Inline callback functions in python

Question

I'd like to be able to define a python function which does some sort of data structure traversal and performs some operation (defined in a callback) at each step. Here's a simple example to illustrate.

def min_value(list):
   m = 0
   for i in range(0, len(list)):
     if list[i] < m: m = list[i]
   return m

This example is trivial, but becomes more complex if I'm traversing an n-dimension array, a numpy array, or I want to define a more complex method of stepping through the data structure.

I'm looking for a "pythonic" analogue to the following JavaScript which does the same thing.

traverse = function(list, callback) {
   for(i = 0; i < list.length; i++) {
      callback(list[i])
   }
}

min_value = function(list) {
   m = 0
   traverse(list, function(element) {
      if(element < m) m = element
   });
}

If I wanted a function max_value(list), it'd be handy to be able to just adjust the one line to if(element > m) m = element without having to to copy/paste my traversal function. What's the best way to do this in python?

EDIT (THE ANSWER): Here I use the accepted answer below to show how my original JS example could be written in python 3. There are clearly better ways to find the minimum value of a list of scalars, but this is the pattern I was looking for.

def traverse(list, callback):
   for i in range(0, len(list)):
      callback(list[i])

def min_value(list):
   m = 0

   def update_min(val):
      nonlocal m
      if(val < m): m = val

   traverse(list, update_min)
   return m

EDIT (COMPLEX EXAMPLE): Here's a more complex and direct example of my use case. I have an image and I want to do some processing on a "focus region" of the image. Let's say, for example, I want to find the darkest 10x10, 20x20, 30x30, etc pixel regions of the image. My image is a numpy array. I'll use a "bastard" multiline lambda to illustrate what I'd like to do, though as far as I know lambdas cannot be used in this way.

In python:

# find the coordinates and of darkest region
def darkest_region(image, fsize):
   darkest_coords = [0, 0]
   darkest_avg = 256

   # illegal lambda
   g = lambda x, y, r:
      # r is an fsize x fsize region
      avg_darkness = # blah blah, traverse and average pixels
      if(avg_darkness < darkest_avg): darkest_coords = [x, y]

   focus_scan(image, fsize, g)
   return darkest_coords

# scan all regions of fsize x fsize
def focus_scan(image, fsize, callback):
   height, width = image.shape[:2]
   for y in range(0, height - fsize):
     for x in range(0, width - fsize):
        focus = image[y:y+fsize, x:x+fsize]
        callback(x, y, focus)

Given that I can't have a multiline lambda like that, I think what I need to do is get a little fancy with the arguments and return value of focus_scan, in order to accommodate a variety of possible inputs/outputs. In this example, I could pass focus scan darkest_avg and have it return the coordinates I'm looking for, but I might want to pass it something other than a scalar, or have it return a more complex data structure. I imagine I'll have to do something like:

def focus_scan(image, fsize, callback, args):
   ... # loopy loops
      some_result = callback([x, y, focus], args)

   return some_result

I like the inline function approach I can use in other programming languages, but it doesn't look like it's a common pattern in python.

Answer 1

Just use a normal function instead of the multiline lambda. To get access to the darkest_coords variable you can use the nonlocal keyword in Python 3. In Python 2 you have to mutate the list, because the nonlocal keyword is not available.

def darkest_region(image, fsize):
    """Find the coordinates of darkest region."""
    darkest_coords = [0, 0]
    darkest_avg = 256

    def g(x, y, r):
        nonlocal darkest_coords
        # r is an fsize x fsize region
        avg_darkness = # blah blah, traverse and average pixels
        if avg_darkness < darkest_avg:
            darkest_coords = [x, y]

    focus_scan(image, fsize, g)
    return darkest_coords


def focus_scan(image, fsize, callback):
    """Scan all regions of fsize x fsize."""
    height, width = image.shape[:2]
    for y in range(0, height - fsize):
        for x in range(0, width - fsize):
            focus = image[y:y+fsize, x:x+fsize]
            callback(x, y, focus)