On Tue, Apr 16, 2013 at 6:09 PM, Dave Angel <da...@davea.name> wrote:
> If the conversion itself will catch the errors, then you can just use
> try/catch to make such errors more polite to your users. That's the case if
> they enter "charlie" when you're asking for a salary value. But if they
> were to put 7.500 when the value is supposed to be in dollars and cents,
> then by the time it's turned into a float or decimal, the extra zero is long
> gone.
With Decimal it's not really 'long gone'. It affects the number of
significant digits retained in arithmetic operations:
>>> from decimal import Decimal
>>> context = decimal.getcontext()
>>> Decimal('7.500') + Decimal('0.50')
Decimal('8.000')
You could look specifically at the _exp attribute, or quantize with a
trap for inexact rounding:
>>> Decimal('7.500')._exp
-3
def round_money(value):
with decimal.localcontext() as ctx:
ctx.traps[decimal.Inexact] = 1
return Decimal(value).quantize(Decimal('1.00'))
>>> try: round_money('7.500')
... except decimal.Inexact: print 'ack!'
...
Decimal('7.50')
>>> try: round_money('7.501')
... except decimal.Inexact: print 'ack!'
...
ack!
>>> try: round_money('charlie')
... except decimal.InvalidOperation as e: print 'ack!'
...
ack!
Please don't print a useless error message like 'ack'; it was just a
silly example. Handle the error within the logical context of your
application.
_______________________________________________
Tutor maillist - Tutor@python.org
To unsubscribe or change subscription options:
http://mail.python.org/mailman/listinfo/tutor
