6543/CommunityID
1
0
Fork 0
Code Issues 3 Pull Requests Releases Wiki Activity

import v1.1.0_beta1 | 2009-08-21

Browse Source
This commit is contained in:
6543
2019-07-17 22:16:19 +02:00
parent 2c1152f0d3
commit 8dee6b1a10
2306 changed files with 251360 additions and 23428 deletions
Download Patch File Download Diff File Expand all files Collapse all files

192
libs/Zend/Search/Lucene/Document/Docx.php
View File

@ -22,98 +22,122 @@
/** Zend_Search_Lucene_Document_OpenXml */
require_once 'Zend/Search/Lucene/Document/OpenXml.php';
if (class_exists ( 'ZipArchive' )) {
if (class_exists('ZipArchive', false)) {
/**
* Docx document.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Document
* @copyright Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenXml {
/**
* Docx document.
* Xml Schema - WordprocessingML
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Document
* @copyright Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @var string
*/
class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenXml {
/**
* Xml Schema - WordprocessingML
*
* @var string
*/
const SCHEMA_WORDPROCESSINGML = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
/**
* Object constructor
*
* @param string $fileName
* @param boolean $storeContent
*/
private function __construct($fileName, $storeContent) {
// Document data holders
$documentBody = array ( );
$coreProperties = array ( );
// Open OpenXML package
$package = new ZipArchive ( );
$package->open ( $fileName );
// Read relations and search for officeDocument
$relations = simplexml_load_string ( $package->getFromName ( "_rels/.rels" ) );
foreach ( $relations->Relationship as $rel ) {
if ($rel ["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
// Found office document! Read in contents...
$contents = simplexml_load_string ( $package->getFromName ( $this->absoluteZipPath ( dirname ( $rel ["Target"] ) . "/" . basename ( $rel ["Target"] ) ) ) );
$contents->registerXPathNamespace ( "w", Zend_Search_Lucene_Document_Docx::SCHEMA_WORDPROCESSINGML );
$paragraphs = $contents->xpath ( '//w:body/w:p' );
foreach ( $paragraphs as $paragraph ) {
$runs = $paragraph->xpath ( '//w:r/w:t' );
foreach ( $runs as $run ) {
$documentBody [] = ( string ) $run;
}
const SCHEMA_WORDPROCESSINGML = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
/**
* Object constructor
*
* @param string $fileName
* @param boolean $storeContent
*/
private function __construct($fileName, $storeContent) {
// Document data holders
$documentBody = array();
$coreProperties = array();
// Open OpenXML package
$package = new ZipArchive();
$package->open($fileName);
// Read relations and search for officeDocument
$relations = simplexml_load_string($package->getFromName('_rels/.rels'));
foreach($relations->Relationship as $rel) {
if ($rel ["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
// Found office document! Read in contents...
$contents = simplexml_load_string($package->getFromName(
$this->absoluteZipPath(dirname($rel['Target'])
. '/'
. basename($rel['Target']))
));
$contents->registerXPathNamespace('w', Zend_Search_Lucene_Document_Docx::SCHEMA_WORDPROCESSINGML);
$paragraphs = $contents->xpath('//w:body/w:p');
foreach ($paragraphs as $paragraph) {
$runs = $paragraph->xpath('.//w:r/*[name() = "w:t" or name() = "w:br"]');
if ($runs === false) {
// Paragraph doesn't contain any text or breaks
continue;
}
break;
foreach ($runs as $run) {
if ($run->getName() == 'br') {
// Break element
$documentBody[] = ' ';
} else {
$documentBody[] = (string)$run;
}
}
// Add space after each paragraph. So they are not bound together.
$documentBody[] = ' ';
}
}
// Read core properties
$coreProperties = $this->extractMetaData ( $package );
// Close file
$package->close ();
// Store filename
$this->addField ( Zend_Search_Lucene_Field::Text ( 'filename', $fileName ) );
// Store contents
if ($storeContent) {
$this->addField ( Zend_Search_Lucene_Field::Text ( 'body', implode ( ' ', $documentBody ) ) );
} else {
$this->addField ( Zend_Search_Lucene_Field::UnStored ( 'body', implode ( ' ', $documentBody ) ) );
}
// Store meta data properties
foreach ( $coreProperties as $key => $value ) {
$this->addField ( Zend_Search_Lucene_Field::Text ( $key, $value ) );
}
// Store title (if not present in meta data)
if (! isset ( $coreProperties ['title'] )) {
$this->addField ( Zend_Search_Lucene_Field::Text ( 'title', $fileName ) );
break;
}
}
/**
* Load Docx document from a file
*
* @param string $fileName
* @param boolean $storeContent
* @return Zend_Search_Lucene_Document_Docx
*/
public static function loadDocxFile($fileName, $storeContent = false) {
return new Zend_Search_Lucene_Document_Docx ( $fileName, $storeContent );
// Read core properties
$coreProperties = $this->extractMetaData($package);
// Close file
$package->close();
// Store filename
$this->addField(Zend_Search_Lucene_Field::Text('filename', $fileName, 'UTF-8'));
// Store contents
if ($storeContent) {
$this->addField(Zend_Search_Lucene_Field::Text('body', implode('', $documentBody), 'UTF-8'));
} else {
$this->addField(Zend_Search_Lucene_Field::UnStored('body', implode('', $documentBody), 'UTF-8'));
}
// Store meta data properties
foreach ($coreProperties as $key => $value) {
$this->addField(Zend_Search_Lucene_Field::Text($key, $value, 'UTF-8'));
}
// Store title (if not present in meta data)
if (! isset($coreProperties['title'])) {
$this->addField(Zend_Search_Lucene_Field::Text('title', $fileName, 'UTF-8'));
}
}
/**
* Load Docx document from a file
*
* @param string $fileName
* @param boolean $storeContent
* @return Zend_Search_Lucene_Document_Docx
* @throws Zend_Search_Lucene_Document_Exception
*/
public static function loadDocxFile($fileName, $storeContent = false) {
if (!is_readable($fileName)) {
require_once 'Zend/Search/Lucene/Document/Exception.php';
throw new Zend_Search_Lucene_Document_Exception('Provided file \'' . $fileName . '\' is not readable.');
}
return new Zend_Search_Lucene_Document_Docx($fileName, $storeContent);
}
}
} // end if (class_exists('ZipArchive'))