/**
* Get text search field
* @param string $name
* @param string $value
* @return Field
*/
public function text($name, $value = null)
{
if ($value === null) {
$value = $this->entity->getValue($name);
}
return Field::text($name, $value, $this->config()->getCharset());
}
public function createDocument(EntityInterface $entity) : Document
{
$document = new Document();
$context = $this->createContext($entity);
$fieldsCollection = $context->getFieldsCollection();
$document->addField(Field::unIndexed('identifier', $entity->getId()));
$fieldsCollection->map(function (SearchField $field) use($document) {
$document->addField(Field::text($field->getName(), $field->getValue()));
});
return $document;
}
/**
* {@inheritdoc}
*/
public function addProduct(ProductInterface $product, $indexName = ProductIndexerInterface::DEFAULT_INDEX_NAME)
{
$index = $this->searchIndexManager->getIndex(ProductIndexerInterface::DEFAULT_INDEX_NAME);
$document = new Document();
$document->addField(Field::unIndexed('identifier', $product->getId()));
$document->addField(Field::text('name', $product->translate('en')->getName()));
$document->addField(Field::text('shortDescription', $product->translate()->getShortDescription()));
$document->addField(Field::text('description', $product->translate()->getDescription()));
$index->addDocument($document);
$index->commit();
}
/**
* Create or update an indexed document
*
* @param object $object
*/
public function index($object)
{
// create property accessor
$accessor = PropertyAccess::createPropertyAccessor();
// delete existing documents with same id
foreach ($this->index->find('id:' . $accessor->getValue($object, 'id')) as $hit) {
$this->index->delete($hit->id);
}
// create new Lucene document
$doc = new Document();
// add primary key to identify it in the search results
$doc->addField(Field::keyword('id', $accessor->getValue($object, 'id')));
// add entity class reference to identify it in the search results
$doc->addField(Field::unIndexed('entityClass', get_class($object)));
// analyze each property's annotations to see which ones must be add to the document
$reflClass = new ReflectionClass($object);
foreach ($reflClass->getProperties() as $property) {
$reflProperty = new \ReflectionProperty($object, $property->name);
$annotation = $this->reader->getPropertyAnnotation($reflProperty, '\\Keratine\\Lucene\\Mapping\\Annotation\\DocumentField');
if ($annotation) {
$value = $accessor->getValue($object, $property->name);
$value = $this->ensureString($value);
// use the appropriate indexing strategy for the field
switch ($annotation->type) {
case 'keyword':
$doc->addField(Field::keyword($property->name, $value, 'UTF-8'));
break;
case 'unIndexed':
$doc->addField(Field::unIndexed($property->name, $value, 'UTF-8'));
break;
case 'binary':
$doc->addField(Field::binary($property->name, $value));
break;
case 'text':
$doc->addField(Field::text($property->name, $value, 'UTF-8'));
break;
case 'unStored':
default:
$doc->addField(Field::unStored($property->name, $value, 'UTF-8'));
break;
}
}
}
// add the document to the index and commit it
$this->index->addDocument($doc);
$this->index->commit();
}
function wavProcess($index, $doc, $documentPath)
{
$fileName = basename($documentPath);
$fileNameField = \ZendSearch\Lucene\Document\Field::text('filename', $fileName);
//APPEL DU FICHIER PYTHON
exec('python ../xmpParse.py ' . $fileName, $res, $retcode);
// Title
$title = $res[0];
$titleField = \ZendSearch\Lucene\Document\Field::text('title', $title);
// echo "Title : " . $title[0] . PHP_EOL;
// Subject
for ($i = 1; $i < count($res); $i++) {
$subject = $res[$i];
$subjectField = \ZendSearch\Lucene\Document\Field::text($subject, $subject);
$doc->addField($subjectField);
}
// echo "Subject : " . $subject[0] . PHP_EOL;
$doc->addField($fileNameField);
$doc->addField($titleField);
$index->addDocument($doc);
}
/**
* 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]);
}
public function generateSearchAction()
{
$searchIndexLocation = $this->getIndexLocation();
$index = Lucene\Lucene::create($searchIndexLocation);
$allUsers = $this->getUserTable()->fetchAll(false);
foreach ($allUsers as $user) {
$id = Document\Field::keyword('userId', $user->userId);
$firstName = Document\Field::text('firstName', $user->firstName);
$lastName = Document\Field::text('lastName', $user->lastName);
$email = Document\Field::text('email', $user->email);
$role = Document\Field::text('role', $user->role);
$activated = Document\Field::keyword('activated', $user->activated);
$indexDoc = new Lucene\Document();
$indexDoc->addField($id);
$indexDoc->addField($firstName);
$indexDoc->addField($lastName);
$indexDoc->addField($email);
$indexDoc->addField($role);
$indexDoc->addField($activated);
$index->addDocument($indexDoc);
}
$index->commit();
}
/**
* create document from configured fields within extracted data
* @param string $url
* @param array $page
* @return Document
*/
protected function createDocument($url, $page)
{
$document = new Document();
if (!isset($page['status_code'])) {
$page['status_code'] = 00;
//tmp
}
setlocale(LC_ALL, "cs_CZ.UTF-8");
$document->addField(Field::keyword('url', $url));
// ancestor URLs to search by URL
$urlParts = parse_url($url);
if (isset($urlParts['path']) && $urlParts['path'] && strlen($urlParts['path']) > 1) {
$uri = $urlParts['path'];
$uris = array($uri);
do {
$uri = substr($uri, 0, strrpos($uri, '/'));
$uris[] = $uri;
} while (strrpos($uri, '/') > 1);
$document->addField(Field::text(Page::URIS_KEY, implode(' ', $uris)));
}
foreach (array(Page::TITLE_KEY, Page::DESCRIPTION_KEY, Page::BODY_KEY, Page::IMAGE_KEY) as $fieldName) {
$fieldValue = isset($page[$fieldName]) ? $page[$fieldName] : '';
switch ($fieldName) {
case Page::TITLE_KEY:
case Page::DESCRIPTION_KEY:
case Page::BODY_KEY:
$field = Field::text($fieldName, $fieldValue);
// translit
$fieldTranslit = Field::text($fieldName . '_translit', str_replace("'", '', iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $fieldValue)));
$fieldTranslit->boost = isset($this->parameters[self::BOOST_PARAM][$fieldName]) ? $this->parameters[self::BOOST_PARAM][$fieldName] : 1.25;
$document->addField($fieldTranslit);
break;
case Page::IMAGE_KEY:
$field = Field::unIndexed($fieldName, $fieldValue);
break;
default:
$translitValue = str_replace("'", '', iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $fieldValue));
$field = Field::text($fieldName, $fieldValue . ($translitValue != $fieldValue ? ' ' . $translitValue : ''));
}
$field->boost = isset($this->parameters[self::BOOST_PARAM][$fieldName]) ? $this->parameters[self::BOOST_PARAM][$fieldName] : 1.25;
$document->addField($field);
}
// title tags as configured i.e. h1, h2, ...
foreach ($this->parameters[self::TITLE_TAGS_PARAM] as $fieldName) {
$fieldValue = Page::hasHeadlineType($page, $fieldName) ? Page::getHeadline($page, $fieldName) : '';
$field = Field::text($fieldName, $fieldValue);
$field->boost = isset($this->parameters[self::BOOST_PARAM][$fieldName]) ? $this->parameters[self::BOOST_PARAM][$fieldName] : 1;
$document->addField($field);
$fieldTranslit = Field::text($fieldName . '_translit', str_replace("'", '', iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $fieldValue)));
$fieldTranslit->boost = isset($this->parameters[self::BOOST_PARAM][$fieldName]) ? $this->parameters[self::BOOST_PARAM][$fieldName] : 1.25;
$document->addField($fieldTranslit);
}
// page ID if selector defined
if ($this->parameters[self::PAGE_ID_PARAM]) {
$fieldValue = isset($page[Page::PAGE_ID_KEY]) ? $page[Page::PAGE_ID_KEY] : '';
$field = Field::unIndexed(Page::PAGE_ID_KEY, $fieldValue);
$document->addField($field);
}
// route name if selector defined
if ($this->parameters[self::ROUTE_NAME_PARAM]) {
$fieldValue = isset($page[Page::ROUTE_NAME_KEY]) ? $page[Page::ROUTE_NAME_KEY] : '';
$field = Field::unIndexed(Page::ROUTE_NAME_KEY, $fieldValue);
$document->addField($field);
}
return $document;
}
/**
* Add a new document to the index.
* Any existing document with the given $id should be deleted first.
* $fields should be indexed but not necessarily stored in the index.
* $parameters should be stored in the index but not necessarily indexed.
*
* @param mixed $id
* @param array $fields
* @param array $parameters
*
* @return bool
*/
public function insert($id, array $fields, array $parameters = array())
{
// Remove any existing documents.
$this->delete($id);
// Create new document.
$doc = new \ZendSearch\Lucene\Document();
// Add id parameters.
$doc->addField(\ZendSearch\Lucene\Document\Field::keyword('xref_id', $id));
// Add fields to document to be indexed and stored.
foreach ($fields as $field => $value) {
if (is_array($value)) {
$value = implode(' ', $value);
}
$doc->addField(\ZendSearch\Lucene\Document\Field::text(trim($field), trim($value)));
}
// Add parameters to document to be stored (but not indexed).
$doc->addField(\ZendSearch\Lucene\Document\Field::unIndexed('_parameters', base64_encode(json_encode($parameters))));
// Add document to index.
$this->getIndex()->addDocument($doc);
return true;
}
protected function createDocument($title, $body, $url)
{
$document = new Document();
$document->addField(Field::text('title', $title));
$document->addField(Field::text('body', strip_tags($body)));
$document->addField(Field::keyword('url', $url));
return $document;
}
private function appendToDocument(Lucene\Document $document, $name, $value)
{
if (is_string($value)) {
$document->addField(Lucene\Document\Field::keyword($name, utf8_decode($value)));
} elseif ($value instanceof \DateTime) {
$document->addField(Lucene\Document\Field::keyword($name, $value->getTimestamp()));
} elseif (is_array($value)) {
$document->addField(Lucene\Document\Field::text($name, implode(',', $value)));
} else {
$document->addField(Lucene\Document\Field::binary($name, $value));
}
}
