please open a ticket about this. I will take care of it asap. On Thursday, 1 November 2012 15:54:53 UTC-5, Joe Barnhart wrote: > > This proposal solves a problem I have when formatting Field values. I > created a custom Validator class which maps the value None into a string > (such as "N/A" or "NT"), but the mapping was ignored because the current > Field processing provides an early exit if the value of the Field is None, > i.e. it never reaches the Validator object for formatting. > > Proposal: > > I propose creating a new keyword variable for Field objects, called > "map_none" which is initialized to None for backwards compatibility: > > > map_none = None, > > . . . > > self.map_none = map_none > > > > Next, we use the value of map_none as the return value for None from the > formatter function: > > > def formatter(self, value): > requires = self.requires > if value is None or not self.requires: > return value or self.map_none > > > Last, for symmetry, we alter the validator to replace any occurrence of > map_none with None when processing items to go back into the database. In > my case it will accept any entered value of "NT" and map it back into None > when storing row data in the database. I would also use it to replace > empty strings with None (which I prefer over empty strings in the database > when values are not entered). > > > def validate(self, value): > if not self.requires: > return ((value if value!=self.map_none else None), None) > requires = self.requires > if not isinstance(requires, (list, tuple)): > requires = [requires] > for validator in requires: > (value, error) = validator(value) > if error: > return (value, error) > return ((value if value!=self.map_none else None), None) > > This approach allows any mapping of None into an object of the user's > choosing, and provides the mapping in a consistent and bi-directional > manner. It preserves all existing behavior of Validators. The only cost > is an additional variable carried in Field objects and the time expended on > the trivial "if" tests to map the value. > > Any thoughts? > > -- Joe B.
--
