/**
* {@inheritdoc}
*/
public function highlight($words)
{
$this->_doc->highlightExtended($words, [$this, 'wrapWords'], []);
}
/**
* Highlight matches in $inputHTMLFragment and return it (without HTML header and body tag)
*
* @param string $inputHTMLFragment
* @param string $encoding Input HTML string encoding
* @param Highlighter|null $highlighter
* @return string
*/
public function htmlFragmentHighlightMatches($inputHTMLFragment, $encoding = 'UTF-8', $highlighter = null)
{
if ($highlighter === null) {
$highlighter = new DefaultHighlighter();
}
$inputHTML = '<html><head><META HTTP-EQUIV="Content-type" CONTENT="text/html; charset=UTF-8"/></head><body>' . iconv($encoding, 'UTF-8//IGNORE', $inputHTMLFragment) . '</body></html>';
$doc = Document\HTML::loadHTML($inputHTML);
$highlighter->setDocument($doc);
$this->_highlightMatches($highlighter);
return $doc->getHTMLBody();
}
/**
* index a file
*
* @param File $file the file to be indexed
* @param bool $commit
*
* @return bool true when something was stored in the index, false otherwise (eg, folders are not indexed)
* @throws NotIndexedException when an unsupported file type is encountered
*/
public function indexFile(File $file, $commit = true)
{
// we decide how to index on mime type or file extension
$mimeType = $file->getMimeType();
$fileExtension = strtolower(pathinfo($file->getName(), PATHINFO_EXTENSION));
// initialize plain lucene document
$doc = new Document();
// index content for local files only
$storage = $file->getStorage();
if ($storage->isLocal()) {
$path = $storage->getLocalFile($file->getInternalPath());
//try to use special lucene document types
if ('text/html' === $mimeType) {
//TODO could be indexed, even if not local
$doc = HTML::loadHTML($file->getContent());
} else {
if ('text/' === substr($mimeType, 0, 5) || 'application/x-tex' === $mimeType) {
$body = $file->getContent();
if ($body != '') {
$doc->addField(Document\Field::UnStored('body', $body));
}
} else {
if ('application/pdf' === $mimeType) {
$doc = Pdf::loadPdf($file->getContent());
// the zend classes only understand docx and not doc files
} else {
if ($fileExtension === 'docx') {
$doc = Document\Docx::loadDocxFile($path);
//} else if ('application/msexcel' === $mimeType) {
} else {
if ($fileExtension === 'xlsx') {
$doc = Document\Xlsx::loadXlsxFile($path);
//} else if ('application/mspowerpoint' === $mimeType) {
} else {
if ($fileExtension === 'pptx') {
$doc = Document\Pptx::loadPptxFile($path);
} else {
if ($fileExtension === 'odt') {
$doc = Odt::loadOdtFile($path);
} else {
if ($fileExtension === 'ods') {
$doc = Ods::loadOdsFile($path);
} else {
throw new NotIndexedException();
}
}
}
}
}
}
}
}
}
// Store filecache id as unique id to lookup by when deleting
$doc->addField(Document\Field::Keyword('fileId', $file->getId()));
// Store document path for the search results
$doc->addField(Document\Field::Text('path', $file->getPath(), 'UTF-8'));
$doc->addField(Document\Field::unIndexed('mtime', $file->getMTime()));
$doc->addField(Document\Field::unIndexed('size', $file->getSize()));
$doc->addField(Document\Field::unIndexed('mimetype', $mimeType));
$this->index->updateFile($doc, $file->getId(), $commit);
return true;
}
/**
* @dataProvider dataProviderXHTML
*/
function testParseXHTMLSplit($term, $field, $descriptiveLocation, $skipped)
{
if ($skipped) {
$this->markTestSkipped('TODO search ' . $descriptiveLocation . ' in ' . $field);
}
$data = file_get_contents(__DIR__ . '/data/libreoffice/document split.xhtml');
$doc = HTML::loadHTML($data, true);
$value = $doc->getFieldValue($field);
$containsTestTerm = is_string(stristr($value, $term));
$this->assertTrue($containsTestTerm, $field . '/' . $descriptiveLocation . ' does not contain "' . $term . '" in ' . $value);
}
public function testHtmlNoFollowLinks()
{
$html = '<HTML>' . '<HEAD><TITLE>Page title</TITLE></HEAD>' . '<BODY>' . 'Document body.' . '<a href="link1.html">Link 1</a>.' . '<a href="link2.html" rel="nofollow">Link 1</a>.' . '</BODY>' . '</HTML>';
$oldNoFollowValue = Document\HTML::getExcludeNoFollowLinks();
Document\HTML::setExcludeNoFollowLinks(false);
$doc1 = Document\HTML::loadHTML($html);
$this->assertTrue($doc1 instanceof Document\HTML);
$this->assertTrue(array_values($doc1->getLinks()) == array('link1.html', 'link2.html'));
Document\HTML::setExcludeNoFollowLinks(true);
$doc2 = Document\HTML::loadHTML($html);
$this->assertTrue($doc2 instanceof Document\HTML);
$this->assertTrue(array_values($doc2->getLinks()) == array('link1.html'));
}
/**
* Highlight specified words
*
* @param string|array $words Words to highlight. They could be organized using the array or string.
*/
public function highlight($words)
{
$color = $this->_highlightColors[$this->_currentColorIndex];
$this->_currentColorIndex = ($this->_currentColorIndex + 1) % count($this->_highlightColors);
$this->_doc->highlight($words, $color);
}
/**
* {@inheritdoc}
*/
public function highlight($words)
{
$this->doc->highlightExtended($words, array($this, 'applyColour'), array());
}
