/**
* Index a file
*
* @param string $filePath The file path
*/
public function index($filePath)
{
$content = file_get_contents($filePath);
$modificationTime = filemtime($filePath);
$checksum = md5($content);
// Get the document
$hits = $this->_data->find('path:' . $filePath);
if (count($hits) > 0) {
$hit = $hits[0];
$document = $hit->getDocument();
// If the checksums are the same, no need to update
if ($checksum === $document->checksum) {
return;
}
// Delete the document
$this->_data->delete($hit);
}
// Create a new document
$document = new Zend_Search_Lucene_Document();
$document->addField(Zend_Search_Lucene_Field::keyword('path', $filePath));
$document->addField(Zend_Search_Lucene_Field::keyword('modificationTime', $modificationTime));
$document->addField(Zend_Search_Lucene_Field::keyword('checksum', $checksum));
$document->addField(Zend_Search_Lucene_Field::unStored('content', $content, 'utf-8'));
$this->_data->addDocument($document);
// Commit the changes
$this->_data->commit();
$this->_data->optimize();
}
public static function addPageToIndex($page, $toasterSearchIndex = false)
{
if (!self::initIndex()) {
return false;
}
if ($page instanceof Application_Model_Models_Page) {
$page = $page->toArray();
$containers = Application_Model_Mappers_ContainerMapper::getInstance()->findByPageId($page['id']);
$page['content'] = '';
if (!empty($containers)) {
foreach ($containers as $container) {
$page['content'] .= $container->getContent();
}
}
}
$document = new Zend_Search_Lucene_Document();
$document->addField(Zend_Search_Lucene_Field::keyword('pageId', $page['id']));
$document->addField(Zend_Search_Lucene_Field::unStored('metaKeyWords', $page['metaKeywords'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::unStored('metaDescription', $page['metaDescription'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::unStored('headerTitle', $page['headerTitle'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::unStored('content', $page['content'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::text('draft', $page['draft'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::text('teaserText', $page['teaserText'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::text('url', $page['url'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::text('navName', $page['navName'], 'UTF-8'));
$document->addField(Zend_Search_Lucene_Field::text('h1', $page['h1'], 'UTF-8'));
// $document->addField(Zend_Search_Lucene_Field::text('previewImage', $page['previewImage']));
self::$_index->addDocument($document);
}
/**
* Add a post to the index.
*
* @param Post $post the post being inserted
*/
public function index_post($post)
{
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Text('url', $post->permalink));
$title = Zend_Search_Lucene_Field::Text('title', strtolower($post->title), 'utf-8');
$title->boost = 50;
$doc->addField($title);
$doc->addField(Zend_Search_Lucene_Field::UnStored('contents', strtolower($post->content), 'utf-8'));
// Add tags
$tags = $post->tags;
$tagstring = '';
foreach ($tags as $tag) {
$tagstring .= $tag . ' ';
}
$dtag = Zend_Search_Lucene_Field::UnStored('tags', strtolower($tagstring), 'utf-8');
$dtag->boost = 10;
$doc->addField($dtag);
// Add ID
$doc->addField(Zend_Search_Lucene_Field::keyword('postid', $post->id));
$this->_index->addDocument($doc);
}
/**
* Index an Zym_Search_Lucene_Indexable_Interface
*
* @throws Zym_Search_Lucene_Exception
* @param Zym_Search_Lucene_Indexable_Interface|array $indexables
* @param boolean $update
* @param string $searchField
* @return Zym_Search_Lucene_Index
*/
public function index($indexables, $update = true, $searchField = null)
{
if (!is_array($indexables)) {
$indexables = array($indexables);
}
if (!$searchField) {
$searchField = $this->_idKey;
}
foreach ($indexables as $indexable) {
if (!$indexable instanceof Zym_Search_Lucene_IIndexable) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('The object needs to have Zym_Search_Lucene_Indexable_Interface implemented.');
}
if ($update) {
$recordId = $indexable->getRecordID();
if (!$recordId) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('The record ID must not be null');
}
$this->delete($recordId, $searchField);
}
$document = $indexable->getSearchDocument();
if (!$document instanceof Zend_Search_Lucene_Document) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('The document is not an instance of Zend_Search_Lucene_Document.');
}
$this->_searchIndex->addDocument($document);
}
return $this;
}
/**
* Adds a document to this index.
*
* @param Zend_Search_Lucene_Document $document
*/
public function addDocument(Zend_Search_Lucene_Document $document)
{
$this->_index->addDocument($document);
}
/**
* @param AJXP_Node $ajxpNode
* @param Zend_Search_Lucene_Interface $index
* @throws Exception
* @return Zend_Search_Lucene_Document
*/
public function createIndexedDocument($ajxpNode, &$index)
{
if (!empty($this->metaFields)) {
$ajxpNode->loadNodeInfo(false, false, "all");
} else {
$ajxpNode->loadNodeInfo();
}
$ext = strtolower(pathinfo($ajxpNode->getLabel(), PATHINFO_EXTENSION));
$parseContent = $this->indexContent;
if ($parseContent && $ajxpNode->bytesize > $this->getFilteredOption("PARSE_CONTENT_MAX_SIZE")) {
$parseContent = false;
}
if ($parseContent && in_array($ext, explode(",", $this->getFilteredOption("PARSE_CONTENT_HTML")))) {
$doc = @Zend_Search_Lucene_Document_Html::loadHTMLFile($ajxpNode->getUrl());
} elseif ($parseContent && $ext == "docx" && class_exists("Zend_Search_Lucene_Document_Docx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Docx::loadDocxFile($realFile);
} elseif ($parseContent && $ext == "docx" && class_exists("Zend_Search_Lucene_Document_Pptx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Pptx::loadPptxFile($realFile);
} elseif ($parseContent && $ext == "xlsx" && class_exists("Zend_Search_Lucene_Document_Xlsx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Xlsx::loadXlsxFile($realFile);
} else {
$doc = new Zend_Search_Lucene_Document();
}
if ($doc == null) {
throw new Exception("Could not load document");
}
$doc->addField(Zend_Search_Lucene_Field::Keyword("node_url", $ajxpNode->getUrl()), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("node_path", str_replace("/", "AJXPFAKESEP", $ajxpNode->getPath())), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Text("basename", basename($ajxpNode->getPath())), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_node", "yes"), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_scope", "shared"));
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_modiftime", date("Ymd", $ajxpNode->ajxp_modiftime)));
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_bytesize", $ajxpNode->bytesize));
$ajxpMime = $ajxpNode->ajxp_mime;
if (empty($ajxpMime)) {
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_mime", pathinfo($ajxpNode->getLabel(), PATHINFO_EXTENSION)));
} else {
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_mime", $ajxpNode->ajxp_mime));
}
// Store a cached copy of the metadata
$serializedMeta = base64_encode(serialize($ajxpNode->metadata));
$doc->addField(Zend_Search_Lucene_Field::Binary("serialized_metadata", $serializedMeta));
if (isset($ajxpNode->indexableMetaKeys["shared"])) {
foreach ($ajxpNode->indexableMetaKeys["shared"] as $sharedField) {
if ($ajxpNode->{$sharedField}) {
$doc->addField(Zend_search_Lucene_Field::keyword($sharedField, $ajxpNode->{$sharedField}));
}
}
}
foreach ($this->metaFields as $field) {
if ($ajxpNode->{$field} != null) {
$doc->addField(Zend_Search_Lucene_Field::Text("ajxp_meta_{$field}", $ajxpNode->{$field}), SystemTextEncoding::getEncoding());
}
}
if (isset($ajxpNode->indexableMetaKeys["user"]) && count($ajxpNode->indexableMetaKeys["user"]) && AuthService::usersEnabled() && AuthService::getLoggedUser() != null) {
$privateDoc = new Zend_Search_Lucene_Document();
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("node_url", $ajxpNode->getUrl(), SystemTextEncoding::getEncoding()));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("node_path", str_replace("/", "AJXPFAKESEP", $ajxpNode->getPath()), SystemTextEncoding::getEncoding()));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_scope", "user"));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_user", AuthService::getLoggedUser()->getId()));
foreach ($ajxpNode->indexableMetaKeys["user"] as $userField) {
if ($ajxpNode->{$userField}) {
$privateDoc->addField(Zend_search_Lucene_Field::keyword($userField, $ajxpNode->{$userField}));
}
}
$privateDoc->addField(Zend_Search_Lucene_Field::Binary("serialized_metadata", $serializedMeta));
$index->addDocument($privateDoc);
}
if ($parseContent) {
$body = $this->extractIndexableContent($ajxpNode);
if (!empty($body)) {
$doc->addField(Zend_Search_Lucene_Field::unStored("body", $body));
}
}
$index->addDocument($doc);
return $doc;
}
/**
* @param AJXP_Node $ajxpNode
* @param Zend_Search_Lucene_Interface $index
* @throws Exception
* @return Zend_Search_Lucene_Document
*/
public function createIndexedDocument($ajxpNode, &$index)
{
$ajxpNode->loadNodeInfo();
$ext = strtolower(pathinfo($ajxpNode->getLabel(), PATHINFO_EXTENSION));
$parseContent = $this->indexContent;
if ($parseContent && $ajxpNode->bytesize > $this->getFilteredOption("PARSE_CONTENT_MAX_SIZE")) {
$parseContent = false;
}
if ($parseContent && in_array($ext, explode(",", $this->getFilteredOption("PARSE_CONTENT_HTML")))) {
$doc = @Zend_Search_Lucene_Document_Html::loadHTMLFile($ajxpNode->getUrl());
} elseif ($parseContent && $ext == "docx" && class_exists("Zend_Search_Lucene_Document_Docx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Docx::loadDocxFile($realFile);
} elseif ($parseContent && $ext == "docx" && class_exists("Zend_Search_Lucene_Document_Pptx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Pptx::loadPptxFile($realFile);
} elseif ($parseContent && $ext == "xlsx" && class_exists("Zend_Search_Lucene_Document_Xlsx")) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$doc = @Zend_Search_Lucene_Document_Xlsx::loadXlsxFile($realFile);
} else {
$doc = new Zend_Search_Lucene_Document();
}
if ($doc == null) {
throw new Exception("Could not load document");
}
$doc->addField(Zend_Search_Lucene_Field::Keyword("node_url", $ajxpNode->getUrl()), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("node_path", str_replace("/", "AJXPFAKESEP", $ajxpNode->getPath())), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Text("basename", basename($ajxpNode->getPath())), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_node", "yes"), SystemTextEncoding::getEncoding());
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_scope", "shared"));
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_modiftime", date("Ymd", $ajxpNode->ajxp_modiftime)));
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_bytesize", $ajxpNode->bytesize));
$ajxpMime = $ajxpNode->ajxp_mime;
if (empty($ajxpMime)) {
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_mime", pathinfo($ajxpNode->getLabel(), PATHINFO_EXTENSION)));
} else {
$doc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_mime", $ajxpNode->ajxp_mime));
}
// Store a cached copy of the metadata
$serializedMeta = base64_encode(serialize($ajxpNode->metadata));
$doc->addField(Zend_Search_Lucene_Field::Binary("serialized_metadata", $serializedMeta));
if (isset($ajxpNode->indexableMetaKeys["shared"])) {
foreach ($ajxpNode->indexableMetaKeys["shared"] as $sharedField) {
if ($ajxpNode->{$sharedField}) {
$doc->addField(Zend_search_Lucene_Field::keyword($sharedField, $ajxpNode->{$sharedField}));
}
}
}
foreach ($this->metaFields as $field) {
if ($ajxpNode->{$field} != null) {
$doc->addField(Zend_Search_Lucene_Field::Text("ajxp_meta_{$field}", $ajxpNode->{$field}), SystemTextEncoding::getEncoding());
}
}
if (isset($ajxpNode->indexableMetaKeys["user"]) && count($ajxpNode->indexableMetaKeys["user"]) && AuthService::usersEnabled() && AuthService::getLoggedUser() != null) {
$privateDoc = new Zend_Search_Lucene_Document();
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("node_url", $ajxpNode->getUrl(), SystemTextEncoding::getEncoding()));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("node_path", str_replace("/", "AJXPFAKESEP", $ajxpNode->getPath()), SystemTextEncoding::getEncoding()));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_scope", "user"));
$privateDoc->addField(Zend_Search_Lucene_Field::Keyword("ajxp_user", AuthService::getLoggedUser()->getId()));
foreach ($ajxpNode->indexableMetaKeys["user"] as $userField) {
if ($ajxpNode->{$userField}) {
$privateDoc->addField(Zend_search_Lucene_Field::keyword($userField, $ajxpNode->{$userField}));
}
}
$privateDoc->addField(Zend_Search_Lucene_Field::Binary("serialized_metadata", $serializedMeta));
$index->addDocument($privateDoc);
}
if ($parseContent && in_array($ext, explode(",", $this->getFilteredOption("PARSE_CONTENT_TXT")))) {
$doc->addField(Zend_Search_Lucene_Field::unStored("body", file_get_contents($ajxpNode->getUrl())));
}
$unoconv = $this->getFilteredOption("UNOCONV");
$pipe = false;
if ($parseContent && !empty($unoconv) && in_array($ext, array("doc", "odt", "xls", "ods"))) {
$targetExt = "txt";
if (in_array($ext, array("xls", "ods"))) {
$targetExt = "csv";
} else {
if (in_array($ext, array("odp", "ppt"))) {
$targetExt = "pdf";
$pipe = true;
}
}
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
$unoconv = "HOME=" . AJXP_Utils::getAjxpTmpDir() . " " . $unoconv . " --stdout -f {$targetExt} " . escapeshellarg($realFile);
if ($pipe) {
$newTarget = str_replace(".{$ext}", ".pdf", $realFile);
$unoconv .= " > {$newTarget}";
register_shutdown_function("unlink", $newTarget);
}
$output = array();
exec($unoconv, $output, $return);
if (!$pipe) {
$out = implode("\n", $output);
$enc = 'ISO-8859-1';
$asciiString = iconv($enc, 'ASCII//TRANSLIT//IGNORE', $out);
$doc->addField(Zend_Search_Lucene_Field::unStored("body", $asciiString));
} else {
$ext = "pdf";
}
}
$pdftotext = $this->getFilteredOption("PDFTOTEXT");
if ($parseContent && !empty($pdftotext) && in_array($ext, array("pdf"))) {
$realFile = call_user_func(array($ajxpNode->wrapperClassName, "getRealFSReference"), $ajxpNode->getUrl());
if ($pipe && isset($newTarget) && is_file($newTarget)) {
$realFile = $newTarget;
}
$cmd = $pdftotext . " " . escapeshellarg($realFile) . " -";
$output = array();
exec($cmd, $output, $return);
$out = implode("\n", $output);
$enc = 'UTF8';
$asciiString = iconv($enc, 'ASCII//TRANSLIT//IGNORE', $out);
$doc->addField(Zend_Search_Lucene_Field::unStored("body", $asciiString));
}
$index->addDocument($doc);
return $doc;
}
/**
* Index an Zym_Search_Lucene_Indexable_Interface
*
* @throws Zym_Search_Lucene_Exception
* @param Zym_Search_Lucene_Indexable_Interface|array $indexables
* @param boolean $update
* @param string $searchField
* @return Zym_Search_Lucene_Index
*/
public function index($indexables, $update = true, $searchField = null)
{
if (!is_array($indexables)) {
$indexables = array($indexables);
}
if (!$searchField) {
$searchField = $this->_idKey;
}
foreach ($indexables as $indexable) {
if (!$indexable instanceof Zym_Search_Lucene_IIndexable) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('The object of type "' . get_class($indexable) . '" ' . 'is not an instance of Zym_Search_Lucene_Indexable_Interface.');
}
if ($update) {
$recordId = $indexable->getRecordId();
if (!$recordId) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('You must provide a valid record ID.');
}
$this->delete($recordId, $searchField);
}
$document = $indexable->getSearchDocument();
if (!$document instanceof Zend_Search_Lucene_Document) {
/**
* @see Zym_Search_Lucene_Exception
*/
require_once 'Zym/Search/Lucene/Exception.php';
throw new Zym_Search_Lucene_Exception('The provided search-document is not ' . 'an instance of Zend_Search_Lucene_Document.');
}
$this->_searchIndex->addDocument($document);
}
return $this;
}
