Report similarity on property labels

DefaultSettings.php

@@ -1394,4 +1394,21 @@
 	'smwgEditProtectionRight' => false,
 	##
 
+	##
+	# Similarity lookup exemption property
+	#
+	# The listed property is used to exclude a property from the similarity
+	# lookup in case the comparing property contains an annotation value with the
+	# exemption property.
+	#
+	# For example, the property `Governance level` may define
+	# [[owl:differentFrom::Governance level of]] which would result in a suppressed
+	# similarity lookup for both `Governance level` and `Governance level of`
+	# property when compared to each other.
+	#
+	# @since 2.5
+	##
+	'smwgSimilarityLookupExemptionProperty' => 'owl:differentFrom',
+	##
+
 );

i18n/en.json

@@ -217,6 +217,7 @@
 	"smw-ask-query-search-info": "The query <code><nowiki>$1</nowiki></code> was answered by the {{PLURAL:$3|1=<code>$2</code> (from cache)|<code>$2</code> (from cache)|<code>$2</code>}} in $4 {{PLURAL:$4|second|seconds}}.",
 	"searchbyproperty": "Search by property",
 	"processingerrorlist": "Processing error list",
+	"propertylabelsimilarity": "Property label similarity report",
 	"smw-processingerrorlist-intro": "The following list provides an overview about the processing errors that appeared in connection with [https://www.semantic-mediawiki.org/ Semantic MediaWiki]. It is recommended to monitor this list on a regular basis and correct invalid value annotations.",
 	"smw_sbv_docu": "Search for all pages that have a given property and value.",
 	"smw_sbv_novalue": "Enter a valid value for the property, or view all property values for \"$1\".",
@@ -303,6 +304,12 @@
 	"smw-admin-supplementary-operational-statistics-intro": "<u>$1</u> displays an extended set of statistics",
 	"smw-admin-supplementary-idlookup-title": "Object ID lookup and disposal",
 	"smw-admin-supplementary-idlookup-intro": "<u>$1</u> contains functions to lookup and dispose individual ID's",
+	"smw-property-label-similarity-title": "Property label similarity report",
+	"smw-property-label-similarity-intro": "<u>$1</u> calculates similarties for existing property labels",
+	"smw-property-label-similarity-threshold": "Threshold:",
+	"smw-property-label-similarity-type": "Display Type ID",
+	"smw-property-label-similarity-noresult": "No results were found for the selected options.",
+	"smw-property-label-similarity-docu": "Compares and reports the [https://www.semantic-mediawiki.org/wiki/Property_similarity syntactic similarity] (not the semantic similarity) between two property labels which may help filter misspelled or equivalent properties that represent the same concept (see [[Special:Properties]] to clarify concept and usage of reported properties). The threshold can be adjusted to widen or narrow the similarity distance and <code>[[Property:$1|$1]]</code> is set to be used to exempt properties from the process.",
 	"smw-admin-operational-statistics": "This page contains operational statistics collected in or from Semantic MediaWiki related functions. An extended list of wiki specific statistics can be found [[Special:Statistics|<b>here</b>]].",
 	"smw_adminlinks_datastructure": "Data structure",
 	"smw_adminlinks_displayingdata": "Data display",

i18n/extra/SemanticMediaWiki.alias.php

@@ -22,6 +22,7 @@
 	'SMWAdmin' => array( 'SemanticMediaWiki', 'SMWAdmin' ),
 	'SearchByProperty' => array( 'SearchByProperty' ),
 	'ProcessingErrorList' => array( 'ProcessingErrorList' ),
+	'PropertyLabelSimilarity' => array( 'PropertyLabelSimilarity' ),
 	'SemanticStatistics' => array( 'SemanticStatistics' ),
 	'Types' => array( 'Types' ),
 	'URIResolver' => array( 'URIResolver' ),

includes/Settings.php

@@ -162,6 +162,7 @@ public static function newFromGlobals() {
 			'smwgQueryResultNonEmbeddedCacheLifetime' => $GLOBALS['smwgQueryResultNonEmbeddedCacheLifetime'],
 			'smwgQueryResultCacheRefreshOnPurge' => $GLOBALS['smwgQueryResultCacheRefreshOnPurge'],
 			'smwgEditProtectionRight' => $GLOBALS['smwgEditProtectionRight'],
+			'smwgSimilarityLookupExemptionProperty' => $GLOBALS['smwgSimilarityLookupExemptionProperty']
 		);
 
 		\Hooks::run( 'SMW::Config::BeforeCompletion', array( &$configuration ) );

includes/Setup.php

@@ -287,6 +287,10 @@ private function registerSpecialPages() {
 				'page' => 'SMW\MediaWiki\Specials\SpecialProcessingErrorList',
 				'group' => 'smw_group'
 			),
+			'PropertyLabelSimilarity' => array(
+				'page' => 'SMW\MediaWiki\Specials\SpecialPropertyLabelSimilarity',
+				'group' => 'smw_group'
+			),
 			'SMWAdmin' => array(
 				'page' => 'SMW\MediaWiki\Specials\SpecialAdmin',
 				'group' => 'smw_group'

src/MediaWiki/Renderer/HtmlFormRenderer.php

@@ -380,7 +380,7 @@ public function addOptionSelectList( $label, $inputName, $inputValue, $options,
 	 *
 	 * @return HtmlFormRenderer
 	 */
-	public function addCheckbox( $label, $inputName, $inputValue, $isChecked = false, $id = null ) {
+	public function addCheckbox( $label, $inputName, $inputValue, $isChecked = false, $id = null, $attributes = array() ) {
 
 		if ( $id === null ) {
 			$id = $inputName;
@@ -396,10 +396,11 @@ public function addCheckbox( $label, $inputName, $inputValue, $isChecked = false
 			array(
 				'id' => $id,
 				'class' => $this->defaultPrefix . '-checkbox',
-				'value' => $inputValue ) + ( $isChecked ? array( 'checked' => 'checked' ) : array() )
+				'value' => $inputValue
+			) + ( $isChecked ? array( 'checked' => 'checked' ) : array() )
 		);
 
-		$this->content[] = $html;
+		$this->content[] = Html::rawElement( 'span', $attributes, $html );
 		return $this;
 	}
 
@@ -409,21 +410,26 @@ public function addCheckbox( $label, $inputName, $inputValue, $isChecked = false
 	 * @note Encapsulate as closure to ensure that the build contains all query
 	 * parameters that are necessary to build the paging links
 	 *
-	 * @param integer $limit,
-	 * @param integer $offset,
-	 * @param integer $count,
+	 * @param integer $limit
+	 * @param integer $offset
+	 * @param integer $count
+	 * @param integer|null $messageCount
 	 *
 	 * @return HtmlFormRenderer
 	 */
-	public function addPaging( $limit, $offset, $count ) {
+	public function addPaging( $limit, $offset, $count, $messageCount = null ) {
 
 		$title = $this->title;
 
-		$this->content[] = function( $instance ) use ( $title, $limit, $offset, $count ) {
+		$this->content[] = function( $instance ) use ( $title, $limit, $offset, $count, $messageCount ) {
+
+			if ( $messageCount === null ) {
+				$messageCount = ( $count > $limit ? $count - 1 : $count );
+			}
 
 			$resultCount = $instance->getMessageBuilder()
 				->getMessage( 'showingresults' )
-				->numParams( ( $count > $limit ? $count - 1 : $count ), $offset + 1 )
+				->numParams( $messageCount, $offset + 1 )
 				->parse();
 
 			$paging = $instance->getMessageBuilder()->prevNextToText(

src/MediaWiki/Specials/PropertyLabelSimilarity/ContentsBuilder.php

@@ -0,0 +1,130 @@
+<?php
+
+namespace SMW\MediaWiki\Specials\PropertyLabelSimilarity;
+
+use SMW\MediaWiki\Renderer\HtmlFormRenderer;
+use SMW\Message;
+use SMW\RequestOptions;
+use SMW\SQLStore\Lookup\PropertyLabelSimilarityLookup;
+use Html;
+
+/**
+ * @license GNU GPL v2+
+ * @since   2.5
+ *
+ * @author mwjames
+ */
+class ContentsBuilder {
+
+	/**
+	 * @var PropertyLabelSimilarityLookup
+	 */
+	private $propertyLabelSimilarityLookup;
+
+	/**
+	 * @var HtmlFormRenderer
+	 */
+	private $htmlFormRenderer;
+
+	/**
+	 * @since 2.5
+	 *
+	 * @param PropertyLabelSimilarityLookup $propertyLabelSimilarityLookup
+	 * @param HtmlFormRenderer $htmlFormRenderer
+	 */
+	public function __construct( PropertyLabelSimilarityLookup $propertyLabelSimilarityLookup, HtmlFormRenderer $htmlFormRenderer ) {
+		$this->propertyLabelSimilarityLookup = $propertyLabelSimilarityLookup;
+		$this->htmlFormRenderer = $htmlFormRenderer;
+	}
+
+	/**
+	 * @since 2.5
+	 *
+	 * @param RequestOptions $requestOption
+	 */
+	public function getHtml( RequestOptions $requestOptions ) {
+
+		$threshold = 90;
+		$type = '';
+
+		foreach ( $requestOptions->getExtraConditions() as $extraCondition ) {
+			if ( isset( $extraCondition['type'] ) ) {
+				$type = $extraCondition['type'];
+			}
+
+			if ( isset( $extraCondition['threshold'] ) ) {
+				$threshold = $extraCondition['threshold'];
+			}
+		}
+
+		$this->propertyLabelSimilarityLookup->setThreshold(
+			$threshold
+		);
+
+		$result = $this->propertyLabelSimilarityLookup->compareAndFindLabels(
+			$requestOptions
+		);
+
+		$count = $this->propertyLabelSimilarityLookup->getLookupCount();
+
+		$html = $this->getForm(
+			$requestOptions->getLimit(),
+			$requestOptions->getOffset(),
+			$count,
+			count( $result ),
+			$threshold,
+			$type
+		);
+
+		if ( $result !== array() ) {
+			 $html .= '<pre>' . json_encode( $result, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE ) . '</pre>';
+		} else {
+			 $html .= Message::get( 'smw-property-label-similarity-noresult' );
+		}
+
+		return $html;
+	}
+
+	private function getForm( $limit, $offset, $count, $resultCount, $threshold, $type ) {
+
+		$exemptionProperty = $this->propertyLabelSimilarityLookup->getExemptionProperty();
+
+		$html = Message::get( array( 'smw-property-label-similarity-docu', $exemptionProperty ), Message::PARSE );
+
+		$html .= $this->htmlFormRenderer
+			->setName( 'smw-property-label-similarity-title' )
+			->setMethod( 'get' )
+			->withFieldset()
+			->addPaging(
+				$limit,
+				$offset,
+				$count,
+				$resultCount )
+			->addHiddenField( 'limit', $limit )
+			->addHiddenField( 'offset', $offset )
+			->addInputField(
+				 Message::get( 'smw-property-label-similarity-threshold' ),
+				'threshold',
+				$threshold,
+				'',
+				5
+			)
+			->addNonBreakingSpace()
+			->addCheckbox(
+				 Message::get( 'smw-property-label-similarity-type', Message::ESCAPED ),
+				'type',
+				'yes',
+				$type === 'yes',
+				null,
+				array(
+					'style' => 'float:right'
+				)
+			)
+			->addQueryParameter( 'type', $type )
+			->addSubmitButton( Message::get( 'allpagessubmit' ) )
+			->getForm();
+
+		return Html::rawElement( 'div', array( 'class' => 'plainlinks'), $html ) . Html::element( 'p', array(), '' );
+	}
+
+}

src/MediaWiki/Specials/SpecialAdmin.php

@@ -77,7 +77,7 @@ public function execute( $query ) {
 			$idActionHandler,
 			$supportWidget
 		) = $this->getHandlers(
-			$applicationFactory->getStore(),
+			$applicationFactory->getStore( '\SMW\SQLStore\SQLStore' ),
 			$applicationFactory->getSettings(),
 			$applicationFactory->newMwCollaboratorFactory()
 		);

src/MediaWiki/Specials/SpecialPropertyLabelSimilarity.php

@@ -0,0 +1,87 @@
+<?php
+
+namespace SMW\MediaWiki\Specials;
+
+use SMW\ApplicationFactory;
+use SpecialPage;
+use SMW\Message;
+use SMW\SQLStore\Lookup\PropertyLabelSimilarityLookup;
+use SMW\MediaWiki\Specials\PropertyLabelSimilarity\ContentsBuilder;
+
+/**
+ * @license GNU GPL v2+
+ * @since 2.5
+ *
+ * @author mwjames
+ */
+class SpecialPropertyLabelSimilarity extends SpecialPage {
+
+	/**
+	 * @codeCoverageIgnore
+	 */
+	public function __construct() {
+		parent::__construct( 'PropertyLabelSimilarity' );
+	}
+
+	/**
+	 * @see SpecialPage::execute
+	 */
+	public function execute( $query ) {
+
+		$this->setHeaders();
+		$output = $this->getOutput();
+		$webRequest = $this->getRequest();
+
+		$store = ApplicationFactory::getInstance()->getStore( '\SMW\SQLStore\SQLStore' );
+
+		$propertyLabelSimilarityLookup = new PropertyLabelSimilarityLookup(
+			$store
+		);
+
+		$propertyLabelSimilarityLookup->setExemptionProperty(
+			ApplicationFactory::getInstance()->getSettings()->get( 'smwgSimilarityLookupExemptionProperty' )
+		);
+
+		$htmlFormRenderer = ApplicationFactory::getInstance()->newMwCollaboratorFactory()->newHtmlFormRenderer(
+			$this->getContext()->getTitle(),
+			$this->getLanguage()
+		);
+
+		$contentsBuilder = new ContentsBuilder(
+			$propertyLabelSimilarityLookup,
+			$htmlFormRenderer
+		);
+
+		$threshold = (int)$webRequest->getText( 'threshold', 90 );
+		$type = $webRequest->getText( 'type', false );
+
+		$offset = (int)$webRequest->getText( 'offset', 0 );
+		$limit = (int)$webRequest->getText( 'limit', 50 );
+
+		$requestOptions = ApplicationFactory::getInstance()->getQueryFactory()->newRequestOptions();
+		$requestOptions->setLimit( $limit );
+		$requestOptions->setOffset( $offset );
+
+		$requestOptions->addExtraCondition(
+			array(
+				'type' => $type,
+				'threshold' => $threshold
+			)
+		);
+
+		$output->addHtml(
+			$contentsBuilder->getHtml( $requestOptions )
+		);
+
+		return true;
+	}
+
+
+	/**
+	 * @see SpecialPage::getGroupName
+	 */
+	protected function getGroupName() {
+		return 'smw_group';
+	}
+
+}