Python sorting str price with two decimal points

Question

My goal: Sort a list of Products (dict) first by Price, then by Name. My problem: Str values with numbers in them aren't sorted properly (AKA "Human sorting" or "Natural Sorting").

I found this function from a similar question: Python sorting list of dictionaries by multiple keys

def multikeysort(items, columns):
    from operator import itemgetter
    comparers = [((itemgetter(col[1:].strip()), -1) if col.startswith('-') else
                  (itemgetter(col.strip()), 1)) for col in columns]
    def comparer(left, right):
        for fn, mult in comparers:
            result = cmp(fn(left), fn(right))
            if result:
                return mult * result
        else:
            return 0
    return sorted(items, cmp=comparer)

The problem is that my Prices are str type, like this:

products = [
    {'name': 'Product 200', 'price': '3000.00'},
    {'name': 'Product 4', 'price': '100.10'},
    {'name': 'Product 15', 'price': '20.00'},
    {'name': 'Product 1', 'price': '5.05'},
    {'name': 'Product 2', 'price': '4.99'},
]

So they're getting sorted alphabetically, like this:

'100.10'
'20.10'
'3000.00'
'4.99'
'5.05'

Similarly, when I sort by name, I get this:

'Product 1'
'Product 15'
'Product 2'
'Product 200'
'Product 4'

The names should be listed in "human" order (1,2,15 instead of 1,15,2). Is it possible to fix this? I'm pretty new to python, so maybe I'm missing something vital. Thanks.

EDIT

More Info: I'm sending the list of products to a Django template, which requires the numbers to be properly formatted. If I float the prices and then un-float them, I have to iterate through the list of products twice, which seems like overkill.

Do your prices have to be strings? The best thing to do would be to cast them to floats — Tyler, Aug 08 '16 at 23:24
`int` type removes the decimal places ('1.00' = '1'), and `float` removes one decimal ('1.00' = '1.0') which doesn't look good in the shopping cart! — ThePloki, Aug 08 '16 at 23:26

score 3 · Answer 1 · answered Aug 08 '16 at 23:32

3

Your sort function is overkill. Try this simple approach:

from pprint import pprint

products = [
    {'name': 'Product 200', 'price': '3000.00'},
    {'name': 'Product 4', 'price': '100.10'},
    {'name': 'Product 15', 'price': '20.00'},
    {'name': 'Product 1', 'price': '5.05'},
    {'name': 'Product 2', 'price': '4.99'},
]

sorted_products = sorted(products, key=lambda x: (float(x['price']), x['name']))
pprint(sorted_products)

Result:

[{'name': 'Product 2', 'price': '4.99'},
 {'name': 'Product 1', 'price': '5.05'},
 {'name': 'Product 15', 'price': '20.00'},
 {'name': 'Product 4', 'price': '100.10'},
 {'name': 'Product 200', 'price': '3000.00'}]

The essence of my solution is to have the key function return a tuple of the sort conditions. Tuples always compare lexicographically, so the first item is the primary sort, the second is the secondary sort, and so on.

answered Aug 08 '16 at 23:32

Robᵩ

163,533
20
239
308

I like this! Only problem is that the sort function I posted allows for an unlimited amount of sort columns and negative sorting. Is it easy to allow for those two things using your method? – ThePloki Aug 08 '16 at 23:37
1

Also, this doesn't fix the second problem of sorting by 'name'. – ThePloki Aug 08 '16 at 23:44
Yes, `sorted()` supports an arbitrary number of columns and negative sorting. – Robᵩ Aug 08 '16 at 23:48
What is the problem of sorting by `name`? In your question, you describe the result of sorting by name, but you don't describe what problem you perceive. – Robᵩ Aug 08 '16 at 23:49
Updated my question to reflect the name problem. They should be listed in "human" order (1,2,15 instead of 1,15,2). – ThePloki Aug 08 '16 at 23:55
What do we know about the format of the product names? Will they always start with the literal word "Product" followed by a space? – Robᵩ Aug 09 '16 at 00:21
No, the user can write anything at all in there. – ThePloki Aug 09 '16 at 00:27

score 2 · Answer 2 · answered Aug 08 '16 at 23:31

2

I think your best bet is to parse the prices as floats (so you can sort them):

float("1.00")
# output: 1.0

Then output them with two decimal places:

"{:.2f}".format(1.0)
# output: "1.00"

answered Aug 08 '16 at 23:31

Tyler

17,669
10
51
89

If I do it this way then I need to iterate through all of the results after sorting them just to format it correctly. I'll review my code and see if I can minimize iteration. Thanks. – ThePloki Aug 08 '16 at 23:38

score 1 · Answer 3 · answered Aug 08 '16 at 23:31

1

Try typecasting them to floats in the question and when you need to print 2 decimal places, you can easily format the output like so:

float_num = float("110.10")
print "{0:.2f}".format(float_num) # prints 110.10

answered Aug 08 '16 at 23:31

Vaibhav Bajaj

1,934
16
29

score 0 · Answer 4 · answered Aug 09 '16 at 00:17

To break ties should there be any sorting the strings using the integer value from the product, you can return a tuple:

products = [
    {'name': 'Product 200', 'price': '2.99'},
    {'name': 'Product 4', 'price': '4.99'},
    {'name': 'Product 15', 'price': '4.99'},
    {'name': 'Product 1', 'price': '9.99'},
    {'name': 'Product 2', 'price': '4.99'},
]

def key(x):
    p, i = x["name"].rsplit(None, 1)
    return float(x["price"]), p, int(i)

sorted_products = sorted(products, key=key)

Which would give you:

[{'name': 'Product 200', 'price': '2.99'},
 {'name': 'Product 2', 'price': '4.99'},
 {'name': 'Product 4', 'price': '4.99'},
 {'name': 'Product 15', 'price': '4.99'},
 {'name': 'Product 1', 'price': '9.99'}]

As opposed to:

[{'name': 'Product 200', 'price': '2.99'},
 {'name': 'Product 15', 'price': '4.99'},
 {'name': 'Product 2', 'price': '4.99'},
 {'name': 'Product 4', 'price': '4.99'},
 {'name': 'Product 1', 'price': '9.99'}]

using just float(x['price']), x['name']

Python sorting str price with two decimal points

4 Answers4