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'))

36
libs/Zend/Search/Lucene/Document/Exception.php Normal file
View File

@ -0,0 +1,36 @@
<?php
/**
* Zend Framework
*
* LICENSE
*
* This source file is subject to the new BSD license that is bundled
* with this package in the file LICENSE.txt.
* It is also available through the world-wide-web at this URL:
* http://framework.zend.com/license/new-bsd
* If you did not receive a copy of the license and are unable to
* obtain it through the world-wide-web, please send an email
* to license@zend.com so we can send you a copy immediately.
*
* @category Zend
* @package Zend_Search_Lucene
* @copyright Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
/**
* Framework base exception
*/
require_once 'Zend/Search/Lucene/Exception.php';
/**
* @category Zend
* @package Zend_Search_Lucene
* @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_Exception extends Zend_Search_Lucene_Exception
{}

170
libs/Zend/Search/Lucene/Document/Html.php
View File

@ -69,11 +69,12 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
/**
* Object constructor
*
* @param string $data
* @param string $data HTML string (may be HTML fragment, )
* @param boolean $isFile
* @param boolean $storeContent
* @param string $defaultEncoding HTML encoding, is used if it's not specified using Content-type HTTP-EQUIV meta tag.
*/
private function __construct($data, $isFile, $storeContent)
private function __construct($data, $isFile, $storeContent, $defaultEncoding = '')
{
$this->_doc = new DOMDocument();
$this->_doc->substituteEntities = true;
@ -85,6 +86,37 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
}
@$this->_doc->loadHTML($htmlData);
if ($this->_doc->encoding === null) {
// Document encoding is not recognized
/** @todo improve HTML vs HTML fragment recognition */
if (preg_match('/<html>/i', $htmlData, $matches, PREG_OFFSET_CAPTURE)) {
// It's an HTML document
// Add additional HEAD section and recognize document
$htmlTagOffset = $matches[0][1] + strlen($matches[0][1]);
@$this->_doc->loadHTML(iconv($defaultEncoding, 'UTF-8//IGNORE', substr($htmlData, 0, $htmlTagOffset))
. '<head><META HTTP-EQUIV="Content-type" CONTENT="text/html; charset=UTF-8"/></head>'
. iconv($defaultEncoding, 'UTF-8//IGNORE', substr($htmlData, $htmlTagOffset)));
// Remove additional HEAD section
$xpath = new DOMXPath($this->_doc);
$head = $xpath->query('/html/head')->item(0);
$head->parentNode->removeChild($head);
} else {
// It's an HTML fragment
@$this->_doc->loadHTML('<html><head><META HTTP-EQUIV="Content-type" CONTENT="text/html; charset=UTF-8"/></head><body>'
. iconv($defaultEncoding, 'UTF-8//IGNORE', $htmlData)
. '</body></html>');
}
}
/** @todo Add correction of wrong HTML encoding recognition processing
* The case is:
* Content-type HTTP-EQUIV meta tag is presented, but ISO-8859-5 encoding is actually used,
* even $this->_doc->encoding demonstrates another recognized encoding
*/
$xpath = new DOMXPath($this->_doc);
$docTitle = '';
@ -93,13 +125,13 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
// title should always have only one entry, but we process all nodeset entries
$docTitle .= $titleNode->nodeValue . ' ';
}
$this->addField(Zend_Search_Lucene_Field::Text('title', $docTitle, $this->_doc->actualEncoding));
$this->addField(Zend_Search_Lucene_Field::Text('title', $docTitle, 'UTF-8'));
$metaNodes = $xpath->query('/html/head/meta[@name]');
foreach ($metaNodes as $metaNode) {
$this->addField(Zend_Search_Lucene_Field::Text($metaNode->getAttribute('name'),
$metaNode->getAttribute('content'),
$this->_doc->actualEncoding));
'UTF-8'));
}
$docBody = '';
@ -109,9 +141,9 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
$this->_retrieveNodeText($bodyNode, $docBody);
}
if ($storeContent) {
$this->addField(Zend_Search_Lucene_Field::Text('body', $docBody, $this->_doc->actualEncoding));
$this->addField(Zend_Search_Lucene_Field::Text('body', $docBody, 'UTF-8'));
} else {
$this->addField(Zend_Search_Lucene_Field::UnStored('body', $docBody, $this->_doc->actualEncoding));
$this->addField(Zend_Search_Lucene_Field::UnStored('body', $docBody, 'UTF-8'));
}
$linkNodes = $this->_doc->getElementsByTagName('a');
@ -196,25 +228,27 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
/**
* Load HTML document from a string
*
* @param string $data
* @param string $data
* @param boolean $storeContent
* @param string $defaultEncoding HTML encoding, is used if it's not specified using Content-type HTTP-EQUIV meta tag.
* @return Zend_Search_Lucene_Document_Html
*/
public static function loadHTML($data, $storeContent = false)
public static function loadHTML($data, $storeContent = false, $defaultEncoding = '')
{
return new Zend_Search_Lucene_Document_Html($data, false, $storeContent);
return new Zend_Search_Lucene_Document_Html($data, false, $storeContent, $defaultEncoding);
}
/**
* Load HTML document from a file
*
* @param string $file
* @param string $file
* @param boolean $storeContent
* @param string $defaultEncoding HTML encoding, is used if it's not specified using Content-type HTTP-EQUIV meta tag.
* @return Zend_Search_Lucene_Document_Html
*/
public static function loadHTMLFile($file, $storeContent = false)
public static function loadHTMLFile($file, $storeContent = false, $defaultEncoding = '')
{
return new Zend_Search_Lucene_Document_Html($file, true, $storeContent);
return new Zend_Search_Lucene_Document_Html($file, true, $storeContent, $defaultEncoding);
}
@ -223,12 +257,14 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
*
* @param DOMText $node
* @param array $wordsToHighlight
* @param string $color
* @param callback $callback Callback method, used to transform (highlighting) text.
* @param array $params Array of additionall callback parameters (first non-optional parameter is a text to transform)
* @throws Zend_Search_Lucene_Exception
*/
public function _highlightTextNode(DOMText $node, $wordsToHighlight, $color)
protected function _highlightTextNode(DOMText $node, $wordsToHighlight, $callback, $params)
{
$analyzer = Zend_Search_Lucene_Analysis_Analyzer::getDefault();
$analyzer->setInput($node->nodeValue, $this->_doc->encoding);
$analyzer->setInput($node->nodeValue, 'UTF-8');
$matchedTokens = array();
@ -251,10 +287,32 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
// Cut matched node
$matchedWordNode = $node->splitText($token->getStartOffset());
$highlightedNode = $this->_doc->createElement('b', $matchedWordNode->nodeValue);
$highlightedNode->setAttribute('style', 'color:black;background-color:' . $color);
// Retrieve HTML string representation for highlihted word
$fullCallbackparamsList = $params;
array_unshift($fullCallbackparamsList, $matchedWordNode->nodeValue);
$highlightedWordNodeSetHtml = call_user_func_array($callback, $fullCallbackparamsList);
$node->parentNode->replaceChild($highlightedNode, $matchedWordNode);
// Transform HTML string to a DOM representation and automatically transform retrieved string
// into valid XHTML (It's automatically done by loadHTML() method)
$highlightedWordNodeSetDomDocument = new DOMDocument('1.0', 'UTF-8');
$success = @$highlightedWordNodeSetDomDocument->
loadHTML('<html><head><meta http-equiv="Content-type" content="text/html; charset=UTF-8"/></head><body>'
. $highlightedWordNodeSetHtml
. '</body></html>');
if (!$success) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception("Error occured while loading highlighted text fragment: '$highlightedNodeHtml'.");
}
$highlightedWordNodeSetXpath = new DOMXPath($highlightedWordNodeSetDomDocument);
$highlightedWordNodeSet = $highlightedWordNodeSetXpath->query('/html/body')->item(0)->childNodes;
for ($count = 0; $count < $highlightedWordNodeSet->length; $count++) {
$nodeToImport = $highlightedWordNodeSet->item($count);
$node->parentNode->insertBefore($this->_doc->importNode($nodeToImport, true /* deep copy */),
$matchedWordNode);
}
$node->parentNode->removeChild($matchedWordNode);
}
}
@ -264,9 +322,10 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
*
* @param DOMNode $contextNode
* @param array $wordsToHighlight
* @param string $color
* @param callback $callback Callback method, used to transform (highlighting) text.
* @param array $params Array of additionall callback parameters (first non-optional parameter is a text to transform)
*/
public function _highlightNode(DOMNode $contextNode, $wordsToHighlight, $color)
protected function _highlightNodeRecursive(DOMNode $contextNode, $wordsToHighlight, $callback, $params)
{
$textNodes = array();
@ -279,38 +338,66 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
// process node later to leave childNodes structure untouched
$textNodes[] = $childNode;
} else {
// Skip script nodes
// Process node if it's not a script node
if ($childNode->nodeName != 'script') {
$this->_highlightNode($childNode, $wordsToHighlight, $color);
$this->_highlightNodeRecursive($childNode, $wordsToHighlight, $callback, $params);
}
}
}
foreach ($textNodes as $textNode) {
$this->_highlightTextNode($textNode, $wordsToHighlight, $color);
$this->_highlightTextNode($textNode, $wordsToHighlight, $callback, $params);
}
}
/**
* Standard callback method used to highlight words.
*
* @param string $stringToHighlight
* @return string
* @internal
*/
public function applyColour($stringToHighlight, $colour)
{
return '<b style="color:black;background-color:' . $colour . '">' . $stringToHighlight . '</b>';
}
/**
* Highlight text with specified color
*
* @param string|array $words
* @param string $color
* @param string $colour
* @return string
*/
public function highlight($words, $color = '#66ffff')
public function highlight($words, $colour = '#66ffff')
{
return $this->highlightExtended($words, array($this, 'applyColour'), array($colour));
}
/**
* Highlight text using specified View helper or callback function.
*
* @param string|array $words Words to highlight. Words could be organized using the array or string.
* @param callback $callback Callback method, used to transform (highlighting) text.
* @param array $params Array of additionall callback parameters passed through into it
* (first non-optional parameter is an HTML fragment for highlighting)
* @return string
* @throws Zend_Search_Lucene_Exception
*/
public function highlightExtended($words, $callback, $params = array())
{
if (!is_array($words)) {
$words = array($words);
}
$wordsToHighlight = array();
$wordsToHighlightList = array();
$analyzer = Zend_Search_Lucene_Analysis_Analyzer::getDefault();
foreach ($words as $wordString) {
$wordsToHighlight = array_merge($wordsToHighlight, $analyzer->tokenize($wordString));
$wordsToHighlightList[] = $analyzer->tokenize($wordString);
}
$wordsToHighlight = call_user_func_array('array_merge', $wordsToHighlightList);
if (count($wordsToHighlight) == 0) {
return $this->_doc->saveHTML();
@ -321,15 +408,20 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
$wordsToHighlightFlipped[$token->getTermText()] = $id;
}
if (!is_callable($callback)) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('$viewHelper parameter mast be a View Helper name, View Helper object or callback.');
}
$xpath = new DOMXPath($this->_doc);
$matchedNodes = $xpath->query("/html/body");
foreach ($matchedNodes as $matchedNode) {
$this->_highlightNode($matchedNode, $wordsToHighlightFlipped, $color);
$this->_highlightNodeRecursive($matchedNode, $wordsToHighlightFlipped, $callback, $params);
}
}
/**
* Get HTML
*
@ -339,5 +431,23 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
{
return $this->_doc->saveHTML();
}
/**
* Get HTML body
*
* @return string
*/
public function getHtmlBody()
{
$xpath = new DOMXPath($this->_doc);
$bodyNodes = $xpath->query('/html/body')->item(0)->childNodes;
$outputFragments = array();
for ($count = 0; $count < $bodyNodes->length; $count++) {
$outputFragments[] = $this->_doc->saveXML($bodyNodes->item($count));
}
return implode($outputFragments);
}
}

2
libs/Zend/Search/Lucene/Document/OpenXml.php
View File

@ -23,7 +23,7 @@
/** Zend_Search_Lucene_Document */
require_once 'Zend/Search/Lucene/Document.php';
if (class_exists('ZipArchive')) {
if (class_exists('ZipArchive', false)) {
/**
* OpenXML document.

38
libs/Zend/Search/Lucene/Document/Pptx.php
View File

@ -23,7 +23,7 @@
/** Zend_Search_Lucene_Document_OpenXml */
require_once 'Zend/Search/Lucene/Document/OpenXml.php';
if (class_exists('ZipArchive')) {
if (class_exists('ZipArchive', false)) {
/**
* Pptx document.
@ -42,14 +42,14 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
* @var string
*/
const SCHEMA_PRESENTATIONML = 'http://schemas.openxmlformats.org/presentationml/2006/main';
/**
* Xml Schema - DrawingML
*
* @var string
*/
const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
/**
* Xml Schema - Slide relation
*
@ -63,7 +63,7 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
* @var string
*/
const SCHEMA_SLIDENOTESRELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/notesSlide';
/**
* Object constructor
*
@ -94,7 +94,7 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
$slides[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
$package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . basename($slideRel["Target"])) )
);
// Search for slide notes
$slideNotesRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/_rels/" . basename($slideRel["Target"]) . ".rels")) );
foreach ($slideNotesRelations->Relationship as $slideNoteRel) {
@ -103,27 +103,27 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
$slideNotes[ str_replace( 'rId', '', (string)$slideRel["Id"] ) ] = simplexml_load_string(
$package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . dirname($slideRel["Target"]) . "/" . dirname($slideNoteRel["Target"]) . "/" . basename($slideNoteRel["Target"])) )
);
break;
}
}
}
}
break;
}
}
// Sort slides
ksort($slides);
ksort($slideNotes);
// Extract contents from slides
foreach ($slides as $slideKey => $slide) {
// Register namespaces
$slide->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
$slide->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
// Fetch all text
$textElements = $slide->xpath('//a:t');
foreach ($textElements as $textElement) {
@ -138,15 +138,15 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
// Register namespaces
$slideNote->registerXPathNamespace("p", Zend_Search_Lucene_Document_Pptx::SCHEMA_PRESENTATIONML);
$slideNote->registerXPathNamespace("a", Zend_Search_Lucene_Document_Pptx::SCHEMA_DRAWINGML);
// Fetch all text
$textElements = $slideNote->xpath('//a:t');
foreach ($textElements as $textElement) {
$documentBody[] = (string)$textElement;
}
}
}
}
// Read core properties
$coreProperties = $this->extractMetaData($package);
@ -154,25 +154,25 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
$package->close();
// Store filename
$this->addField(Zend_Search_Lucene_Field::Text('filename', $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)));
$this->addField(Zend_Search_Lucene_Field::Text('body', implode(' ', $documentBody), 'UTF-8'));
} else {
$this->addField(Zend_Search_Lucene_Field::UnStored('body', implode(' ', $documentBody)));
$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));
$this->addField(Zend_Search_Lucene_Field::Text($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));
$this->addField(Zend_Search_Lucene_Field::Text('title', $fileName, 'UTF-8'));
}
}

50
libs/Zend/Search/Lucene/Document/Xlsx.php
View File

@ -23,7 +23,7 @@
/** Zend_Search_Lucene_Document_OpenXml */
require_once 'Zend/Search/Lucene/Document/OpenXml.php';
if (class_exists('ZipArchive')) {
if (class_exists('ZipArchive', false)) {
/**
* Xlsx document.
@ -42,21 +42,21 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
* @var string
*/
const SCHEMA_SPREADSHEETML = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main';
/**
* Xml Schema - DrawingML
*
* @var string
*/
const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
const SCHEMA_DRAWINGML = 'http://schemas.openxmlformats.org/drawingml/2006/main';
/**
* Xml Schema - Shared Strings
*
* @var string
*/
const SCHEMA_SHAREDSTRINGS = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/sharedStrings';
/**
* Xml Schema - Worksheet relation
*
@ -70,7 +70,7 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
* @var string
*/
const SCHEMA_SLIDENOTESRELATION = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/notesSlide';
/**
* Object constructor
*
@ -84,7 +84,7 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
$worksheets = array();
$documentBody = array();
$coreProperties = array();
// Open OpenXML package
$package = new ZipArchive();
$package->open($fileName);
@ -96,11 +96,11 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
// Found office document! Read relations for workbook...
$workbookRelations = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/_rels/" . basename($rel["Target"]) . ".rels")) );
$workbookRelations->registerXPathNamespace("rel", Zend_Search_Lucene_Document_OpenXml::SCHEMA_RELATIONSHIP);
// Read shared strings
$sharedStringsPath = $workbookRelations->xpath("rel:Relationship[@Type='" . Zend_Search_Lucene_Document_Xlsx::SCHEMA_SHAREDSTRINGS . "']");
$sharedStringsPath = (string)$sharedStringsPath[0]['Target'];
$xmlStrings = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . $sharedStringsPath)) );
$sharedStringsPath = (string)$sharedStringsPath[0]['Target'];
$xmlStrings = simplexml_load_string($package->getFromName( $this->absoluteZipPath(dirname($rel["Target"]) . "/" . $sharedStringsPath)) );
if (isset($xmlStrings) && isset($xmlStrings->si)) {
foreach ($xmlStrings->si as $val) {
if (isset($val->t)) {
@ -119,14 +119,14 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
);
}
}
break;
}
}
// Sort worksheets
ksort($worksheets);
// Extract contents from worksheets
foreach ($worksheets as $sheetKey => $worksheet) {
foreach ($worksheet->sheetData->row as $row) {
@ -143,7 +143,7 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
}
break;
case "b":
// Value is boolean
$value = (string)$c->v;
@ -156,13 +156,13 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
}
break;
case "inlineStr":
// Value is rich text inline
$value = $this->_parseRichText($c->is);
break;
case "e":
// Value is an error message
if ((string)$c->v != '') {
@ -184,11 +184,11 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
elseif ($value == (double)$value) $value = (double)$value;
}
}
$documentBody[] = $value;
}
}
}
}
// Read core properties
$coreProperties = $this->extractMetaData($package);
@ -197,28 +197,28 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
$package->close();
// Store filename
$this->addField(Zend_Search_Lucene_Field::Text('filename', $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)));
$this->addField(Zend_Search_Lucene_Field::Text('body', implode(' ', $documentBody), 'UTF-8'));
} else {
$this->addField(Zend_Search_Lucene_Field::UnStored('body', implode(' ', $documentBody)));
$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));
$this->addField(Zend_Search_Lucene_Field::Text($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));
$this->addField(Zend_Search_Lucene_Field::Text('title', $fileName, 'UTF-8'));
}
}
/**
* Parse rich text XML
*