Java replace characters with uppercase around (before and after) specific character

Question

I have this kind of input

word w'ord wo'rd

I need to convert to uppercase both characters at the starts of the word and right after the ' character (which can exists multiple times).

The output I need (using the previous example) is

word W'Ord Wo'Rd

I tried with a simple pattern

s.replaceAll("(\\w)(\\w*)'(\\w)", "$1");

but I'm unable to convert the group 1 and 3 to uppercase

---

EDIT: After I discovered a little mistake in the main question, I edited @Wiktor Stribizew code in order to include the case I missed.

Matcher m = Pattern.compile("(\\w)(\\w*)'(\\w)").matcher(s);
StringBuffer result = new StringBuffer();
while (m.find()) {
    m.appendReplacement(result, m.group(1).toUpperCase() + m.group(2) + "'" + m.group(3).toUpperCase());
}
m.appendTail(result);
s = result.toString();

Answer 1

You need to use Matcher#appendReplacement in Java to be able to process the match. Here is an example:

String s = "word w'ord wo'rd";
StringBuffer result = new StringBuffer();
Matcher m = Pattern.compile("\\b(\\w)(\\w*)'(\\w(?:'\\w)*)").matcher(s);
while (m.find()) {
    m.appendReplacement(result, 
        m.group(1).toUpperCase()+m.group(2) + "'" + m.group(3).toUpperCase());
}
m.appendTail(result);
System.out.println(result.toString());
// => word W'Ord Wo'Rd

See the Java demo

Java 9+ equivalent (demo):

String s = "wo'rd w'ord wo'r'd";
Matcher m = Pattern.compile("\\b(\\w)(\\w*)'(\\w(?:'\\w)*)").matcher(s);
System.out.println(
    m.replaceAll(r -> r.group(1).toUpperCase()+r.group(2) + "'" + r.group(3).toUpperCase())
);
//wo'rd w'ord wo'r'd => Wo'Rd W'Ord Wo'R'D
//word w'ord wo'rd => word W'Ord Wo'Rd

Pattern break-down:

• \b - a leading word boundary
• (\w) - Group 1: a single word char
• (\w*) - Group 2: zero or more word chars
• ' - a single quote
• (\w(?:'\w)*) - Group 3:
  • \w - a word char
  • (?:'\w)* - zero or more sequences of:
    • ' - a single quote
    • \w - a word char.

Now, if you want to make the pattern more precise, you can change the \w that are supposed to match lowercase letters with \p{Ll} and the \w that is supposed to match any letter with \p{L}. The pattern would look like "(?U)\\b(\\p{Ll})(\\p{L}*)'(\\p{Ll}(?:'\\p{Ll})*)" - however, you risk to leave letters in lowercase (those after ') if there are uppercase before lowercase ones (like in wo'r'D's -> Wo'R'D's). (?U) is a Pattern.UNICODE_CHARACTER_CLASS inline modifier that makes \b word boundary Unicode-aware.