/**
* get_complete_data1 function.
* Jetzt noch die verlinkten Dokumente auslesen
* @access private
* @return void
*/
private function get_complete_data1($data_org = array(), $db)
{
foreach ($data_org as $key => $value) {
if ($this->count >= MAX_DOK) {
continue;
}
//Checken ob Dokument schon in DB, wenn ja weiter, wenn nein scannen und neu laden
if ($this->check_ob_dokument_vorhanden($value['id'], $db) == true or $this->noch_nicht_vorhanden == true) {
continue;
} else {
//Wenn x Dokumente erreicht sind
if ($this->count > MAX_DOK) {
$this->noch_nicht_vorhanden = true;
}
$this->count = $this->count + 1;
}
$data[$key] = $value;
//debug::print_d($value);
//Zuerst das Dokument selber aber mit Fallunterscheidung
if (!empty($value['id_link_html'])) {
//Fall 1 . htm
$dok = class_methods::get_site('http://www2.bonn.de/bo_ris/' . $value['id_link_html']);
$row['html'] = $dok;
$row['html_text'] = class_methods::get_clean_text($dok);
$row['html_meta'] = $this->get_extra_infos_dokument($dok);
}
if (!empty($value['id_link_pdf'])) {
$dok = class_methods::get_site('http://www2.bonn.de/bo_ris/' . $value['id_link_pdf']);
$row['pdf'] = $dok;
//PDF Daten auslesen wenn möglich / geht natürlich nicht bei Bildern usw.
$convert = new pdf2txt();
$result = $convert->convert($dok);
$row['pfd_text'] = class_methods::get_clean_text($result);
}
if (!empty($value['id_link_rtf'])) {
$dok = class_methods::get_site('http://www2.bonn.de/bo_ris/' . $value['id_link_rtf']);
$row['sonstiges_dokument'] = $dok;
$row['sonstiges_text'] = class_methods::get_clean_text($dok);
}
//Daten in Hauptarray übergeben
$data[$key]['id_data'] = $row;
//Dann die Metainformationen zum Dokument
if (!empty($value['meta_link'])) {
//echo 'http://www2.bonn.de/bo_ris/ris_sql/'.$value['meta_link'];
$dok = class_methods::get_site('http://www2.bonn.de/bo_ris/ris_sql/' . $value['meta_link']);
$row['meta_data'] = $dok;
$row['meta_data_text'] = class_methods::get_clean_text($dok);
$row['meta_data_extra'] = $this->get_extra_infos($dok);
//Daten in Hauptarray übergeben
$data[$key]['id_data'] = $row;
}
//$data[]=$value;
}
return $data;
//print_r($data);
}
/**
* Indexes a certain file.
* Returns A587_FILE_GENERATED or an error code.
*
* @param string $_filename
* @param mixed $_clang
* @param mixed $_doPlaintext
* @param mixed $_articleData
*
* @return mixed
*/
function indexFile($_filename, $_doPlaintext = false, $_clang = false, $_fid = false, $_catid = false)
{
// extract file-extension
$filenameArray = explode('.', $_filename);
$fileext = $filenameArray[count($filenameArray) - 1];
// check file-extension
if (!in_array($fileext, $this->fileExtensions) and !empty($this->fileExtensions) and !$this->indexUnknownFileExtensions and !$this->indexMissingFileExtensions) {
return A587_FILE_FORBIDDEN_EXTENSION;
}
// delete cache
$delete = new rex_sql();
$where = sprintf(" `filename` = '%s' AND `texttype` = 'file'", $delete->escape($_filename));
if (is_int($_clang)) {
$where .= sprintf(' AND clang = %d', $_clang);
}
if (is_int($_fid)) {
$where .= sprintf(' AND fid = %d', $_fid);
} elseif (is_array($_fid)) {
$where .= sprintf(" AND fid = '%s'", $delete->escape(json_encode($_fid)));
}
if (is_int($_catid)) {
$where .= sprintf(' AND catid = %d', $_catid);
}
// delete from cache
$select = new rex_sql();
$select->setTable($this->tablePrefix . '587_searchindex');
$select->setWhere($where);
$indexIds = array();
if ($select->select('id')) {
foreach ($select->getArray() as $result) {
$indexIds[] = $result['id'];
}
$this->deleteCache($indexIds);
}
// delete old data
$delete->setTable($this->tablePrefix . '587_searchindex');
$delete->setWhere($where);
$delete->delete();
// index file
$text = '';
$plaintext = '';
switch ($fileext) {
// pdf-files
case 'pdf':
// try XPDF
$return = 0;
$xpdf = false;
$error = false;
if (function_exists('exec')) {
$tempFile = tempnam($this->generatedPath . '/files/', 'rexsearch');
$encoding = 'UTF-8';
exec('pdftotext ' . escapeshellarg($this->documentRoot . '/' . $_filename) . ' ' . escapeshellarg($tempFile) . ' -enc ' . $encoding, $dummy, $return);
if ($return > 0) {
if ($return == 1) {
$error = A587_FILE_XPDFERR_OPENSRC;
}
if ($return == 2) {
$error = A587_FILE_XPDFERR_OPENDEST;
}
if ($return == 3) {
$error = A587_FILE_XPDFERR_PERM;
}
if ($return == 99) {
$error = A587_FILE_XPDFERR_OTHER;
}
} else {
if (false === ($text = @file_get_contents($tempFile))) {
$error = A587_FILE_NOEXIST;
} else {
$xpdf = true;
}
}
unlink($tempFile);
}
if (!$xpdf) {
// if xpdf returned an error, try pdf2txt via php
if (false === ($pdfContent = @file_get_contents($this->documentRoot . '/' . $_filename))) {
$error = A587_FILE_NOEXIST;
} else {
require_once 'class.pdf2txt.inc.php';
$text = pdf2txt::directConvert($pdfContent);
$error = false;
}
}
if ($error !== false) {
return $error;
} elseif (trim($text) == '') {
return A587_FILE_EMPTY;
}
$plaintext = $this->getPlaintext($text);
break;
// html- or php-file
// html- or php-file
case 'htm':
case 'html':
case 'php':
if (false === ($text = @file_get_contents($this->documentRoot . '/' . $_filename))) {
return A587_FILE_NOEXIST;
}
$plaintext = $this->getPlaintext($text);
// other filetype
// other filetype
default:
if (false === ($text = @file_get_contents($this->documentRoot . '/' . $_filename))) {
return A587_FILE_NOEXIST;
}
}
$text = @iconv(mb_detect_encoding($text), 'UTF-8', $text);
// Plaintext
if (empty($plaintext)) {
if ($_doPlaintext) {
$plaintext = $this->getPlaintext($text);
} else {
$plaintext = $text;
}
}
// index file-content
$insert = new rex_sql();
$fileData['texttype'] = 'file';
if ($_fid !== false) {
$fileData['ftable'] = $this->tablePrefix . 'file';
}
$fileData['filename'] = $insert->escape($_filename);
$fileData['fileext'] = $insert->escape($fileext);
if ($_clang !== false) {
$fileData['clang'] = intval($_clang);
}
if ($_fid !== false) {
$fileData['fid'] = intval($_fid);
} else {
$fileData['fid'] = NULL;
}
if (is_null($fileData['fid'])) {
$fileData['fid'] = $this->getMinFID();
}
if ($_catid !== false) {
$fileData['catid'] = intval($_catid);
}
$fileData['unchangedtext'] = $insert->escape($text);
$fileData['plaintext'] = $insert->escape($plaintext);
$keywords = array();
foreach (preg_split($this->encodeRegex('~[[:punct:][:space:]]+~ism'), $plaintext) as $keyword) {
if ($this->significantCharacterCount <= mb_strlen($keyword, 'UTF-8')) {
$keywords[] = array('search' => $keyword, 'clang' => !isset($fileData['clang']) ? false : $fileData['clang']);
}
}
$this->storeKeywords($keywords, false);
$fileData['teaser'] = $insert->escape($this->getTeaserText($plaintext));
$insert->setTable($this->tablePrefix . '587_searchindex');
$insert->setValues($fileData);
$insert->insert();
return A587_FILE_GENERATED;
}
function directConvert($_data)
{
$pdf2txt = new pdf2txt();
return $pdf2txt->convert($_data);
}
