web/lib/Zend/Search/Lucene/Document/Pptx.php
author Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
Mon, 19 Dec 2011 00:56:02 +0100
changeset 428 8b51d23b758f
parent 68 ecaf28ffe26e
child 207 621fa6caec0c
permissions -rw-r--r--
correct sync
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     1
<?php
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     2
/**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     3
 * Zend Framework
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     4
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     5
 * LICENSE
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     6
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     7
 * This source file is subject to the new BSD license that is bundled
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     8
 * with this package in the file LICENSE.txt.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     9
 * It is also available through the world-wide-web at this URL:
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    10
 * http://framework.zend.com/license/new-bsd
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    11
 * If you did not receive a copy of the license and are unable to
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    12
 * obtain it through the world-wide-web, please send an email
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    13
 * to license@zend.com so we can send you a copy immediately.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    14
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    15
 * @category   Zend
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    16
 * @package    Zend_Search_Lucene
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    17
 * @subpackage Document
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    18
 * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    19
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    20
 * @version    $Id: Pptx.php 20096 2010-01-06 02:05:09Z bkarwin $
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    21
 */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    22
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    23
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    24
/** Zend_Search_Lucene_Document_OpenXml */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    25
require_once 'Zend/Search/Lucene/Document/OpenXml.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    26
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    27
/**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    28
 * Pptx document.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    29
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    30
 * @category   Zend
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    31
 * @package    Zend_Search_Lucene
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    32
 * @subpackage Document
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    33
 * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    34
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    35
 */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    36
class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenXml
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    37
{
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    38
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    39
     * Xml Schema - PresentationML
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    40
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    41
     * @var string
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    42
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    43
    const SCHEMA_PRESENTATIONML = 'http://schemas.openxmlformats.org/presentationml/2006/main';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    44
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    45
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    46
     * Xml Schema - DrawingML
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    47
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    48
     * @var string
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    49
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    50
    const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    51
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    52
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    53
     * Xml Schema - Slide relation
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    54
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    55
     * @var string
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    56
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    57
    const SCHEMA_SLIDERELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/slide';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    58
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    59
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    60
     * Xml Schema - Slide notes relation
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    61
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    62
     * @var string
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    63
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    64
    const SCHEMA_SLIDENOTESRELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/notesSlide';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    65
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    66
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    67
     * Object constructor
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    68
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    69
     * @param string  $fileName
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    70
     * @param boolean $storeContent
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    71
     * @throws Zend_Search_Lucene_Exception
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    72
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    73
    private function __construct($fileName, $storeContent)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    74
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    75
        if (!class_exists('ZipArchive', false)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    76
            require_once 'Zend/Search/Lucene/Exception.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    77
            throw new Zend_Search_Lucene_Exception('MS Office documents processing functionality requires Zip extension to be loaded');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    78
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    79
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    80
        // Document data holders
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    81
        $slides = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    82
        $slideNotes = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    83
        $documentBody = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    84
        $coreProperties = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    85
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    86
        // Open OpenXML package
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    87
        $package = new ZipArchive();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    88
        $package->open($fileName);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    89
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    90
        // Read relations and search for officeDocument
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    91
        $relationsXml = $package->getFromName('_rels/.rels');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    92
        if ($relationsXml === false) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    93
            require_once 'Zend/Search/Lucene/Exception.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    94
            throw new Zend_Search_Lucene_Exception('Invalid archive or corrupted .pptx file.');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    95
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    96
        $relations = simplexml_load_string($relationsXml);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    97
        foreach ($relations->Relationship as $rel) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    98
            if ($rel["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    99
                // Found office document! Search for slides...
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   100
                $slideRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/_rels/" . basename($rel["Target"]) . ".rels")) );
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   101
                foreach ($slideRelations->Relationship as $slideRel) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   102
                    if ($slideRel["Type"] == Zend_Search_Lucene_Document_Pptx::SCHEMA_SLIDERELATION) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   103
                        // Found slide!
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   104
                        $slides[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   105
                            $package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . basename($slideRel["Target"])) )
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   106
                        );
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   107
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   108
                        // Search for slide notes
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   109
                        $slideNotesRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/_rels/" . basename($slideRel["Target"]) . ".rels")) );
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   110
                        foreach ($slideNotesRelations->Relationship as $slideNoteRel) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   111
                            if ($slideNoteRel["Type"] == Zend_Search_Lucene_Document_Pptx::SCHEMA_SLIDENOTESRELATION) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   112
                                // Found slide notes!
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   113
                                $slideNotes[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   114
                                    $package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . dirname($slideNoteRel["Target"]) . "/" . basename($slideNoteRel["Target"])) )
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   115
                                );
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   116
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   117
                                break;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   118
                            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   119
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   120
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   121
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   122
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   123
                break;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   124
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   125
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   126
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   127
        // Sort slides
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   128
        ksort($slides);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   129
        ksort($slideNotes);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   130
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   131
        // Extract contents from slides
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   132
        foreach ($slides as $slideKey => $slide) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   133
            // Register namespaces
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   134
            $slide->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   135
            $slide->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   136
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   137
            // Fetch all text
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   138
            $textElements = $slide->xpath('//a:t');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   139
            foreach ($textElements as $textElement) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   140
                $documentBody[] = (string)$textElement;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   141
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   142
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   143
            // Extract contents from slide notes
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   144
            if (isset($slideNotes[$slideKey])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   145
                // Fetch slide note
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   146
                $slideNote = $slideNotes[$slideKey];
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   147
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   148
                // Register namespaces
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   149
                $slideNote->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   150
                $slideNote->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   151
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   152
                // Fetch all text
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   153
                $textElements = $slideNote->xpath('//a:t');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   154
                foreach ($textElements as $textElement) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   155
                    $documentBody[] = (string)$textElement;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   156
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   157
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   158
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   159
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   160
        // Read core properties
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   161
        $coreProperties = $this->extractMetaData($package);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   162
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   163
        // Close file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   164
        $package->close();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   165
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   166
        // Store filename
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   167
        $this->addField(Zend_Search_Lucene_Field::Text('filename', $fileName, 'UTF-8'));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   168
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   169
            // Store contents
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   170
        if ($storeContent) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   171
            $this->addField(Zend_Search_Lucene_Field::Text('body', implode(' ', $documentBody), 'UTF-8'));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   172
        } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   173
            $this->addField(Zend_Search_Lucene_Field::UnStored('body', implode(' ', $documentBody), 'UTF-8'));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   174
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   175
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   176
        // Store meta data properties
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   177
        foreach ($coreProperties as $key => $value)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   178
        {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   179
            $this->addField(Zend_Search_Lucene_Field::Text($key, $value, 'UTF-8'));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   180
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   181
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   182
        // Store title (if not present in meta data)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   183
        if (!isset($coreProperties['title']))
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   184
        {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   185
            $this->addField(Zend_Search_Lucene_Field::Text('title', $fileName, 'UTF-8'));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   186
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   187
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   188
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   189
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   190
     * Load Pptx document from a file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   191
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   192
     * @param string  $fileName
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   193
     * @param boolean $storeContent
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   194
     * @return Zend_Search_Lucene_Document_Pptx
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   195
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   196
    public static function loadPptxFile($fileName, $storeContent = false)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   197
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   198
        return new Zend_Search_Lucene_Document_Pptx($fileName, $storeContent);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   199
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   200
}