Can Python have support for number separators?

I see that when I call something like int(2,345,565), it gives an error. Interestingly, in Java, this is supported. Is it a good idea to have something like this in Python?

You’re a few years too late:

% python
Python 3.11.1 (main, Dec 21 2022, 16:19:04) [Clang 14.0.0 (clang-1400.0.29.202)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> 2_345_678

Added in Python 3.6 by PEP 515:


Ouch :joy::joy::joy::joy::grin::grin:. I never knew this. I was revisiting Java and the thought crossed my mind. It’s good that I am now getting ideas to contribute to open source though. Lol

Thanks a bunch

1 Like

How does the Java code look like? I was curious and googled it but can’t find it.

I can’t find any evidence that Java supports any kind of thousands separator in the literal syntax for integers. Of course it’s more than capable of parsing a string which contains thousands separators to create an integer (or floating-point) value.

Excuse me for chiming in. But since @Daveson217 specifically asked for a more natural number conversion to int using thousands comma separators (in the west and period for European based system), is there a number conversion built-in to int not using underscores whose syntax is not generally natural ?

*** Update ***

I suppose you can do your own conversion manually assuming you enter the number as a string, however:

from re import sub

def str_to_int(num):
    remv_comma = sub(',', '', num)
    return int(remv_comma)

# test function    
add_nums = 123123 + str_to_int('4,123,123')

1 Like

You are right @kknechtel.
However, the code I speak of is in the Java source code for processing input, Scanner; Python’s equivalent of the input function.

See below:

import java.util.Scanner;

public class AskAge {
    public static void main(String[] args) {
        Scanner sc = new Scanner(;
        System.out.println("Enter your age");
        int age = sc.nextInt();

Then somewhere down the rabbit hole of calling various methods (starting from the sc.nextInt(), the following method is called:

private String processIntegerToken(String token) {
        String result = token.replaceAll(""+groupSeparator, "");
        boolean isNegative = false;
        int preLen = negativePrefix.length();
        if ((preLen > 0) && result.startsWith(negativePrefix)) {
            isNegative = true;
            result = result.substring(preLen);
        int sufLen = negativeSuffix.length();
        if ((sufLen > 0) && result.endsWith(negativeSuffix)) {
            isNegative = true;
            result = result.substring(result.length() - sufLen,
        if (isNegative)
            result = "-" + result;
        return result;

From the docs, it says:

The integer token must be stripped of prefixes, group separators, and suffixes, non ascii digits must be converted into ascii digits before parse will accept it.

cc: @Stefan2

This works perfectly as well. @onePythonUser

Not really.

input in Python gets a string, no matter what is typed. It’s the equivalent of the readline method on a DataInputStream. You need to use other code to interpret the contents of the string to get other values of other types. java.util.Scanner is a completely different interface. You can make something work like that in Python, but it isn’t built-in and isn’t the normal way to do things in Python. Especially not when reading input from the user - because of how the terminal works, it will naturally be much simpler and less error-prone to read the input a line at a time consistently and do separate parsing on each line.

If you want to parse a string that is expected to represent a number that uses a thousands separator, see:


In addition to the locale module, you may want to look into the babel package, which provides more advanced facilities for handling specific languages and cultures.

1 Like

Thanks for letting me know @kknechtel.

I have not used the locale package before.