Regular expression removing all words shorter than n

Question

Well, I'm looking for a regexp in Java that deletes all words shorter than 3 characters. I thought something like \s\w{1,2}\s would grab all the 1 and 2 letter words (a whitespace, one to two word characters and another whitespace), but it just doesn't work. Where am I wrong?

Answer 1

I've got it working fairly well, but it took two passes.

public static void main(String[] args) {
    String passage = "Well, I'm looking for a regexp in Java that deletes all words shorter than 3 characters.";
    System.out.println(passage);

    passage = passage.replaceAll("\\b[\\w']{1,2}\\b", "");
    passage = passage.replaceAll("\\s{2,}", " ");

    System.out.println(passage);
}

The first pass replaces all words containing less than three characters with a single space. Note that I had to include the apostrophe in the character class to eliminate because the word "I'm" was giving me trouble without it. You may find other special characters in your text that you also need to include here.

The second pass is necessary because the first pass left a few spots where there were double spaces. This just collapses all occurrences of 2 or more spaces down to one. It's up to you whether you need to keep this or not, but I think it's better with the spaces collapsed.

Output:

Well, I'm looking for a regexp in Java that deletes all words shorter than 3 characters.

Well, looking for regexp Java that deletes all words shorter than characters.

Answer 2

If you don't want the whitespace matched, you might want to use

\b\w{1,2}\b

to get the word boundaries.

That's working for me in RegexBuddy using the Java flavor; for the test string

"The dog is fun a cat"

it highlights "is" and "a". Similarly for words at the beginning/end of a line.

You might want to post a code sample.

(And, as GameFreak just posted, you'll still end up with double spaces.)

EDIT:

\b\w{1,2}\b\s?

is another option. This will partially fix the space-stripping issue, although words at the end of a string or followed by punctuation can still cause issues. For example, "A dog is fun no?" becomes "dog fun ?" In any case, you're still going to have issues with capitalization (dog should now be Dog).

Answer 3

Try: \b\w{1,2}\b although you will still have to get rid of the double spaces that will show up.