Peter Abel p-abel at
Sat Jan 18 20:49:22 CET 2003

Martin Maney <maney at> wrote in message news:<b0aggh$e7m$4 at>...
> Peter Abel <p-abel at> wrote:
> > Sorry, when i didn't read the thread quite exactly,
> > but I think your basic problem is something like this, which can
> > be easily solved.
> > I'm using PYTHON 2.2.2:
> >>>> l1=['zero','one','two','three']
> >>>> l2=range(5)
> >>>> l1
>  ['zero', 'one', 'two', 'three']
> >>>> l2
>  [0, 1, 2, 3, 4]
> >>>> di=dict(zip(l1,l2))
> >>>> di
> > {'zero': 0, 'three': 3, 'two': 2, 'one': 1}
> It is something like that, but the difference is sort of crucial.  For
> the dictionary in your example, the source would be the single sequence
> (('zero',0),('one',1),('two',2)('three',3))
> Except that in practice, the key may be more than one field of the
> nested sequences, as may the value.  So this doesn't reduce to a simple
> dict(zip(...)) for any simple '...'
> Besides, the dict() constructor seems not to exist prior to 2.2, and at
> this time I still need most of my code to coexist with 2.1. 
> Furthermore, when I ran some simple benchmarks, calling a separate
> function ran significantly faster than an inline dict(zip(...)) for the
> small (five to ten items) cases I'm most frequently building.  I think
> I emailed the test proggy home before I left the lab - have to look for
> it tomorrow.
> Here's the bare sketch:
> def dzip(keys, values):
>     res = {}
>     for k,v in zip(keys, values):
>         res[k] = v
>     return res
> foo = dzip(field_names, field_values)
> runs faster than
> foo = dict(zip(field_names, field_values))
> for cases where there are a few to a few tens of names & values.  Hmmm,
> just occurred to me that, although I tried making local bindings for
> other things, I never did that to dict or zip...
> Okay, definitely need to repoen this can of worms... later.

Well, in order not to post some stupid thoughts I read the whole thread again,
and I think having understood that there may be 2 basic problems:
1) to generate the keys and the values by a special algo what is
   nicely solved by Martin v. Löwis in my opinion
2) optimizing for time as you showed with the dzip-function
   which can be optimized with some effort.
>>> def myzip(keys,values):
... 	res={}
... 	map(lambda x,y:res.setdefault(x,y),keys,values)
... 	return res

   And I think it goes with python 2.1. or earlier.
   A little time_test(number_of_keys) - function which generates
   simple 5-random-letters-keys and a value-list as range(number_of_keys)
   will show an increase in speed.
>>> time_test(50000)
dzip  :  0.28100001812
myzip :  0.140999913216
>>> time_test(500000)
dzip  :  3.31200003624
myzip :  1.59399998188

PS. Though I like the map-function, because it's short and pithy,
    I don't like it very much in situations like this, because it's
    used like a procedure (as in pascal or fortran), but it's a function
    that returns a list, which is thrown in the wind.

More information about the Python-list mailing list