py2/py3: add str2unicode() utility function, that does nothing in Python 3.

product/ERP5Type/Utils.py

@@ -567,6 +567,8 @@ def checkPythonSourceCode(source_code_str, portal_type=None):
 
 if str is bytes:
   bytes2str = str2bytes = lambda s: s
+  def str2unicode(s):
+    return s.decode('utf-8')
   def unicode2str(s):
     return s.encode('utf-8')
 else:
@@ -574,6 +576,8 @@ else:
     return s.decode()
   def str2bytes(s):
     return s.encode()
+  def str2unicode(s):
+    return s
   def unicode2str(s):
     return s
 

product/ERP5Type/__init__.py

@@ -189,7 +189,7 @@ ModuleSecurityInfo('Products.ERP5Type.Utils').declarePublic(
     'int2letter', 'getMessageIdWithContext', 'getTranslationStringWithContext',
     'Email_parseAddressHeader', 'guessEncodingFromText',
     'isValidTALESExpression',
-    'ensure_list', 'bytes2str', 'str2bytes', 'unicode2str',
+    'ensure_list', 'bytes2str', 'str2bytes', 'str2unicode', 'unicode2str',
 )
 
 allow_module('Products.ERP5Type.Message')

[Kazuhiko Shiozaki]

@jerome which is better / which do you prefer ?

def foo(s):
  return s

or

foo = lambda s: s

[Jérome Perrin]

maybe the lambda form causes some pylint warning, but in any case, I think both forms are OK.

I tried timeit and it seems equivalent from a performance point of view,

python3 -m timeit -s 'foo = lambda s: s' 'foo("")' ;  python3  -m timeit -s 'def foo(s):' -s '  return s' 'foo("")'

Looking again at this patch, it would be a bit nicer to be consistent here, do you have a preference ? one point where function is a bit better, is that they have a meaningful repr <function str2bytes at 0x7f996ad848c0> vs <function <lambda> at 0x7f996ad84848>