Question

I have a simple line of text which might include numbers like “12.3” or “1983” or “5/8”.
Whenever any number appears, I just need to replace with a fixed character, say the digit “8”.

I’ve been fiddling about with Regex in Java, with things like this:

String line = str.replaceAll("[0-9]+/*.*[0-9]*", "8");

but to no avail.

Any idea what the correct pattern should be?

Answer 1

Try this expression: (?>-?\d+(?:[\./]\d+)?), keep in mind that in Java strings you need to escape the backslashes, i.e. you’d get "(?>-?\\d+(?:[\\./]\\d+)?)"

Here’s a breakdown of the expression:

1. The encloseing (?>...) is an atomic group to prevent catastrophic backtracking. For simple or short strings it would work without as well.

2. -? a potential minus for negative numbers

3. \d+ any sequence of digits (at least one)

4. (?:[\./]\d+)? an optional non-capturing group consisting of either a dot (note that you don’t need to escape it here, it’s just for consistency) or a slash followed by at least one more digit.

Update

If you don’t want to replace “numbers” like .1234, 1234. /1 or 5/ (a digit is missing either left or right), try this expression: (?>(?<![\d\./])-?\d+(?:(?:[\./]\d+)|(?![\d\./])))

Here’s a breakdown again:

1. The encloseing (?>...) is an atomic group to prevent catastrophic backtracking. For simple or short strings it would work without as well.

2. (?<![\d\./]) the match must not directly follow a digit, dot or slash – note that the not follow a digit constraint is needed to match at the start of the number, otherwise you’d match 234 in .1234

3. -? a potential minus for negative numbers

4. \\d+ any sequence of digits (at least one)

5. (?:(?:[\./]\d+)|(?![\d\./])) the match must either have a dot or slash followed by at least one digit or must not be followed by a digit, dot or slash, this would match 1.0 but not 1. – note that the not to be followed by a digit constraint is needed to prevent matching 123 in 1234.