public function modify(eZContentObject $contentObject, &$docList) { $isEvent = $from = $to = false; $attributes = $contentObject->fetchAttributesByIdentifier(array('from_time', 'to_time')); if (empty($attributes)) { return false; } foreach ($attributes as $attribute) { if ($attribute instanceof eZContentObjectAttribute) { if ($attribute->attribute('contentclass_attribute_identifier') == 'from_time' && $attribute->hasContent()) { $from = $attribute->toString(); } if ($attribute->attribute('contentclass_attribute_identifier') == 'to_time' && $attribute->hasContent()) { $to = $attribute->toString(); } } } if ($from && $to) { $isEvent = true; } if ($isEvent) { $duration = $to - $from; $version = $contentObject->currentVersion(); if ($version === false) { return; } $availableLanguages = $version->translationList(false, false); foreach ($availableLanguages as $languageCode) { if ($docList[$languageCode] instanceof eZSolrDoc) { if ($docList[$languageCode]->Doc instanceof DOMDocument) { $xpath = new DomXpath($docList[$languageCode]->Doc); if ($xpath->evaluate('//field[@name="extra_event_duration_s"]')->length == 0) { $docList[$languageCode]->addField('extra_event_duration_s', $duration); $docList[$languageCode]->addField('extra_event_duration_si', intval($duration)); } } elseif (is_array($docList[$languageCode]->Doc) && !isset($docList[$languageCode]->Doc['extra_event_duration_s'])) { $docList[$languageCode]->addField('extra_event_duration_s', $duration); $docList[$languageCode]->addField('extra_event_duration_si', intval($duration)); } } } } }
/** * Parses the XML for the attributes in $classAttributeIdentifiers, and fixes the relations for $object * @param eZContentObject $object * @param array $classAttributeIdentifiers * @return int The number of created relations */ function restoreXmlRelations(eZContentObject $object, array $classAttributeIdentifiers) { $currentVersion = $object->currentVersion(); $langMask = $currentVersion->attribute('language_mask'); $languageList = eZContentLanguage::decodeLanguageMask($langMask, true); $languageList = $languageList['language_list']; // nothing to do if the object isn't translated if (count($languageList) < 2) { return 0; } $attributeArray = $object->fetchAttributesByIdentifier($classAttributeIdentifiers, $currentVersion->attribute('version'), $languageList); $embedRelationsCount = $object->relatedContentObjectCount(false, 0, array('AllRelations' => eZContentObject::RELATION_EMBED)); $linkRelationsCount = $object->relatedContentObjectCount(false, 0, array('AllRelations' => eZContentObject::RELATION_LINK)); $embeddedObjectIdArray = $linkedObjectIdArray = array(); foreach ($attributeArray as $attribute) { $xmlText = eZXMLTextType::rawXMLText($attribute); $dom = new DOMDocument('1.0', 'utf-8'); if (!$dom->loadXML($xmlText)) { continue; } // linked objects $linkedObjectIdArray = array_merge($linkedObjectIdArray, getRelatedObjectList($dom->getElementsByTagName('link'))); // embedded objects $embeddedObjectIdArray = array_merge($embeddedObjectIdArray, getRelatedObjectList($dom->getElementsByTagName('embed')), getRelatedObjectList($dom->getElementsByTagName('embed-inline'))); } $doCommit = false; $restoredRelations = 0; if (!empty($embeddedObjectIdArray)) { $object->appendInputRelationList($embeddedObjectIdArray, eZContentObject::RELATION_EMBED); $restoredRelations += count($embeddedObjectIdArray) - $embedRelationsCount; $doCommit = true; } if (!empty($linkedObjectIdArray)) { $object->appendInputRelationList($linkedObjectIdArray, eZContentObject::RELATION_LINK); $restoredRelations += count($linkedObjectIdArray) - $linkRelationsCount; $doCommit = true; } if ($doCommit) { $object->commitInputRelations($currentVersion->attribute('version')); } return $restoredRelations; }
/** * Adds object $contentObject to the search database. * * @param eZContentObject $contentObject Object to add to search engine * @param bool $commit Whether to commit after adding the object * @return bool True if the operation succeed. */ public function addObject($contentObject, $commit = true) { $contentObjectID = $contentObject->attribute('id'); $currentVersion = $contentObject->currentVersion(); if (!$currentVersion) { $errCurrentVersion = $contentObject->attribute('current_version'); eZDebug::writeError("Failed to fetch \"current version\" ({$errCurrentVersion})" . " of content object (ID: {$contentObjectID})", 'eZSearchEngine'); return false; } $indexArray = array(); $indexArrayOnlyWords = array(); $wordCount = 0; $placement = 0; $previousWord = ''; eZContentObject::recursionProtectionStart(); foreach ($currentVersion->contentObjectAttributes() as $attribute) { $metaData = array(); $classAttribute = $attribute->contentClassAttribute(); if ($classAttribute->attribute("is_searchable") == 1) { // Fetch attribute translations $attributeTranslations = $attribute->fetchAttributeTranslations(); foreach ($attributeTranslations as $translation) { $tmpMetaData = $translation->metaData(); if (!is_array($tmpMetaData)) { $tmpMetaData = array(array('id' => $attribute->attribute('contentclass_attribute_identifier'), 'text' => $tmpMetaData)); } $metaData = array_merge($metaData, $tmpMetaData); } foreach ($metaData as $metaDataPart) { $text = eZSearchEngine::normalizeText(htmlspecialchars($metaDataPart['text'], ENT_NOQUOTES, 'UTF-8'), true); // Split text on whitespace if (is_numeric(trim($text))) { $integerValue = (int) $text; } else { $integerValue = 0; } $wordArray = explode(' ', $text); foreach ($wordArray as $word) { if (trim($word) != "") { // words stored in search index are limited to 150 characters if (strlen($word) > 150) { $word = substr($word, 0, 150); } $indexArray[] = array('Word' => $word, 'ContentClassAttributeID' => $attribute->attribute('contentclassattribute_id'), 'identifier' => $metaDataPart['id'], 'integer_value' => $integerValue); $indexArrayOnlyWords[$word] = 1; $wordCount++; //if we have "www." before word than //treat it as url and add additional entry to the index if (substr(strtolower($word), 0, 4) == 'www.') { $additionalUrlWord = substr($word, 4); $indexArray[] = array('Word' => $additionalUrlWord, 'ContentClassAttributeID' => $attribute->attribute('contentclassattribute_id'), 'identifier' => $metaDataPart['id'], 'integer_value' => $integerValue); $indexArrayOnlyWords[$additionalUrlWord] = 1; $wordCount++; } } } } } } eZContentObject::recursionProtectionEnd(); $wordIDArray = $this->buildWordIDArray(array_keys($indexArrayOnlyWords)); $db = eZDB::instance(); $db->begin(); for ($arrayCount = 0; $arrayCount < $wordCount; $arrayCount += 1000) { $placement = $this->indexWords($contentObject, array_slice($indexArray, $arrayCount, 1000), $wordIDArray, $placement); } $db->commit(); return true; }
/** * Method triggered on publish for xml text datatype * * This method makes sure that links from all translations of an xml text * are registered in the ezurl_object_link table, and thus retained, if * previous versions of an object are removed. * * It also checks for embedded objects in other languages xml, and makes * sure the matching object relations are stored for the publish version. * * @param eZContentObjectAttribute $contentObjectAttribute * @param eZContentObject $object * @param array $publishedNodes * @return boolean */ function onPublish($contentObjectAttribute, $object, $publishedNodes) { $currentVersion = $object->currentVersion(); $langMask = $currentVersion->attribute('language_mask'); // We find all translations present in the current version. We calculate // this from the language mask already present in the fetched version, // so no further round-trip to the DB is required. $languageList = eZContentLanguage::decodeLanguageMask($langMask, true); $languageList = $languageList['language_list']; // We want to have the class attribute identifier of the attribute // containing the current ezxmltext, as we then can use the more efficient // eZContentObject->fetchAttributesByIdentifier() to get the data $identifier = $contentObjectAttribute->attribute('contentclass_attribute_identifier'); $attributeArray = $object->fetchAttributesByIdentifier(array($identifier), $currentVersion->attribute('version'), $languageList); foreach ($attributeArray as $attribute) { $xmlText = eZXMLTextType::rawXMLText($attribute); $dom = new DOMDocument('1.0', 'utf-8'); if (!$dom->loadXML($xmlText)) { continue; } // urls $urlIdArray = array(); foreach ($dom->getElementsByTagName('link') as $link) { // We are looking for external 'http://'-style links, not the internal // object or node links. if ($link->hasAttribute('url_id')) { $urlIdArray[] = $link->getAttribute('url_id'); } } if (count($urlIdArray) > 0) { eZSimplifiedXMLInput::updateUrlObjectLinks($attribute, $urlIdArray); } // linked objects $linkedObjectIdArray = $this->getRelatedObjectList($dom->getElementsByTagName('link')); // embedded objects $embeddedObjectIdArray = array_merge($this->getRelatedObjectList($dom->getElementsByTagName('embed')), $this->getRelatedObjectList($dom->getElementsByTagName('embed-inline'))); if (!empty($embeddedObjectIdArray)) { $object->appendInputRelationList($embeddedObjectIdArray, eZContentObject::RELATION_EMBED); } if (!empty($linkedObjectIdArray)) { $object->appendInputRelationList($linkedObjectIdArray, eZContentObject::RELATION_LINK); } if (!empty($linkedObjectIdArray) || !empty($embeddedObjectIdArray)) { $object->commitInputRelations($currentVersion->attribute('version')); } } }
/** * Adds object $contentObject to the search database. * * @param eZContentObject $contentObject Object to add to search engine * @param bool $commit Whether to commit after adding the object. If set, run optimize() as well every 1000nd time this function is run. * @return bool True if the operation succeed. */ function addObject( $contentObject, $commit = true ) { // Add all translations to the document list $docList = array(); // Check if we need to index this object after all // Exclude if class identifier is in the exclude list for classes $excludeClasses = $this->FindINI->variable( 'IndexExclude', 'ClassIdentifierList' ); if ( $excludeClasses && in_array( $contentObject->attribute( 'class_identifier' ), $excludeClasses ) ) { return true; } // Get global object values $mainNode = $contentObject->attribute( 'main_node' ); if ( !$mainNode ) { eZDebug::writeError( 'Unable to fetch main node for object: ' . $contentObject->attribute( 'id' ), __METHOD__ ); return false; } $mainNodePathArray = $mainNode->attribute( 'path_array' ); $mainNodeID = $mainNode->attribute( 'node_id' ); // initialize array of parent node path ids, needed for multivalued path field and subtree filters $nodePathArray = array(); //included in $nodePathArray //$pathArray = $mainNode->attribute( 'path_array' ); $currentVersion = $contentObject->currentVersion(); // Get object meta attributes. $metaAttributeValues = self::getMetaAttributesForObject( $contentObject ); // Get node attributes. $nodeAttributeValues = array(); foreach ( $contentObject->attribute( 'assigned_nodes' ) as $contentNode ) { $nodeID = $contentNode->attribute( 'node_id' ); foreach ( eZSolr::nodeAttributes() as $attributeName => $fieldType ) { $nodeAttributeValues[$nodeID][] = array( 'name' => $attributeName, 'value' => $contentNode->attribute( $attributeName ), 'fieldType' => $fieldType ); } $nodePathArray[] = $contentNode->attribute( 'path_array' ); } // Check anonymous user access. if ( $this->FindINI->variable( 'SiteSettings', 'IndexPubliclyAvailable' ) == 'enabled' ) { $anonymousUserID = $this->SiteINI->variable( 'UserSettings', 'AnonymousUserID' ); $currentUserID = eZUser::currentUserID(); $user = eZUser::instance( $anonymousUserID ); eZUser::setCurrentlyLoggedInUser( $user, $anonymousUserID ); $anonymousAccess = $contentObject->attribute( 'can_read' ); $user = eZUser::instance( $currentUserID ); eZUser::setCurrentlyLoggedInUser( $user, $currentUserID ); $anonymousAccess = $anonymousAccess ? 'true' : 'false'; } else { $anonymousAccess = 'false'; } // Load index time boost factors if any //$boostMetaFields = $this->FindINI->variable( "IndexBoost", "MetaField" ); $boostClasses = $this->FindINI->variable( 'IndexBoost', 'Class' ); $boostAttributes = $this->FindINI->variable( 'IndexBoost', 'Attribute' ); $boostDatatypes = $this->FindINI->variable( 'IndexBoost', 'Datatype' ); $reverseRelatedScale = $this->FindINI->variable( 'IndexBoost', 'ReverseRelatedScale' ); // Initialise default doc boost $docBoost = 1.0; $contentClassIdentifier = $contentObject->attribute( 'class_identifier' ); // Just test if the boost factor is defined by checking if it has a numeric value if ( isset( $boostClasses[$contentClassIdentifier] ) && is_numeric( $boostClasses[$contentClassIdentifier] ) ) { $docBoost += $boostClasses[$contentClassIdentifier]; } // Google like boosting, using eZ Publish reverseRelatedObjectCount $reverseRelatedObjectCount = $contentObject->reverseRelatedObjectCount(); $docBoost += $reverseRelatedScale * $reverseRelatedObjectCount; // Create the list of available languages for this version : $availableLanguages = $currentVersion->translationList( false, false ); // Loop over each language version and create an eZSolrDoc for it foreach ( $availableLanguages as $languageCode ) { $doc = new eZSolrDoc( $docBoost ); // Set global unique object ID $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'guid' ), $this->guid( $contentObject, $languageCode ) ); // Set installation identifier $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'installation_id' ), self::installationID() ); $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'installation_url' ), $this->FindINI->variable( 'SiteSettings', 'URLProtocol' ) . $this->SiteINI->variable( 'SiteSettings', 'SiteURL' ) . '/' ); // Set Object attributes $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'name' ), $contentObject->name( false, $languageCode ) ); // Also add value to the "sort_name" field as "name" is unsortable, due to Solr limitation (tokenized field) $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'sort_name' ), $contentObject->name( false, $languageCode ) ); $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'anon_access' ), $anonymousAccess ); $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'language_code' ), $languageCode ); $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'available_language_codes' ), $availableLanguages ); if ( $owner = $contentObject->attribute( 'owner' ) ) { // Set owner name $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'owner_name' ), $owner->name( false, $languageCode ) ); // Set owner group ID foreach ( $owner->attribute( 'parent_nodes' ) as $groupID ) { $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'owner_group_id' ), $groupID ); } } // from eZ Publish 4.1 only: object states // so let's check if the content object has it if ( method_exists( $contentObject, 'stateIDArray' ) ) { $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'object_states' ), $contentObject->stateIDArray() ); } // Set content object meta attribute values. foreach ( $metaAttributeValues as $metaInfo ) { $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( $metaInfo['name'] ), ezfSolrDocumentFieldBase::preProcessValue( $metaInfo['value'], $metaInfo['fieldType'] ) ); } // Set content node meta attribute values. foreach ( $nodeAttributeValues as $nodeID => $metaInfoArray ) { foreach( $metaInfoArray as $metaInfo) { $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( $metaInfo['name'] ), ezfSolrDocumentFieldBase::preProcessValue( $metaInfo['value'], $metaInfo['fieldType'] ) ); } } // Main node gets single valued fields for sorting, using a dedicated prefix foreach ( $nodeAttributeValues[$mainNodeID] as $metaInfo ) { $fieldName = 'main_node_' . ezfSolrDocumentFieldBase::generateMetaFieldName( $metaInfo['name'] ); $doc->addField( $fieldName, ezfSolrDocumentFieldBase::preProcessValue( $metaInfo['value'], $metaInfo['fieldType'] ) ); } // Add main url_alias $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'main_url_alias' ), $mainNode->attribute( 'url_alias' ) ); // Add main path_string $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'main_path_string' ), $mainNode->attribute( 'path_string' ) ); // add nodeid of all parent nodes path elements foreach ( $nodePathArray as $pathArray ) { foreach ( $pathArray as $pathNodeID) { $doc->addField( ezfSolrDocumentFieldBase::generateMetaFieldName( 'path' ), $pathNodeID ); } } // Since eZ Fnd 2.3 // cannot call metafield field bame constructor as we are creating multiple fields foreach ( $mainNodePathArray as $key => $pathNodeID ) { $doc->addField( 'meta_main_path_element_' . $key . '_si', $pathNodeID ); } eZContentObject::recursionProtectionStart(); // Loop through all eZContentObjectAttributes and add them to the Solr document. // @since eZ Find 2.3: look for the attribute storage setting $doAttributeStorage = ( ( $this->FindINI->variable( 'IndexOptions', 'EnableSolrAttributeStorage' ) ) === 'true' ) ? true : false; if ( $doAttributeStorage ) { $allAttributeData = array(); } foreach ( $currentVersion->contentObjectAttributes( $languageCode ) as $attribute ) { $metaDataText = ''; $classAttribute = $attribute->contentClassAttribute(); $attributeIdentifier = $classAttribute->attribute( 'identifier' ); $combinedIdentifier = $contentClassIdentifier . '/' . $attributeIdentifier; $boostAttribute = false; if ( isset( $boostAttributes[$attributeIdentifier]) && is_numeric( $boostAttributes[$attributeIdentifier])) { $boostAttribute = $boostAttributes[$attributeIdentifier]; } if ( isset( $boostAttributes[$combinedIdentifier]) && is_numeric( $boostAttributes[$combinedIdentifier])) { $boostAttribute += $boostAttributes[$combinedIdentifier]; } if ( $classAttribute->attribute( 'is_searchable' ) == 1 ) { $documentFieldBase = ezfSolrDocumentFieldBase::getInstance( $attribute ); $this->addFieldBaseToDoc( $documentFieldBase, $doc, $boostAttribute ); } if ( $doAttributeStorage ) { $storageFieldName = ezfSolrStorage::getSolrStorageFieldName( $attributeIdentifier ); $attributeData = ezfSolrStorage::getAttributeData( $attribute ); $allAttributeData['data_map'][$attributeIdentifier] = $attributeData; $doc->addField( $storageFieldName, ezfSolrStorage::serializeData( $attributeData ) ); } } eZContentObject::recursionProtectionEnd(); if ( $doAttributeStorage ) { $doc->addField( 'as_all_bst', ezfSolrStorage::serializeData( $allAttributeData ) ); } $docList[$languageCode] = $doc; } // Since eZFind 2.7: indexhooks $generalPlugins = $this->FindINI->variable( 'IndexPlugins', 'General' ); $classPlugins = $this->FindINI->variable( 'IndexPlugins', 'Class' ); if ( !empty( $generalPlugins ) ) { foreach ( $generalPlugins as $pluginClassString ) { if( !class_exists( $pluginClassString ) ) { eZDebug::writeError( "Unable to find the PHP class '$classname' defined for index time plugins for eZ Find", __METHOD__ ); continue; } $plugin = new $pluginClassString; if ( $plugin instanceof ezfIndexPlugin ) { $plugin->modify( $contentObject, $docList ); } } } if (array_key_exists($contentObject->attribute( 'class_identifier' ), $classPlugins ) ) { $pluginClassString = $classPlugins[$contentObject->attribute( 'class_identifier' )]; if ( class_exists( $pluginClassString ) ) { $plugin = new $pluginClassString; if ($plugin instanceof ezfIndexPlugin) { $plugin->modify( $contentObject, $docList ); } } } $optimize = false; if ( $this->FindINI->variable( 'IndexOptions', 'DisableDirectCommits' ) === 'true' ) { $commit = false; } $commitWithin = 0; if ( $this->FindINI->variable( 'IndexOptions', 'CommitWithin' ) > 0 ) { $commitWithin = $this->FindINI->variable( 'IndexOptions', 'CommitWithin' ); } if ( $commit && ( $this->FindINI->variable( 'IndexOptions', 'OptimizeOnCommit' ) === 'enabled' ) ) { $optimize = true; } if ( $this->UseMultiLanguageCores === true) { $result = true; foreach ( $availableLanguages as $languageCode ) { $languageResult = $this->SolrLanguageShards[$languageCode]->addDocs( array( $docList[$languageCode] ), $commit, $optimize, $commitWithin ); if ( !$languageResult ) { $result = false; } } return $result; } else { return $this->Solr->addDocs( $docList, $commit, $optimize, $commitWithin ); } }