From db3873454253dab13b8ab5c6207c00c0c38463d7 Mon Sep 17 00:00:00 2001
From: Rotem Liss <rotem@users.mediawiki.org>
Date: Sat, 5 Aug 2006 17:13:55 +0000
Subject: [PATCH] =?utf8?q?Adding=20a=20check=20for=20non-XHTML=20messages?=
 =?utf8?q?=20(mostly=20written=20by=20Niklas=20Laxstr=C3=B6m),=20and=20a?=
 =?utf8?q?=20fix.?=
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

---
 maintenance/checkLanguage.php |  5 +++++
 maintenance/languages.inc     | 28 +++++++++++++++++++++++++++-
 2 files changed, 32 insertions(+), 1 deletion(-)

diff --git a/maintenance/checkLanguage.php b/maintenance/checkLanguage.php
index 7c5a50bccb..1681ff02c2 100644
--- a/maintenance/checkLanguage.php
+++ b/maintenance/checkLanguage.php
@@ -70,4 +70,9 @@ $messagesWithWhitespace = $wgLanguages->getMessagesWithWhitespace( $code );
 $messagesWithWhitespaceNumber = count( $messagesWithWhitespace );
 $wgLanguages->outputMessagesList( $messagesWithWhitespace, "\n$messagesWithWhitespaceNumber messages of $localMessagesNumber in $code have a trailing whitespace:", $wgHideMessages );
 
+# Non-XHTML messages
+$nonXHTMLMessages = $wgLanguages->getNonXHTMLMessages( $code );
+$nonXHTMLMessagesNumber = count( $nonXHTMLMessages );
+$wgLanguages->outputMessagesList( $nonXHTMLMessages, "\n$nonXHTMLMessagesNumber messages of $localMessagesNumber in $code are not well-formed XHTML:", $wgHideMessages );
+
 ?>
diff --git a/maintenance/languages.inc b/maintenance/languages.inc
index f35220934e..82292ecec3 100644
--- a/maintenance/languages.inc
+++ b/maintenance/languages.inc
@@ -231,13 +231,39 @@ class languages {
 		$messagesWithWhitespace = array();
 		foreach ( $this->mMessages[$code] as $key => $value ) {
 			if ( isset( $this->mMessages['en'][$key] ) && $this->mMessages['en'][$key] !== '' &&
-				$this->mMessages[$code][$key] !== rtrim( $this->mMessages[$code][$key] ) ) {
+				$value !== rtrim( $value ) ) {
 				$messagesWithWhitespace[$key] = $value;
 			}
 		}
 		return $messagesWithWhitespace;
 	}
 
+	/**
+	 * Get the non-XHTML messages.
+	 *
+	 * @param $code The langauge code.
+	 *
+	 * @return The non-XHTML messages for this language.
+	 */
+	public function getNonXHTMLMessages( $code ) {
+		$this->loadMessages( 'en' );
+		$this->loadMessages( $code );
+		$wrongPhrases = array(
+			'<hr *\\?>',
+			'<br *\\?>',
+			'<hr/>',
+			'<br/>',
+		);
+		$wrongPhrases = '~(' . implode( '|', $wrongPhrases ) . ')~sDu';
+		$nonXHTMLMessages = array();
+		foreach ( $this->mMessages[$code] as $key => $value ) {
+			if ( isset( $this->mMessages['en'][$key] ) && preg_match( $wrongPhrases, $value ) ) {
+				$nonXHTMLMessages[$key] = $value;
+			}
+		}
+		return $nonXHTMLMessages;
+	}
+
 	/**
 	 * Output a messages list.
 	 *
-- 
2.20.1