function index()
{
$this->load->library('zend', 'Zend/Feed');
$this->load->library('zend', 'Zend/Search/Lucene');
$this->load->library('zend');
$this->zend->load('Zend/Feed');
$this->zend->load('Zend/Search/Lucene');
//Create index.
$index = new Zend_Search_Lucene('C:\\xampp\\xampp\\htdocs\\controle_frota\\lucene\\feeds_index', true);
$feeds = array('http://oglobo.globo.com/rss.xml?limite=50');
//grab each feed.
foreach ($feeds as $feed) {
$channel = Zend_Feed::import($feed);
echo $channel->title() . '<br />';
//index each item.
foreach ($channel->items as $item) {
if ($item->link() && $item->title() && $item->description()) {
//create an index doc.
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Keyword('link', $this->sanitize($item->link())));
$doc->addField(Zend_Search_Lucene_Field::Text('title', $this->sanitize($item->title())));
$doc->addField(Zend_Search_Lucene_Field::Unstored('contents', $this->sanitize($item->description())));
echo "\tAdding: " . $item->title() . '<br />';
$index->addDocument($doc);
}
}
}
$index->commit();
echo $index->count() . ' Documents indexed.<br />';
}
public function add($content, $section, $mtime)
{
foreach ($this->split_headings($content) as $headers) {
$doc = new Zend_Search_Lucene_Document();
$link = "index.php?page=" . preg_replace('/\\/|\\\\/', '.', $section);
$link = str_replace('.page', '', $link) . '#' . $headers['section'];
//unsearchable text
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('link', $link));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('mtime', $mtime));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('title', $headers['title']));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('text', $headers['content']));
//searchable text
$doc->addField(Zend_Search_Lucene_Field::Keyword('page', strtolower($headers['title'])));
$body = strtolower($this->sanitize($headers['content'])) . ' ' . strtolower($headers['title']);
$doc->addField(Zend_Search_Lucene_Field::Unstored('contents', $body));
$this->_index->addDocument($doc);
}
}
function create_index()
{
echo "Building search index...\n";
$files = $this->get_files($this->_api);
$count = 0;
foreach ($files as $file) {
$content = $this->get_details($file, $this->_api);
$doc = new Zend_Search_Lucene_Document();
$title = $content['class'];
echo " Adding " . $title . "\n";
//unsearchable text
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('link', $content['link']));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('title', $title));
//$doc->addField(Zend_Search_Lucene_Field::UnIndexed('text', $content['content']));
//searchable
$body = strtolower($this->sanitize($content['content'])) . ' ' . strtolower($title);
$doc->addField(Zend_Search_Lucene_Field::Keyword('page', strtolower(str_replace('.', ' ', $title))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('contents', $body));
$this->_index->addDocument($doc);
$count++;
}
$this->_index->commit();
echo "\n {$count} files indexed.\n";
}
public function indexAction()
{
// action body
$this->_helper->viewRenderer->setNoRender(true);
$jobs = new Application_Model_DbTable_JobPortal();
$index = Zend_Search_Lucene::create('C:\\indexed');
$maxBufferedDocs = 100;
$index->setMaxBufferedDocs($maxBufferedDocs);
$users = $jobs->fetchAll();
Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num());
foreach ($users as $user) {
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Keyword('pri', $user->id));
$doc->addField(Zend_Search_Lucene_Field::Text('title', $user->title));
//here field = ur database column
$doc->addField(Zend_Search_Lucene_Field::Text('shortd', $user->shortd));
$doc->addField(Zend_Search_Lucene_Field::Unstored('longd', $user->longd));
$doc->addField(Zend_Search_Lucene_Field::Text('exp', $user->exp));
$doc->addField(Zend_Search_Lucene_Field::Text('location', $user->location));
$index->addDocument($doc);
}
$index->commit();
$this->_helper->redirector('search');
}
public function actionFulltext()
{
// get dirty
$this->template->num_docs = fulltext::index()->numDocs();
$this->template->dirty = fulltext::dirty();
$this->template->num_dirty = count($this->template->dirty);
// index
$index = fulltext::index();
$this->template->update_now = array_slice($this->template->dirty, 0, 50);
if (!empty($this->template->update_now)) {
adminlog::log(__('Attempt to update fulltext'));
}
foreach (mapper::products()->findByIds($this->template->update_now) as $product) {
// delete old
foreach ($index->termDocs(new Zend_Search_Lucene_Index_Term($product->getId(), 'id')) as $id) {
$index->delete($id);
}
// add
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Keyword('id', $product->getId()));
$doc->addField(Zend_Search_Lucene_Field::UnStored('name', $product->getName()));
$doc->addField(Zend_Search_Lucene_Field::UnStored('nice_name', $product->getNiceName()));
$doc->addField(Zend_Search_Lucene_Field::Unstored('code', $product->getCode()));
$doc->addField(Zend_Search_Lucene_Field::UnStored('meta_keywords', $product->getMetaKeywords()));
$doc->addField(Zend_Search_Lucene_Field::UnStored('meta_description', $product->getMetaDescription()));
$description = '';
if (strlen($product->getDescription()) < 1) {
if (strlen($product->getMetaDescription()) < 1) {
$description = $product->getName();
} else {
$description = $product->getMetaDescription();
}
} else {
$description = $product->getDescription();
}
if ($manufacturer = mapper::products()->findManufacturerOf($product->getId())) {
$doc->addField(Zend_Search_Lucene_Field::UnStored('manufacturer', $manufacturer->getName()));
$description .= ' ' . $manufacturer->getName();
$description .= ' ' . $manufacturer->getDescription();
}
if ($category = mapper::products()->findCategoryOf($product->getId())) {
$doc->addField(Zend_Search_Lucene_Field::UnStored('category', $category->getName()));
$description .= ' ' . $category->getName();
$description .= ' ' . $category->getDescription();
}
$description .= ' ' . $product->getName();
$doc->addField(Zend_Search_Lucene_Field::UnStored('description', $description));
$index->addDocument($doc);
}
// undirty updated
foreach ($this->template->update_now as $id) {
fulltext::dirty($id, FALSE);
}
// log
adminlog::log(__('Successfully updated %d fulltext items, %d remains'), count($this->template->update_now), $this->template->num_dirty - count($this->template->update_now));
// refresh
$s = 5;
Environment::getHttpResponse()->setHeader('Refresh', $s . '; ' . (string) Environment::getHttpRequest()->getOriginalUri());
$this->template->next_update = $s;
}
$cindex->addDocument($doc);
$changed = true;
}
mysql_free_result($thread_result);
if($changed)
$cindex->commit();
//Do the upload:
$term = new Zend_Search_Lucene_Index_Term($fileid, 'fileid');
$docIds = $uindex->termDocs($term);
if(count($docIds) == 0)
{
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Keyword('fileid', $fileid));
$doc->addField(Zend_Search_Lucene_Field::Unstored('filename', $filename));
$doc->addField(Zend_Search_Lucene_Field::Unstored('uploader', $uploader));
$uindex->addDocument($doc);
$uindex->commit();
}
echo "\n";
flush();
}
echo "Done.\noptimizing indices... ";
flush();
$cindex->optimize();
$uindex->optimize();
echo "Done (".number_format($timer->reset(), 5)."s)\n\nUpload index size is ".$uindex->count().", ".$uindex->numDocs()." documents.\n\nComment index size is ".$cindex->count().", ".$cindex->numDocs()." documents.\n\n";
public static function saveSearchIndex()
{
$index = new Zend_Search_Lucene(sfConfig::get('sf_lib_dir') . '/modules/search/tmp/lucene.user.index', true);
$doc = new Zend_Search_Lucene_Document();
$doc->addField(Zend_Search_Lucene_Field::Keyword('id', $this->pageDocument->getId()));
$doc->addField(Zend_Search_Lucene_Field::Keyword('pageid', $this->pageDocument->getId()));
$doc->addField(Zend_Search_Lucene_Field::Keyword('title', $this->pageDocument->getNavigationTitle()));
$blobData = $this->pageDocument->getContent();
// $blockContents = $blobData->__toString();
$blockContents = $blobData;
$doc->addField(Zend_Search_Lucene_Field::Unstored('contents', $blockContents . ' ' . $this->pageDocument->getNavigationTitle()));
$index->addDocument($doc);
$index->commit();
$hits = $index->find(strtolower('maquette'));
foreach ($hits as $hit) {
// echo $hit->score.'<br/>';
// echo $hit->id;
// echo $hit->contents.'<br/>';
echo $hit->pageid;
}
}
public static function addInformationObjectIndex(QubitInformationObject $informationObject, $language, $options = array())
{
// Only ROOT node should have no parent, don't index
if (null === $informationObject->parent) {
return;
}
$doc = new Zend_Search_Lucene_Document();
// Reference elements
$doc->addField(Zend_Search_Lucene_Field::Keyword('id', $informationObject->id));
$doc->addField(Zend_Search_Lucene_Field::Keyword('slug', $informationObject->slug));
$doc->addField(Zend_Search_Lucene_Field::Keyword('culture', $language));
$doc->addField(Zend_Search_Lucene_Field::Keyword('className', 'QubitInformationObject'));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('parentId', $informationObject->parentId));
$doc->addField(Zend_Search_Lucene_Field::Keyword('parent', $informationObject->parent->slug));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('collectionRootId', $informationObject->getCollectionRoot()->id));
$doc->addField(Zend_Search_Lucene_Field::Keyword('collectionRootSlug', $informationObject->getCollectionRoot()->slug));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('collectionRootTitle', $informationObject->getCollectionRoot()->getTitle()));
// Digital object information
if ($digitalObject = $informationObject->getDigitalObject()) {
$doc->addField(Zend_Search_Lucene_Field::Keyword('hasDigitalObject', 'true'));
$doc->addField(Zend_Search_Lucene_Field::Keyword('do_mediaTypeId', $digitalObject->mediaTypeId));
if (null !== $digitalObject->thumbnail) {
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('do_thumbnail_FullPath', $digitalObject->thumbnail->getFullPath()));
}
// $doc->addField(Zend_Search_Lucene_Field::Unstored('mediatype', $digitalObject->getMediaType()->getName(array('culture' => $language))));
// $doc->addField(Zend_Search_Lucene_Field::Unstored('filename', $digitalObject->getName()));
// $doc->addField(Zend_Search_Lucene_Field::Unstored('mimetype', $digitalObject->mimeType));
} else {
$doc->addField(Zend_Search_Lucene_Field::Keyword('hasDigitalObject', 'false'));
}
// Title
// include an i18n fallback for proper search result display in case the title field was not translated
if (0 < strlen($informationObject->getTitle(array('culture' => $language)))) {
$titleField = Zend_Search_Lucene_Field::Text('title', $informationObject->getTitle(array('culture' => $language)));
} else {
$titleField = Zend_Search_Lucene_Field::Text('title', $informationObject->getTitle(array('sourceCulture' => true)));
}
// Boost the hit relevance for the title field
$titleField->boost = 10;
$doc->addField($titleField);
// Publication status
$doc->addField(Zend_Search_Lucene_Field::Text('publicationStatusId', $informationObject->getPublicationStatus()->status->id));
$doc->addField(Zend_Search_Lucene_Field::Text('scopeAndContent', $informationObject->getScopeAndContent(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Text('referenceCode', $informationObject->referenceCode));
// Store dates as serialized array
$dates = array();
foreach ($informationObject->getDates() as $date) {
$save_date['id'] = $date->id;
$save_date['rendered'] = Qubit::renderDateStartEnd($date->getDate(array('cultureFallback' => true)), $date->startDate, $date->endDate);
$save_date['type'] = $date->getType(array('cultureFallback' => true))->__toString();
if (isset($date->actor)) {
$save_date['actor'] = $date->actor->__toString();
}
$dates[] = $save_date;
}
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('dates', serialize($dates)));
// CREATOR
$creatorField = Zend_Search_Lucene_Field::Unstored('creator', $informationObject->getCreatorsNameString(array('culture' => $language)));
// Boost the hit relevance for the creator field
$creatorField->boost = 8;
$doc->addField($creatorField);
$doc->addField(Zend_Search_Lucene_Field::Unstored('creatorhistory', $informationObject->getCreatorsHistoryString(array('culture' => $language))));
// Level of Description
if (null !== $informationObject->getLevelOfDescription()) {
$doc->addField(Zend_Search_Lucene_Field::Text('levelOfDescription', $informationObject->getLevelOfDescription()->getName(array('culture' => $language))));
} else {
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('levelOfDescription', null));
}
// Repository
$repository = $informationObject->getRepository(array('inherit' => true));
if (null !== $repository) {
$doc->addField(Zend_Search_Lucene_Field::Keyword('repositoryId', $repository->id));
$doc->addField(Zend_Search_Lucene_Field::Keyword('repositorySlug', $repository->slug));
$doc->addField(Zend_Search_Lucene_Field::Text('repositoryName', $repository->getAuthorizedFormOfName(array('culture' => $language))));
} else {
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('repositoryId', null));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('repositorySlug', null));
$doc->addField(Zend_Search_Lucene_Field::UnIndexed('repositoryName', null));
}
// Identifier
$identifierField = Zend_Search_Lucene_Field::Text('identifier', $informationObject->getIdentifier());
$identifierField->boost = 5;
$doc->addField($identifierField);
// I18n fields
$doc->addField(Zend_Search_Lucene_Field::Unstored('alternatetitle', $informationObject->getAlternateTitle(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('edition', $informationObject->getEdition(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('extentandmedium', $informationObject->getExtentAndMedium(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('archivalhistory', $informationObject->getArchivalHistory(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('acquisition', $informationObject->getAcquisition(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('appraisal', $informationObject->getAppraisal(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('accruals', $informationObject->getAccruals(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('arrangement', $informationObject->getArrangement(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('accessconditions', $informationObject->getAccessConditions(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('reproductionconditions', $informationObject->getReproductionConditions(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('physicalcharacteristics', $informationObject->getPhysicalCharacteristics(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('findingaids', $informationObject->getFindingAids(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('locationoforiginals', $informationObject->getLocationOfOriginals(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('locationofcopies', $informationObject->getLocationOfCopies(array('culture' => $language))));
$doc->addField(Zend_Search_Lucene_Field::Unstored('relatedunitsofdescription', $informationObject->getRelatedUnitsOfDescription(array('culture' => $language))));
// Subjects
$subjectField = Zend_Search_Lucene_Field::Unstored('subject', $informationObject->getAccessPointsString(QubitTaxonomy::SUBJECT_ID, array('culture' => $language)));
// Boost the hit relevance for the subject field
$subjectField->boost = 5;
$doc->addField($subjectField);
// Place
$placeField = Zend_Search_Lucene_Field::Unstored('place', $informationObject->getAccessPointsString(QubitTaxonomy::PLACE_ID, array('culture' => $language)));
// Boost the hit relevance for the place field
$placeField->boost = 3;
$doc->addField($placeField);
// Names
$nameField = Zend_Search_Lucene_Field::Unstored('name', $informationObject->getNameAccessPointsString(array('culture' => $language)));
// Boost the hit relevance for the place field
$nameField->boost = 3;
$doc->addField($nameField);
$cultureInfo = sfCultureInfo::getInstance($language);
$languages = $cultureInfo->getLanguages();
$scripts = $cultureInfo->getScripts();
// Languages
if (0 < count($properties = $informationObject->getProperties($name = 'language'))) {
$languageCodes = unserialize($properties->offsetGet(0)->getValue(array('sourceCulture' => true)));
if (0 < count($languageCodes)) {
$languageString = '';
foreach ($languageCodes as $languageCode) {
$languageString .= $languages[$languageCode] . ' ';
}
$doc->addField(Zend_Search_Lucene_Field::Unstored('language', rtrim($languageString)));
}
}
// Scripts
if (0 < count($properties = $informationObject->getProperties($name = 'script'))) {
$scriptCodes = unserialize($properties->offsetGet(0)->getValue(array('sourceCulture' => true)));
if (0 < count($scriptCodes)) {
$scriptString = '';
foreach ($scriptCodes as $scriptCode) {
$scriptString .= $scripts[$scriptCode] . ' ';
}
$doc->addField(Zend_Search_Lucene_Field::Unstored('script', rtrim($scriptString)));
}
}
// Notes
if (0 < count($notes = $informationObject->getNotes())) {
$noteString = '';
foreach ($notes as $note) {
$noteString .= $note->getContent(array('culture' => $language)) . ' ';
}
$doc->addField(Zend_Search_Lucene_Field::Unstored('notes', $noteString));
}
// Exclude control area fields for now, maybe add a seperate index for administrative data?
// (institution_responsible_identifier, rules, sources, revision_history)
// To come:
// Add all dynamic metadata fields to index
self::getInstance()->getEngine()->getIndex()->addDocument($doc);
}
