From 535fc2af834825e543464463d6774ccfb9fc339f Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Wed, 5 Jul 2017 17:49:05 +0200
Subject: [PATCH] Column rewriter and optimized View getter

---
 python/obitools3/dms/view/view.pxd | 12 +++++--
 python/obitools3/dms/view/view.pyx | 58 +++++++++++++++++++++++-------
 2 files changed, 56 insertions(+), 14 deletions(-)

diff --git a/python/obitools3/dms/view/view.pxd b/python/obitools3/dms/view/view.pxd
index fc3d057..90b3799 100644
--- a/python/obitools3/dms/view/view.pxd
+++ b/python/obitools3/dms/view/view.pxd
@@ -3,12 +3,14 @@
 
 from ..capi.obiview cimport Obiview_p
 
-from ..capi.obitypes cimport index_t
+from ..capi.obitypes cimport index_t, obitype_t
 
 from ..object cimport OBIWrapper                          
 
 from ..dms cimport DMS
 
+from ..column.column cimport Column
+
 
 cdef dict __OBIDMS_VIEW_CLASS__
 
@@ -25,7 +27,13 @@ cdef class View(OBIWrapper):
     cpdef rename_column(self, 
                         object current_name, 
                         object new_name)
-        
+    
+    cpdef Column rewrite_column_with_diff_attributes(self,
+                                                     object    column_name,
+                                                     obitype_t new_data_type=*,
+                                                     index_t   new_nb_elements_per_line=*,
+                                                     list      new_elements_names=*)
+
     cpdef Line_selection new_selection(self,
                                        list lines=*)
 
diff --git a/python/obitools3/dms/view/view.pyx b/python/obitools3/dms/view/view.pyx
index 58b5b7d..26e3953 100644
--- a/python/obitools3/dms/view/view.pyx
+++ b/python/obitools3/dms/view/view.pyx
@@ -17,19 +17,17 @@ from ..capi.obiview cimport Alias_column_pair_p, \
                            
 from ..capi.obidmscolumn cimport OBIDMS_column_p
 from ..capi.obidms cimport OBIDMS_p
-from ..capi.obitypes cimport obitype_t
 
 from obitools3.utils cimport tobytes, \
                              str2bytes, \
                              bytes2str
    
 from ..object cimport OBIObjectClosedInstance
-          
-from ..column.column cimport Column
 
 from obitools3.dms.view import typed_view
 
-from ..capi.obitypes cimport only_ATGC, \
+from ..capi.obitypes cimport is_a_DNA_seq, \
+                             OBI_VOID, \
                              OBI_BOOL, \
                              OBI_CHAR, \
                              OBI_FLOAT, \
@@ -191,10 +189,10 @@ cdef class View(OBIWrapper) :
        
     def get_column(self,
                    object column_name):
-        
+
         if not self.active() :
             raise OBIObjectClosedInstance()
-
+                
         return Column.open(self, column_name)
     
     
@@ -214,8 +212,6 @@ cdef class View(OBIWrapper) :
         if obi_view_delete_column(self.pointer(), column_name_b) < 0 :
             raise Exception("Problem deleting column %s from a view",
                             bytes2str(column_name_b))
-        
-        
 
 
     cpdef rename_column(self, 
@@ -236,7 +232,45 @@ cdef class View(OBIWrapper) :
                                 bytes2str(current_name_b),
                                 bytes2str(new_name_b)))
 
+
+    # TODO warning, not multithreading compliant
+    cpdef Column rewrite_column_with_diff_attributes(self,
+                                                     object    column_name,
+                                                     obitype_t new_data_type=<obitype_t>OBI_VOID,
+                                                     index_t   new_nb_elements_per_line=0,
+                                                     list      new_elements_names=None) :
         
+        cdef Column  old_column                                             
+        cdef Column  new_column
+        cdef index_t length = len(self)
+        
+        old_column = self.get_column(column_name)
+
+        if new_data_type == 0 :
+            new_data_type = old_column.data_type
+        
+        if new_nb_elements_per_line == 0 :
+            new_nb_elements_per_line = old_column.nb_elements_per_line
+        
+        if new_elements_names is None :
+            new_elements_names = old_column.elements_names
+
+        new_column = Column.new_column(self, old_column.pointer().header.name, new_data_type, 
+                                       nb_elements_per_line=new_nb_elements_per_line, elements_names=new_elements_names, 
+                                       comments=old_column.comments, alias=tobytes(column_name)+tobytes('___new___'))
+        
+        for i in range(length) :
+            new_column[i] = old_column[i]
+
+        # Remove old column from view
+        self.delete_column(column_name)
+
+        # Rename new
+        new_column.name = column_name
+        
+        return new_column
+
+
     cpdef Line_selection new_selection(self,list lines=None):
         return Line_selection(self, lines)
 
@@ -255,10 +289,10 @@ cdef class View(OBIWrapper) :
 
 
     def __getitem__(self, object item) :
-        if type(item) == str :
-            return self.get_column(item)    # TODO hyper lent dans la pratique
-        elif type(item) == int :
+        if type(item) == int :
             return Line(self, item)
+        else :  # TODO assume str or bytes for optimization?
+            return self.get_column(item)    # TODO hyper lent dans la pratique            
 
 
     def __contains__(self, str column_name):
@@ -341,7 +375,7 @@ cdef class Line :
                     value_b = str2bytes(value)
                 else :
                     value_b = value
-                if only_ATGC(value_b) :    # TODO detect IUPAC
+                if is_a_DNA_seq(value_b) :
                     value_obitype = OBI_SEQ
                 elif len(value) == 1 :
                     value_obitype = OBI_CHAR