_elements, 
+#                                       version_number, type, 
-#                                        nb_elements_per_line, elements_names)
+#                                       nb_elements, nb_elements_per_line, 
+#                                       elements_names)
 
         elif type == 5 :
             column = OBIDMS_column_idx(dms_name, column_name, 
-                                       create, version_number, 
+                                       create, clone, clone_data, 
-                                       type, nb_elements, 
+                                       version_number, type, 
-                                       nb_elements_per_line, elements_names)
+                                       nb_elements, nb_elements_per_line, 
+                                       elements_names)
         
         else :
             "Problem with the data type"

src/obidmscolumn.c

@@ -723,6 +723,7 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
 	header->data_type     		 = type;
 	header->creation_date 		 = time(NULL);
 	header->version       		 = version_number;
+	header->cloned_from    		 = -1;
 	header->comments[0]   		 = 0x0;
 
 	obi_column_set_elements_names(new_column, elements_names);
@@ -813,7 +814,7 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms, const char* column_name, obiversio
 	column->header 			 = mmap(NULL,
 			                  	  	header_size,
 			                  	  	PROT_READ,
-			                  	  	MAP_SHARED,
+			                  	  	MAP_PRIVATE,
 			                  	  	column_file_descriptor,
 			                  	  	0
 			                 	 	);
@@ -836,7 +837,7 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms, const char* column_name, obiversio
 	column->data   			 = mmap(NULL,
 			                 	 	data_size,
 			                 	 	PROT_READ,
-			                 	 	MAP_SHARED,
+									MAP_PRIVATE,
 			                 	 	column_file_descriptor,
 			                 	 	header_size
 			                	    );
@@ -861,9 +862,59 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms, const char* column_name, obiversio
 }
 
 
+OBIDMS_column_p obi_clone_column(OBIDMS_p dms, const char* column_name, obiversion_t version_number, bool clone_data)
+{
+	OBIDMS_column_p   column_to_clone;
+	OBIDMS_column_p   new_column;
+	size_t 			  nb_lines;
+	size_t			  nb_elements_per_line;
+	OBIType_t		  data_type;
+
+	column_to_clone = obi_open_column(dms, column_name, version_number);
+	if (column_to_clone == NULL)
+	{
+		obi_set_errno(OBICOL_UNKNOWN_ERROR);
+		obidebug(1, "\nError opening the column to clone");
+		return NULL;
+	}
+
+	if (clone_data)
+		nb_lines = (column_to_clone->header)->line_count;
+	else
+		nb_lines = INITIAL_LINE_COUNT;
+
+	nb_elements_per_line = (column_to_clone->header)->nb_elements_per_line;
+	data_type = (column_to_clone->header)->data_type;
+
+	new_column = obi_create_column(dms,
+								   column_name,
+								   data_type,
+								   nb_lines,
+								   nb_elements_per_line,
+								   (column_to_clone->header)->elements_names);
+
+	if (new_column == NULL)
+	{
+		obi_set_errno(OBICOL_UNKNOWN_ERROR);
+		obidebug(1, "\nError creating the new column when cloning a column");
+		// TODO the new file should be deleted
+		return NULL;
+	}
+
+	(new_column->header)->cloned_from = version_number;
+	// TODO copy header->comments?
+
+	if (clone_data)
+		memcpy(new_column->data, column_to_clone->data, nb_lines*nb_elements_per_line*sizeof(data_type));
+
+	return new_column;
+}
+
+
 int obi_close_column(OBIDMS_column_p column)
 {
 	//munmap? TODO
+	//truncate to lines used TODO
 	free(column);
 	return 0;
 }

src/obidmscolumn.h

@@ -17,6 +17,7 @@
 #include <unistd.h>
 #include <stdbool.h>
 #include <time.h>
+#include <stdbool.h>
 
 #include "obidms.h"
 #include "obitypes.h"
@@ -25,6 +26,7 @@
 #include "obidmscolumndir.h"
 
 #define ELEMENTS_NAMES_MAX (2048)
+#define INITIAL_LINE_COUNT (1000)
 
 
 typedef int32_t obiversion_t; 	/**< Used to store the column version number
@@ -49,6 +51,7 @@ typedef struct OBIDMS_column_header {
 	OBIType_t		data_type;				    /**< type of the data */
 	time_t			creation_date;			    /**< date of creation of the file */
 	obiversion_t	version;				    /**< version of the OBIColumn */
+	obiversion_t	cloned_from;			    /**< version of the OBIColumn from which the column was cloned from (-1 if it does not come from cloning).*/
 	char            name[OBIDMS_MAX_COLNAME+1]; /**< The column name as a NULL
 	                                             *   terminated string.
 	                                             */
@@ -148,7 +151,7 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
  *
  * @param dms a pointer on an OBIDMS
  * @param column_name the name of the column
- * @param version_number the version of the column that should be opened
+ * @param version_number the version of the column that should be opened (if -1, the latest version number is retrieved)
  *
  * @return a pointer to the opened column
  *
@@ -158,6 +161,22 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
 OBIDMS_column_p obi_open_column(OBIDMS_p dms, const char* column_name, obiversion_t version_number);
 
 
+/**
+ * @brief Clones a column, and returns a pointer to the writable new column.
+ *
+ * @param dms a pointer on an OBIDMS
+ * @param column_name the name of the column to clone
+ * @param version_number the version of the column that should be cloned (if -1, the latest version number is retrieved)
+ * @param clone_data whether the data should be copied or not
+ *
+ * @return a pointer to the created column
+ *
+ * @since August 2015
+ * @author Celine Mercier (celine.mercier@metabarcoding.org)
+ */
+OBIDMS_column_p obi_clone_column(OBIDMS_p dms, const char* column_name, obiversion_t version_number, bool clone_data);
+
+
 /**
  * @brief Closes a column.
  *