Fast sorting many lists together...?

[Alex Martelli]

"William Park" <parkw at better.net> wrote in message
news:20000915134034.B2932 at better.net...
> Hi,
>
> I need to sort about 50 lists (each about 500 elements) together, using
few
> of them as keys.  That is, sort the key lists first; look at which indexes
> have changed; then, shuffle the rest of list to reflect this new order.
>
> Does anyone know a fast fast routine for doing this?

If I understand you correctly, what you want is very well offered
by Numerical Python (which is not only for numerical stuff: its
arrays can hold generic Python objects).  If you do need to have
your stuff in lists, then moving it back and forth list<->array
might be slow, though.

So, as mentioned on another thread (yesterday, I think): zip
the list your sorting with the indices, then use those.  It's easier
in Python 2.0, currently in beta:

def sortone_changemany(firstlist, *theothers):
    zipped = zip(firstlist, range(len(firstlist)))
    zipped.sort()
    results = [ [i for i, j in zipped] ]
    indices = [j for i, j in zipped]
    for alist in theothers:
        results.append([alist[j] for j in indices])
    return results

With older Pythons, you can use map instead of zip, and more
explicit loops instead of the list comprehensions, but the
general idea is similar.


Alex