Wikimedia Bugzilla is closed!

Code says:

                # Context links: [[|name]] and [[name (context)|]]
                $tc = '[' . Title::legalChars() . ']';
                $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!

                # ... omitted ...

                $t = $this->mTitle->getText();
                $m = array();
                if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
                        $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
                } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
                        $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
                } else {
                        # if there's no context, don't bother duplicating the title
                        $text = preg_replace( $p2, '[[\\1]]', $text );
                }

In 'if' and 'elseif' line, regexes begins with ($nc+:|) so I guess the original author intended to include namespace name automatically, but $t = $this->mTitle->getText(); should be written as $t = $this->mTitle->getPrefixedText(); in this case.