Duncan Booth wrote:
> [EMAIL PROTECTED] wrote:
>
> > As for the (k,v) vs (v,k), I still don't think it is a good example. I
> > can always use index to access the tuple elements and most other
> > functions expect the first element to be the key. For example :
> >
> > a=d.items()
> > do something about a
> > b = dict(a)
> >
> > But using the imaginary example :
> >
> > a = zip(d.values(), d.keys())
> > do something about a
> > b = dict(a) # probably not what I want.
> >
>
> The typical use case for the (v,k) tuple would be when you want sort the
> contents of a dictionary by value. e.g. counting the number of occurrences
> of each word in a document.
>
> a = zip(d.values(), d.keys())
> a.sort()
> a.reverse()
> print "Today's top 10:"
> print str.join(',', [ k for (v,k) in a[:10]])
>
> Ok, so today you can do that another way, but before there was a key
> parameter to sort you had to build the tuple somehow:
>
> print str.join(',', sorted(a.iterkeys(),
> key=a.__getitem__, reverse=True)[:10])
>
> and the advantage of the latter is of course that it works even when you've
> been counting the number of occurences of complex numbers in a document :)
Thanks, so for newer python(2.3+?, don't know when the sort has been
enhanced), this (v,k) is again there for historical reason but not a
real use case any more.
--
http://mail.python.org/mailman/listinfo/python-list
