Java regular expression to validate numeric comma separated values

Question

I need to make a regular expression with such numbers valid:

 "+1", "1.0", "1,233", "1,233,456.34", "-1", ".34", "1,345,234,122,123"

and these invalid:

 "++1", "1.0.0", "1,23,3", "+-1233456.34", "002", "1.", "a1", "1,,2", "1 2", "1,2", ",2".

I've tried different variants of this regexp:

 "[\\+\\-]?[1-9]{0,3}([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\.][\\d]*)?"

Code for testing:

import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class REGEX {
private static final String REGEX = "[\\+\\-]?[1-9]{0,3}([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\.][\\d]*)?";

private static String[] validNumbers = { "+1", "1.0", "1,233",
        "1,233,456.34", "-1", ".34", "1,345,234,122,123" };

private static String[] invalidNumbers = { "++1", "1.0.0", "1,23,3",
        "+-1233456.34", "002", "1.", "a1", "1,,2", "1 2", "1,2", ",2" };

public static void main(String[] args) {
    Pattern pattern = Pattern.compile(REGEX);

    for (String number : validNumbers) {
        Matcher matcher = pattern.matcher(number);
        if (!matcher.matches()) {
            System.out.println("Valid number is detected as invalid: "
                    + number);
        }
    }
    for (String number : invalidNumbers) {
        Matcher matcher = pattern.matcher(number);
        if (matcher.matches()) {
            System.out.println("Invalid number is detected as valid: "
                    + number);
        }
    }
}

}

When the console will be empty then the task is done. Now I've such problems:

Valid number is detected as invalid: 1,233

Valid number is detected as invalid: 1,233,456.34

Valid number is detected as invalid: 1,345,234,122,123

Invalid number is detected as valid: 1.

Regards. Sorry for big size.

Update. Thanks to Noob UnChained, I've progressed to this regexp:

^([\\+\\-]?[1-9]\\d{0,2})*(\\,\\d{3})*([\\.][\\d*])?$

and now there is fewer problems:

Valid number is detected as invalid: 1,233,456.34

Valid number is detected as invalid: .34

Update.

regexp:

"([\\+\\-]?[1-9]\\d{0,2})*(\\,\\d{3})*([.][\\d]*)?"

problems:

Invalid number is detected as valid: 1.

FINISHED

The final result is:

"(?!$)[\\+-]?([1-9]\\d{0,2}|0)?(\\,\\d{3})*(\\.\\d+)?"

Change since the last update is:([.][\d]+)?

plus in this block makes impossible situation when user inputs number then puts dot but nothing after it.

New update: added 2 "\" into last block to avoid entering comma or other symbol instead of dot.

New update: Thanks to user2266098 and nhahtdh.

user2266098 pointed my attention on the uncharted problem of "0.1" number and showed solution with adding "|0" to the second block. But his regexp doesn't work correctly with "+" and "-" for my data (because of "()" instead of "[]"). And I don't like the quantifier "{0,}" instead of "*" because of it's size.

nhahtdh pointed my attention on the uncharted problem of empty string and showed solution with "(?!$)".

Thanks to everyone!

Update

There are new conditions for this case:

I need to make a regular expression with such numbers valid:

"+1", "1.0", "1,233", "1,233,456.34", "-1", ".34", "1,345,234,122,123", "0.1"

and these invalid:

"++1", "1.0.0", "1,23,3", "+-1233456.34", "002", "1.", "a1", "1,,2", "1 2", "1,2", ",2", "", "0,123"

I still can't get "perfect" regexp =)

"(?!$)[\\+-]?([1-9]\\d{0,2}|0)?(\\,\\d{3})*(\\.\\d+)?"

gives: Invalid number is detected as valid: 0,123

This may be of help : http://stackoverflow.com/questions/2811031/decimal-or-numeric-values-in-regular-expression-validation — AllTooSir, Apr 16 '13 at 15:55
You can replace `([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?([\\d]{3}[\\,]{1})?` with `(\\d{3}\\,){0,5}` (they're the same). — Bernhard Barker, Apr 16 '13 at 15:59
Do you need to do this with a regex? There are methods that you can call to do this. — Rob Watts, Apr 16 '13 at 16:01
@YaroslavSelivanov You are very close to achieve. For `.34` replace `^([\\+\\-]?` with `^([+-\\.]?` — Smit, Apr 16 '13 at 16:19
Thanks. Sorry for simple questions, I'm a newbie to this site. — Yaroslav Selivanov, Apr 16 '13 at 16:30

Yaroslav Selivanov · Answer 1 · 2013-04-17T11:57:32.243

FINISHED

The final result is:

"(?!$)[\\+-]?([1-9]\\d{0,2}|0)?(\\,\\d{3})*(\\.\\d+)?"

Change since the last update is:([.][\d]+)?

plus in this block makes impossible situation when user inputs number then puts dot but nothing after it.

New update: added 2 "\" into last block to avoid entering comma or other symbol instead of dot.

New update: Thanks to user2266098 and nhahtdh.

user2266098 pointed my attention on the uncharted problem of "0.1" number and showed solution with adding "|0" to the second block. But his regexp doesn't work correctly with "+" and "-" for my data. And I don't like the quantifier "{0,}" instead of "*" because of it's size.

nhahtdh pointed my attention on the uncharted problem of empty string and showed solution with "(?!$)".

Thanks to everyone!

Update

There are new conditions for this case:

I need to make a regular expression with such numbers valid:

"+1", "1.0", "1,233", "1,233,456.34", "-1", ".34", "1,345,234,122,123", "0.1"

and these invalid:

"++1", "1.0.0", "1,23,3", "+-1233456.34", "002", "1.", "a1", "1,,2", "1 2", "1,2", ",2", "", "0,123"

I still can't get "perfect" regexp =)

"(?!$)[\\+-]?([1-9]\\d{0,2}|0)?(\\,\\d{3})*(\\.\\d+)?"

gives: Invalid number is detected as valid: 0,123

score 3 · Answer 2 · answered Apr 16 '13 at 16:50

3

This one works for me on all your test data:

^(\\+|-)?([1-9]\\d{0,2}|0)?(,\\d{3}){0,}(\\.\\d+)?

btw I expect you also want 0.1 to be a match, but your regexp doesn't work on it. I can't comment, so I'm writing this in an answer

answered Apr 16 '13 at 16:50

NeplatnyUdaj

6,052
6
43
76

1

Your regex will allow empty string. – nhahtdh Apr 16 '13 at 17:27
@user2266098 Did you lost one symbol in the quantifier? ^(\\+|-)?([1-9]\\d{0,2}|0)?(,\\d{3})**{0,}**(\\.\\d+)? – Yaroslav Selivanov Apr 16 '13 at 17:31
@nhahtdh Is there any option to avoid this? – Yaroslav Selivanov Apr 16 '13 at 17:37
Yep... will allow empty string. Anyway the question was not about perfect regexp, but sadly just to pass the homework test. Yaroslav: I didn't miss the quantifier. There can be infinite number of (\,\d{3}) groups I think. – NeplatnyUdaj Apr 16 '13 at 17:41
1

@YaroslavSelivanov: I haven't looked at your question in details, so I don't know. A crude and dirty fix is to put an assertion `(?!$)` at the beginning. – nhahtdh Apr 16 '13 at 17:44
@user2266098 Yes, infinite number but why not to use "*"? – Yaroslav Selivanov Apr 16 '13 at 17:51
@nhahtdh Why you call this solution crude and dirty? It works but i'd like to know more. – Yaroslav Selivanov Apr 16 '13 at 17:53
@user2266098 "()" in the first block aren't working correctly with my data. – Yaroslav Selivanov Apr 16 '13 at 17:56
yaroslav-selivanov: I used data from the question and it works on those. {0,} is the same as *. I don't remember why I did that :) – NeplatnyUdaj Apr 16 '13 at 19:27
@YaroslavSelivanov: Well, I personally consider it as last resort, when there is no better way to write it. – nhahtdh Apr 16 '13 at 19:37
@nhahtdh: This post is still relevant on 2022. I get that `(?!$)` is used to check for empty string. But, will you please explain briefly the logic behind it? – priyamtheone Mar 07 '22 at 16:00

Java regular expression to validate numeric comma separated values

2 Answers2