Incorrect mapping for RAW Oracle column types.

There is some inconsistency in the mapping of OracleRaw and OracleBinary type to the corresponding sqltype which throws the following error with both version 0.4.1 and SVN HEAD when running the attached test case which attempts to map a Oracle System View which has columns of type RAW:

Traceback (most recent call last):
  File "sqlalchemy-svn\lib\sa-oracle-map-tc.py", line 21, in <mo
dule>
    create_session().query(MetaType).all()
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 676, i
n all
    return list(self)
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 737, i
n __iter__
    return self._execute_and_instances(context)
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 742, i
n _execute_and_instances
    return iter(self.instances(result, querycontext=querycontext))
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 808, i
n instances
    rows.append(main(context, row))
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 766, i
n main
    extension=context.extension, only_load_props=context.only_load_props, refres
h_instance=context.refresh_instance
  File "sqlalchemy-svn\lib\sqlalchemy\orm\mapper.py", line 1356,
 in _instance
    self.populate_instance(context, instance, row, only_load_props=only_load_pro
ps, instancekey=identitykey, isnew=isnew)
  File "sqlalchemy-svn\lib\sqlalchemy\orm\mapper.py", line 1444,
 in populate_instance
    selectcontext.exec_with_path(self, key, populator, instance, row, ispostsele
ct=ispostselect, isnew=isnew, **flags)
  File "sqlalchemy-svn\lib\sqlalchemy\orm\query.py", line 1336,
in exec_with_path
    return func(*args, **kwargs)
  File "sqlalchemy-svn\lib\sqlalchemy\orm\strategies.py", line 7
9, in new_execute
    instance.__dict__[self.key](self.key) = row[self.columns[0](self.columns[0)]
  File "sqlalchemy-svn\lib\sqlalchemy\engine\base.py", line 1256
, in __getitem__
    return self.__parent._get_col(self.__row, key)
  File "sqlalchemy-svn\lib\sqlalchemy\engine\base.py", line 1510
, in _get_col
    return processor(row[index](index))
  File "sqlalchemy-svn\lib\sqlalchemy\databases\oracle.py", line
 137, in process
    return value.read()
AttributeError: 'str' object has no attribute 'read'

The attached patch, against SVN, which maps sqltypes.Binary to OracleRaw and sqltypes.BLOB to OracleBinary seems to solve the problem.

Comments (6)