Regexp with word-boundary followed by unicode character doesn't work in 19, 21

[Raffaello Giulietti]

By default, a word boundary only considers ASCII letters and digits. See 
"Predefined character classes" in the documentation.

To add Unicode support, you have a choice between adding a flag as a 2nd 
argument to the compile() method

Pattern p = Pattern.compile("(\\b" + word + "\\b)", 
Pattern.UNICODE_CHARACTER_CLASS);

or add a flag in the regex pattern, as documented in "Special constructs 
(named-capturing and non-capturing)"

Pattern p = Pattern.compile("(?U)(\\b" + word + "\\b)");


Greetings
Raffaello


On 2023-12-15 20:07, Stefan Norberg wrote:
> The following test works in 17 but fails in 19.0.2, and 21.0.1 on the 
> last assertion. Bug or feature?
> 
> import org.junit.jupiter.api.Assertions;
> import org.junit.jupiter.api.Test;
> 
> import java.util.ArrayList;
> import java.util.regex.Matcher;
> import java.util.regex.Pattern;
> 
> /**
> * Tests passes in JDK 17 but fails in JDK 19, 21.
> *
> * The combination of a \b "word boundary" and a unicode char doesn't 
> seem to work in 19, 21.
> *
> */
> public class UnicodeTest {
> @Test
> public void testRegexp() throws Exception {
> var text = "En sak som ökas och sedan minskas. Bra va!";
> var word = "ökas";
> Assertions.assertTrue(text.contains(word));
> 
> Pattern p = Pattern.compile("(\\b" + word + "\\b)");
> Matcher m = p.matcher(text);
> var matches = new ArrayList<>();
> 
> while (m.find()) {
> String matchString = m.group();
> System.out.println(matchString);
> matches.add(matchString);
> }
> Assertions.assertEquals(1, matches.size());
> }
> }
> 
> 
> 
> openjdk version "21.0.1" 2023-10-17 LTS
> 
> OpenJDK Runtime Environment Corretto-21.0.1.12.1 (build 21.0.1+12-LTS)
> 
> OpenJDK 64-Bit Server VM Corretto-21.0.1.12.1 (build 21.0.1+12-LTS, 
> mixed mode, sharing)
> 
> 
> System Version: macOS 14.2 (23C64)
> 
> Kernel Version: Darwin 23.2.0
> 
> 
> Thanks!
> 
> 
> /Stefan
>