Philippe Niquille wrote:
> Tanks a mill!
>
> I don't know why I searched so far..
>
> Anyway, I wrapped the django custom SQL call and built a nice dictionary
> out of the resulting rows (which is similar to querysets).
> See http://www.djangosnippets.org/snippets/207/
> <http://www.djangosnippets.org/snippets/207/> for details.
That code could be much simpler. You wrote,
fdicts = []
for row in rows:
i = 0
cur_row = {}
for key in qkeys:
cur_row[key] = row[i]
i = i+1
fdicts.append(cur_row)
So qkeys is a list of names for the values in each row. You can convert
the list of names and row into a dict with
cur_row = dict(zip(qkeys, row))
zip(qkeys, row) converts the two list into a list of key, value pairs
which can be used directly to create the dict.
This is simple enough to use in a list comprehension, so the outer loop
can be written as
fdicts = [ dict(zip(qkeys, row)) for row in rows ]
To answer your original question, the best way to sort on multiple
attributes is to make a key function that returns a tuple of the desired
attributes and provide that to sort. In Python 2.5, the function
operator.attrgetter() will create the key function for you, so you can write
import operator
newscore.sort(key=operator.attrgetter('score', 'owner'))
or whatever the correct attributes are.
In older Python attrgetter only takes a single argument so you have to
create the key function yourself:
def key_func(item):
return (item.score, item.owner)
newscore.sort(key=key_func)
Kent
_______________________________________________
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
