/**
* Get index
* @return \ZendSearch\Lucene\Index
*/
private function index()
{
if (!isset(self::$index)) {
$analyzer = new CaseInsensitive();
if ($this->config()->exists('zend_search', 'stop_words')) {
$stop_word_filter = new StopWords();
$words = $this->getRealPath($this->config()->get('zend_search', 'stop_words'));
if ($words !== false) {
$stop_word_filter->loadFromFile($words);
} else {
throw new \InvalidArgumentException('Path not found');
}
$analyzer->addFilter($stop_word_filter);
}
if ($this->config()->exists('zend_search', 'morphy_dicts')) {
$morphy_dicts = $this->getRealPath($this->config()->get('zend_search', 'morphy_dicts'));
if ($morphy_dicts !== false) {
$analyzer->addFilter(new Morphy($morphy_dicts, $this->config()->getCharset()));
} else {
throw new \InvalidArgumentException('Path not found');
}
}
Analyzer::setDefault($analyzer);
Lucene::setResultSetLimit($this->limit);
QueryParser::setDefaultEncoding($this->config()->getCharset());
$index = $this->config() - get('zend_search', 'index');
$path = $this->getRealPath($index);
self::$index = $path ? Lucene::open($path) : Lucene::create($index);
}
return self::$index;
}
/**
* {@inheritdoc}
*/
public function register(Application $app)
{
Analyzer::setDefault(new CaseInsensitive());
QueryParser::setDefaultEncoding('UTF-8');
$app['zendsearch.indices_path'] = array();
$app['zendsearch.indices.initializer'] = $app->protect(function () use($app) {
static $initialized = false;
if ($initialized) {
return;
}
$initialized = true;
$indices = array();
foreach ($app['zendsearch.indices_path'] as $name => $index) {
$indices[$name] = file_exists($index) ? Lucene::open($index) : Lucene::create($index);
}
$app['zendsearch.indices_collection'] = $indices;
});
$app['zendsearch.indices'] = $app->share(function ($app) {
$app['zendsearch.indices.initializer']();
return $app['zendsearch.indices_collection'];
});
$app['zendsearch.multisearcher'] = $app->share(function ($app) {
$app['zendsearch.indices.initializer']();
$multi = new MultiSearcher();
foreach ($app['zendsearch.indices'] as $index) {
$multi->addIndex($index);
}
return $multi;
});
$app['zendsearch'] = $app->share(function ($app) {
return $app['zendsearch.multisearcher'];
});
}
/**
* @param DB $db
* @param Core $core
* @param string $indexesBasePath
*/
public function __construct(DB $db, Core $core, $indexesBasePath)
{
$this->db = $db;
$this->core = $core;
$this->indexesBasePath = $indexesBasePath;
$this->indexes = array();
Analyzer::setDefault(new CaseInsensitive());
}
/**
* @param string $path Path to search index
*/
public function __construct($path, NodeTypeManagerInterface $nodeTypeManager = null, $hideDestructException = false)
{
$this->path = $path;
$this->filesystem = new Filesystem();
$this->nodeTypeManager = $nodeTypeManager;
$this->hideDestructException = $hideDestructException;
Analyzer::setDefault(new ExactMatchAnalyzer());
Wildcard::setMinPrefixLength(0);
}
/**
* Set analyzer for words highlighting (not for indexing).
*/
public function setHighlighterAnalyzer()
{
/** @var AbstractCommon $analyzer */
$analyzer = App::make('ZendSearch\\Lucene\\Analysis\\Analyzer\\Common\\AbstractCommon');
foreach ($this->filterClasses as $filterClass) {
$analyzer->addFilter(App::make($filterClass));
}
Analyzer::setDefault($analyzer);
}
/**
* Set analyzer for words highlighting (not for indexing).
*/
public function setHighlighterAnalyzer()
{
/** @var AbstractCommon $analyzer */
$analyzer = App::make(AbstractCommon::class);
foreach ($this->filterClasses as $filterClass) {
$analyzer->addFilter(App::make($filterClass));
}
Analyzer::setDefault($analyzer);
}
public function init()
{
if ($this->caseSensitivity) {
Analyzer::setDefault(new Utf8());
} else {
Analyzer::setDefault(new CaseInsensitive());
}
$this->indexDirectory = FileHelper::normalizePath(Yii::getAlias($this->indexDirectory));
$this->luceneIndex = $this->getLuceneIndex($this->indexDirectory);
}
public function init()
{
QueryParser::setDefaultEncoding('UTF-8');
if ($this->caseSensitivity) {
Analyzer::setDefault($this->parseNumeric ? new Utf8Num() : new Utf8());
} else {
Analyzer::setDefault($this->parseNumeric ? new CaseInsensitiveNum() : new CaseInsensitive());
}
$this->indexDirectory = FileHelper::normalizePath(Yii::getAlias($this->indexDirectory));
$this->luceneIndex = $this->getLuceneIndex($this->indexDirectory);
}
public function testAnalyzer()
{
$currentAnalyzer = Analyzer::getDefault();
$this->assertTrue($currentAnalyzer instanceof AnalyzerInterface);
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num */
$newAnalyzer = new Common\Utf8Num();
Analyzer::setDefault($newAnalyzer);
$this->assertTrue(Analyzer::getDefault() === $newAnalyzer);
// Set analyzer to the default value (used in other tests)
Analyzer::setDefault($currentAnalyzer);
}
public function __construct(FrontendController $frontendController, $moduleName)
{
parent::__construct($frontendController, $moduleName);
$this->controllerRoutes = array('/' => array('GET' => 'showSearchResults'), '/rebuild-index' => array('GET' => 'rebuildIndex'));
Analyzer::setDefault(new CaseInsensitive());
QueryParser::setDefaultEncoding('UTF-8');
$this->registerService('index', 'generateIndex');
$cmsPage = $frontendController->getCmsPage();
if ($cmsPage !== null) {
$cmsPage->setLastModified(date('Y-m-d H:i:s'));
}
}
/**
* opens or creates the given lucene index
*
* @throws SetUpException
*/
public function openOrCreate()
{
$indexFolder = $this->files->setUpIndexFolder();
$storage = $indexFolder->getStorage();
$localPath = $storage->getLocalFolder($indexFolder->getInternalPath());
//let lucene search for numbers as well as words
Analyzer::setDefault(new CaseInsensitive());
// can we use the index?
if ($indexFolder->nodeExists('v0.6.0')) {
// correct index present
$this->index = Lucene::open($localPath);
} else {
$this->logger->info('recreating outdated lucene index');
$indexFolder->delete();
$this->index = Lucene::create($localPath);
$indexFolder->newFile('v0.6.0');
}
}
/**
* Get the ZendSearch lucene index instance associated with this instance.
*
* @return \ZendSearch\Lucene\Index
*/
protected function getIndex()
{
if (!$this->index) {
$path = rtrim(Config::get('search.connections.zend.path'), '/') . '/' . $this->name;
try {
$this->index = \ZendSearch\Lucene\Lucene::open($path);
} catch (\ZendSearch\Exception\ExceptionInterface $e) {
$this->index = \ZendSearch\Lucene\Lucene::create($path);
} catch (\ErrorException $e) {
if (!file_exists($path)) {
throw new \Exception("'path' directory does not exist for the 'zend' search driver: '" . rtrim(Config::get('search.connections.zend.path'), '/') . "'");
}
throw $e;
}
\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\Utf8Num\CaseInsensitive());
}
return $this->index;
}
public function search($expression, $page = 1, $conditions = null, $indexName = null)
{
if (!$expression) {
throw new NotFoundHttpException('Empty expression');
}
if (!$indexName) {
$indexName = $this->kernel->getContainer()->getParameter('symbio_fulltext_search.' . Crawler::DEFAULT_INDEX_PARAM);
}
if (mb_strlen($expression, 'utf-8') > 2) {
$index = $this->indexManager->getIndex($indexName);
Analyzer::setDefault(new CaseInsensitive());
$query = $this->prepareQuery($expression, $conditions);
$results = $index->find($query);
// strankovani
$paginator = $this->kernel->getContainer()->get('knp_paginator');
$pagination = $paginator->paginate($results, $page, $this->kernel->getContainer()->getParameter('symbio_fulltext_search.items_on_page'));
return array('expression' => $expression, 'pagination' => $pagination, 'pgdata' => $pagination->getPaginationData());
}
return false;
}
/**
* Gets the index mapped by the given lucene identifier.
*
* @param string $identifier The lucene identifier.
*
* @return \ZendSearch\Lucene\Index The lucene index.
*/
public function getIndex($identifier)
{
$config = $this->getConfig($identifier);
$path = $config['path'];
if (!$this->checkPath($path)) {
$this->indexes[$identifier] = Lucene::create($path);
} else {
$this->indexes[$identifier] = Lucene::open($path);
}
Analyzer::setDefault(new $config['analyzer']());
$this->indexes[$identifier]->setMaxBufferedDocs($config['max_buffered_docs']);
$this->indexes[$identifier]->setMaxMergeDocs($config['max_merge_docs']);
$this->indexes[$identifier]->setMergeFactor($config['merge_factor']);
ZfFilesystem::setDefaultFilePermissions($config['permissions']);
if ($config['auto_optimized']) {
$this->indexes[$identifier]->optimize();
}
QueryParser::setDefaultEncoding($config['query_parser_encoding']);
return $this->indexes[$identifier];
}
/**
* インデックスファイルを生成
*/
public static function updateIndex()
{
if (empty(self::$igo)) {
self::$igo = new Tagger(array('dict_dir' => LIB_DIR . 'ipadic', 'reduce_mode' => true));
}
Analyzer::setDefault(new Utf8());
// 索引の作成
$index = Lucene::create(CACHE_DIR . self::INDEX_NAME);
foreach (Listing::pages() as $page) {
if (empty($page)) {
continue;
}
$wiki = Factory::Wiki($page);
// 読む権限がない場合スキップ
if (!$wiki->isReadable() || $wiki->isHidden()) {
continue;
}
/*
// HTML出力
$html[] = '<html><head>';
$html[] = '<meta http-equiv="Content-type" content="text/html; charset=UTF-8"/>';
$html[] = '<title>' . $wiki->title() . '</title>';
$html[] = '</head>';
$html[] = '<body>' . $wiki->render() . '</body>';
$html[] = '</html>';
*/
$doc = new LuceneDoc();
$doc->addField(Field::Text('title', $wiki->title()));
// Store document URL to identify it in the search results
$doc->addField(Field::Text('url', $wiki->uri()));
// Index document contents
//$contents = join(" ", self::$igo->wakati(strip_tags($wiki->render())));
$contents = strip_tags($wiki->render());
$doc->addField(Field::UnStored('contents', $contents));
// 索引へ文書の登録
$index->addDocument($doc);
}
$index->optimize();
}
/**
* Lists all Post models.
* @return mixed
*/
public function actionIndex()
{
$searchModel = new PostSearch();
$dataProvider = $searchModel->search(Yii::$app->request->post());
//setlocale(LC_ALL, 'en_US.UTF-8');
setlocale(LC_CTYPE, 'ru_RU.UTF-8');
//Lucene\Lucene::setDefaultSearchField('contents');
Lucene\Search\QueryParser::setDefaultEncoding('UTF-8');
Lucene\Analysis\Analyzer\Analyzer::setDefault(new Lucene\Analysis\Analyzer\Common\Utf8\CaseInsensitive());
Lucene\Lucene::setResultSetLimit(10);
// create blog posts index located in /data/posts_index ,make sure the folder is writable
$index = Lucene\Lucene::create('data/posts_index');
$posts = Post::find()->all();
//var_dump($posts);die();
// iterate through posts and build the index
foreach ($posts as $p) {
$doc = new Lucene\Document();
$doc->addField(Lucene\Document\Field::UnIndexed('entry_id', $p->id));
$doc->addField(Lucene\Document\Field::Keyword('title', $p->title));
$doc->addField(Lucene\Document\Field::text('contents', $p->content));
$index->addDocument($doc);
}
// commit the index
$index->commit();
//Lucene\Analysis\Analyzer\Analyzer::setDefault(new Lucene\Analysis\Analyzer\Common\Utf8\CaseInsensitive());
// explode the search query to individual words
$words = explode(' ', urldecode(Yii::$app->getRequest()->getQueryParam('q')));
// start a search query and add a term for each word to it
$query = new Lucene\Search\Query\MultiTerm();
foreach ($words as $w) {
$query->addTerm(new Lucene\Index\Term($w));
}
// open and query the index
$index = Lucene\Lucene::open('data/posts_index');
$results = $index->find($query);
// the search results
//var_dump($results);
return $this->render('index', ['searchModel' => $searchModel, 'dataProvider' => $dataProvider, 'search' => $results, 'query' => $query]);
}
protected function getIndex()
{
if ($this->index != null) {
return $this->index;
}
\ZendSearch\Lucene\Search\QueryParser::setDefaultEncoding('utf-8');
\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\Utf8Num\CaseInsensitive());
\ZendSearch\Lucene\Search\QueryParser::setDefaultOperator(\ZendSearch\Lucene\Search\QueryParser::B_AND);
try {
$index = \ZendSearch\Lucene\Lucene::open($this->getIndexPath());
} catch (\ZendSearch\Lucene\Exception\RuntimeException $ex) {
$index = \ZendSearch\Lucene\Lucene::create($this->getIndexPath());
}
$this->index = $index;
return $index;
}
public function luceneSearchAetCommunications($index, $searchKeyWord)
{
\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\Text\CaseInsensitive());
$dbIds = array();
$searchValue = SearchHelper::utf8_to_ascii(mb_strtolower($searchKeyWord, "UTF-8"));
$em = $this->getDoctrine()->getManager();
/*
$term1 = new \ZendSearch\Lucene\Index\Term($searchValue, 'firstname');
//$subquery1 = new \ZendSearch\Lucene\Search\Query\Term($term1);
$term2 = new \ZendSearch\Lucene\Index\Term($searchValue, 'title');
//$subquery2 = new \ZendSearch\Lucene\Search\Query\Term($term2);
$term3 = new \ZendSearch\Lucene\Index\Term($searchValue, 'shortdesc');
//$subquery3 = new \ZendSearch\Lucene\Search\Query\Term($term3);
$term4 = new \ZendSearch\Lucene\Index\Term($searchValue, 'body');
//$subquery4 = new \ZendSearch\Lucene\Search\Query\Term($term4);
$term5 = new \ZendSearch\Lucene\Index\Term($searchValue, 'author');
//$subquery5 = new \ZendSearch\Lucene\Search\Query\Term($term5);
$terms = array($term1, $term2, $term3, $term4, $term5);
//$subqueries = array($subquery1, $subquery2, $subquery3, $subquery4, $subquery5);
$signs = array(null, null, null, null, null);
$termsQuery = new \ZendSearch\Lucene\Search\Query\MultiTerm($terms,$signs);
*/
\ZendSearch\Lucene\Search\QueryParser::setDefaultOperator(\ZendSearch\Lucene\Search\QueryParser::B_OR);
$query = \ZendSearch\Lucene\Search\QueryParser::parse($searchValue, 'UTF-8');
$foundDocuments = $index->find($query);
//$docNum = count($foundDocuments);
foreach ($foundDocuments as $foundDoc) {
$dbIds[] = $foundDoc->dbId;
}
$results = $em->getRepository('AetCommunicationBundle:Communication')->findById($dbIds);
return $results;
}
private function getIndex() : SearchIndexInterface
{
$path = $this->getIndexPath();
if (!$this->checkIndexPath($path)) {
$index = Lucene::create($path);
} else {
$index = Lucene::open($path);
}
Analyzer::setDefault(new CaseInsensitive());
LuceneFilesystem::setDefaultFilePermissions(0775);
QueryParser::setDefaultEncoding('UTF-8');
$index->setMaxBufferedDocs($this->options['max_buffered_docs']);
$index->setMaxMergeDocs($this->options['max_merge_docs']);
$index->setMergeFactor($this->options['merge_factor']);
$index->optimize();
return $index;
}
public function luceneSearchAetUsers($index, $searchKeyWord)
{
\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\Text\CaseInsensitive());
$dbIds = array();
$searchValue = SearchHelper::utf8_to_ascii(mb_strtolower($searchKeyWord, "UTF-8"));
$em = $this->getDoctrine()->getManager();
/*
$term1 = new \ZendSearch\Lucene\Index\Term($searchValue, 'firstname');
//$subquery1 = new \ZendSearch\Lucene\Search\Query\Term($term1);
$term2 = new \ZendSearch\Lucene\Index\Term($searchValue, 'lastname');
//$subquery2 = new \ZendSearch\Lucene\Search\Query\Term($term2);
$term3 = new \ZendSearch\Lucene\Index\Term($searchValue, 'activiteprincipale');
//$subquery3 = new \ZendSearch\Lucene\Search\Query\Term($term3);
$term4 = new \ZendSearch\Lucene\Index\Term($searchValue, 'codepostal');
//$subquery4 = new \ZendSearch\Lucene\Search\Query\Term($term4);
$term5 = new \ZendSearch\Lucene\Index\Term($searchValue, 'email');
//$subquery5 = new \ZendSearch\Lucene\Search\Query\Term($term5);
$term6 = new \ZendSearch\Lucene\Index\Term($searchValue, 'matricule');
//$subquery6 = new \ZendSearch\Lucene\Search\Query\Term($term6);
$term7 = new \ZendSearch\Lucene\Index\Term($searchValue, 'pays');
//$subquery7 = new \ZendSearch\Lucene\Search\Query\Term($term7);
$term8 = new \ZendSearch\Lucene\Index\Term($searchValue, 'promotion');
//$subquery8 = new \ZendSearch\Lucene\Search\Query\Term($term8);
$term9 = new \ZendSearch\Lucene\Index\Term($searchValue, 'telephone');
//$subquery9 = new \ZendSearch\Lucene\Search\Query\Term($term9);
$term10 = new \ZendSearch\Lucene\Index\Term($searchValue, 'ville');
//$subquery10 = new \ZendSearch\Lucene\Search\Query\Term($term10);
$term11 = new \ZendSearch\Lucene\Index\Term($searchValue, 'whoami');
//$subquery11 = new \ZendSearch\Lucene\Search\Query\Term($term11);
//$subqueries = array($subquery1, $subquery2, $subquery3, $subquery4, $subquery5, $subquery6, $subquery7, $subquery8, $subquery9, $subquery10, $subquery11);
$terms = array($term1, $term2, $term3, $term4, $term5, $term6, $term7, $term8, $term9, $term10, $term11);
$signs = array(null, null, null, null, null, null, null, null, null, null, null);
$termsQuery = new \ZendSearch\Lucene\Search\Query\MultiTerm($terms, $signs);
//$boolQuery = new \ZendSearch\Lucene\Search\Query\Boolean($subqueries, $signs);
*/
\ZendSearch\Lucene\Search\QueryParser::setDefaultOperator(\ZendSearch\Lucene\Search\QueryParser::B_OR);
$query = \ZendSearch\Lucene\Search\QueryParser::parse($searchValue, 'UTF-8');
$foundDocuments = $index->find($query);
//$docNum = count($foundDocuments);
foreach ($foundDocuments as $foundDoc) {
$dbIds[] = $foundDoc->dbId;
}
$results = $em->getRepository('AetAnnuaireBundle:User')->findById($dbIds);
return $results;
}
/**
* Подсветка результата поиска в html-фрагменте
*
* @param string $inputHTMLFragment исходный фрагмента html
* @param string $inputEncoding Кодировка исходного фрагмента html
* @param string $outputEncoding Кодировка резульрирующего фрагмента html
* @return string html фрагмент с подсвеченными результатами поиска
*/
public function highlightMatches($inputHTMLFragment, $inputEncoding = 'utf-8', $outputEncoding = 'utf-8')
{
$highlightedHTMLFragment = '';
if (!empty($this->lastQuery)) {
$queryParser = QueryParser::parse($this->lastQuery);
/**
* Убираем фильтры стоп-слов для подсветки слов с псевдокорнями типа 'под' и т.п.
*/
Analyzer::setDefault($this->analyzerForHighlighter);
$highlightedHTMLFragment = $queryParser->htmlFragmentHighlightMatches($inputHTMLFragment, $inputEncoding, new Highlighter());
Analyzer::setDefault($this->defaultAnalyzer);
$highlightedHTMLFragment = mb_convert_encoding($highlightedHTMLFragment, $outputEncoding, 'utf-8');
}
return $highlightedHTMLFragment;
}
