/**
* Extract data from a PDF document and add this to the Lucene index.
*
* @param string $pdfPath The path to the PDF document.
* @param Zend_Search_Lucene_Proxy $luceneIndex The Lucene index object.
* @return Zend_Search_Lucene_Proxy
*/
public static function index($pdfPath, $luceneIndex)
{
// Load the PDF document.
$pdf = Zend_Pdf::load($pdfPath);
$key = md5($pdfPath);
/**
* Set up array to contain the document index data.
* The Filename will be used to retrive the document if it is found in
* the search resutls.
* The Key will be used to uniquely identify the document so we can
* delete it from the search index.
*/
$indexValues = array('Filename' => $pdfPath, 'Key' => $key, 'Title' => '', 'Author' => '', 'Subject' => '', 'Keywords' => '', 'Creator' => '', 'Producer' => '', 'CreationDate' => '', 'ModDate' => '', 'Contents' => '');
// Go through each meta data item and add to index array.
foreach ($pdf->properties as $meta => $metaValue) {
switch ($meta) {
case 'Title':
$indexValues['Title'] = $pdf->properties['Title'];
break;
case 'Subject':
$indexValues['Subject'] = $pdf->properties['Subject'];
break;
case 'Author':
$indexValues['Author'] = $pdf->properties['Author'];
break;
case 'Keywords':
$indexValues['Keywords'] = $pdf->properties['Keywords'];
break;
case 'CreationDate':
$dateCreated = $pdf->properties['CreationDate'];
$distance = substr($dateCreated, 16, 2);
if (!is_long($distance)) {
$distance = null;
}
// Convert date from the PDF format of D:20090731160351+01'00'
$dateCreated = mktime(substr($dateCreated, 10, 2), substr($dateCreated, 12, 2), substr($dateCreated, 14, 2), substr($dateCreated, 6, 2), substr($dateCreated, 8, 2), substr($dateCreated, 2, 4), $distance);
//distance
$indexValues['CreationDate'] = date('Ymd', $dateCreated);
break;
case 'Date':
$indexValues['Date'] = $pdf->properties['Date'];
break;
}
}
/**
* Parse the contents of the PDF document and pass the text to the
* contents item in the $indexValues array.
*/
$pdfParse = new App_Search_Helper_PdfParser();
$indexValues['Contents'] = $pdfParse->pdf2txt($pdf->render());
// Create the document using the values
$doc = new App_Search_Lucene_Document($indexValues);
if ($doc !== false) {
// If the document creation was sucessful then add it to our index.
$luceneIndex->addDocument($doc);
}
// Return the Lucene index object.
return $luceneIndex;
}
public function __construct()
{
try {
parent::__construct(new Zend_Search_Lucene(self::INDEX_DIR, false));
} catch (Zend_Search_Lucene_Exception $e) {
parent::__construct(new Zend_Search_Lucene(self::INDEX_DIR, true));
}
Zend_Search_Lucene_Search_QueryParser::setDefaultEncoding("UTF-8");
}
/**
* Add a document to the index.
*
* @return void
* @access private
* @static
*
* @author Etienne de Longeaux <*****@*****.**>
* @since 2012-06-11
*/
public static function addDocument()
{
// Search for documents with the same Key value.
$term = new \Zend_Search_Lucene_Index_Term(self::$_doc->Key, 'Key');
$docIds = self::$_index->termDocs($term);
// Delete any documents found.
foreach ($docIds as $id) {
self::$_index->delete($id);
}
if (self::$_doc instanceof \Zend_Search_Lucene_Document) {
self::$_index->addDocument(self::$_doc);
}
}
/**
*
* @since 5-24-11
*/
protected function find(Zend_Search_Lucene_Proxy $lucene, array $where_criteria)
{
$ret_list = array();
Zend_Search_Lucene::setResultSetLimit($where_criteria['limit'][0]);
if (empty($where_criteria['sort'])) {
$ret_list = $lucene->find($where_criteria['query']);
} else {
// http://framework.zend.com/manual/en/zend.search.lucene.searching.html#zend.search.lucene.searching.sorting
$args = $where_criteria['sort'];
array_unshift($args, $where_criteria['query']);
$ret_list = call_user_func_array(array($lucene, 'find'), $args);
}
//if/else
return $lucene->find($where_criteria['query']);
}
/**
* Object constructor
*
* @param Zend_Search_Lucene_Interface $index
* @param string $module the Drupal module managing the index
*/
public function __construct(Zend_Search_Lucene_Interface $index, $module)
{
parent::__construct($index);
$this->_module = (string) $module;
}
/**
* Destructor
*
* @return void
*/
public function __destruct()
{
if ($this->_index instanceof \Zend_Search_Lucene_Interface) {
$this->_index->removeReference();
}
}
/**
* Search all pages that match the query.
*
* <code>
* //$query = '(pi AND groupe AND partner*) OR pi-groupe';
* $query = " travers projet ference coin";
* $options = array(
* 'searchBool' => true,
* 'searchBoolType' => 'AND',
* 'searchByMotif' => true,
* 'setMinPrefixLength'=> 0,
* 'getResultSetLimit' => 0,
* 'searchFields' => array(
* 0=> array('sortField'=>'Contents', 'sortType'=> SORT_STRING, 'sortOrder' => SORT_ASC),
* 1=> array('sortField'=>'Key', 'sortType'=> SORT_NUMERIC, 'sortOrder' => SORT_DESC)
* ),
* );
* $result = $this->container->get('pi_app_admin.manager.search_lucene')->searchPagesByQuery($query, $options);
* </code>
*
* @link http://framework.zend.com/manual/fr/zend.search.lucene.searching.html
* @link http://framework.zend.com/manual/fr/learning.lucene.queries.html
* @link http://framework.zend.com/manual/1.12/fr/zend.search.lucene.query-api.html
* @param string $query The search query index file
* @param array $options Options of the search query of the index file
* @return array All Etags from pages that match the query.
* @access public
*
* @author Etienne de Longeaux <*****@*****.**>
* @since 2012-06-11
*/
public function searchPagesByQuery($query = "Key:*", $options = null, $locale = '')
{
try {
if (isset($options) && is_array($options) && count($options) >= 1) {
$options_values = array_map(function ($key, $value) {
if (in_array($value, array("true"))) {
return 1;
} elseif (in_array($value, array("false"))) {
return 0;
} elseif (!is_array($value) && preg_match_all("/[0-9]+/", $value, $nbrs, PREG_SET_ORDER)) {
return intval($value);
} else {
return $value;
}
}, array_keys($options), array_values($options));
$options = array_combine(array_keys($options), $options_values);
}
if (empty($query)) {
return null;
} else {
$query = $this->container->get('sfynx.tool.string_manager')->minusculesSansAccents($query);
}
if (empty($locale)) {
$locale = $this->container->get('request')->getLocale();
}
$options_default = array('searchBool' => true, 'searchBoolType' => 'OR', 'searchByMotif' => true, 'setMinPrefixLength' => 0, 'getResultSetLimit' => 0, 'searchFields' => '*', 'searchMaxResultByWord' => 5);
if (is_array($options)) {
$options = array_merge($options_default, $options);
} else {
$options = $options_default;
}
if ($options['searchBool']) {
$q_string = $this->container->get('sfynx.tool.string_manager')->cleanWhitespace($query);
$q_array = explode(' ', $q_string);
if ($options['searchByMotif']) {
$q_array = array_map(function ($value) {
return $value . '*';
}, array_values($q_array));
}
switch ($options['searchBoolType']) {
case 'OR':
$new_query = implode(' OR ', $q_array);
break;
case 'AND':
$new_query = implode(' AND ', $q_array);
break;
default:
break;
}
} else {
$new_query = $query;
}
// Open the index.
self::open($this->_indexPath);
// Set minimum prefix length.
\Zend_Search_Lucene_Search_Query_Wildcard::setMinPrefixLength($options['setMinPrefixLength']);
// Set result set limit.
\Zend_Search_Lucene::setResultSetLimit($options['getResultSetLimit']);
// Performs a query against the index.
if (is_array($options['searchFields']) && $query != "Key:*") {
$fields_vars = "\$hits = self::\$_index->find(\$new_query,";
$i = 0;
foreach ($options['searchFields'] as $key => $valuesField) {
$sortField = $valuesField["sortField"];
if (isset($valuesField["sortType"]) && !empty($valuesField["sortType"])) {
$sortType = $valuesField["sortType"];
} else {
$sortType = SORT_STRING;
}
if (isset($valuesField["sortOrder"]) && !empty($valuesField["sortOrder"])) {
$sortOrder = $valuesField["sortOrder"];
} else {
$sortOrder = $valuesField["sortOrder"];
}
if ($i == 0) {
$fields_vars .= " \"{$sortField}\", {$sortType}, {$sortOrder}";
} else {
$fields_vars .= ", \"{$sortField}\", {$sortType}, {$sortOrder}";
}
$i++;
}
$fields_vars .= ");";
try {
setlocale(LC_ALL, $locale);
eval($fields_vars);
// print_r($options);
// print_r($new_query);
// print_r('<br />');
// print_r($fields_vars);
// //exit;
} catch (\Exception $e) {
setlocale(LC_ALL, 'fr_FR');
eval($fields_vars);
}
//eval("\$hits = self::\$_index->find(\$query, \"\$sortField\", \$sortType, \$sortOrder);");
//$hits = self::$_index->find($query, "Contents", SORT_STRING, SORT_DESC);
//$hits = self::$_index->find(' *"férence"* ', "Contents", SORT_STRING, SORT_ASC);
//$hits = self::$_index->find(' *"MOTIVTelecommunication"* OR *"Sophisticated"* ', "Contents", SORT_STRING, SORT_ASC);
} else {
try {
setlocale(LC_ALL, $locale);
$hits = self::$_index->find($new_query);
} catch (\Exception $e) {
setlocale(LC_ALL, 'fr_FR');
$hits = self::$_index->find($new_query);
}
}
$result_search = null;
if (isset($hits) && is_array($hits)) {
foreach ($hits as $hit) {
$field = $hit->getDocument()->getFieldNames();
if (in_array('Key', $field)) {
$data['Key'] = $hit->getDocument()->Key;
} else {
$data['Key'] = "";
}
if (in_array('Route', $field)) {
$data['Route'] = $hit->getDocument()->Route;
} else {
$data['Route'] = "";
}
if (in_array('Title', $field)) {
$data['Title'] = utf8_decode($hit->getDocument()->Title);
} else {
$data['Title'] = "";
}
if (in_array('Keywords', $field)) {
$data['Keywords'] = utf8_decode($hit->getDocument()->Keywords);
} else {
$data['Keywords'] = "";
}
if (in_array('ModDate', $field)) {
$data['ModDate'] = $hit->getDocument()->ModDate;
} else {
$data['ModDate'] = "";
}
$data['MaxResultByWord'] = $options['searchMaxResultByWord'];
$result_search[] = $data;
}
}
return $result_search;
} catch (\Exception $e) {
return array();
}
}
