python/numpy versatile max function -

in python (possibly including numpy) there various maximum, max , amax functions, extent covered in stackoverflow questions, none of them seems solve rather obvious need of computing the maximum of mixed arguments, i.e. matching following requirements:

• any argument iterable (tuple, list, numpy array) or number
• ... , recursively, i.e. iterable in turn contain numbers , iterables
• iterables turning out empty, , none items, should not affect max
• only if arguments empty output should given default
• if arguments contain nan, should propagated (i think automatic)

i seeking solution versatile, compact , proper pythonic point of view.

nb. tried code myself little success. not attach many attempts because ugly, , don't work either, , above all, believe misleading , time waste of possibly can help. question long.

anyway, here 1 of attempts, doesn't work:

<!-- language: python -->  def flexmax1(aa):     '''     works on 1 argument, breaks into:     - empties: eliminated     - single numbers: replace max (or none)     - iterables (if any): deal recursively, don't forget add prev calc max     '''     # progressively replaces iterables max (or none if no iterables)     m=none     while aa:         if m not none:             aa=aa.append(m)         # eliminates empties , none, preserves true zeros         aa= [a in aa if or a==0]          try:             # max of non iterables             m=max( in aa if not iterable(a) )         except valueerror: # empty sequence              # there weren't bare numbers, iterables, reduce them             aa=[ flexmax1(a) in aa if iterable(a) ]         else:             # same, append max of iterables             aa=[ (flexmax1(a)) in aa if iterable(a) ]      return m  if __name__ == '__main__':     print(flexmax1([1]))     print(flexmax1([1,2]))     print(flexmax1([1,[2,3]])) 

ps. know that, standpoint, maximum of items should 1 of items, , solved using max's key=function optional argument. need return overall maximum number

you can first flatten down , compute max of that:

import collections  def flatted(data):     if isinstance(data, collections.iterable):         element in data:             yield flatted(data)     else:         yield data  def versatile_max(data, max_function=max):     max_function(flatted(data))