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

import v2.0.0.0_RC3 | 2012-07-01

Browse Source 
https://github.com/lucanos/CommunityID -> http://www.itadmins.net/archives/357
This commit is contained in:
6543
2019-07-17 22:31:04 +02:00
parent 38c146901c
commit 2f397f01f7
2677 changed files with 296182 additions and 45159 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
libs/Zend/Search/Lucene/Analysis/Analyzer.php
View File

@ -17,42 +17,37 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Analyzer.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Analyzer.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Analysis_Token */
require_once 'Zend/Search/Lucene/Analysis/Token.php';
/** User land classes and interfaces turned on by Zend/Search/Analyzer.php file inclusion. */
/** @todo Section should be removed with ZF 2.0 release as obsolete */
if (!defined('ZEND_SEARCH_LUCENE_COMMON_ANALYZER_PROCESSED')) {
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8 */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8 */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8Num.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8Num.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8Num/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8Num/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Text */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Text.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Text */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Text.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Text_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Text/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Text_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Text/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/TextNum.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/TextNum.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/TextNum/CaseInsensitive.php';
/** Zend_Search_Lucene_Analysis_TokenFilter_StopWords */
require_once 'Zend/Search/Lucene/Analysis/TokenFilter/StopWords.php';
/** Zend_Search_Lucene_Analysis_TokenFilter_ShortWords */
require_once 'Zend/Search/Lucene/Analysis/TokenFilter/ShortWords.php';
/** Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/TextNum/CaseInsensitive.php';
}
/**
@ -167,6 +162,9 @@ abstract class Zend_Search_Lucene_Analysis_Analyzer
*/
public static function getDefault()
{
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Text_CaseInsensitive */
require_once 'Zend/Search/Lucene/Analysis/Analyzer/Common/Text/CaseInsensitive.php';
if (!self::$_defaultImpl instanceof Zend_Search_Lucene_Analysis_Analyzer) {
self::$_defaultImpl = new Zend_Search_Lucene_Analysis_Analyzer_Common_Text_CaseInsensitive();
}

13
libs/Zend/Search/Lucene/Analysis/Analyzer/Common.php
View File

@ -17,13 +17,24 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Common.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Common.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Define constant used to provide correct file processing order */
/** @todo Section should be removed with ZF 2.0 release as obsolete */
define('ZEND_SEARCH_LUCENE_COMMON_ANALYZER_PROCESSED', true);
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
/** Zend_Search_Lucene_Analysis_Token */
require_once 'Zend/Search/Lucene/Analysis/Token.php';
/** Zend_Search_Lucene_Analysis_TokenFilter */
require_once 'Zend/Search/Lucene/Analysis/TokenFilter.php';
/**
* Common implementation of the Zend_Search_Lucene_Analysis_Analyzer interface.

10
libs/Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8.php
View File

@ -17,7 +17,7 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Utf8.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Utf8.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -48,7 +48,7 @@ class Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8 extends Zend_Search_Lucen
* @var integer
*/
private $_bytePosition;
/**
* Object constructor
*
@ -101,12 +101,12 @@ class Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8 extends Zend_Search_Lucen
// matched string
$matchedWord = $match[0][0];
// binary position of the matched word in the input stream
$binStartPos = $match[0][1];
// character position of the matched word in the input stream
$startPos = $this->_position +
$startPos = $this->_position +
iconv_strlen(substr($this->_input,
$this->_bytePosition,
$binStartPos - $this->_bytePosition),

4
libs/Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8/CaseInsensitive.php
View File

@ -17,7 +17,7 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: CaseInsensitive.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: CaseInsensitive.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -37,7 +37,7 @@ require_once 'Zend/Search/Lucene/Analysis/TokenFilter/LowerCaseUtf8.php';
*/
class Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive extends Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8
class Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive extends Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8
{
public function __construct()
{

8
libs/Zend/Search/Lucene/Analysis/Analyzer/Common/Utf8Num.php
View File

@ -17,7 +17,7 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Utf8Num.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Utf8Num.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -101,12 +101,12 @@ class Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8Num extends Zend_Search_Lu
// matched string
$matchedWord = $match[0][0];
// binary position of the matched word in the input stream
$binStartPos = $match[0][1];
// character position of the matched word in the input stream
$startPos = $this->_position +
$startPos = $this->_position +
iconv_strlen(substr($this->_input,
$this->_bytePosition,
$binStartPos - $this->_bytePosition),

3
libs/Zend/Search/Lucene/Analysis/TokenFilter.php
View File

@ -17,7 +17,7 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: TokenFilter.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: TokenFilter.php 18947 2009-11-12 11:57:17Z alexander $
*/
@ -34,7 +34,6 @@ require_once 'Zend/Search/Lucene/Analysis/Token.php';
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
abstract class Zend_Search_Lucene_Analysis_TokenFilter
{
/**

4
libs/Zend/Search/Lucene/Analysis/TokenFilter/LowerCaseUtf8.php
View File

@ -17,7 +17,7 @@
* @subpackage Analysis
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: LowerCaseUtf8.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: LowerCaseUtf8.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -48,7 +48,7 @@ class Zend_Search_Lucene_Analysis_TokenFilter_LowerCaseUtf8 extends Zend_Search_
throw new Zend_Search_Lucene_Exception('Utf8 compatible lower case filter needs mbstring extension to be enabled.');
}
}
/**
* Normalize Token or remove it (if null is returned)
*

22
libs/Zend/Search/Lucene/Document/Docx.php
View File

@ -17,7 +17,7 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Docx.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Docx.php 19035 2009-11-19 14:34:11Z alexander $
*/
/** Zend_Search_Lucene_Document_OpenXml */
@ -47,6 +47,7 @@ class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenX
*
* @param string $fileName
* @param boolean $storeContent
* @throws Zend_Search_Lucene_Exception
*/
private function __construct($fileName, $storeContent) {
// Document data holders
@ -58,7 +59,12 @@ class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenX
$package->open($fileName);
// Read relations and search for officeDocument
$relations = simplexml_load_string($package->getFromName('_rels/.rels'));
$relationsXml = $package->getFromName('_rels/.rels');
if ($relationsXml === false) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Invalid archive or corrupted .docx file.');
}
$relations = simplexml_load_string($relationsXml);
foreach($relations->Relationship as $rel) {
if ($rel ["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
// Found office document! Read in contents...
@ -75,8 +81,8 @@ class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenX
$runs = $paragraph->xpath('.//w:r/*[name() = "w:t" or name() = "w:br"]');
if ($runs === false) {
// Paragraph doesn't contain any text or breaks
continue;
// Paragraph doesn't contain any text or breaks
continue;
}
foreach ($runs as $run) {
@ -84,7 +90,7 @@ class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenX
// Break element
$documentBody[] = ' ';
} else {
$documentBody[] = (string)$run;
$documentBody[] = (string)$run;
}
}
@ -133,11 +139,11 @@ class Zend_Search_Lucene_Document_Docx extends Zend_Search_Lucene_Document_OpenX
*/
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.');
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, $storeContent);
return new Zend_Search_Lucene_Document_Docx($fileName, $storeContent);
}
}

54
libs/Zend/Search/Lucene/Document/Html.php
View File

@ -17,7 +17,7 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Html.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Html.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -88,28 +88,28 @@ 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
// 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]);
/** @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][0]);
@$this->_doc->loadHTML(iconv($defaultEncoding, 'UTF-8//IGNORE', substr($htmlData, 0, $htmlTagOffset))
@$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>');
}
$head->parentNode->removeChild($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
@ -264,6 +264,9 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
*/
protected function _highlightTextNode(DOMText $node, $wordsToHighlight, $callback, $params)
{
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$analyzer = Zend_Search_Lucene_Analysis_Analyzer::getDefault();
$analyzer->setInput($node->nodeValue, 'UTF-8');
@ -301,16 +304,16 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
. $highlightedWordNodeSetHtml
. '</body></html>');
if (!$success) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception("Error occured while loading highlighted text fragment: '$highlightedNodeHtml'.");
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception("Error occured while loading highlighted text fragment: '$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);
$nodeToImport = $highlightedWordNodeSet->item($count);
$node->parentNode->insertBefore($this->_doc->importNode($nodeToImport, true /* deep copy */),
$matchedWordNode);
}
$node->parentNode->removeChild($matchedWordNode);
@ -372,7 +375,7 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
*/
public function highlight($words, $colour = '#66ffff')
{
return $this->highlightExtended($words, array($this, 'applyColour'), array($colour));
return $this->highlightExtended($words, array($this, 'applyColour'), array($colour));
}
@ -389,6 +392,9 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
*/
public function highlightExtended($words, $callback, $params = array())
{
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
if (!is_array($words)) {
$words = array($words);
}
@ -410,8 +416,8 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
}
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.');
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);
@ -445,7 +451,7 @@ class Zend_Search_Lucene_Document_Html extends Zend_Search_Lucene_Document
$outputFragments = array();
for ($count = 0; $count < $bodyNodes->length; $count++) {
$outputFragments[] = $this->_doc->saveXML($bodyNodes->item($count));
$outputFragments[] = $this->_doc->saveXML($bodyNodes->item($count));
}
return implode($outputFragments);

9
libs/Zend/Search/Lucene/Document/OpenXml.php
View File

@ -17,13 +17,14 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: OpenXml.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: OpenXml.php 18951 2009-11-12 16:26:19Z alexander $
*/
/** Zend_Search_Lucene_Document */
require_once 'Zend/Search/Lucene/Document.php';
if (class_exists('ZipArchive', false)) {
/**
@ -82,7 +83,7 @@ abstract class Zend_Search_Lucene_Document_OpenXml extends Zend_Search_Lucene_Do
{
// Data holders
$coreProperties = array();
// Read relations and search for core properties
$relations = simplexml_load_string($package->getFromName("_rels/.rels"));
foreach ($relations->Relationship as $rel) {
@ -103,10 +104,10 @@ abstract class Zend_Search_Lucene_Document_OpenXml extends Zend_Search_Lucene_Do
}
}
}
return $coreProperties;
}
/**
* Determine absolute zip path
*

10
libs/Zend/Search/Lucene/Document/Pptx.php
View File

@ -17,7 +17,7 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Pptx.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Pptx.php 19035 2009-11-19 14:34:11Z alexander $
*/
@ -70,6 +70,7 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
*
* @param string $fileName
* @param boolean $storeContent
* @throws Zend_Search_Lucene_Exception
*/
private function __construct($fileName, $storeContent)
{
@ -84,7 +85,12 @@ class Zend_Search_Lucene_Document_Pptx extends Zend_Search_Lucene_Document_OpenX
$package->open($fileName);
// Read relations and search for officeDocument
$relations = simplexml_load_string($package->getFromName("_rels/.rels"));
$relationsXml = $package->getFromName('_rels/.rels');
if ($relationsXml === false) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Invalid archive or corrupted .pptx file.');
}
$relations = simplexml_load_string($relationsXml);
foreach ($relations->Relationship as $rel) {
if ($rel["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
// Found office document! Search for slides...

10
libs/Zend/Search/Lucene/Document/Xlsx.php
View File

@ -17,7 +17,7 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Xlsx.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Xlsx.php 19035 2009-11-19 14:34:11Z alexander $
*/
@ -77,6 +77,7 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
*
* @param string $fileName
* @param boolean $storeContent
* @throws Zend_Search_Lucene_Exception
*/
private function __construct($fileName, $storeContent)
{
@ -91,7 +92,12 @@ class Zend_Search_Lucene_Document_Xlsx extends Zend_Search_Lucene_Document_OpenX
$package->open($fileName);
// Read relations and search for officeDocument
$relations = simplexml_load_string($package->getFromName("_rels/.rels"));
$relationsXml = $package->getFromName('_rels/.rels');
if ($relationsXml === false) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Invalid archive or corrupted .xlsx file.');
}
$relations = simplexml_load_string($relationsXml);
foreach ($relations->Relationship as $rel) {
if ($rel["Type"] == Zend_Search_Lucene_Document_OpenXml::SCHEMA_OFFICEDOCUMENT) {
// Found office document! Read relations for workbook...

20
libs/Zend/Search/Lucene/Field.php
View File

@ -17,7 +17,7 @@
* @subpackage Document
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Field.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Field.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -46,41 +46,41 @@ class Zend_Search_Lucene_Field
/**
* Field value
*
*
* @var boolean
*/
public $value;
/**
* Field is to be stored in the index for return with search hits.
*
*
* @var boolean
*/
public $isStored = false;
/**
* Field is to be indexed, so that it may be searched on.
*
*
* @var boolean
*/
public $isIndexed = true;
/**
* Field should be tokenized as text prior to indexing.
*
*
* @var boolean
*/
public $isTokenized = true;
/**
* Field is stored as binary.
*
*
* @var boolean
*/
public $isBinary = false;
/**
* Field are stored as a term vector
*
*
* @var boolean
*/
public $storeTermVector = false;
@ -218,7 +218,7 @@ class Zend_Search_Lucene_Field
strcasecmp($this->encoding, 'utf-8') == 0 ) {
return $this->value;
} else {
return (PHP_OS != 'AIX') ? iconv($this->encoding, 'UTF-8', $this->value) : iconv('ISO8859-1', 'UTF-8', $this->value);
}
}

103
libs/Zend/Search/Lucene/Index/SegmentInfo.php
View File

@ -17,18 +17,25 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: SegmentInfo.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: SegmentInfo.php 18951 2009-11-12 16:26:19Z alexander $
*/
/** Zend_Search_Lucene_Index_DictionaryLoader */
require_once 'Zend/Search/Lucene/Index/DictionaryLoader.php';
/** Zend_Search_Lucene_Index_DocsFilter */
require_once 'Zend/Search/Lucene/Index/DocsFilter.php';
/** Zend_Search_Lucene_Index_TermsStream_Interface */
require_once 'Zend/Search/Lucene/Index/TermsStream/Interface.php';
/** Zend_Search_Lucene_Search_Similarity */
require_once 'Zend/Search/Lucene/Search/Similarity.php';
/** Zend_Search_Lucene_Index_FieldInfo */
require_once 'Zend/Search/Lucene/Index/FieldInfo.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Index_TermInfo */
require_once 'Zend/Search/Lucene/Index/TermInfo.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -297,6 +304,7 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
$fieldNames = array();
$fieldNums = array();
$this->_fields = array();
for ($count=0; $count < $fieldsCount; $count++) {
$fieldName = $fnmFile->readString();
$fieldBits = $fnmFile->readByte();
@ -318,8 +326,8 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
$this->_fieldsDicPositions = array_flip($fieldNums);
if ($this->_delGen == -2) {
// SegmentInfo constructor is invoked from index writer
// Autodetect current delete file generation number
// SegmentInfo constructor is invoked from index writer
// Autodetect current delete file generation number
$this->_delGen = $this->_detectLatestDelGen();
}
@ -436,18 +444,18 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
if (extension_loaded('bitset')) {
for ($bit = 0; $bit < 8; $bit++) {
if ($nonZeroByte & (1<<$bit)) {
for ($bit = 0; $bit < 8; $bit++) {
if ($nonZeroByte & (1<<$bit)) {
bitset_incl($deletions, $byteNum*8 + $bit);
}
}
}
}
return $deletions;
} else {
for ($bit = 0; $bit < 8; $bit++) {
if ($nonZeroByte & (1<<$bit)) {
for ($bit = 0; $bit < 8; $bit++) {
if ($nonZeroByte & (1<<$bit)) {
$deletions[$byteNum*8 + $bit] = 1;
}
}
}
}
return (count($deletions) > 0) ? $deletions : null;
}
@ -785,6 +793,9 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
$tiiFile = $this->openCompoundFile('.tii');
$tiiFileData = $tiiFile->readBytes($this->compoundFileLength('.tii'));
/** Zend_Search_Lucene_Index_DictionaryLoader */
require_once 'Zend/Search/Lucene/Index/DictionaryLoader.php';
// Load dictionary index data
list($this->_termDictionary, $this->_termDictionaryInfos) =
Zend_Search_Lucene_Index_DictionaryLoader::load($tiiFileData);
@ -1549,35 +1560,35 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
$latestDelGen = $this->_detectLatestDelGen();
if (!$this->_deletedDirty) {
// There was no deletions by current process
// There was no deletions by current process
if ($latestDelGen == $this->_delGen) {
// Delete file hasn't been updated by any concurrent process
return;
// Delete file hasn't been updated by any concurrent process
return;
} else if ($latestDelGen > $this->_delGen) {
// Delete file has been updated by some concurrent process
// Reload deletions file
$this->_delGen = $latestDelGen;
$this->_deleted = $this->_loadDelFile();
// Delete file has been updated by some concurrent process
// Reload deletions file
$this->_delGen = $latestDelGen;
$this->_deleted = $this->_loadDelFile();
return;
return;
} else {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Delete file processing workflow is corrupted for the segment \'' . $this->_name . '\'.');
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Delete file processing workflow is corrupted for the segment \'' . $this->_name . '\'.');
}
}
if ($latestDelGen > $this->_delGen) {
// Merge current deletions with latest deletions file
$this->_delGen = $latestDelGen;
// Merge current deletions with latest deletions file
$this->_delGen = $latestDelGen;
$latestDelete = $this->_loadDelFile();
$latestDelete = $this->_loadDelFile();
if (extension_loaded('bitset')) {
$this->_deleted = bitset_union($this->_deleted, $latestDelete);
} else {
$this->_deleted += $latestDelete;
}
if (extension_loaded('bitset')) {
$this->_deleted = bitset_union($this->_deleted, $latestDelete);
} else {
$this->_deleted += $latestDelete;
}
}
if (extension_loaded('bitset')) {
@ -1756,19 +1767,19 @@ class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_T
*/
public function resetTermsStream(/** $startId = 0, $mode = self::SM_TERMS_ONLY */)
{
/**
* SegmentInfo->resetTermsStream() method actually takes two optional parameters:
* $startId (default value is 0)
* $mode (default value is self::SM_TERMS_ONLY)
*/
$argList = func_get_args();
if (count($argList) > 2) {
/**
* SegmentInfo->resetTermsStream() method actually takes two optional parameters:
* $startId (default value is 0)
* $mode (default value is self::SM_TERMS_ONLY)
*/
$argList = func_get_args();
if (count($argList) > 2) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wrong number of arguments');
} else if (count($argList) == 2) {
$startId = $argList[0];
$mode = $argList[1];
} else if (count($argList) == 1) {
} else if (count($argList) == 2) {
$startId = $argList[0];
$mode = $argList[1];
} else if (count($argList) == 1) {
$startId = $argList[0];
$mode = self::SM_TERMS_ONLY;
} else {

12
libs/Zend/Search/Lucene/Index/SegmentMerger.php
View File

@ -17,17 +17,12 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: SegmentMerger.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: SegmentMerger.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
/** Zend_Search_Lucene_Index_SegmentWriter_StreamWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter/StreamWriter.php';
/** Zend_Search_Lucene_Index_TermsPriorityQueue */
require_once 'Zend/Search/Lucene/Index/TermsPriorityQueue.php';
/**
* @category Zend
@ -87,6 +82,8 @@ class Zend_Search_Lucene_Index_SegmentMerger
*/
public function __construct($directory, $name)
{
/** Zend_Search_Lucene_Index_SegmentWriter_StreamWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter/StreamWriter.php';
$this->_writer = new Zend_Search_Lucene_Index_SegmentWriter_StreamWriter($directory, $name);
}
@ -226,6 +223,9 @@ class Zend_Search_Lucene_Index_SegmentMerger
*/
private function _mergeTerms()
{
/** Zend_Search_Lucene_Index_TermsPriorityQueue */
require_once 'Zend/Search/Lucene/Index/TermsPriorityQueue.php';
$segmentInfoQueue = new Zend_Search_Lucene_Index_TermsPriorityQueue();
$segmentStartId = 0;

13
libs/Zend/Search/Lucene/Index/SegmentWriter.php
View File

@ -17,11 +17,18 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: SegmentWriter.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: SegmentWriter.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
/** Zend_Search_Lucene_Index_FieldInfo */
require_once 'Zend/Search/Lucene/Index/FieldInfo.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Index_TermInfo */
require_once 'Zend/Search/Lucene/Index/TermInfo.php';
/**
* @category Zend

14
libs/Zend/Search/Lucene/Index/SegmentWriter/DocumentWriter.php
View File

@ -17,12 +17,9 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: DocumentWriter.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: DocumentWriter.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
/** Zend_Search_Lucene_Index_SegmentWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter.php';
@ -74,6 +71,9 @@ class Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter extends Zend_Search_
*/
public function addDocument(Zend_Search_Lucene_Document $document)
{
/** Zend_Search_Lucene_Search_Similarity */
require_once 'Zend/Search/Lucene/Search/Similarity.php';
$storedFields = array();
$docNorms = array();
$similarity = Zend_Search_Lucene_Search_Similarity::getDefault();
@ -92,6 +92,9 @@ class Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter extends Zend_Search_
if ($field->isIndexed) {
if ($field->isTokenized) {
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$analyzer = Zend_Search_Lucene_Analysis_Analyzer::getDefault();
$analyzer->setInput($field->value, $field->encoding);
@ -201,6 +204,9 @@ class Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter extends Zend_Search_
$this->_generateCFS();
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
return new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
$this->_name,
$this->_docCount,

8
libs/Zend/Search/Lucene/Index/SegmentWriter/StreamWriter.php
View File

@ -17,12 +17,9 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: StreamWriter.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: StreamWriter.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
/** Zend_Search_Lucene_Index_SegmentWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter.php';
@ -82,6 +79,9 @@ class Zend_Search_Lucene_Index_SegmentWriter_StreamWriter extends Zend_Search_Lu
$this->_dumpFNM();
$this->_generateCFS();
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
return new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
$this->_name,
$this->_docCount,

4
libs/Zend/Search/Lucene/Index/TermsPriorityQueue.php
View File

@ -17,10 +17,10 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: TermsPriorityQueue.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: TermsPriorityQueue.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene */
/** Zend_Search_Lucene_PriorityQueue */
require_once 'Zend/Search/Lucene/PriorityQueue.php';
/**

132
libs/Zend/Search/Lucene/Index/TermsStream/Interface.php
View File

@ -1,66 +1,66 @@
<?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
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Interface.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
interface Zend_Search_Lucene_Index_TermsStream_Interface
{
/**
* Reset terms stream.
*/
public function resetTermsStream();
/**
* Skip terms stream up to specified term preffix.
*
* Prefix contains fully specified field info and portion of searched term
*
* @param Zend_Search_Lucene_Index_Term $prefix
*/
public function skipTo(Zend_Search_Lucene_Index_Term $prefix);
/**
* Scans terms dictionary and returns next term
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function nextTerm();
/**
* Returns term in current position
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function currentTerm();
/**
* Close terms stream
*
* Should be used for resources clean up if stream is not read up to the end
*/
public function closeTermsStream();
}
<?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
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Interface.php 18946 2009-11-12 11:48:58Z alexander $
*/
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
interface Zend_Search_Lucene_Index_TermsStream_Interface
{
/**
* Reset terms stream.
*/
public function resetTermsStream();
/**
* Skip terms stream up to specified term preffix.
*
* Prefix contains fully specified field info and portion of searched term
*
* @param Zend_Search_Lucene_Index_Term $prefix
*/
public function skipTo(Zend_Search_Lucene_Index_Term $prefix);
/**
* Scans terms dictionary and returns next term
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function nextTerm();
/**
* Returns term in current position
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function currentTerm();
/**
* Close terms stream
*
* Should be used for resources clean up if stream is not read up to the end
*/
public function closeTermsStream();
}

20
libs/Zend/Search/Lucene/Index/Writer.php
View File

@ -17,24 +17,14 @@
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Writer.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Writer.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter/DocumentWriter.php';
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
/** Zend_Search_Lucene_Index_SegmentMerger */
require_once 'Zend/Search/Lucene/Index/SegmentMerger.php';
/** Zend_Search_Lucene_LockManager */
require_once 'Zend/Search/Lucene/LockManager.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -244,6 +234,9 @@ class Zend_Search_Lucene_Index_Writer
*/
public function addDocument(Zend_Search_Lucene_Document $document)
{
/** Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter */
require_once 'Zend/Search/Lucene/Index/SegmentWriter/DocumentWriter.php';
if ($this->_currentSegment === null) {
$this->_currentSegment =
new Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter($this->_directory, $this->_newSegmentName());
@ -379,6 +372,9 @@ class Zend_Search_Lucene_Index_Writer
private function _mergeSegments($segments)
{
$newName = $this->_newSegmentName();
/** Zend_Search_Lucene_Index_SegmentMerger */
require_once 'Zend/Search/Lucene/Index/SegmentMerger.php';
$merger = new Zend_Search_Lucene_Index_SegmentMerger($this->_directory,
$newName);
foreach ($segments as $segmentInfo) {
@ -521,6 +517,8 @@ class Zend_Search_Lucene_Index_Writer
$isCompound = true;
}
/** Zend_Search_Lucene_Index_SegmentInfo */
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
$this->_segmentInfos[$segName] =
new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
$segName,

15
libs/Zend/Search/Lucene/Interface.php
View File

@ -16,13 +16,26 @@
* @package Zend_Search_Lucene
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Interface.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Interface.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Index_TermsStream_Interface */
require_once 'Zend/Search/Lucene/Index/TermsStream/Interface.php';
/** Classes used within Zend_Search_Lucene_Interface API */
/** Zend_Search_Lucene_Document */
require_once 'Zend/Search/Lucene/Document.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Index_DocsFilter */
require_once 'Zend/Search/Lucene/Index/DocsFilter.php';
/**
* @category Zend
* @package Zend_Search_Lucene

1931
libs/Zend/Search/Lucene/MultiSearcher.php
View File

File diff suppressed because it is too large Load Diff

14
libs/Zend/Search/Lucene/Search/BooleanExpressionRecognizer.php
View File

@ -17,19 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: BooleanExpressionRecognizer.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: BooleanExpressionRecognizer.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_FSM */
require_once 'Zend/Search/Lucene/FSM.php';
/** Zend_Search_Lucene_Search_QueryToken */
require_once 'Zend/Search/Lucene/Search/QueryToken.php';
/** Zend_Search_Lucene_Search_QueryParser */
require_once 'Zend/Search/Lucene/Search/QueryParser.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -220,6 +214,9 @@ class Zend_Search_Lucene_Search_BooleanExpressionRecognizer extends Zend_Search_
*/
public function emptyOperatorAction()
{
/** Zend_Search_Lucene_Search_QueryParser */
require_once 'Zend/Search/Lucene/Search/QueryParser.php';
if (Zend_Search_Lucene_Search_QueryParser::getDefaultOperator() == Zend_Search_Lucene_Search_QueryParser::B_AND) {
// Do nothing
} else {
@ -235,6 +232,9 @@ class Zend_Search_Lucene_Search_BooleanExpressionRecognizer extends Zend_Search_
*/
public function emptyNotOperatorAction()
{
/** Zend_Search_Lucene_Search_QueryParser */
require_once 'Zend/Search/Lucene/Search/QueryParser.php';
if (Zend_Search_Lucene_Search_QueryParser::getDefaultOperator() == Zend_Search_Lucene_Search_QueryParser::B_AND) {
// Do nothing
} else {

188
libs/Zend/Search/Lucene/Search/Highlighter/Default.php
View File

@ -1,94 +1,94 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Default.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/** Zend_Search_Lucene_Search_Highlighter_Interface */
require_once 'Zend/Search/Lucene/Search/Highlighter/Interface.php';
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Highlighter_Default implements Zend_Search_Lucene_Search_Highlighter_Interface
{
/**
* List of colors for text highlighting
*
* @var array
*/
protected $_highlightColors = array('#66ffff', '#ff66ff', '#ffff66',
'#ff8888', '#88ff88', '#8888ff',
'#88dddd', '#dd88dd', '#dddd88',
'#aaddff', '#aaffdd', '#ddaaff',
'#ddffaa', '#ffaadd', '#ffddaa');
/**
* Index of current color for highlighting
*
* Index is increased at each highlight() call, so terms matching different queries are highlighted using different colors.
*
* @var integer
*/
protected $_currentColorIndex = 0;
/**
* HTML document for highlighting
*
* @var Zend_Search_Lucene_Document_Html
*/
protected $_doc;
/**
* Set document for highlighting.
*
* @param Zend_Search_Lucene_Document_Html $document
*/
public function setDocument(Zend_Search_Lucene_Document_Html $document)
{
$this->_doc = $document;
}
/**
* Get document for highlighting.
*
* @return Zend_Search_Lucene_Document_Html $document
*/
public function getDocument()
{
return $this->_doc;
}
/**
* Highlight specified words
*
* @param string|array $words Words to highlight. They could be organized using the array or string.
*/
public function highlight($words)
{
$color = $this->_highlightColors[$this->_currentColorIndex];
$this->_currentColorIndex = ($this->_currentColorIndex + 1) % count($this->_highlightColors);
$this->_doc->highlight($words, $color);
}
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Default.php 18946 2009-11-12 11:48:58Z alexander $
*/
/** @see Zend_Search_Lucene_Search_Highlighter_Interface */
require_once 'Zend/Search/Lucene/Search/Highlighter/Interface.php';
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Highlighter_Default implements Zend_Search_Lucene_Search_Highlighter_Interface
{
/**
* List of colors for text highlighting
*
* @var array
*/
protected $_highlightColors = array('#66ffff', '#ff66ff', '#ffff66',
'#ff8888', '#88ff88', '#8888ff',
'#88dddd', '#dd88dd', '#dddd88',
'#aaddff', '#aaffdd', '#ddaaff',
'#ddffaa', '#ffaadd', '#ffddaa');
/**
* Index of current color for highlighting
*
* Index is increased at each highlight() call, so terms matching different queries are highlighted using different colors.
*
* @var integer
*/
protected $_currentColorIndex = 0;
/**
* HTML document for highlighting
*
* @var Zend_Search_Lucene_Document_Html
*/
protected $_doc;
/**
* Set document for highlighting.
*
* @param Zend_Search_Lucene_Document_Html $document
*/
public function setDocument(Zend_Search_Lucene_Document_Html $document)
{
$this->_doc = $document;
}
/**
* Get document for highlighting.
*
* @return Zend_Search_Lucene_Document_Html $document
*/
public function getDocument()
{
return $this->_doc;
}
/**
* Highlight specified words
*
* @param string|array $words Words to highlight. They could be organized using the array or string.
*/
public function highlight($words)
{
$color = $this->_highlightColors[$this->_currentColorIndex];
$this->_currentColorIndex = ($this->_currentColorIndex + 1) % count($this->_highlightColors);
$this->_doc->highlight($words, $color);
}
}

106
libs/Zend/Search/Lucene/Search/Highlighter/Interface.php
View File

@ -1,53 +1,53 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Interface.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
interface Zend_Search_Lucene_Search_Highlighter_Interface
{
/**
* Set document for highlighting.
*
* @param Zend_Search_Lucene_Document_Html $document
*/
public function setDocument(Zend_Search_Lucene_Document_Html $document);
/**
* Get document for highlighting.
*
* @return Zend_Search_Lucene_Document_Html $document
*/
public function getDocument();
/**
* Highlight specified words (method is invoked once per subquery)
*
* @param string|array $words Words to highlight. They could be organized using the array or string.
*/
public function highlight($words);
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Interface.php 18946 2009-11-12 11:48:58Z alexander $
*/
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
interface Zend_Search_Lucene_Search_Highlighter_Interface
{
/**
* Set document for highlighting.
*
* @param Zend_Search_Lucene_Document_Html $document
*/
public function setDocument(Zend_Search_Lucene_Document_Html $document);
/**
* Get document for highlighting.
*
* @return Zend_Search_Lucene_Document_Html $document
*/
public function getDocument();
/**
* Highlight specified words (method is invoked once per subquery)
*
* @param string|array $words Words to highlight. They could be organized using the array or string.
*/
public function highlight($words);
}

23
libs/Zend/Search/Lucene/Search/Query.php
View File

@ -17,18 +17,9 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Query.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Query.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Document_Html */
require_once 'Zend/Search/Lucene/Document/Html.php';
/** Zend_Search_Lucene_Index_DocsFilter */
require_once 'Zend/Search/Lucene/Index/DocsFilter.php';
/** Zend_Search_Lucene_Search_Highlighter_Default */
require_once 'Zend/Search/Lucene/Search/Highlighter/Default.php';
/**
* @category Zend
@ -195,9 +186,13 @@ abstract class Zend_Search_Lucene_Search_Query
public function highlightMatches($inputHTML, $defaultEncoding = '', $highlighter = null)
{
if ($highlighter === null) {
$highlighter = new Zend_Search_Lucene_Search_Highlighter_Default();
require_once 'Zend/Search/Lucene/Search/Highlighter/Default.php';
$highlighter = new Zend_Search_Lucene_Search_Highlighter_Default();
}
/** Zend_Search_Lucene_Document_Html */
require_once 'Zend/Search/Lucene/Document/Html.php';
$doc = Zend_Search_Lucene_Document_Html::loadHTML($inputHTML, false, $defaultEncoding);
$highlighter->setDocument($doc);
@ -217,13 +212,17 @@ abstract class Zend_Search_Lucene_Search_Query
public function htmlFragmentHighlightMatches($inputHtmlFragment, $encoding = 'UTF-8', $highlighter = null)
{
if ($highlighter === null) {
require_once 'Zend/Search/Lucene/Search/Highlighter/Default.php';
$highlighter = new Zend_Search_Lucene_Search_Highlighter_Default();
}
$inputHTML = '<html><head><META HTTP-EQUIV="Content-type" CONTENT="text/html; charset=UTF-8"/></head><body>'
. iconv($encoding, 'UTF-8//IGNORE', $inputHtmlFragment) . '</body></html>';
$doc = Zend_Search_Lucene_Document_Html::loadHTML($inputHTML);
/** Zend_Search_Lucene_Document_Html */
require_once 'Zend/Search/Lucene/Document/Html.php';
$doc = Zend_Search_Lucene_Document_Html::loadHTML($inputHTML);
$highlighter->setDocument($doc);
$this->_highlightMatches($highlighter);

17
libs/Zend/Search/Lucene/Search/Query/Boolean.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Boolean.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Boolean.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Weight_Boolean */
require_once 'Zend/Search/Lucene/Search/Weight/Boolean.php';
/**
* @category Zend
@ -177,6 +174,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
}
if (count($subqueries) == 0) {
// Boolean query doesn't has non-insignificant subqueries
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
// Check if all non-insignificant subqueries are prohibited
@ -188,6 +186,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
}
}
if ($allProhibited) {
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
@ -197,6 +196,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
if ($subquery instanceof Zend_Search_Lucene_Search_Query_Empty) {
if ($signs[$id] === true) {
// Matching is required, but is actually empty
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else {
// Matching is optional or prohibited, but is empty
@ -209,6 +209,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
// Check, if reduced subqueries list is empty
if (count($subqueries) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -221,6 +222,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
}
}
if ($allProhibited) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -355,6 +357,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
// Check, if all subqueries have been decomposed and all terms has the same boost factor
if (count($subqueries) == 0 && count(array_unique($boostFactors)) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$optimizedQuery = new Zend_Search_Lucene_Search_Query_MultiTerm($terms, $tsigns);
$optimizedQuery->setBoost(reset($boostFactors)*$this->getBoost());
@ -378,6 +381,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
}
if (count($terms) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$clause = new Zend_Search_Lucene_Search_Query_Term(reset($terms));
$clause->setBoost(reset($boostFactors));
@ -387,6 +391,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
// Clear terms list
$terms = array();
} else if (count($terms) > 1 && count(array_unique($boostFactors)) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$clause = new Zend_Search_Lucene_Search_Query_MultiTerm($terms, $tsigns);
$clause->setBoost(reset($boostFactors));
@ -400,6 +405,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
if (count($prohibitedTerms) == 1) {
// (boost factors are not significant for prohibited clauses)
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$subqueries[] = new Zend_Search_Lucene_Search_Query_Term(reset($prohibitedTerms));
$signs[] = false;
@ -414,6 +420,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
}
// (boost factors are not significant for prohibited clauses)
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$subqueries[] = new Zend_Search_Lucene_Search_Query_MultiTerm($prohibitedTerms, $prohibitedSigns);
// Clause sign is 'prohibited'
$signs[] = false;
@ -464,6 +471,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/Boolean.php';
$this->_weight = new Zend_Search_Lucene_Search_Weight_Boolean($this, $reader);
return $this->_weight;
}
@ -686,6 +694,7 @@ class Zend_Search_Lucene_Search_Query_Boolean extends Zend_Search_Lucene_Search_
if ($docsFilter === null) {
// Create local documents filter if it's not provided by upper query
require_once 'Zend/Search/Lucene/Index/DocsFilter.php';
$docsFilter = new Zend_Search_Lucene_Index_DocsFilter();
}

6
libs/Zend/Search/Lucene/Search/Query/Empty.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Empty.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Empty.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Weight_Empty */
require_once 'Zend/Search/Lucene/Search/Weight/Empty.php';
/**
* @category Zend
@ -68,6 +65,7 @@ class Zend_Search_Lucene_Search_Query_Empty extends Zend_Search_Lucene_Search_Qu
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/Empty.php';
return new Zend_Search_Lucene_Search_Weight_Empty();
}

19
libs/Zend/Search/Lucene/Search/Query/Fuzzy.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Fuzzy.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Fuzzy.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Query_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
/**
* @category Zend
@ -196,6 +193,7 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
$fields = array($this->_term->field);
}
require_once 'Zend/Search/Lucene/Index/Term.php';
$prefix = Zend_Search_Lucene_Index_Term::getPrefix($this->_term->text, $this->_prefixLength);
$prefixByteLength = strlen($prefix);
$prefixUtf8Length = Zend_Search_Lucene_Index_Term::getLength($prefix);
@ -208,10 +206,12 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
$scaleFactor = 1/(1 - $this->_minimumSimilarity);
require_once 'Zend/Search/Lucene.php';
$maxTerms = Zend_Search_Lucene::getTermsPerQueryLimit();
foreach ($fields as $field) {
$index->resetTermsStream();
require_once 'Zend/Search/Lucene/Index/Term.php';
if ($prefix != '') {
$index->skipTo(new Zend_Search_Lucene_Index_Term($prefix, $field));
@ -298,10 +298,13 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
}
if (count($this->_matches) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else if (count($this->_matches) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
return new Zend_Search_Lucene_Search_Query_Term(reset($this->_matches));
} else {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$rewrittenQuery = new Zend_Search_Lucene_Search_Query_Boolean();
array_multisort($this->_scores, SORT_DESC, SORT_NUMERIC,
@ -309,6 +312,7 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
$this->_matches);
$termCount = 0;
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
foreach ($this->_matches as $id => $matchedTerm) {
$subquery = new Zend_Search_Lucene_Search_Query_Term($matchedTerm);
$subquery->setBoost($this->_scores[$id]);
@ -418,6 +422,7 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
{
$words = array();
require_once 'Zend/Search/Lucene/Index/Term.php';
$prefix = Zend_Search_Lucene_Index_Term::getPrefix($this->_term->text, $this->_prefixLength);
$prefixByteLength = strlen($prefix);
$prefixUtf8Length = Zend_Search_Lucene_Index_Term::getLength($prefix);
@ -430,13 +435,13 @@ class Zend_Search_Lucene_Search_Query_Fuzzy extends Zend_Search_Lucene_Search_Qu
$scaleFactor = 1/(1 - $this->_minimumSimilarity);
$docBody = $highlighter->getDocument()->getFieldUtf8Value('body');
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($docBody, 'UTF-8');
foreach ($tokens as $token) {
$termText = $token->getTermText();
$termText = $token->getTermText();
if (substr($termText, 0, $prefixByteLength) == $prefix) {
if (substr($termText, 0, $prefixByteLength) == $prefix) {
// Calculate similarity
$target = substr($termText, $prefixByteLength);

6
libs/Zend/Search/Lucene/Search/Query/Insignificant.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Insignificant.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Insignificant.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Weight_Empty */
require_once 'Zend/Search/Lucene/Search/Weight/Empty.php';
/**
* The insignificant query returns empty result, but doesn't limit result set as a part of other queries
@ -69,6 +66,7 @@ class Zend_Search_Lucene_Search_Query_Insignificant extends Zend_Search_Lucene_S
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/Empty.php';
return new Zend_Search_Lucene_Search_Weight_Empty();
}

15
libs/Zend/Search/Lucene/Search/Query/MultiTerm.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: MultiTerm.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: MultiTerm.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Weight_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Weight/MultiTerm.php';
/**
* @category Zend
@ -109,6 +106,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
public function __construct($terms = null, $signs = null)
{
if (is_array($terms)) {
require_once 'Zend/Search/Lucene.php';
if (count($terms) > Zend_Search_Lucene::getTermsPerQueryLimit()) {
throw new Zend_Search_Lucene_Exception('Terms per query limit is reached.');
}
@ -165,6 +163,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if (count($this->_terms) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -181,9 +180,11 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
return $this;
} else {
/** transform multiterm query to boolean and apply rewrite() method to subqueries. */
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->setBoost($this->getBoost());
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
foreach ($this->_terms as $termId => $term) {
$subquery = new Zend_Search_Lucene_Search_Query_Term($term);
@ -210,6 +211,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
if (!$index->hasTerm($term)) {
if ($signs === null || $signs[$id] === true) {
// Term is required
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else {
// Term is optional or prohibited
@ -233,6 +235,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
}
}
if ($allProhibited) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -245,6 +248,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
// It's already checked, that it's not a prohibited term
// It's one term query with one required or optional element
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$optimizedQuery = new Zend_Search_Lucene_Search_Query_Term(reset($terms));
$optimizedQuery->setBoost($this->getBoost());
@ -252,6 +256,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
}
if (count($terms) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -303,6 +308,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/MultiTerm.php';
$this->_weight = new Zend_Search_Lucene_Search_Weight_MultiTerm($this, $reader);
return $this->_weight;
}
@ -333,6 +339,7 @@ class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Searc
$ids, SORT_ASC, SORT_NUMERIC,
$this->_terms);
require_once 'Zend/Search/Lucene/Index/DocsFilter.php';
$docsFilter = new Zend_Search_Lucene_Index_DocsFilter();
foreach ($this->_terms as $termId => $term) {
$termDocs = $reader->termDocs($term, $docsFilter);

25
libs/Zend/Search/Lucene/Search/Query/Phrase.php
View File

@ -17,20 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Phrase.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Phrase.php 18954 2009-11-12 20:01:33Z alexander $
*/
/**
* Zend_Search_Lucene_Search_Query
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/**
* Zend_Search_Lucene_Search_Weight_Phrase
*/
require_once 'Zend/Search/Lucene/Search/Weight/Phrase.php';
/**
* A Query that matches documents containing a particular sequence of terms.
@ -109,6 +102,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
if (is_array($terms)) {
$this->_terms = array();
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($terms as $termId => $termText) {
$this->_terms[$termId] = ($field !== null)? new Zend_Search_Lucene_Index_Term($termText, $field):
new Zend_Search_Lucene_Index_Term($termText);
@ -116,11 +110,13 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
} else if ($terms === null) {
$this->_terms = array();
} else {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('terms argument must be array of strings or null');
}
if (is_array($offsets)) {
if (count($this->_terms) != count($offsets)) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('terms and offsets arguments must have the same size.');
}
$this->_offsets = $offsets;
@ -131,6 +127,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
$this->_offsets[$termId] = $position;
}
} else {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('offsets argument must be array of strings or null');
}
}
@ -167,6 +164,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
*/
public function addTerm(Zend_Search_Lucene_Index_Term $term, $position = null) {
if ((count($this->_terms) != 0)&&(end($this->_terms)->field != $term->field)) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('All phrase terms must be in the same field: ' .
$term->field . ':' . $term->text);
}
@ -191,10 +189,12 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if (count($this->_terms) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else if ($this->_terms[0]->field !== null) {
return $this;
} else {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->setBoost($this->getBoost());
@ -202,6 +202,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
$subquery = new Zend_Search_Lucene_Search_Query_Phrase();
$subquery->setSlop($this->getSlop());
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($this->_terms as $termId => $term) {
$qualifiedTerm = new Zend_Search_Lucene_Index_Term($term->text, $fieldName);
@ -226,12 +227,14 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
// Check, that index contains all phrase terms
foreach ($this->_terms as $term) {
if (!$index->hasTerm($term)) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
}
if (count($this->_terms) == 1) {
// It's one term query
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$optimizedQuery = new Zend_Search_Lucene_Search_Query_Term(reset($this->_terms));
$optimizedQuery->setBoost($this->getBoost());
@ -239,6 +242,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
}
if (count($this->_terms) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -277,6 +281,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/Phrase.php';
$this->_weight = new Zend_Search_Lucene_Search_Weight_Phrase($this, $reader);
return $this->_weight;
}
@ -543,7 +548,7 @@ class Zend_Search_Lucene_Search_Query_Phrase extends Zend_Search_Lucene_Search_Q
if (isset($this->_terms[0]) && $this->_terms[0]->field !== null) {
$query = $this->_terms[0]->field . ':';
} else {
$query = '';
$query = '';
}
$query .= '"';

261
libs/Zend/Search/Lucene/Search/Query/Preprocessing.php
View File

@ -1,134 +1,127 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Preprocessing.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/**
* Zend_Search_Lucene_Search_Query
*/
require_once 'Zend/Search/Lucene/Search/Query.php';
/**
* Zend_Search_Lucene_Search_Weight
*/
require_once 'Zend/Search/Lucene/Search/Weight.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
abstract class Zend_Search_Lucene_Search_Query_Preprocessing extends Zend_Search_Lucene_Search_Query
{
/**
* Matched terms.
*
* Matched terms list.
* It's filled during rewrite operation and may be used for search result highlighting
*
* Array of Zend_Search_Lucene_Index_Term objects
*
* @var array
*/
protected $_matches = null;
/**
* Optimize query in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function optimize(Zend_Search_Lucene_Interface $index)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Constructs an appropriate Weight implementation for this query.
*
* @param Zend_Search_Lucene_Interface $reader
* @return Zend_Search_Lucene_Search_Weight
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Execute query in context of index reader
* It also initializes necessary internal structures
*
* @param Zend_Search_Lucene_Interface $reader
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter
*/
public function execute(Zend_Search_Lucene_Interface $reader, $docsFilter = null)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Get document ids likely matching the query
*
* It's an array with document ids as keys (performance considerations)
*
* @return array
*/
public function matchedDocs()
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Score specified document
*
* @param integer $docId
* @param Zend_Search_Lucene_Interface $reader
* @return float
*/
public function score($docId, Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Return query terms
*
* @return array
*/
public function getQueryTerms()
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Rewrite operation has to be done before retrieving query terms.');
}
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Preprocessing.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
abstract class Zend_Search_Lucene_Search_Query_Preprocessing extends Zend_Search_Lucene_Search_Query
{
/**
* Matched terms.
*
* Matched terms list.
* It's filled during rewrite operation and may be used for search result highlighting
*
* Array of Zend_Search_Lucene_Index_Term objects
*
* @var array
*/
protected $_matches = null;
/**
* Optimize query in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function optimize(Zend_Search_Lucene_Interface $index)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Constructs an appropriate Weight implementation for this query.
*
* @param Zend_Search_Lucene_Interface $reader
* @return Zend_Search_Lucene_Search_Weight
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Execute query in context of index reader
* It also initializes necessary internal structures
*
* @param Zend_Search_Lucene_Interface $reader
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter
*/
public function execute(Zend_Search_Lucene_Interface $reader, $docsFilter = null)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Get document ids likely matching the query
*
* It's an array with document ids as keys (performance considerations)
*
* @return array
*/
public function matchedDocs()
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Score specified document
*
* @param integer $docId
* @param Zend_Search_Lucene_Interface $reader
* @return float
*/
public function score($docId, Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('This query is not intended to be executed.');
}
/**
* Return query terms
*
* @return array
*/
public function getQueryTerms()
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Rewrite operation has to be done before retrieving query terms.');
}
}

574
libs/Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php
View File

@ -1,287 +1,287 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Fuzzy.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/** Zend_Search_Lucene_Search_Query_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
/** Zend_Search_Lucene_Search_Query_Insignificant */
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
/** Zend_Search_Lucene_Search_Query_Empty */
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
/** Zend_Search_Lucene_Search_Query_Term */
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* word (query parser lexeme) to find.
*
* @var string
*/
private $_word;
/**
* Word encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_encoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* A value between 0 and 1 to set the required similarity
* between the query term and the matching terms. For example, for a
* _minimumSimilarity of 0.5 a term of the same length
* as the query term is considered similar to the query term if the edit distance
* between both terms is less than length(term)*0.5
*
* @var float
*/
private $_minimumSimilarity;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $word Non-tokenized word (query parser lexeme) to search.
* @param string $encoding Word encoding.
* @param string $fieldName Field name.
* @param float $minimumSimilarity minimum similarity
*/
public function __construct($word, $encoding, $fieldName, $minimumSimilarity)
{
$this->_word = $word;
$this->_encoding = $encoding;
$this->_field = $fieldName;
$this->_minimumSimilarity = $minimumSimilarity;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if ($this->_field === null) {
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$hasInsignificantSubqueries = false;
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_word,
$this->_encoding,
$fieldName,
$this->_minimumSimilarity);
$rewrittenSubquery = $subquery->rewrite($index);
if ( !($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant ||
$rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Empty) ) {
$query->addSubquery($rewrittenSubquery);
}
if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
$hasInsignificantSubqueries = true;
}
}
$subqueries = $query->getSubqueries();
if (count($subqueries) == 0) {
$this->_matches = array();
if ($hasInsignificantSubqueries) {
return new Zend_Search_Lucene_Search_Query_Insignificant();
} else {
return new Zend_Search_Lucene_Search_Query_Empty();
}
}
if (count($subqueries) == 1) {
$query = reset($subqueries);
}
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
$term = new Zend_Search_Lucene_Index_Term($this->_word, $this->_field);
if ($index->hasTerm($term)) {
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$subPatterns = preg_split('/[*?]/u', iconv($this->_encoding, 'UTF-8', $this->_word));
} else {
$subPatterns = preg_split('/[*?]/', $this->_word);
}
if (count($subPatterns) > 1) {
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search doesn\'t support wildcards (except within Keyword fields).');
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
$this->_matches = array();
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// Word is tokenized into several tokens
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search is supported only for non-multiple word terms');
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$subPatterns = preg_split('/[*?]/u', iconv($this->_encoding, 'UTF-8', $this->_word));
} else {
$subPatterns = preg_split('/[*?]/', $this->_word);
}
if (count($subPatterns) > 1) {
// Do nothing
return;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->_highlightMatches($highlighter);
return;
}
// Word is tokenized into several tokens
// But fuzzy search is supported only for non-multiple word terms
// Do nothing
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= $this->_word;
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Fuzzy.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* word (query parser lexeme) to find.
*
* @var string
*/
private $_word;
/**
* Word encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_encoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* A value between 0 and 1 to set the required similarity
* between the query term and the matching terms. For example, for a
* _minimumSimilarity of 0.5 a term of the same length
* as the query term is considered similar to the query term if the edit distance
* between both terms is less than length(term)*0.5
*
* @var float
*/
private $_minimumSimilarity;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $word Non-tokenized word (query parser lexeme) to search.
* @param string $encoding Word encoding.
* @param string $fieldName Field name.
* @param float $minimumSimilarity minimum similarity
*/
public function __construct($word, $encoding, $fieldName, $minimumSimilarity)
{
$this->_word = $word;
$this->_encoding = $encoding;
$this->_field = $fieldName;
$this->_minimumSimilarity = $minimumSimilarity;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if ($this->_field === null) {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$hasInsignificantSubqueries = false;
require_once 'Zend/Search/Lucene.php';
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php';
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_word,
$this->_encoding,
$fieldName,
$this->_minimumSimilarity);
$rewrittenSubquery = $subquery->rewrite($index);
if ( !($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant ||
$rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Empty) ) {
$query->addSubquery($rewrittenSubquery);
}
if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
$hasInsignificantSubqueries = true;
}
}
$subqueries = $query->getSubqueries();
if (count($subqueries) == 0) {
$this->_matches = array();
if ($hasInsignificantSubqueries) {
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
} else {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
}
if (count($subqueries) == 1) {
$query = reset($subqueries);
}
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($this->_word, $this->_field);
if ($index->hasTerm($term)) {
require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php';
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$subPatterns = preg_split('/[*?]/u', iconv($this->_encoding, 'UTF-8', $this->_word));
} else {
$subPatterns = preg_split('/[*?]/', $this->_word);
}
if (count($subPatterns) > 1) {
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search doesn\'t support wildcards (except within Keyword fields).');
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
$this->_matches = array();
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php';
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// Word is tokenized into several tokens
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search is supported only for non-multiple word terms');
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$subPatterns = preg_split('/[*?]/u', iconv($this->_encoding, 'UTF-8', $this->_word));
} else {
$subPatterns = preg_split('/[*?]/', $this->_word);
}
if (count($subPatterns) > 1) {
// Do nothing
return;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php';
$query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity);
$query->_highlightMatches($highlighter);
return;
}
// Word is tokenized into several tokens
// But fuzzy search is supported only for non-multiple word terms
// Do nothing
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= $this->_word;
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}

544
libs/Zend/Search/Lucene/Search/Query/Preprocessing/Phrase.php
View File

@ -1,274 +1,270 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Phrase.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/** Zend_Search_Lucene_Search_Query_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
/** Zend_Search_Lucene_Search_Query_Insignificant */
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
/** Zend_Search_Lucene_Search_Query_Empty */
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
/** Zend_Search_Lucene_Search_Query_Term */
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Phrase extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* Phrase to find.
*
* @var string
*/
private $_phrase;
/**
* Phrase encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_phraseEncoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* Sets the number of other words permitted between words in query phrase.
* If zero, then this is an exact phrase search. For larger values this works
* like a WITHIN or NEAR operator.
*
* The slop is in fact an edit-distance, where the units correspond to
* moves of terms in the query phrase out of position. For example, to switch
* the order of two words requires two moves (the first move places the words
* atop one another), so to permit re-orderings of phrases, the slop must be
* at least two.
* More exact matches are scored higher than sloppier matches, thus search
* results are sorted by exactness.
*
* The slop is zero by default, requiring exact matches.
*
* @var integer
*/
private $_slop;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $phrase Phrase to search.
* @param string $phraseEncoding Phrase encoding.
* @param string $fieldName Field name.
*/
public function __construct($phrase, $phraseEncoding, $fieldName)
{
$this->_phrase = $phrase;
$this->_phraseEncoding = $phraseEncoding;
$this->_field = $fieldName;
}
/**
* Set slop
*
* @param integer $slop
*/
public function setSlop($slop)
{
$this->_slop = $slop;
}
/**
* Get slop
*
* @return integer
*/
public function getSlop()
{
return $this->_slop;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
// Allow to use wildcards within phrases
// They are either removed by text analyzer or used as a part of keyword for keyword fields
//
// if (strpos($this->_phrase, '?') !== false || strpos($this->_phrase, '*') !== false) {
// require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
// throw new Zend_Search_Lucene_Search_QueryParserException('Wildcards are only allowed in a single terms.');
// }
// Split query into subqueries if field name is not specified
if ($this->_field === null) {
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->setBoost($this->getBoost());
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Phrase($this->_phrase,
$this->_phraseEncoding,
$fieldName);
$subquery->setSlop($this->getSlop());
$query->addSubquery($subquery->rewrite($index));
}
$this->_matches = $query->getQueryTerms();
return $query;
}
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
$term = new Zend_Search_Lucene_Index_Term($this->_phrase, $this->_field);
if ($index->hasTerm($term)) {
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// tokenize phrase using current analyzer and process it as a phrase query
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_phrase, $this->_phraseEncoding);
if (count($tokens) == 0) {
$this->_matches = array();
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's non-trivial phrase query
$position = -1;
$query = new Zend_Search_Lucene_Search_Query_Phrase();
foreach ($tokens as $token) {
$position += $token->getPositionIncrement();
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, $position);
$query->setSlop($this->getSlop());
}
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
/** Skip wildcard queries recognition. Supported wildcards are removed by text analyzer */
// tokenize phrase using current analyzer and process it as a phrase query
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_phrase, $this->_phraseEncoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
$highlighter->highlight($tokens[0]->getTermText());
return;
}
//It's non-trivial phrase query
$words = array();
foreach ($tokens as $token) {
$words[] = $token->getTermText();
}
$highlighter->highlight($words);
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= '"' . $this->_phrase . '"';
if ($this->_slop != 0) {
$query .= '~' . $this->_slop;
}
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Phrase.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Phrase extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* Phrase to find.
*
* @var string
*/
private $_phrase;
/**
* Phrase encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_phraseEncoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* Sets the number of other words permitted between words in query phrase.
* If zero, then this is an exact phrase search. For larger values this works
* like a WITHIN or NEAR operator.
*
* The slop is in fact an edit-distance, where the units correspond to
* moves of terms in the query phrase out of position. For example, to switch
* the order of two words requires two moves (the first move places the words
* atop one another), so to permit re-orderings of phrases, the slop must be
* at least two.
* More exact matches are scored higher than sloppier matches, thus search
* results are sorted by exactness.
*
* The slop is zero by default, requiring exact matches.
*
* @var integer
*/
private $_slop;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $phrase Phrase to search.
* @param string $phraseEncoding Phrase encoding.
* @param string $fieldName Field name.
*/
public function __construct($phrase, $phraseEncoding, $fieldName)
{
$this->_phrase = $phrase;
$this->_phraseEncoding = $phraseEncoding;
$this->_field = $fieldName;
}
/**
* Set slop
*
* @param integer $slop
*/
public function setSlop($slop)
{
$this->_slop = $slop;
}
/**
* Get slop
*
* @return integer
*/
public function getSlop()
{
return $this->_slop;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
// Allow to use wildcards within phrases
// They are either removed by text analyzer or used as a part of keyword for keyword fields
//
// if (strpos($this->_phrase, '?') !== false || strpos($this->_phrase, '*') !== false) {
// require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
// throw new Zend_Search_Lucene_Search_QueryParserException('Wildcards are only allowed in a single terms.');
// }
// Split query into subqueries if field name is not specified
if ($this->_field === null) {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->setBoost($this->getBoost());
require_once 'Zend/Search/Lucene.php';
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Phrase($this->_phrase,
$this->_phraseEncoding,
$fieldName);
$subquery->setSlop($this->getSlop());
$query->addSubquery($subquery->rewrite($index));
}
$this->_matches = $query->getQueryTerms();
return $query;
}
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($this->_phrase, $this->_field);
if ($index->hasTerm($term)) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// tokenize phrase using current analyzer and process it as a phrase query
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_phrase, $this->_phraseEncoding);
if (count($tokens) == 0) {
$this->_matches = array();
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's non-trivial phrase query
$position = -1;
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
$query = new Zend_Search_Lucene_Search_Query_Phrase();
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($tokens as $token) {
$position += $token->getPositionIncrement();
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, $position);
$query->setSlop($this->getSlop());
}
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
/** Skip wildcard queries recognition. Supported wildcards are removed by text analyzer */
// tokenize phrase using current analyzer and process it as a phrase query
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_phrase, $this->_phraseEncoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
$highlighter->highlight($tokens[0]->getTermText());
return;
}
//It's non-trivial phrase query
$words = array();
foreach ($tokens as $token) {
$words[] = $token->getTermText();
}
$highlighter->highlight($words);
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= '"' . $this->_phrase . '"';
if ($this->_slop != 0) {
$query .= '~' . $this->_slop;
}
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}

676
libs/Zend/Search/Lucene/Search/Query/Preprocessing/Term.php
View File

@ -1,335 +1,341 @@
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Term.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/** Zend_Search_Lucene_Search_Query_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
/** Zend_Search_Lucene_Search_Query_Insignificant */
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
/** Zend_Search_Lucene_Search_Query_Empty */
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
/** Zend_Search_Lucene_Search_Query_Term */
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Term extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* word (query parser lexeme) to find.
*
* @var string
*/
private $_word;
/**
* Word encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_encoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $word Non-tokenized word (query parser lexeme) to search.
* @param string $encoding Word encoding.
* @param string $fieldName Field name.
*/
public function __construct($word, $encoding, $fieldName)
{
$this->_word = $word;
$this->_encoding = $encoding;
$this->_field = $fieldName;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if ($this->_field === null) {
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
$query->setBoost($this->getBoost());
$hasInsignificantSubqueries = false;
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_word,
$this->_encoding,
$fieldName);
$rewrittenSubquery = $subquery->rewrite($index);
foreach ($rewrittenSubquery->getQueryTerms() as $term) {
$query->addTerm($term);
}
if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
$hasInsignificantSubqueries = true;
}
}
if (count($query->getTerms()) == 0) {
$this->_matches = array();
if ($hasInsignificantSubqueries) {
return new Zend_Search_Lucene_Search_Query_Insignificant();
} else {
return new Zend_Search_Lucene_Search_Query_Empty();
}
}
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
$term = new Zend_Search_Lucene_Index_Term($this->_word, $this->_field);
if ($index->hasTerm($term)) {
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$word = iconv($this->_encoding, 'UTF-8', $this->_word);
$wildcardsPattern = '/[*?]/u';
$subPatternsEncoding = 'UTF-8';
} else {
$word = $this->_word;
$wildcardsPattern = '/[*?]/';
$subPatternsEncoding = $this->_encoding;
}
$subPatterns = preg_split($wildcardsPattern, $word, -1, PREG_SPLIT_OFFSET_CAPTURE);
if (count($subPatterns) > 1) {
// Wildcard query is recognized
$pattern = '';
foreach ($subPatterns as $id => $subPattern) {
// Append corresponding wildcard character to the pattern before each sub-pattern (except first)
if ($id != 0) {
$pattern .= $word[ $subPattern[1] - 1 ];
}
// Check if each subputtern is a single word in terms of current analyzer
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($subPattern[0], $subPatternsEncoding);
if (count($tokens) > 1) {
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Wildcard search is supported only for non-multiple word terms');
}
foreach ($tokens as $token) {
$pattern .= $token->getTermText();
}
}
$term = new Zend_Search_Lucene_Index_Term($pattern, $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Wildcard($term);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
$this->_matches = array();
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's not insignificant or one term query
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
/**
* @todo Process $token->getPositionIncrement() to support stemming, synonyms and other
* analizer design features
*/
foreach ($tokens as $token) {
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, true); // all subterms are required
}
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$word = iconv($this->_encoding, 'UTF-8', $this->_word);
$wildcardsPattern = '/[*?]/u';
$subPatternsEncoding = 'UTF-8';
} else {
$word = $this->_word;
$wildcardsPattern = '/[*?]/';
$subPatternsEncoding = $this->_encoding;
}
$subPatterns = preg_split($wildcardsPattern, $word, -1, PREG_SPLIT_OFFSET_CAPTURE);
if (count($subPatterns) > 1) {
// Wildcard query is recognized
$pattern = '';
foreach ($subPatterns as $id => $subPattern) {
// Append corresponding wildcard character to the pattern before each sub-pattern (except first)
if ($id != 0) {
$pattern .= $word[ $subPattern[1] - 1 ];
}
// Check if each subputtern is a single word in terms of current analyzer
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($subPattern[0], $subPatternsEncoding);
if (count($tokens) > 1) {
// Do nothing (nothing is highlighted)
return;
}
foreach ($tokens as $token) {
$pattern .= $token->getTermText();
}
}
$term = new Zend_Search_Lucene_Index_Term($pattern, $this->_field);
$query = new Zend_Search_Lucene_Search_Query_Wildcard($term);
$query->_highlightMatches($highlighter);
return;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
$highlighter->highlight($tokens[0]->getTermText());
return;
}
//It's not insignificant or one term query
$words = array();
foreach ($tokens as $token) {
$words[] = $token->getTermText();
}
$highlighter->highlight($words);
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= $this->_word;
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}
<?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
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Term.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query_Processing */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing.php';
/**
* It's an internal abstract class intended to finalize ase a query processing after query parsing.
* This type of query is not actually involved into query execution.
*
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Search
* @internal
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_Search_Query_Preprocessing_Term extends Zend_Search_Lucene_Search_Query_Preprocessing
{
/**
* word (query parser lexeme) to find.
*
* @var string
*/
private $_word;
/**
* Word encoding (field name is always provided using UTF-8 encoding since it may be retrieved from index).
*
* @var string
*/
private $_encoding;
/**
* Field name.
*
* @var string
*/
private $_field;
/**
* Class constructor. Create a new preprocessing object for prase query.
*
* @param string $word Non-tokenized word (query parser lexeme) to search.
* @param string $encoding Word encoding.
* @param string $fieldName Field name.
*/
public function __construct($word, $encoding, $fieldName)
{
$this->_word = $word;
$this->_encoding = $encoding;
$this->_field = $fieldName;
}
/**
* Re-write query into primitive queries in the context of specified index
*
* @param Zend_Search_Lucene_Interface $index
* @return Zend_Search_Lucene_Search_Query
*/
public function rewrite(Zend_Search_Lucene_Interface $index)
{
if ($this->_field === null) {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
$query->setBoost($this->getBoost());
$hasInsignificantSubqueries = false;
require_once 'Zend/Search/Lucene.php';
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Term.php';
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_word,
$this->_encoding,
$fieldName);
$rewrittenSubquery = $subquery->rewrite($index);
foreach ($rewrittenSubquery->getQueryTerms() as $term) {
$query->addTerm($term);
}
if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
$hasInsignificantSubqueries = true;
}
}
if (count($query->getTerms()) == 0) {
$this->_matches = array();
if ($hasInsignificantSubqueries) {
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
} else {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
}
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize exact term matching (it corresponds to Keyword fields stored in the index)
// encoding is not used since we expect binary matching
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($this->_word, $this->_field);
if ($index->hasTerm($term)) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$word = iconv($this->_encoding, 'UTF-8', $this->_word);
$wildcardsPattern = '/[*?]/u';
$subPatternsEncoding = 'UTF-8';
} else {
$word = $this->_word;
$wildcardsPattern = '/[*?]/';
$subPatternsEncoding = $this->_encoding;
}
$subPatterns = preg_split($wildcardsPattern, $word, -1, PREG_SPLIT_OFFSET_CAPTURE);
if (count($subPatterns) > 1) {
// Wildcard query is recognized
$pattern = '';
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
foreach ($subPatterns as $id => $subPattern) {
// Append corresponding wildcard character to the pattern before each sub-pattern (except first)
if ($id != 0) {
$pattern .= $word[ $subPattern[1] - 1 ];
}
// Check if each subputtern is a single word in terms of current analyzer
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($subPattern[0], $subPatternsEncoding);
if (count($tokens) > 1) {
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Wildcard search is supported only for non-multiple word terms');
}
foreach ($tokens as $token) {
$pattern .= $token->getTermText();
}
}
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($pattern, $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Wildcard.php';
$query = new Zend_Search_Lucene_Search_Query_Wildcard($term);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
$this->_matches = array();
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's not insignificant or one term query
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
/**
* @todo Process $token->getPositionIncrement() to support stemming, synonyms and other
* analizer design features
*/
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($tokens as $token) {
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, true); // all subterms are required
}
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
/** Skip fields detection. We don't need it, since we expect all fields presented in the HTML body and don't differentiate them */
/** Skip exact term matching recognition, keyword fields highlighting is not supported */
// -------------------------------------
// Recognize wildcard queries
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
if (@preg_match('/\pL/u', 'a') == 1) {
$word = iconv($this->_encoding, 'UTF-8', $this->_word);
$wildcardsPattern = '/[*?]/u';
$subPatternsEncoding = 'UTF-8';
} else {
$word = $this->_word;
$wildcardsPattern = '/[*?]/';
$subPatternsEncoding = $this->_encoding;
}
$subPatterns = preg_split($wildcardsPattern, $word, -1, PREG_SPLIT_OFFSET_CAPTURE);
if (count($subPatterns) > 1) {
// Wildcard query is recognized
$pattern = '';
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
foreach ($subPatterns as $id => $subPattern) {
// Append corresponding wildcard character to the pattern before each sub-pattern (except first)
if ($id != 0) {
$pattern .= $word[ $subPattern[1] - 1 ];
}
// Check if each subputtern is a single word in terms of current analyzer
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($subPattern[0], $subPatternsEncoding);
if (count($tokens) > 1) {
// Do nothing (nothing is highlighted)
return;
}
foreach ($tokens as $token) {
$pattern .= $token->getTermText();
}
}
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($pattern, $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Wildcard.php';
$query = new Zend_Search_Lucene_Search_Query_Wildcard($term);
$query->_highlightMatches($highlighter);
return;
}
// -------------------------------------
// Recognize one-term multi-term and "insignificant" queries
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding);
if (count($tokens) == 0) {
// Do nothing
return;
}
if (count($tokens) == 1) {
$highlighter->highlight($tokens[0]->getTermText());
return;
}
//It's not insignificant or one term query
$words = array();
foreach ($tokens as $token) {
$words[] = $token->getTermText();
}
$highlighter->highlight($words);
}
/**
* Print a query
*
* @return string
*/
public function __toString()
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_field !== null) {
$query = $this->_field . ':';
} else {
$query = '';
}
$query .= $this->_word;
if ($this->getBoost() != 1) {
$query .= '^' . round($this->getBoost(), 4);
}
return $query;
}
}

25
libs/Zend/Search/Lucene/Search/Query/Range.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Range.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Range.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Query_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
/**
* @category Zend
@ -162,10 +159,12 @@ class Zend_Search_Lucene_Search_Query_Range extends Zend_Search_Lucene_Search_Qu
$fields = array($this->_field);
}
require_once 'Zend/Search/Lucene.php';
$maxTerms = Zend_Search_Lucene::getTermsPerQueryLimit();
foreach ($fields as $field) {
$index->resetTermsStream();
require_once 'Zend/Search/Lucene/Index/Term.php';
if ($this->_lowerTerm !== null) {
$lowerTerm = new Zend_Search_Lucene_Index_Term($this->_lowerTerm->text, $field);
@ -220,10 +219,13 @@ class Zend_Search_Lucene_Search_Query_Range extends Zend_Search_Lucene_Search_Qu
}
if (count($this->_matches) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else if (count($this->_matches) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
return new Zend_Search_Lucene_Search_Query_Term(reset($this->_matches));
} else {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$rewrittenQuery = new Zend_Search_Lucene_Search_Query_MultiTerm();
foreach ($this->_matches as $matchedTerm) {
@ -328,19 +330,20 @@ class Zend_Search_Lucene_Search_Query_Range extends Zend_Search_Lucene_Search_Qu
$words = array();
$docBody = $highlighter->getDocument()->getFieldUtf8Value('body');
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($docBody, 'UTF-8');
$lowerTermText = ($this->_lowerTerm !== null)? $this->_lowerTerm->text : null;
$upperTermText = ($this->_upperTerm !== null)? $this->_upperTerm->text : null;
if ($this->_inclusive) {
foreach ($tokens as $token) {
$termText = $token->getTermText();
if (($lowerTermText == null || $lowerTermText <= $termText) &&
($upperTermText == null || $termText <= $upperTermText)) {
$words[] = $termText;
}
}
foreach ($tokens as $token) {
$termText = $token->getTermText();
if (($lowerTermText == null || $lowerTermText <= $termText) &&
($upperTermText == null || $termText <= $upperTermText)) {
$words[] = $termText;
}
}
} else {
foreach ($tokens as $token) {
$termText = $token->getTermText();

15
libs/Zend/Search/Lucene/Search/Query/Term.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Term.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Term.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Weight_Term */
require_once 'Zend/Search/Lucene/Search/Weight/Term.php';
/**
* @category Zend
@ -82,9 +79,11 @@ class Zend_Search_Lucene_Search_Query_Term extends Zend_Search_Lucene_Search_Que
if ($this->_term->field != null) {
return $this;
} else {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
$query->setBoost($this->getBoost());
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($index->getFieldNames(true) as $fieldName) {
$term = new Zend_Search_Lucene_Index_Term($this->_term->text, $fieldName);
@ -105,6 +104,7 @@ class Zend_Search_Lucene_Search_Query_Term extends Zend_Search_Lucene_Search_Que
{
// Check, that index contains specified term
if (!$index->hasTerm($this->_term)) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
}
@ -120,6 +120,7 @@ class Zend_Search_Lucene_Search_Query_Term extends Zend_Search_Lucene_Search_Que
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Search/Weight/Term.php';
$this->_weight = new Zend_Search_Lucene_Search_Weight_Term($this->_term, $this, $reader);
return $this->_weight;
}
@ -198,7 +199,7 @@ class Zend_Search_Lucene_Search_Query_Term extends Zend_Search_Lucene_Search_Que
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
$highlighter->highlight($this->_term->text);
$highlighter->highlight($this->_term->text);
}
/**
@ -210,9 +211,9 @@ class Zend_Search_Lucene_Search_Query_Term extends Zend_Search_Lucene_Search_Que
{
// It's used only for query visualisation, so we don't care about characters escaping
if ($this->_term->field !== null) {
$query = $this->_term->field . ':';
$query = $this->_term->field . ':';
} else {
$query = '';
$query = '';
}
$query .= $this->_term->text;

27
libs/Zend/Search/Lucene/Search/Query/Wildcard.php
View File

@ -17,16 +17,13 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Wildcard.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Wildcard.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Query */
require_once 'Zend/Search/Lucene/Search/Query.php';
/** Zend_Search_Lucene_Search_Query_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
/**
* @category Zend
@ -84,7 +81,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public static function getMinPrefixLength()
{
return self::$_minPrefixLength;
return self::$_minPrefixLength;
}
/**
@ -94,7 +91,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public static function setMinPrefixLength($minPrefixLength)
{
self::$_minPrefixLength = $minPrefixLength;
self::$_minPrefixLength = $minPrefixLength;
}
/**
@ -144,7 +141,8 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
$matchExpression = '/^' . str_replace(array('\\?', '\\*'), array('.', '.*') , preg_quote($this->_pattern->text, '/')) . '$/';
if ($prefixLength < self::$_minPrefixLength) {
throw new Zend_Search_Lucene_Exception('At least ' . self::$_minPrefixLength . ' non-wildcard characters are required at the beginning of pattern.');
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('At least ' . self::$_minPrefixLength . ' non-wildcard characters are required at the beginning of pattern.');
}
/** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */
@ -158,6 +156,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
foreach ($fields as $field) {
$index->resetTermsStream();
require_once 'Zend/Search/Lucene/Index/Term.php';
if ($prefix != '') {
$index->skipTo(new Zend_Search_Lucene_Index_Term($prefix, $field));
@ -168,7 +167,8 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
$this->_matches[] = $index->currentTerm();
if ($maxTerms != 0 && count($this->_matches) > $maxTerms) {
throw new Zend_Search_Lucene_Exception('Terms per query limit is reached.');
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Terms per query limit is reached.');
}
}
@ -182,6 +182,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
$this->_matches[] = $index->currentTerm();
if ($maxTerms != 0 && count($this->_matches) > $maxTerms) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Terms per query limit is reached.');
}
}
@ -194,10 +195,13 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
}
if (count($this->_matches) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
return new Zend_Search_Lucene_Search_Query_Empty();
} else if (count($this->_matches) == 1) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
return new Zend_Search_Lucene_Search_Query_Term(reset($this->_matches));
} else {
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$rewrittenQuery = new Zend_Search_Lucene_Search_Query_MultiTerm();
foreach ($this->_matches as $matchedTerm) {
@ -216,6 +220,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public function optimize(Zend_Search_Lucene_Interface $index)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wildcard query should not be directly used for search. Use $query->rewrite($index)');
}
@ -240,6 +245,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
public function getQueryTerms()
{
if ($this->_matches === null) {
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Search has to be performed first to get matched terms');
}
@ -255,6 +261,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public function createWeight(Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wildcard query should not be directly used for search. Use $query->rewrite($index)');
}
@ -269,6 +276,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public function execute(Zend_Search_Lucene_Interface $reader, $docsFilter = null)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wildcard query should not be directly used for search. Use $query->rewrite($index)');
}
@ -282,6 +290,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public function matchedDocs()
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wildcard query should not be directly used for search. Use $query->rewrite($index)');
}
@ -295,6 +304,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
*/
public function score($docId, Zend_Search_Lucene_Interface $reader)
{
require_once 'Zend/Search/Lucene/Exception.php';
throw new Zend_Search_Lucene_Exception('Wildcard query should not be directly used for search. Use $query->rewrite($index)');
}
@ -315,6 +325,7 @@ class Zend_Search_Lucene_Search_Query_Wildcard extends Zend_Search_Lucene_Search
}
$docBody = $highlighter->getDocument()->getFieldUtf8Value('body');
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($docBody, 'UTF-8');
foreach ($tokens as $token) {
if (preg_match($matchExpression, $token->getTermText()) === 1) {

14
libs/Zend/Search/Lucene/Search/QueryEntry.php
View File

@ -17,21 +17,9 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: QueryEntry.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: QueryEntry.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_QueryEntry_Term */
require_once 'Zend/Search/Lucene/Search/QueryEntry/Term.php';
/** Zend_Search_Lucene_Search_QueryEntry_Phrase */
require_once 'Zend/Search/Lucene/Search/QueryEntry/Phrase.php';
/** Zend_Search_Lucene_Search_QueryEntry_Subquery */
require_once 'Zend/Search/Lucene/Search/QueryEntry/Subquery.php';
/**
* @category Zend
* @package Zend_Search_Lucene

20
libs/Zend/Search/Lucene/Search/QueryEntry/Phrase.php
View File

@ -17,18 +17,12 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Phrase.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Phrase.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_QueryEntry */
require_once 'Zend/Search/Lucene/Search/QueryEntry.php';
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -103,11 +97,13 @@ class Zend_Search_Lucene_Search_QueryEntry_Phrase extends Zend_Search_Lucene_Sea
*/
public function getQuery($encoding)
{
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Phrase($this->_phrase,
$encoding,
($this->_field !== null)?
iconv($encoding, 'UTF-8', $this->_field) :
null);
/** Zend_Search_Lucene_Search_Query_Preprocessing_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Phrase.php';
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Phrase($this->_phrase,
$encoding,
($this->_field !== null)?
iconv($encoding, 'UTF-8', $this->_field) :
null);
if ($this->_proximityQuery) {
$query->setSlop($this->_wordsDistance);

5
libs/Zend/Search/Lucene/Search/QueryEntry/Subquery.php
View File

@ -17,12 +17,9 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Subquery.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Subquery.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_QueryEntry */
require_once 'Zend/Search/Lucene/Search/QueryEntry.php';

36
libs/Zend/Search/Lucene/Search/QueryEntry/Term.php
View File

@ -17,18 +17,12 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Term.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: Term.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_QueryEntry */
require_once 'Zend/Search/Lucene/Search/QueryEntry.php';
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -92,6 +86,8 @@ class Zend_Search_Lucene_Search_QueryEntry_Term extends Zend_Search_Lucene_Searc
if ($parameter !== null) {
$this->_similarity = $parameter;
} else {
/** Zend_Search_Lucene_Search_Query_Fuzzy */
require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php';
$this->_similarity = Zend_Search_Lucene_Search_Query_Fuzzy::DEFAULT_MIN_SIMILARITY;
}
}
@ -105,26 +101,30 @@ class Zend_Search_Lucene_Search_QueryEntry_Term extends Zend_Search_Lucene_Searc
*/
public function getQuery($encoding)
{
if ($this->_fuzzyQuery) {
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_term,
$encoding,
($this->_field !== null)?
if ($this->_fuzzyQuery) {
/** Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php';
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_term,
$encoding,
($this->_field !== null)?
iconv($encoding, 'UTF-8', $this->_field) :
null,
$this->_similarity
);
$this->_similarity
);
$query->setBoost($this->_boost);
return $query;
}
}
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_term,
$encoding,
($this->_field !== null)?
/** Zend_Search_Lucene_Search_Query_Preprocessing_Term */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_term,
$encoding,
($this->_field !== null)?
iconv($encoding, 'UTF-8', $this->_field) :
null
);
$query->setBoost($this->_boost);
$query->setBoost($this->_boost);
return $query;
}
}

3
libs/Zend/Search/Lucene/Search/QueryHit.php
View File

@ -17,7 +17,7 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: QueryHit.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: QueryHit.php 18954 2009-11-12 20:01:33Z alexander $
*/
@ -64,6 +64,7 @@ class Zend_Search_Lucene_Search_QueryHit
public function __construct(Zend_Search_Lucene_Interface $index)
{
require_once 'Zend/Search/Lucene/Proxy.php';
$this->_index = new Zend_Search_Lucene_Proxy($index);
}

63
libs/Zend/Search/Lucene/Search/QueryParser.php
View File

@ -17,50 +17,19 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: QueryParser.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: QueryParser.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_Query_Term */
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
/** Internally used classes */
/** Zend_Search_Lucene_Search_Query_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
/** Zend_Search_Lucene_Analysis_Analyzer */
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
/** Zend_Search_Lucene_Search_Query_Boolean */
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
/** Zend_Search_Lucene_Search_QueryToken */
require_once 'Zend/Search/Lucene/Search/QueryToken.php';
/** Zend_Search_Lucene_Search_Query_Preprocessing_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Phrase.php';
/** Zend_Search_Lucene_Search_Query_Preprocessing_Term */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Term.php';
/** Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy */
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php';
/** Zend_Search_Lucene_Search_Query_Wildcard */
require_once 'Zend/Search/Lucene/Search/Query/Wildcard.php';
/** Zend_Search_Lucene_Search_Query_Range */
require_once 'Zend/Search/Lucene/Search/Query/Range.php';
/** Zend_Search_Lucene_Search_Query_Fuzzy */
require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php';
/** Zend_Search_Lucene_Search_Query_Empty */
require_once 'Zend/Search/Lucene/Search/Query/Empty.php';
/** Zend_Search_Lucene_Search_Query_Insignificant */
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
/** Zend_Search_Lucene_Search_QueryLexer */
require_once 'Zend/Search/Lucene/Search/QueryLexer.php';
/** Zend_Search_Lucene_Search_QueryParserContext */
require_once 'Zend/Search/Lucene/Search/QueryParserContext.php';
/** Zend_Search_Lucene_FSM */
require_once 'Zend/Search/Lucene/FSM.php';
@ -281,7 +250,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
$this->addEntryAction(self::ST_CLOSEDINT_RQ_LAST_TERM, $closedRQLastTermAction);
require_once 'Zend/Search/Lucene/Search/QueryLexer.php';
$this->_lexer = new Zend_Search_Lucene_Search_QueryLexer();
}
@ -380,6 +349,8 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
try {
require_once 'Zend/Search/Lucene/Search/QueryParserContext.php';
self::$_instance->_encoding = ($encoding !== null) ? $encoding : self::$_instance->_defaultEncoding;
self::$_instance->_lastToken = null;
self::$_instance->_context = new Zend_Search_Lucene_Search_QueryParserContext(self::$_instance->_encoding);
@ -388,6 +359,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
// Empty query
if (count(self::$_instance->_tokens) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
@ -416,10 +388,12 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
if (self::$_instance->_suppressQueryParsingExceptions) {
$queryTokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($strQuery, self::$_instance->_encoding);
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
$termsSign = (self::$_instance->_defaultOperator == self::B_AND) ? true /* required term */ :
null /* optional term */;
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($queryTokens as $token) {
$query->addTerm(new Zend_Search_Lucene_Index_Term($token->getTermText()), $termsSign);
}
@ -443,6 +417,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
*/
public function addTermEntry()
{
require_once 'Zend/Search/Lucene/Search/QueryEntry/Term.php';
$entry = new Zend_Search_Lucene_Search_QueryEntry_Term($this->_currentToken->text, $this->_context->getField());
$this->_context->addEntry($entry);
}
@ -452,6 +427,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
*/
public function addPhraseEntry()
{
require_once 'Zend/Search/Lucene/Search/QueryEntry/Phrase.php';
$entry = new Zend_Search_Lucene_Search_QueryEntry_Phrase($this->_currentToken->text, $this->_context->getField());
$this->_context->addEntry($entry);
}
@ -515,6 +491,8 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
*/
public function subqueryStart()
{
require_once 'Zend/Search/Lucene/Search/QueryParserContext.php';
$this->_contextStack[] = $this->_context;
$this->_context = new Zend_Search_Lucene_Search_QueryParserContext($this->_encoding, $this->_context->getField());
}
@ -532,6 +510,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
$query = $this->_context->getQuery();
$this->_context = array_pop($this->_contextStack);
require_once 'Zend/Search/Lucene/Search/QueryEntry/Subquery.php';
$this->_context->addEntry(new Zend_Search_Lucene_Search_QueryEntry_Subquery($query));
}
@ -563,6 +542,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Range query boundary terms must be non-multiple word terms');
} else if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$from = new Zend_Search_Lucene_Index_Term(reset($tokens)->getTermText(), $this->_context->getField());
} else {
$from = null;
@ -573,6 +553,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Range query boundary terms must be non-multiple word terms');
} else if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$to = new Zend_Search_Lucene_Index_Term(reset($tokens)->getTermText(), $this->_context->getField());
} else {
$to = null;
@ -583,7 +564,9 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
throw new Zend_Search_Lucene_Search_QueryParserException('At least one range query boundary term must be non-empty term');
}
require_once 'Zend/Search/Lucene/Search/Query/Range.php';
$rangeQuery = new Zend_Search_Lucene_Search_Query_Range($from, $to, false);
require_once 'Zend/Search/Lucene/Search/QueryEntry/Subquery.php';
$entry = new Zend_Search_Lucene_Search_QueryEntry_Subquery($rangeQuery);
$this->_context->addEntry($entry);
}
@ -608,6 +591,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Range query boundary terms must be non-multiple word terms');
} else if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$from = new Zend_Search_Lucene_Index_Term(reset($tokens)->getTermText(), $this->_context->getField());
} else {
$from = null;
@ -618,6 +602,7 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Range query boundary terms must be non-multiple word terms');
} else if (count($tokens) == 1) {
require_once 'Zend/Search/Lucene/Index/Term.php';
$to = new Zend_Search_Lucene_Index_Term(reset($tokens)->getTermText(), $this->_context->getField());
} else {
$to = null;
@ -628,7 +613,9 @@ class Zend_Search_Lucene_Search_QueryParser extends Zend_Search_Lucene_FSM
throw new Zend_Search_Lucene_Search_QueryParserException('At least one range query boundary term must be non-empty term');
}
require_once 'Zend/Search/Lucene/Search/Query/Range.php';
$rangeQuery = new Zend_Search_Lucene_Search_Query_Range($from, $to, true);
require_once 'Zend/Search/Lucene/Search/QueryEntry/Subquery.php';
$entry = new Zend_Search_Lucene_Search_QueryEntry_Subquery($rangeQuery);
$this->_context->addEntry($entry);
}

31
libs/Zend/Search/Lucene/Search/QueryParserContext.php
View File

@ -17,35 +17,12 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: QueryParserContext.php 16971 2009-07-22 18:05:45Z mikaelkael $
* @version $Id: QueryParserContext.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_FSM */
require_once 'Zend/Search/Lucene/FSM.php';
/** Zend_Search_Lucene_Index_Term */
require_once 'Zend/Search/Lucene/Index/Term.php';
/** Zend_Search_Lucene_Search_QueryToken */
require_once 'Zend/Search/Lucene/Search/QueryToken.php';
/** Zend_Search_Lucene_Search_Query_Term */
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
/** Zend_Search_Lucene_Search_Query_MultiTerm */
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
/** Zend_Search_Lucene_Search_Query_Boolean */
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
/** Zend_Search_Lucene_Search_Query_Phrase */
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
/** Zend_Search_Lucene_Search_BooleanExpressionRecognizer */
require_once 'Zend/Search/Lucene/Search/BooleanExpressionRecognizer.php';
/** Zend_Search_Lucene_Search_QueryEntry */
require_once 'Zend/Search/Lucene/Search/QueryEntry.php';
/**
* @category Zend
@ -277,8 +254,10 @@ class Zend_Search_Lucene_Search_QueryParserContext
*/
public function _signStyleExpressionQuery()
{
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
require_once 'Zend/Search/Lucene/Search/QueryParser.php';
if (Zend_Search_Lucene_Search_QueryParser::getDefaultOperator() == Zend_Search_Lucene_Search_QueryParser::B_AND) {
$defaultSign = true; // required
} else {
@ -314,6 +293,7 @@ class Zend_Search_Lucene_Search_QueryParserContext
* one or more query entries
*/
require_once 'Zend/Search/Lucene/Search/BooleanExpressionRecognizer.php';
$expressionRecognizer = new Zend_Search_Lucene_Search_BooleanExpressionRecognizer();
require_once 'Zend/Search/Lucene/Exception.php';
@ -373,6 +353,7 @@ class Zend_Search_Lucene_Search_QueryParserContext
if (count($conjuction) == 1) {
$subqueries[] = $conjuction[0][0]->getQuery($this->_encoding);
} else {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$subquery = new Zend_Search_Lucene_Search_Query_Boolean();
foreach ($conjuction as $conjuctionEntry) {
@ -384,6 +365,7 @@ class Zend_Search_Lucene_Search_QueryParserContext
}
if (count($subqueries) == 0) {
require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php';
return new Zend_Search_Lucene_Search_Query_Insignificant();
}
@ -392,6 +374,7 @@ class Zend_Search_Lucene_Search_QueryParserContext
}
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
foreach ($subqueries as $subquery) {

7
libs/Zend/Search/Lucene/Search/Similarity.php
View File

@ -17,14 +17,10 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Similarity.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Similarity.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Search_Similarity_Default */
require_once 'Zend/Search/Lucene/Search/Similarity/Default.php';
/**
* @category Zend
* @package Zend_Search_Lucene
@ -326,6 +322,7 @@ abstract class Zend_Search_Lucene_Search_Similarity
public static function getDefault()
{
if (!self::$_defaultImpl instanceof Zend_Search_Lucene_Search_Similarity) {
require_once 'Zend/Search/Lucene/Search/Similarity/Default.php';
self::$_defaultImpl = new Zend_Search_Lucene_Search_Similarity_Default();
}

3
libs/Zend/Search/Lucene/Search/Weight/MultiTerm.php
View File

@ -17,7 +17,7 @@
* @subpackage Search
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: MultiTerm.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: MultiTerm.php 18954 2009-11-12 20:01:33Z alexander $
*/
@ -76,6 +76,7 @@ class Zend_Search_Lucene_Search_Weight_MultiTerm extends Zend_Search_Lucene_Sear
foreach ($query->getTerms() as $id => $term) {
if ($signs === null || $signs[$id] === null || $signs[$id]) {
require_once 'Zend/Search/Lucene/Search/Weight/Term.php';
$this->_weights[$id] = new Zend_Search_Lucene_Search_Weight_Term($term, $query, $reader);
$query->setWeight($id, $this->_weights[$id]);
}

6
libs/Zend/Search/Lucene/Storage/Directory.php
View File

@ -17,7 +17,7 @@
* @subpackage Storage
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Directory.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Directory.php 18951 2009-11-12 16:26:19Z alexander $
*/
@ -64,14 +64,14 @@ abstract class Zend_Search_Lucene_Storage_Directory
/**
* Purge file if it's cached by directory object
*
*
* Method is used to prevent 'too many open files' error
*
* @param string $filename
* @return void
*/
abstract public function purgeFile($filename);
/**
* Returns true if a file with the given $filename exists.
*

7
libs/Zend/Search/Lucene/Storage/Directory/Filesystem.php
View File

@ -17,16 +17,13 @@
* @subpackage Storage
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: Filesystem.php 16541 2009-07-07 06:59:03Z bkarwin $
* @version $Id: Filesystem.php 18954 2009-11-12 20:01:33Z alexander $
*/
/** Zend_Search_Lucene_Storage_Directory */
require_once 'Zend/Search/Lucene/Storage/Directory.php';
/** Zend_Search_Lucene_Storage_File_Filesystem */
require_once 'Zend/Search/Lucene/Storage/File/Filesystem.php';
/**
* FileSystem implementation of Directory abstraction.
@ -183,6 +180,7 @@ class Zend_Search_Lucene_Storage_Directory_Filesystem extends Zend_Search_Lucene
$this->_fileHandlers[$filename]->close();
}
unset($this->_fileHandlers[$filename]);
require_once 'Zend/Search/Lucene/Storage/File/Filesystem.php';
$this->_fileHandlers[$filename] = new Zend_Search_Lucene_Storage_File_Filesystem($this->_dirPath . '/' . $filename, 'w+b');
// Set file permissions, but don't care about any possible failures, since file may be already
@ -347,6 +345,7 @@ class Zend_Search_Lucene_Storage_Directory_Filesystem extends Zend_Search_Lucene
{
$fullFilename = $this->_dirPath . '/' . $filename;
require_once 'Zend/Search/Lucene/Storage/File/Filesystem.php';
if (!$shareHandler) {
return new Zend_Search_Lucene_Storage_File_Filesystem($fullFilename);
}

352
libs/Zend/Search/Lucene/TermStreamsPriorityQueue.php
View File

@ -1,176 +1,176 @@
<?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
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: TermStreamsPriorityQueue.php 16971 2009-07-22 18:05:45Z mikaelkael $
*/
/** Zend_Search_Lucene_Index_TermsStream_Interface */
require_once 'Zend/Search/Lucene/Index/TermsStream/Interface.php';
/** Zend_Search_Lucene_Index_TermsPriorityQueue */
require_once 'Zend/Search/Lucene/Index/TermsPriorityQueue.php';
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_TermStreamsPriorityQueue implements Zend_Search_Lucene_Index_TermsStream_Interface
{
/**
* Array of term streams (Zend_Search_Lucene_Index_TermsStream_Interface objects)
*
* @var array
*/
protected $_termStreams;
/**
* Terms stream queue
*
* @var Zend_Search_Lucene_Index_TermsPriorityQueue
*/
protected $_termsStreamQueue = null;
/**
* Last Term in a terms stream
*
* @var Zend_Search_Lucene_Index_Term
*/
protected $_lastTerm = null;
/**
* Object constructor
*
* @param array $termStreams array of term streams (Zend_Search_Lucene_Index_TermsStream_Interface objects)
*/
public function __construct(array $termStreams)
{
$this->_termStreams = $termStreams;
$this->resetTermsStream();
}
/**
* Reset terms stream.
*/
public function resetTermsStream()
{
$this->_termsStreamQueue = new Zend_Search_Lucene_Index_TermsPriorityQueue();
foreach ($this->_termStreams as $termStream) {
$termStream->resetTermsStream();
// Skip "empty" containers
if ($termStream->currentTerm() !== null) {
$this->_termsStreamQueue->put($termStream);
}
}
$this->nextTerm();
}
/**
* Skip terms stream up to specified term preffix.
*
* Prefix contains fully specified field info and portion of searched term
*
* @param Zend_Search_Lucene_Index_Term $prefix
*/
public function skipTo(Zend_Search_Lucene_Index_Term $prefix)
{
$termStreams = array();
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
$termStreams[] = $termStream;
}
foreach ($termStreams as $termStream) {
$termStream->skipTo($prefix);
if ($termStream->currentTerm() !== null) {
$this->_termsStreamQueue->put($termStream);
}
}
$this->nextTerm();
}
/**
* Scans term streams and returns next term
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function nextTerm()
{
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
if ($this->_termsStreamQueue->top() === null ||
$this->_termsStreamQueue->top()->currentTerm()->key() !=
$termStream->currentTerm()->key()) {
// We got new term
$this->_lastTerm = $termStream->currentTerm();
if ($termStream->nextTerm() !== null) {
// Put segment back into the priority queue
$this->_termsStreamQueue->put($termStream);
}
return $this->_lastTerm;
}
if ($termStream->nextTerm() !== null) {
// Put segment back into the priority queue
$this->_termsStreamQueue->put($termStream);
}
}
// End of stream
$this->_lastTerm = null;
return null;
}
/**
* Returns term in current position
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function currentTerm()
{
return $this->_lastTerm;
}
/**
* Close terms stream
*
* Should be used for resources clean up if stream is not read up to the end
*/
public function closeTermsStream()
{
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
$termStream->closeTermsStream();
}
$this->_termsStreamQueue = null;
$this->_lastTerm = null;
}
}
<?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
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
* @version $Id: TermStreamsPriorityQueue.php 18947 2009-11-12 11:57:17Z alexander $
*/
/** Zend_Search_Lucene_Index_TermsStream_Interface */
require_once 'Zend/Search/Lucene/Index/TermsStream/Interface.php';
/**
* @category Zend
* @package Zend_Search_Lucene
* @subpackage Index
* @copyright Copyright (c) 2005-2009 Zend Technologies USA Inc. (http://www.zend.com)
* @license http://framework.zend.com/license/new-bsd New BSD License
*/
class Zend_Search_Lucene_TermStreamsPriorityQueue implements Zend_Search_Lucene_Index_TermsStream_Interface
{
/**
* Array of term streams (Zend_Search_Lucene_Index_TermsStream_Interface objects)
*
* @var array
*/
protected $_termStreams;
/**
* Terms stream queue
*
* @var Zend_Search_Lucene_Index_TermsPriorityQueue
*/
protected $_termsStreamQueue = null;
/**
* Last Term in a terms stream
*
* @var Zend_Search_Lucene_Index_Term
*/
protected $_lastTerm = null;
/**
* Object constructor
*
* @param array $termStreams array of term streams (Zend_Search_Lucene_Index_TermsStream_Interface objects)
*/
public function __construct(array $termStreams)
{
$this->_termStreams = $termStreams;
$this->resetTermsStream();
}
/**
* Reset terms stream.
*/
public function resetTermsStream()
{
/** Zend_Search_Lucene_Index_TermsPriorityQueue */
require_once 'Zend/Search/Lucene/Index/TermsPriorityQueue.php';
$this->_termsStreamQueue = new Zend_Search_Lucene_Index_TermsPriorityQueue();
foreach ($this->_termStreams as $termStream) {
$termStream->resetTermsStream();
// Skip "empty" containers
if ($termStream->currentTerm() !== null) {
$this->_termsStreamQueue->put($termStream);
}
}
$this->nextTerm();
}
/**
* Skip terms stream up to specified term preffix.
*
* Prefix contains fully specified field info and portion of searched term
*
* @param Zend_Search_Lucene_Index_Term $prefix
*/
public function skipTo(Zend_Search_Lucene_Index_Term $prefix)
{
$termStreams = array();
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
$termStreams[] = $termStream;
}
foreach ($termStreams as $termStream) {
$termStream->skipTo($prefix);
if ($termStream->currentTerm() !== null) {
$this->_termsStreamQueue->put($termStream);
}
}
$this->nextTerm();
}
/**
* Scans term streams and returns next term
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function nextTerm()
{
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
if ($this->_termsStreamQueue->top() === null ||
$this->_termsStreamQueue->top()->currentTerm()->key() !=
$termStream->currentTerm()->key()) {
// We got new term
$this->_lastTerm = $termStream->currentTerm();
if ($termStream->nextTerm() !== null) {
// Put segment back into the priority queue
$this->_termsStreamQueue->put($termStream);
}
return $this->_lastTerm;
}
if ($termStream->nextTerm() !== null) {
// Put segment back into the priority queue
$this->_termsStreamQueue->put($termStream);
}
}
// End of stream
$this->_lastTerm = null;
return null;
}
/**
* Returns term in current position
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function currentTerm()
{
return $this->_lastTerm;
}
/**
* Close terms stream
*
* Should be used for resources clean up if stream is not read up to the end
*/
public function closeTermsStream()
{
while (($termStream = $this->_termsStreamQueue->pop()) !== null) {
$termStream->closeTermsStream();
}
$this->_termsStreamQueue = null;
$this->_lastTerm = null;
}
}