git.roojs.org Git - web.mtrack/blob - Zend/Search/Lucene/Document/Pptx.php

   1 <?php
   2 /**
   3  * Zend Framework
   4  *
   5  * LICENSE
   6  *
   7  * This source file is subject to the new BSD license that is bundled
   8  * with this package in the file LICENSE.txt.
   9  * It is also available through the world-wide-web at this URL:
  10  * http://framework.zend.com/license/new-bsd
  11  * If you did not receive a copy of the license and are unable to
  12  * obtain it through the world-wide-web, please send an email
  13  * to license@zend.com so we can send you a copy immediately.
  14  *
  15  * @category   Zend
  16  * @package    Zend_Search_Lucene
  17  * @subpackage Document
  18  * @copyright  Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  19  * @license    http://framework.zend.com/license/new-bsd     New BSD License
  20  * @version    $Id: Pptx.php 16971 2009-07-22 18:05:45Z mikaelkael $
  21  */
  22
  23
  24 /** Zend_Search_Lucene_Document_OpenXml */
  25 require_once 'Zend/Search/Lucene/Document/OpenXml.php';
  26
  27 if (class_exists('ZipArchive', false)) {
  28
  29 /**
  30  * Pptx document.
  31  *
  32  * @category   Zend
  33  * @package    Zend_Search_Lucene
  34  * @subpackage Document
  35  * @copyright  Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
  36  * @license    http://framework.zend.com/license/new-bsd     New BSD License
  37  */
  38 class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenXml
  39 {
  40     /**
  41      * Xml Schema - PresentationML
  42      *
  43      * @var string
  44      */
  45     const SCHEMA_PRESENTATIONML = 'http://schemas.openxmlformats.org/presentationml/2006/main';
  46
  47     /**
  48      * Xml Schema - DrawingML
  49      *
  50      * @var string
  51      */
  52     const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
  53
  54     /**
  55      * Xml Schema - Slide relation
  56      *
  57      * @var string
  58      */
  59     const SCHEMA_SLIDERELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/slide';
  60
  61     /**
  62      * Xml Schema - Slide notes relation
  63      *
  64      * @var string
  65      */
  66     const SCHEMA_SLIDENOTESRELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/notesSlide';
  67
  68     /**
  69      * Object constructor
  70      *
  71      * @param string  $fileName
  72      * @param boolean $storeContent
  73      */
  74     private function __construct($fileName, $storeContent)
  75     {
  76         // Document data holders
  77         $slides = array();
  78         $slideNotes = array();
  79         $documentBody = array();
  80         $coreProperties = array();
  81
  82         // Open OpenXML package
  83         $package = new ZipArchive();
  84         $package->open($fileName);
  85
  86         // Read relations and search for officeDocument
  87         $relations = simplexml_load_string($package->getFromName("_rels/.rels"));
  88         foreach ($relations->Relationship as $rel) {
  89             if ($rel["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
  90                 // Found office document! Search for slides...
  91                 $slideRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/_rels/" . basename($rel["Target"]) . ".rels")) );
  92                 foreach ($slideRelations->Relationship as $slideRel) {
  93                     if ($slideRel["Type"] == Zend_Search_Lucene_Document_Pptx::SCHEMA_SLIDERELATION) {
  94                         // Found slide!
  95                         $slides[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
  96                             $package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . basename($slideRel["Target"])) )
  97                         );
  98
  99                         // Search for slide notes
 100                         $slideNotesRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/_rels/" . basename($slideRel["Target"]) . ".rels")) );
 101                         foreach ($slideNotesRelations->Relationship as $slideNoteRel) {
 102                             if ($slideNoteRel["Type"] == Zend_Search_Lucene_Document_Pptx::SCHEMA_SLIDENOTESRELATION) {
 103                                 // Found slide notes!
 104                                 $slideNotes[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
 105                                     $package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . dirname($slideNoteRel["Target"]) . "/" . basename($slideNoteRel["Target"])) )
 106                                 );
 107
 108                                 break;
 109                             }
 110                         }
 111                     }
 112                 }
 113
 114                 break;
 115             }
 116         }
 117
 118         // Sort slides
 119         ksort($slides);
 120         ksort($slideNotes);
 121
 122         // Extract contents from slides
 123         foreach ($slides as $slideKey => $slide) {
 124             // Register namespaces
 125             $slide->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
 126             $slide->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
 127
 128             // Fetch all text
 129             $textElements = $slide->xpath('//a:t');
 130             foreach ($textElements as $textElement) {
 131                 $documentBody[] = (string)$textElement;
 132             }
 133
 134             // Extract contents from slide notes
 135             if (isset($slideNotes[$slideKey])) {
 136                 // Fetch slide note
 137                 $slideNote = $slideNotes[$slideKey];
 138
 139                 // Register namespaces
 140                 $slideNote->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
 141                 $slideNote->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
 142
 143                 // Fetch all text
 144                 $textElements = $slideNote->xpath('//a:t');
 145                 foreach ($textElements as $textElement) {
 146                     $documentBody[] = (string)$textElement;
 147                 }
 148             }
 149         }
 150
 151         // Read core properties
 152         $coreProperties = $this->extractMetaData($package);
 153
 154         // Close file
 155         $package->close();
 156
 157         // Store filename
 158         $this->addField(Zend_Search_Lucene_Field::Text('filename', $fileName, 'UTF-8'));
 159
 160             // Store contents
 161         if ($storeContent) {
 162             $this->addField(Zend_Search_Lucene_Field::Text('body', implode(' ', $documentBody), 'UTF-8'));
 163         } else {
 164             $this->addField(Zend_Search_Lucene_Field::UnStored('body', implode(' ', $documentBody), 'UTF-8'));
 165         }
 166
 167         // Store meta data properties
 168         foreach ($coreProperties as $key => $value)
 169         {
 170             $this->addField(Zend_Search_Lucene_Field::Text($key, $value, 'UTF-8'));
 171         }
 172
 173         // Store title (if not present in meta data)
 174         if (!isset($coreProperties['title']))
 175         {
 176             $this->addField(Zend_Search_Lucene_Field::Text('title', $fileName, 'UTF-8'));
 177         }
 178     }
 179
 180     /**
 181      * Load Pptx document from a file
 182      *
 183      * @param string  $fileName
 184      * @param boolean $storeContent
 185      * @return Zend_Search_Lucene_Document_Pptx
 186      */
 187     public static function loadPptxFile($fileName, $storeContent = false)
 188     {
 189         return new Zend_Search_Lucene_Document_Pptx($fileName, $storeContent);
 190     }
 191 }
 192
 193 } // end if (class_exists('ZipArchive'))