Question

Real world application: categorize bytes into the categories: control, printable, non-printable character (category list will be longer)

I have a list of numbers:

numbers = [1, 1, 2, 3, 3, 3, 3, 4]

Now I want to put them into different categories. To do so, I have to define in which category a number belongs. So far I have two approaches, both using a predefined index/value pair.

category_list = ["apple", "apple", "banana", "melon", "melon", "melon"]
category_dict = {1 : "apple", 2 : "apple", 3 : "banana", 4 : "melon", 5 : "melon", 6 : "melon"}
for number in numbers:
    print category_list[number]
    category_dict[number]

Another option would be a list for every category. This is eventually faster to write/implement but forces me to brute-force the dictionary (see one of the answers):

dict_category = {
    apple : [1, 2],
    banana : [3,],
    melon : [4, 5, 6]
}
for number in numbers:
    for key, val in dict_category.iteritems():
        if number in val:
            print key
            break

Is there a better, more pythonic way to do this? Maybe which doesn’t require me to write a list/dict with 256 entries?

Answer 1

# configuration dict, written by the user
categories = {
    'apple'  : [1, 2],
    'banana' : [3],
    'melon'  : [4, 5, 6]
}

# dynamically generate lookup table
lookup = {}
for cat, nums in categories.iteritems():
    for n in nums: lookup[n] = cat

# ... later:
from collections import defaultdict
categorized = defaultdict(list)
for n in numbers:
    cat = lookup.get(n)
    if not cat:
        continue
    categorized[cat].append(n)
    print "put %d in category %s" % (n, cat)

If your numbers are within a certain range, you could also use a vector for lookup:

lookup = [None]*256
for cat, nums in categories.iteritems():
    for n in nums: lookup[n] = cat

# ... later
categorized = defaultdict(list)
for n in numbers:
    categorized[lookup[n]].append(n)
    print "put %d in category %s" % (n, lookup[n])