Wikimedia Bugzilla is closed!

At some point fairly recently, the behavior of the Message Class parse function change, so that it now leaves extraneous HTML after the parsed text. This creates issues with output created by core Mediawiki functions that use this functionality.

Example: Parsing a message key that should result in the word "Contact" in English.

Message::parseText() returns:

    <p>Contact\n</p><p><br />\n</p>

The Message::toString function applies a Regex to this.

if ( preg_match( '/^<p>(.*)\n?<\/p>\n?$/sU', $string, $m ) ) {
    $string = $m[1];
}

It looks like this may be trying to remove the extraneous HTML, but it doesn't entirely succeed. Instead, the result is:

    Contact\n</p><p><br />

which, of course, creates unnecessary blank lines in the output.