Different behavior of same regular expression in Python and Java

Question

Firstly, my apologies as I don't know regular expressions that well.

I am using a regular expression to match a string. I tested it in Python command line interface, but when I ran it in Java, it produced a different result.

Python execution:

re.search("[0-9]*[\\.[0-9]+]?[^0-9]*D\\([M|W]\\)\\s*US", "9.5 D(M) US");

gives the result as:

<_sre.SRE_Match object; span=(0, 11), match='9.5 D(M) US'>

But the Java code

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

class RegexTest {
    private static final Pattern FALLBACK_MEN_SIZE_PATTERN = Pattern.compile("[0-9]*[\\.[0-9]+]?[^0-9]*D\\([M|W]\\)\\s*US");

    public static void main(String[] args) {
    String strTest = "9.5 D(M) US";
    Matcher matcher = FALLBACK_MEN_SIZE_PATTERN.matcher(strTest);
        if (matcher.find()) {
            System.out.println(matcher.group(0));
        }
    }
}

gives the output as:

5 D(M) US

I don't understand why it is behaving the different way.

Note that you can ditch the extra backslashes in Python with a "raw string* r'[0-9]*[\.[0-9]+]?...', and that you can use \d for [0-9]. — jonrsharpe
– jonrsharpe, Commented May 29, 2015 at 10:27
Well, the regex definitely needs some adjustment. You put alternatives into a character class rather than in a group. — Wiktor Stribiżew
– Wiktor Stribiżew, Commented May 29, 2015 at 10:28
Also the pipe in [M|W] is a literal character to match... have a look at e.g. regex101.com/r/kT9fD4/1 — jonrsharpe
– jonrsharpe, Commented May 29, 2015 at 10:30

Wiktor Stribiżew · Accepted Answer · 2015-05-29 10:56:50Z

5

Here is the pattern that will work the same in Java and Python:

"[0-9]*(?:\\.[0-9]+)?[^0-9]*D\\([MW]\\)\\s*US"

See Python and Java demos.

In Python, [\\.[0-9]+]? is read as 2 subpatterns: [\.[0-9]+ (1 or more .s, [s, or digits) and ]? (0 or 1 ]). See how your regex works in Python here. Or, with more detalization with capturing groups, here.

In Java, it is read as one single character class (i.e. [ and ] inside are ignored as they cannot be parsed correctly by the regex engine, thus the whole subpattern standing for 0 or 1 ., a digit, or +) and since it is optional, it was not capturing anything (you can get a visual hint at Visual Regex Tester, type 123.+[] as input and [\.[0-9]+]? as regex).

And a final touch: [M|W] stands for M, |, or W, while I think you meant [MW] = M or W.

edited May 29, 2015 at 10:56

answered May 29, 2015 at 10:48

Wiktor Stribiżew

631k41 gold badges502 silver badges632 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

RealSkeptic · Accepted Answer · 2015-05-29 12:22:19Z

1

I'm not a Python expert, so I can't tell you why it worked on Python, but in Java, your problem is the [\\.[0-9]+]? part. You probably meant it to be (\\.[0-9]+)?.

As it is, it's a list of characters inside a [] followed by a ?. That is, this part of the expression only matches a single or zero character, so it cannot match the .5 together.

Here is an illustration of the matching attempts:

Graphical demonstration of matching in Java

Now, if your pattern used () instead of [], this would be the result:

enter image description here

edited May 29, 2015 at 12:22

answered May 29, 2015 at 10:45

RealSkeptic

34.8k7 gold badges56 silver badges83 bronze badges

Collectives™ on Stack Overflow

Different behavior of same regular expression in Python and Java

2 Answers 2

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related