Merge "Fix separation of text into sentences."

includes/ExtractFormatter.php

@@ -80,7 +80,7 @@ class ExtractFormatter extends HtmlFormatter {
 	public static function getFirstSentences( $text, $requestedSentenceCount ) {
 		// Based on code from OpenSearchXml by Brion Vibber
 		$endchars = array(
-			'[^\p{Lu}]\.\s', '\!\s', '\?\s', // regular ASCII
+			'[^\p{Lu}]\.[ \n]', '\![ \n]', '\?[ \n]', // regular ASCII
 			'。', // full-width ideographic full-stop
 			'．', '！', '？', // double-width roman forms
 			'｡', // half-width ideographic full stop

tests/ExtractFormatterTest.php

@@ -115,6 +115,12 @@ class ExtractFormatterTest extends MediaWikiTestCase {
 				1,
 				'P.J. Harvey is a singer.',
 			),
+			// Bug T115817 - Non-breaking space is not a delimiter
+			array(
+				html_entity_decode( 'Pigeons (lat. Columbidae) are birds. They primarily feed on seeds.' ),
+				1,
+				html_entity_decode( 'Pigeons (lat. Columbidae) are birds.' ),
+			),
 		);
 	}