protected function useSearchIndex($slug)
{
if (!dmConfig::get('smart_404')) {
return false;
}
try {
$searchIndex = $this->serviceContainer->get('search_engine')->getCurrentIndex();
$queryString = str_replace('/', ' ', dmString::unSlugify($slug));
$query = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
$results = $searchIndex->search($query);
$foundPage = null;
foreach ($results as $result) {
if ($result->getScore() > 0.5) {
if ($foundPage = $result->getPage()) {
break;
}
} else {
break;
}
}
if ($foundPage) {
return $this->serviceContainer->getService('helper')->link($foundPage)->getHref();
}
} catch (Exception $e) {
$this->dispatcher->notify(new sfEvent($this, 'application.log', array('Can not use search index to find redirection for slug ' . $slug, sfLogger::ERR)));
if (sfConfig::get('dm_debug')) {
throw $e;
}
}
}
public static function searchLuceneWithValues(Doctrine_Table $table, $luceneQueryString, $culture = null)
{
// Ugh: UTF8 Lucene is case sensitive work around this
if (function_exists('mb_strtolower')) {
$luceneQueryString = mb_strtolower($luceneQueryString);
} else {
$luceneQueryString = strtolower($luceneQueryString);
}
// We have to register the autoloader before we can use these classes
self::registerZend();
$luceneQuery = Zend_Search_Lucene_Search_QueryParser::parse($luceneQueryString);
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->addSubquery($luceneQuery, true);
if (!is_null($culture)) {
$culture = self::normalizeCulture($culture);
$cultureTerm = new Zend_Search_Lucene_Index_Term($culture, 'culture');
// Oops, this said $aTerm before. Thanks to Quentin Dugauthier
$cultureQuery = new Zend_Search_Lucene_Search_Query_Term($cultureTerm);
$query->addSubquery($cultureQuery, true);
}
$index = $table->getLuceneIndex();
$hits = $index->find($query);
$ids = array();
foreach ($hits as $hit) {
$ids[$hit->primarykey] = $hit;
}
return $ids;
}
/**
* Method to perform a search. This function returns results from Zend_Lucene_Search
* and is unpolished as it doesn't check permission issues, etc.
*
* @access private
* @param string $text Text to search for
* @param string $module Module to search items for
* @param string $context Context in which to search
* @param array $extra Any additional items
* @return object
*/
private function search($text, $module = NULL, $context = NULL, $extra = array())
{
$objIndexData = $this->getObject('indexdata');
$indexer = $objIndexData->checkIndexPath();
// Some cllean up till we can find a better way
$phrase = trim(str_replace(':', ' ', $text));
$phrase = trim(str_replace('+', ' ', $phrase));
$phrase = trim(str_replace('-', ' ', $phrase));
if ($module != NULL) {
if ($text != '') {
$phrase .= ' AND ';
}
$phrase .= ' module:' . $module . ' ';
}
if ($context != NULL) {
if ($phrase != '') {
$phrase .= ' AND ';
}
$phrase .= ' context:' . $context . ' ';
}
if (is_array($extra) && count($extra) > 0) {
foreach ($extra as $item => $value) {
if ($phrase != '') {
$phrase .= ' AND ';
}
$phrase .= $item . ':' . $value . ' ';
}
}
//echo $phrase;
$query = Zend_Search_Lucene_Search_QueryParser::parse($phrase);
return $indexer->find($query);
}
public function index($msg = NULL)
{
$results = NULL;
$results2 = NULL;
$query = NULL;
$view = new View('search_example');
$view->bind("results", $results)->bind("results2", $results2)->bind("query", $query)->bind("msg", $msg);
if (!empty($_GET["q"])) {
try {
$query = $_GET["q"];
$form = $_GET["form"];
if ($form == "artists") {
$results = Search::instance()->find($query);
} else {
Search::instance()->load_search_libs();
$query = Zend_Search_Lucene_Search_QueryParser::parse($query);
$hits = Search::instance()->find($query);
if (sizeof($hits) > 0) {
$results2 = $query->highlightMatches(iconv('UTF-8', 'ASCII//TRANSLIT', $hits[0]->body));
} else {
$results2 = '<p style="color:#f00">No results found</p>';
}
}
} catch (Exception $e) {
Kohana::log("error", $e);
}
}
$view->render(TRUE);
}
public function actionIndex()
{
$this->webpageType = 'SearchResultsPage';
$type = '';
if (isset($_GET['type'])) {
$type = $_GET['type'];
}
if (isset($_GET['q'])) {
// $originalQuery = $_GET['q'];
$queryString = $originalQuery = isset($_GET['q']) ? $_GET['q'] : '';
$index = new Zend_Search_Lucene($this->_indexFile);
//only look for queryString in title and content, well if advanced search techniques aren't used
if (!(strpos(strtolower($queryString), 'and') || strpos(strtolower($queryString), 'or') || strpos(strtolower($queryString), ':'))) {
$queryString = "title:{$queryString} OR content:{$queryString}";
}
if ($type) {
$queryString .= ' AND type:' . $type;
}
$results = $index->find($queryString);
$query = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
$this->render('index', compact('results', 'originalQuery', 'query', 'type'));
} else {
$this->render('advanced', array('type' => $type));
}
}
/**
* Adds a string that is parsed into Zend API queries
* @param string $query The query to parse
* @param string $encoding The encoding to parse query as
*
* @return sfLuceneCriteria
*/
public function addString($query, $encoding = null, $type = true)
{
$this->search->configure();
// setup query parser
$this->add(Zend_Search_Lucene_Search_QueryParser::parse($query, $encoding), $type);
return $this;
}
public function search($keywords, $charset = 'utf-8')
{
$index = new Zend_Search_Lucene(ZY_ROOT . '/index');
$query = Zend_Search_Lucene_Search_QueryParser::parse($keywords, $charset);
$hits = $index->find($query);
return $hits;
}
/**
* Process and render search results. Uses the Lucene_results.ss template to
* render the form.
*
* @access public
* @param array $data The raw request data submitted by user
* @param Form $form The form instance that was submitted
* @param SS_HTTPRequest $request Request generated for this action
* @return String The rendered form, for inclusion into the page template.
*/
public function ZendSearchLuceneResults($data, $form, $request)
{
$querystring = $form->Fields()->dataFieldByName('Search')->dataValue();
$query = Zend_Search_Lucene_Search_QueryParser::parse($querystring);
$hits = ZendSearchLuceneWrapper::find($query);
$data = $this->getDataArrayFromHits($hits, $request);
return $this->owner->customise($data)->renderWith(array('Lucene_results', 'Page'));
}
public function searchTag($tag, $value)
{
if (!$this->_enabled) {
return array();
}
Zend_Search_Lucene::setDefaultSearchField($tag);
$query = Zend_Search_Lucene_Search_QueryParser::parse($value);
return $this->_index->find($query);
}
private function search($terms)
{
Yii::import('application.vendor.*');
require_once 'Zend/Search/Lucene.php';
$index = new Zend_Search_Lucene(Yii::getPathOfAlias('application.index'));
$result = $index->find($terms);
$query = Zend_Search_Lucene_Search_QueryParser::parse($terms);
return $result;
}
// public function init(){
// Yii::import('application.vendors.*');
// require_once('Zend/Search/Lucene.php');
// parent::init();
// }
public function actionSearch()
{
$this->layout = 'column2';
if (($term = Yii::app()->getRequest()->getParam('q', null)) !== null) {
$index = new Zend_Search_Lucene(Yii::getPathOfAlias('application.' . $this->_indexFiles));
public function parse($query)
{
$query = Zend_Search_Lucene_Search_QueryParser::parse($query, 'UTF-8');
if ($query instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
throw new Exception('No search terms specified.');
} elseif ($query instanceof Zend_Search_Lucene_Search_Query_MultiTerm) {
throw new Exception('Error parsing search terms.');
}
return $query;
}
public function searchAction()
{
if ($this->_request->isPost() && Digitalus_Filter_Post::has('submitSearchForm')) {
$index = Zend_Search_Lucene::open('./application/modules/search/data/index');
$queryString = Digitalus_Filter_Post::get('keywords');
$query = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
$this->view->searchResults = $index->find($query);
$this->view->keywords = $queryString;
}
}
public function testQueryParser()
{
$queries = array('title:"The Right Way" AND text:go', 'title:"Do it right" AND right', 'title:Do it right', '"jakarta apache"~10', 'jakarta apache', 'jakarta^4 apache', '"jakarta apache"^4 "Apache Lucene"', '"jakarta apache" jakarta', '"jakarta apache" OR jakarta', '"jakarta apache" || jakarta', '"jakarta apache" AND "Apache Lucene"', '"jakarta apache" && "Apache Lucene"', '+jakarta apache', '"jakarta apache" AND NOT "Apache Lucene"', '"jakarta apache" && !"Apache Lucene"', 'NOT "jakarta apache"', '!"jakarta apache"', '"jakarta apache" -"Apache Lucene"', '(jakarta OR apache) AND website', '(jakarta || apache) && website', 'title:(+return +"pink panther")', 'title:(+re\\turn\\ value +"pink panther\\"" +body:cool)', '+type:1 +id:5', 'type:1 AND id:5', 'f1:word1 f1:word2 and f1:word3', 'f1:word1 not f1:word2 and f1:word3');
$rewritedQueries = array('+(title:"the right way") +(text:go)', '+(title:"do it right") +(path:right modified:right contents:right)', '(title:do) (path:it modified:it contents:it) (path:right modified:right contents:right)', '((path:"jakarta apache"~10) (modified:"jakarta apache"~10) (contents:"jakarta apache"~10))', '(path:jakarta modified:jakarta contents:jakarta) (path:apache modified:apache contents:apache)', '((path:jakarta modified:jakarta contents:jakarta)^4)^4 (path:apache modified:apache contents:apache)', '((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache"))^4 ((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) (path:jakarta modified:jakarta contents:jakarta)', '((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) (path:jakarta modified:jakarta contents:jakarta)', '((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) (path:jakarta modified:jakarta contents:jakarta)', '+((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) +((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '+((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) +((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '+(path:jakarta modified:jakarta contents:jakarta) (path:apache modified:apache contents:apache)', '+((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) -((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '+((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) -((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '<EmptyQuery>', '<EmptyQuery>', '((path:"jakarta apache") (modified:"jakarta apache") (contents:"jakarta apache")) -((path:"apache lucene") (modified:"apache lucene") (contents:"apache lucene"))', '+((path:jakarta modified:jakarta contents:jakarta) (path:apache modified:apache contents:apache)) +(path:website modified:website contents:website)', '+((path:jakarta modified:jakarta contents:jakarta) (path:apache modified:apache contents:apache)) +(path:website modified:website contents:website)', '(+(title:return) +(title:"pink panther"))', '(+(+title:return +title:value) +(title:"pink panther") +(body:cool))', '+(<EmptyQuery>) +(<EmptyQuery>)', '+(<EmptyQuery>) +(<EmptyQuery>)', '(f1:word) (+(f1:word) +(f1:word))', '(f1:word) (-(f1:word) +(f1:word))');
$index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_files/_indexSample');
foreach ($queries as $id => $queryString) {
$query = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
$this->assertTrue($query instanceof Zend_Search_Lucene_Search_Query);
$this->assertEquals($query->rewrite($index)->__toString(), $rewritedQueries[$id]);
}
}
function __construct($query)
{
$qstr = $query->__toString();
// query needs the object_type field removing for highlighting
$qstr = preg_replace('/\\+?\\(\\(object_type.*?\\)\\)/', '', $qstr);
// this is the only way i can find to remove a term form a query
$query = Zend_Search_Lucene_Search_QueryParser::parse($qstr, 'UTF-8');
// rebuild
$this->query = $query;
$this->snippetHelper = new Search_ResultSet_SnippetHelper();
}
public function actionSearch()
{
//working.
$this->layout = 'column2';
if (($term = Yii::app()->getRequest()->getParam('q', null)) !== null) {
Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum_CaseInsensitive());
$index = new Zend_Search_Lucene(Yii::getPathOfAlias('application.' . $this->_indexFiles));
$results = $index->find($term);
$query = Zend_Search_Lucene_Search_QueryParser::parse($term);
$this->render('search', compact('results', 'term', 'query'));
}
}
public function parse()
{
$data = $this->_data;
$query = Zend_Search_Lucene_Search_QueryParser::parse($data['query']);
if ('' != $data['pricefrom'] && '' != $data['priceto']) {
$from = new Zend_Search_Lucene_Index_Term($this->_formatPrice($data['pricefrom']), 'price');
$to = new Zend_Search_Lucene_Index_Term($this->_formatPrice($data['priceto']), 'price');
$q = new Zend_Search_Lucene_Search_Query_Range($from, $to, true);
$query = Zend_Search_Lucene_Search_QueryParser::parse($data['query'] . ' +' . $q);
}
return $query;
}
/**
* php index.php docx search "Licence"
*
* @param $phrase
* @return mixed|void
*/
public function search($phrase)
{
$index = self::open(APP_PATH . '/' . self::INDEX_DIR);
echo sprintf("Keyword : %s\n\n", $phrase);
$phrase = iconv('gbk', 'utf-8', $phrase);
$phrase = Zend_Search_Lucene_Search_QueryParser::parse($phrase, "utf-8");
$results = $index->find($phrase);
foreach ($results as $index => $hit) {
echo sprintf("%s : %s - %s\n", $index, basename($hit->filename), $hit->score);
}
echo "\n###Done###\n";
}
public function indexAction()
{
if ($this->_request->isPost()) {
$keywords = $this->_request->getParam('keywords');
$query = Zend_Search_Lucene_Search_QueryParser::parse($keywords);
$index = Zend_Search_Lucene::open(APPLICATION_PATH . '/indexes');
$hits = $index->find($query);
$this->view->results = $hits;
$this->view->keywords = $keywords;
} else {
$this->view->results = null;
}
}
public function searchAction()
{
$query = $this->getRequest()->getParam('q');
if ($query) {
$qo = Zend_Search_Lucene_Search_QueryParser::parse($query);
$options = $this->getInvokeArg('bootstrap')->getOption('lucene');
$luceneDir = $options['dir'];
$index = Zend_Search_Lucene::open($luceneDir);
$results = $index->find($query);
$this->view->matches = $results;
$this->view->query = $qo;
}
}
public function actionSearch()
{
$indexFiles = Yii::app()->getModule('zendsearch')->indexFiles;
SetLocale(LC_ALL, 'ru_RU.UTF-8');
Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive());
Zend_Search_Lucene_Search_QueryParser::setDefaultEncoding('UTF-8');
if (($term = Yii::app()->getRequest()->getQuery('q', null)) !== null) {
$index = new Zend_Search_Lucene(Yii::getPathOfAlias('application.' . $indexFiles));
$results = $index->find($term);
$query = Zend_Search_Lucene_Search_QueryParser::parse($term);
$this->render('search', compact('results', 'term', 'query'));
}
}
public static function highlightIfSuitable($html, $request_uri, $referer)
{
if (self::suitable($request_uri) && ($query = self::query($referer))) {
$queryObj = Zend_Search_Lucene_Search_QueryParser::parse($query);
$out = $queryObj->highlightMatches($html);
if (!$out) {
return $html;
}
$htmlNice = self::joinHtml($html, $out);
if ($htmlNice) {
return $htmlNice;
}
}
return $html;
}
public function searchAction()
{
$searchForm = new Search_Form();
if ($this->_request->isPost() && $searchForm->isValid($_POST) && Digitalus_Filter_Post::has('submitSearchForm')) {
$index = Zend_Search_Lucene::open(APPLICATION_PATH . '/modules/search/data/index');
$queryString = Digitalus_Filter_Post::get('keywords');
$query = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
$this->view->searchResults = $index->find($query);
if (!empty($queryString)) {
$keywordsElement = $searchForm->getElement('keywords');
$keywordsElement->setValue($queryString);
}
}
$this->view->form = $searchForm;
}
function searchDocsByContent($q)
{
$hits = array();
try {
$this->initLuceneEngine();
$indexer = $this->zend->get_Zend_Search_Lucene();
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$subquery = Zend_Search_Lucene_Search_QueryParser::parse('+(' . $q . ')');
$query->addSubquery($subquery, true);
// $query->addSubquery(self::makeTermQuery('object_type', JS_TEXT_DATA), true);
$hits = $indexer->find($query);
return $hits;
} catch (Exception $e) {
echo $e->getTraceAsString();
}
return $hits;
}
function searchDocsByContent($q, $userid)
{
$hits = array();
try {
$this->initLuceneEngine();
$indexer = $this->zend->get_Zend_Search_Lucene();
echo $indexer->count();
$query = Zend_Search_Lucene_Search_QueryParser::parse("(title:{$q} OR keywords:{$q} OR content:{$q}) AND userid:{$userid}");
// var_dump($query);
$hits = $indexer->find($query);
} catch (Exception $e) {
echo $e->getTraceAsString();
}
// var_dump($indexer);
// var_dump($hits); exit;
return $hits;
}
/**
* DOCUMENT ME
* @param Doctrine_Table $table
* @param mixed $luceneQueryString
* @param mixed $culture
* @param mixed $andLuceneQuery
* @return mixed
*/
public static function searchLuceneWithValues(Doctrine_Table $table, $luceneQueryString, $culture = null, $andLuceneQuery = null)
{
// Ugh: UTF8 Lucene is case sensitive work around this
if (function_exists('mb_strtolower')) {
$luceneQueryString = mb_strtolower($luceneQueryString);
} else {
$luceneQueryString = strtolower($luceneQueryString);
}
// We have to register the autoloader before we can use these classes
self::registerZend();
// Specify character set. Apostrophe is always UTF-8
$luceneQuery = Zend_Search_Lucene_Search_QueryParser::parse($luceneQueryString, 'utf-8');
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->addSubquery($luceneQuery, true);
if (!is_null($culture)) {
$culture = self::normalizeCulture($culture);
$cultureTerm = new Zend_Search_Lucene_Index_Term($culture, 'culture');
// Oops, this said $aTerm before. Thanks to Quentin Dugauthier
$cultureQuery = new Zend_Search_Lucene_Search_Query_Term($cultureTerm);
$query->addSubquery($cultureQuery, true);
}
if (!is_null($andLuceneQuery)) {
$query->addSubquery($andLuceneQuery, true);
}
$index = $table->getLuceneIndex();
$hits = $index->find($query);
// Never look at more than n results, no matter what. This is necessary
// to avoid out of memory errors on large sites. We suggest setting this
// to 1000. Note that if 1,000 locked pages precede the first unlocked page
// and you are logged out, you could theoretically not get your result. In
// practice unlocked pages tend to be prominent and come up early. Thanks to
// recent improvements that avoid holding on to the memory for most of the
// information about a search result indefinitely you can set this higher
// than previously
if (sfConfig::get('app_a_search_hard_limit', false)) {
$hits = array_splice($hits, 0, sfConfig::get('app_a_search_hard_limit'));
}
$ids = array();
foreach ($hits as $hit) {
// New way: don't touch anything but $hit->id directly and you won't force a persistent
// use of memory for the lazy loaded columns http://zendframework.com/issues/browse/ZF-8267
$ids[$index->getDocument($hit->id)->getFieldValue('primarykey')] = $hit;
}
return $ids;
}
/**
* php index.php db search "City:London"
* php index.php db search "Tofu"
* php index.php db search "Seafood"
*
* @param $phrase
* @return mixed|void
*/
public function search($phrase)
{
$index = self::open(APP_PATH . '/' . self::INDEX_DIR);
echo sprintf("Keyword : %s\n\n", $phrase);
$phrase = iconv('gbk', 'utf-8', $phrase);
$phrase = Zend_Search_Lucene_Search_QueryParser::parse($phrase, "utf-8");
$results = $index->find($phrase);
foreach ($results as $index => $hit) {
$fields = $hit->getDocument($hit->id)->getFieldNames();
echo sprintf("[%s] : Score: %s\n", $hit->id, $hit->score);
foreach ($fields as $field) {
if ($field == 'Description' || $field == 'Picture') {
continue;
}
echo sprintf("%s: %s\n", $field, $hit->{$field});
}
echo "----------------------------\n";
}
}
public function indexAction()
{
if (!isset($_POST['search'])) {
$this->_redirect('/');
}
$options = $this->getInvokeArg('bootstrap')->getOptions();
try {
$index = Zend_Search_Lucene::open($options['search']['indexPath']);
} catch (Zend_Search_Lucene_Exception $e) {
$index = Zend_Search_Lucene::create($options['search']['indexPath']);
}
try {
$query = Zend_Search_Lucene_Search_QueryParser::parse($_POST['search'], 'utf-8');
$hits = $index->find($_POST['search']);
} catch (Zend_Search_Lucene_Exception $e) {
$hits = array();
}
$this->view->results = $hits;
}
public function userSearch(Kwf_Component_Data $subroot, $queryString, $offset, $limit, $params = array())
{
$index = Kwf_Util_Fulltext_Lucene::getInstance($subroot);
$error = false;
$userQuery = false;
if ($queryString) {
try {
$userQuery = Zend_Search_Lucene_Search_QueryParser::parse($queryString);
} catch (ErrorException $e) {
//ignore iconv errors that happen with invalid input
}
}
$hits = array();
if ($userQuery) {
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->addSubquery($userQuery, true);
if (isset($params['type'])) {
$pathTerm = new Zend_Search_Lucene_Index_Term($params['type'], 'type');
$pathQuery = new Zend_Search_Lucene_Search_Query_Term($pathTerm);
$query->addSubquery($pathQuery, true);
}
$time = microtime(true);
try {
$hits = $index->find($query);
} catch (Zend_Search_Lucene_Exception $e) {
$error = $subroot->trlKwf('Invalid search terms');
}
}
$ret = array();
if (count($hits)) {
$numStart = $offset;
$numEnd = min(count($hits), $offset + $limit);
for ($i = $numStart; $i < $numEnd; $i++) {
$h = $hits[$i];
$c = Kwf_Component_Data_Root::getInstance()->getComponentById($h->componentId);
if ($c) {
$ret[] = array('data' => $c, 'content' => $h->content);
}
}
}
return array('error' => $error, 'numHits' => count($hits), 'hits' => $ret);
}
public function search($apiKey, $keywords)
{
if (!$this->_validateKey($apiKey)) {
return array('error' => 'invalid api key', 'status' => false);
}
// fetch the index and run the search
$query = Zend_Search_Lucene_Search_QueryParser::parse($keywords);
$index = Zend_Search_Lucene::open(APPLICATION_PATH . '/indexes');
$hits = $index->find($query);
// build the response array
if (is_array($hits) && count($hits) > 0) {
$response['hits'] = count($hits);
foreach ($hits as $page) {
$pageObj = new CMS_Content_Item_Page($page->page_id);
$response['results']['page_' . $page->page_id] = $pageObj->toArray();
}
} else {
$response['hits'] = 0;
}
}
