[Tutor] user defined compare for sorting

[Remco Gerlich]

On  0, Rick Pasotto <rick@niof.net> wrote:
> I have a list of dictionaries. How can I specify at runtime the key to
> use for sorting the list?
> 
> lst[ [ 'A':'a', 'B':'2', 'C':'z' ]
>      [ 'A':'b', 'B':'3', 'C':'x' ]
> 	 [ 'A':'c', 'B':'1', 'C':'y' ] ]
> 
> How can I pass 'A', 'B', or 'C' to 'key' in the compare function?
> 
> def mycmp(a,b): cmp(a[key],b[key])

(The above isn't correct, I assume you meant { curly braces } around the
dicts, and the mycmp needs a "return").

I understand you want to be able to sort the list on different keys.

Make a function that returns a function that sorts on some key:

def make_sort_function(key):
   def sort_func(a, b, key=key):
      return cmp(a[key], b[key])
   return sort_func

cmp_func works as we want it to work because it gets key as a default
argument.

Now you can do, say,
lst.sort(make_sort_function('A'))

'lambda' is a way to make functions on the fly. The same thing could like
like this:
lst.sort(lambda a,b,key='A': cmp(a[key],b[key]))

It's up to you to choose the one you find easier to understand.

If you don't understand how this works, play around with make_sort_function
in the interpreter...

-- 
Remco Gerlich