openerp-sane: Bring sanity back, one step at a time

openerp-sane is a collection of small utilities, making OpenERP development seem less like ancient warfare with blood and guts everywhere, and more like Python.

For now, there's two utilities: @oemeth and s2d/d2s (date conversion). The @oemeth method decorator which straightens out our we manage the famous ids argument in our model methods. Example:

def myaction(self, cr, uid, ids, context=None): pass

Normally, ids is supposed to be a list of int, but sometimes, just, sometimes, we get a naked int, then our method crashes. We have to add code like:

if isinstance(ids, (int, long)):
    ids = [ids]

On top of that, there's another annoyance: why, oh why do all my methods have to handle cases with multiple ids? When I have an action for some button in a form, I know it's only ever going to handle one id at a time. I can do ids[0] easily enough, but if I really want to be on the safe side, I'll make sure that ids is a list first. Aren't you tired of that ridiculous dance? Well, that's why we have @oemeth

Install

You can't wait to start using it in your modules, right? openerp-sane can be installed from PyPI:

$ pip install openerp-sane

When you use it in a module, you can document its dependency to it in your __openerp__.py:

{
    # [...]
    'external_dependencies': {
        'python': ['openerp_sane'],
    },
    # [...]
}

@oemeth

@oemeth is a method decorator that takes 2 (optional, default to False) arguments: single and browse. By default, it simply makes sure that ids is a list:

from openerp_sane import oemeth

# [...]

@oemeth
def myaction(self, cr, uid, ids, context=None):
    # Write code that assumes ids is a list

With single to True, we enforce a single int id:

@oemeth(single=True)
def myaction(self, cr, uid, objid, context=None):
    # objid is an ``int``.
    # WARNING: Use this only when you're sure that you'll only ever have single arguments.
    # If the input is a list with a len() != 1, an exception is raised.

With browse to True, we wrap our id(s) in a self.browse() call:

@oemeth(browse=True)
def myaction(self, cr, uid, objs, context=None):
    # objs is a list of browse records

s2d/d2s

s2d() and d2s() (meaning "string-to-date" and "date-to-string") are there to alleviate the horrible problem we face when we actually have to process and compare dates. Without these helpers, we have to manually convert those with the right date format, which can get heavy quick.

s2d(string_date) takes the string value from a date field and returns a datetime.date. If it can't parse it, it returns False.

d2s(date) takes a datetime.date and returns a string which can be written to a date field. If date isn't a date, we return False (which can also be written to a date field).

In both those functions, we use OE's DEFAULT_SERVER_DATE_FORMAT constant.

Bits of wisdom

Don't use single (which requires to always have exactly one id all the time) on on_change methods. Sure, most of the time you get a single id, but if your call is made on a record that isn't committed yet, you will get zero ids, which will raise an exception. In the future, maybe the single mode will support zero ids situations.