Index: library/Zend/Search/Lucene/Document/Html.php
===================================================================
--- library/Zend/Search/Lucene/Document/Html.php (revision 20197)
+++ library/Zend/Search/Lucene/Document/Html.php (working copy)
@@ -68,6 +68,17 @@
private static $_excludeNoFollowLinks = false;
/**
+ *
+ * List of inline tags
+ *
+ * @var array
+ */
+ private $_inlineTags = array('a', 'abbr', 'acronym', 'dfn', 'em', 'strong', 'code',
+ 'samp', 'kbd', 'var', 'b', 'i', 'big', 'small', 'strike',
+ 'tt', 'u', 'font', 'span', 'bdo', 'cite', 'del', 'ins',
+ 'q', 'sub', 'sup');
+
+ /**
* Object constructor
*
* @param string $data HTML string (may be HTML fragment, )
@@ -197,8 +208,10 @@
private function _retrieveNodeText(DOMNode $node, &$text)
{
if ($node->nodeType == XML_TEXT_NODE) {
- $text .= $node->nodeValue ;
- $text .= ' ';
+ $text .= $node->nodeValue;
+ if(!in_array($node->parentNode->tagName, $this->_inlineTags)) {
+ $text .= ' ';
+ }
} else if ($node->nodeType == XML_ELEMENT_NODE && $node->nodeName != 'script') {
foreach ($node->childNodes as $childNode) {
$this->_retrieveNodeText($childNode, $text);
Index: tests/Zend/Search/Lucene/DocumentTest.php
===================================================================
--- tests/Zend/Search/Lucene/DocumentTest.php (revision 20197)
+++ tests/Zend/Search/Lucene/DocumentTest.php (working copy)
@@ -55,6 +55,23 @@
*/
class Zend_Search_Lucene_DocumentTest extends PHPUnit_Framework_TestCase
{
+
+ private function _clearDirectory($dirName)
+ {
+ if (!file_exists($dirName) || !is_dir($dirName)) {
+ return;
+ }
+
+ // remove files from temporary direcytory
+ $dir = opendir($dirName);
+ while (($file = readdir($dir)) !== false) {
+ if (!is_dir($dirName . '/' . $file)) {
+ @unlink($dirName . '/' . $file);
+ }
+ }
+ closedir($dir);
+ }
+
public function testCreate()
{
$document = new Zend_Search_Lucene_Document();
@@ -171,6 +188,32 @@
'contributing.html'));
}
+
+ /**
+ * @group ZF-4252
+ */
+ public function testHtmlInlineTagsIndexing()
+ {
+ $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
+
+ $htmlString = 'Hello World'
+ . 'ZendFramework' . "\n" . ' Foo
Bar ' . "\n"
+ . ' Test';
+
+ $doc = Zend_Search_Lucene_Document_Html::loadHTML($htmlString);
+
+ $index->addDocument($doc);
+
+ $hits = $index->find('FooBar');
+ $this->assertEquals(count($hits), 0);
+
+ $hits = $index->find('ZendFramework');
+ $this->assertEquals(count($hits), 1);
+
+ unset($index);
+ $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
+ }
+
public function testHtmlNoFollowLinks()
{
$html = ''