\s in the regexp destroys some UTF-8 characters in pingback_ping()

[tenpura]

\s in the regexp destroys some UTF-8 characters in pingback_ping(). Same issue as in #21625.

Steps to reproduce:

1. Pingback with the post title "САПР".
2. It will create a pingback comment with no comment author (Anonymous).

Solution:

Use [\r\n\t ] rather than [\s\r\n\t].

[SergeyBiryukov]

Introduced in [2619].

[azaozz]

Yeah, we shouldn't be using \s in regex that filters user submitted or translatable text as it matches bytes that are part of some multibyte (UTF-8, others) chars. The [\r\n\t ] replacement seems to work properly. In theory best would be to use the u ​modifier, however that leaves installs with charset other than UTF-8 out in a "grey area".

[tenpura]

Replying to azaozz:

In theory best would be to use the u ​modifier, however that leaves installs with charset other than UTF-8 out in a "grey area".

It seems that PCRE is not always compiled with UTF-8 support. That is another drawback of using the u modifier.

[SergeyBiryukov]

In 23952:

Remove \s from regex in pingback_ping() to avoid UTF-8 issues. props tenpura. fixes #24001.