Suggestion for cleaning Item data (removing or opt-in pipelining)

[Hedde]

I don't really like the pipeline cycle as a basic clean cycle because most of the time it requires some sort of differentiation per Item leaf class.

My suggestion is to create something similar as django's forms.Form clean loop. Which does something similar like my usual work-around, e.g.:

class ExtendedItem(Item):
    def _process(self):
        # Runs every '_process_FIELD_NAME' before being reached.
        [getattr(self, func)() for func in dir(self) if func.split('_')[-1] in self.fields and callable(getattr(self, func))]

Now in my class definitions I can do something like (normally extended with optional arguments):

class Necklace(ExtendedItem):
    pearls = Field()
    diamonds = Field()

    def _process_pearls(self):
        # clean field

    def _process(self):
        # override clean all
        super(Necklace, self)._process()

This plays nicely without intervening other classes their data unless you want to.. but that's where the pipeline is a nice opt-in 👍

[pablohoffman]

Where would _process() be called in your proposal?.

If we're gonna add these methods, we'll probably be adding a new SanitizeItem class that subclasses item.

There are probably also mechanism to make the _process() code more efficient using meta-classes.

[Hedde]

I could have a closer look how django implements the clean cycle. I am not proposing code, just an idea, I'll investigate some more