bug#60621: grep -P does not set PCRE2_UCP

[Karl Pettersson]

Hi

Using grep -P for boundary matches yields incorrect results with
non-ASCII letters:

$ echo 'Öst' | grep -P '\bs'
Öst

The output should be nothing in this case, and the culprit seems to be
this line in pcresearch.c:

      flags |= PCRE2_UTF;

If the PCRE2_UCP flag is added according to this, the program behaves
correctly:

      flags |= PCRE2_UTF|PCRE2_UCP;

The pcre2grep test program in the pcre2 has the same problem, and I
filed an issue there too:

https://github.com/PCRE2Project/pcre2/issues/185

A Twitter discussion with more examples:

https://twitter.com/gro_tsen/status/1610972356972875777

Kind regards
-- 
Karl Pettersson
Uppsala, Sverige/Sweden

https://static-dust.klpn.se/