The Eighty-Twenty Solution

I've been doing some Python scripting this week, and discovered that you can put if filters on list comprehensions. I love list comprehensions to begin with and this makes them even cooler. For example, to filter only numbers out of a list, I defined isNumberish based on recipe 3.2 in the Python Cookbook, and then wrote a filter like this:

 def isNumberish(anobj): 
 ... 	try: anobj + 0
 ... 	except: return 0
 ... 	else: return 1
  [x * x for x in ['a', 1, 'b', 3, 'c', 7] if isNumberish(x)]
 [1, 9, 49]
 

In Scheme, SRFI-1 provides filter-map to do effectively the same thing. It's a bit less clear what goes on here; filter-map accumulates only true values (it might be more to the point to say anything not #f).

 (require (lib "1.ss" "srfi"))
  (filter-map (lambda (x) (and (number? x) (* x x))) '("a" 1 "b" 3 "c" 7))
 (1 9 49)
 

filter-map could get you in a bit of trouble if you wanted to work with booleans:

 (filter-map (lambda (x) (and (boolean? x) (not x))) '("a" 1 "b" #t "c" #f))
 (#t)
  (map not (filter boolean? '("a" 1 "b" #t "c" #f)))
 (#f #t)
 

Python can handle this OK, though:

 [not x for x in ['a', 1, 'b', True, 'c', False] if type(x) == type(True)]
 [False, True]