fix: call tokenize with document encoding

nqxcode · nqxcode · commit 73f478aab73c · 2014-06-15T23:48:53.000+01:00
diff --git a/library/ZendSearch/Lucene/Document/HTML.php b/library/ZendSearch/Lucene/Document/HTML.php
@@ -414,7 +414,7 @@ public function highlightExtended($words, $callback, $params = array())
         $wordsToHighlightList = array();
         $analyzer = Analyzer\Analyzer::getDefault();
         foreach ($words as $wordString) {
-            $wordsToHighlightList[] = $analyzer->tokenize($wordString);
+            $wordsToHighlightList[] = $analyzer->tokenize($wordString, $this->_doc->encoding);
         }
         $wordsToHighlight = call_user_func_array('array_merge', $wordsToHighlightList);
 

Original file line number	Diff line number	Diff line change
`@@ -414,7 +414,7 @@ public function highlightExtended($words, $callback, $params = array())`
`414`	`414`	`$wordsToHighlightList = array();`
`415`	`415`	`$analyzer = Analyzer\Analyzer::getDefault();`
`416`	`416`	`foreach ($words as $wordString) {`
`417`		`- $wordsToHighlightList[] = $analyzer->tokenize($wordString);`
	`417`	`+ $wordsToHighlightList[] = $analyzer->tokenize($wordString, $this->_doc->encoding);`
`418`	`418`	`}`
`419`	`419`	`$wordsToHighlight = call_user_func_array('array_merge', $wordsToHighlightList);`
`420`	`420`