/**
* This takes an Omeka_Record instance and returns a populated
* Apache_Solr_Document.
*
* @param Omeka_Record $item The record to index.
*
* @return Apache_Solr_Document
* @author Eric Rochester <*****@*****.**>
**/
public static function itemToDocument($item)
{
$fields = get_db()->getTable('SolrSearchField');
$doc = new Apache_Solr_Document();
$doc->setField('id', "Item_{$item->id}");
$doc->setField('resulttype', 'Item');
$doc->setField('model', 'Item');
$doc->setField('modelid', $item->id);
// Title:
$title = metadata($item, array('Dublin Core', 'Title'));
$doc->setField('title', $title);
// Elements:
self::indexItem($fields, $item, $doc);
// Tags:
foreach ($item->getTags() as $tag) {
$doc->setMultiValue('tag', $tag->name);
}
// Collection:
if ($collection = $item->getCollection()) {
$doc->collection = metadata($collection, array('Dublin Core', 'Title'));
}
// Item type:
if ($itemType = $item->getItemType()) {
$doc->itemtype = $itemType->name;
}
$doc->featured = (bool) $item->featured;
// File metadata
foreach ($item->getFiles() as $file) {
self::indexItem($fields, $file, $doc);
}
return $doc;
}
protected function getMockDocument($id)
{
$document = new Apache_Solr_Document();
$document->setField('id', $id);
$document->setField('title', "Item {$id}");
return $document;
}
/**
* Method for adding an object from the database into the index.
*
* @param DataObject
* @param string
* @param array
*/
protected function _addAs($object, $base, $options)
{
$includeSubs = $options['include_children'];
$doc = new Apache_Solr_Document();
// Always present fields
$doc->setField('_documentid', $this->getDocumentID($object, $base, $includeSubs));
$doc->setField('ID', $object->ID);
$doc->setField('ClassName', $object->ClassName);
foreach (SearchIntrospection::hierarchy(get_class($object), false) as $class) {
$doc->addField('ClassHierarchy', $class);
}
// Add the user-specified fields
foreach ($this->getFieldsIterator() as $name => $field) {
if ($field['base'] == $base) {
$this->_addField($doc, $object, $field);
}
}
// CUSTOM Duplicate index combined fields ("Title" rather than
// "SiteTree_Title").
//
// This allows us to sort on these fields without deeper architectural
// changes to the fulltextsearch module. Note: We can't use <copyField>
// for this purpose because it only writes into multiValue=true
// fields, and those can't be (reliably) sorted on.
$this->_addField($doc, $object, $this->getCustomPropertyFieldData('Title', $object));
$this->_addField($doc, $object, $this->getCustomPropertyFieldData('LastEdited', $object, 'SSDatetime'));
$this->getService()->addDocument($doc);
return $doc;
}
/**
* @test
*/
public function transformsUnixTimestampToIsoDateOnSingleValuedField()
{
$this->documentMock->setField('dateField', '1262343600');
// 2010-01-01 12:00
$configuration = array('dateField' => 'timestampToIsoDate');
$this->service->processDocument($this->documentMock, $configuration);
$value = $this->documentMock->getField('dateField');
$this->assertEquals($value['value'], '2010-01-01T12:00:00Z', 'field was not processed with timestampToIsoDate');
}
/**
* modifies a document according to the given configuration
*
* @param Apache_Solr_Document $document
* @param array $processingConfiguration
*/
public function processDocument(Apache_Solr_Document $document, array $processingConfiguration)
{
foreach ($processingConfiguration as $fieldName => $instruction) {
$fieldInformation = $document->getField($fieldName);
$isSingleValueField = FALSE;
if ($fieldInformation !== FALSE) {
$fieldValue = $fieldInformation['value'];
if (!is_array($fieldValue)) {
// turn single value field into multi value field
$fieldValue = array($fieldValue);
$isSingleValueField = TRUE;
}
switch ($instruction) {
case 'timestampToIsoDate':
$processor = t3lib_div::makeInstance('tx_solr_fieldprocessor_TimestampToIsoDate');
$fieldValue = $processor->process($fieldValue);
break;
case 'uppercase':
$fieldValue = array_map('strtoupper', $fieldValue);
break;
}
if ($isSingleValueField) {
// turn multi value field back into single value field
$fieldValue = $fieldValue[0];
}
$document->setField($fieldName, $fieldValue);
}
}
}
/**
* Adds fields to the document as defined in $indexingConfiguration
*
* @param \Apache_Solr_Document $document base document to add fields to
* @param array $indexingConfiguration Indexing configuration / mapping
* @param array $data Record data
* @return \Apache_Solr_Document Modified document with added fields
*/
protected function addDocumentFieldsFromTyposcript(\Apache_Solr_Document $document, array $indexingConfiguration, array $data)
{
// mapping of record fields => solr document fields, resolving cObj
foreach ($indexingConfiguration as $solrFieldName => $recordFieldName) {
if (is_array($recordFieldName)) {
// configuration for a content object, skipping
continue;
}
if (!self::isAllowedToOverrideField($solrFieldName)) {
throw new InvalidFieldNameException('Must not overwrite field .' . $solrFieldName, 1435441863);
}
$fieldValue = $this->resolveFieldValue($indexingConfiguration, $solrFieldName, $data);
if (is_array($fieldValue)) {
// multi value
foreach ($fieldValue as $multiValue) {
$document->addField($solrFieldName, $multiValue);
}
} else {
if ($fieldValue !== '' && $fieldValue !== null) {
$document->setField($solrFieldName, $fieldValue);
}
}
}
return $document;
}
/**
* Builds an Apache_Solr_Document to pass to Apache_Solr_Service
**/
public function buildDocument(array $document)
{
$doc = new Apache_Solr_Document();
foreach ($document as $fieldName => $field) {
$value = $field['value'];
// Apache_Solr_Document always expect arrays
if (!is_array($value)) {
$value = array($value);
}
if (isset($field['boost'])) {
$doc->setField($fieldName, $value, $field['boost']);
} else {
$doc->setField($fieldName, $value);
}
}
return $doc;
}
public function testAddFieldWithBoostMultipliesWithAPreexistingBoost()
{
$field = 'field';
$boost = 0.5;
// set a field with a boost
$this->_fixture->setField($field, 'value1', $boost);
// now add another value with the same boost
$this->_fixture->addField($field, 'value2', $boost);
// new boost should be $boost * $boost
$this->assertEquals($boost * $boost, $this->_fixture->getFieldBoost($field));
}
/**
* Build a Solr document for a specific file
*
* @param integer $storeId Store ID the file belongs to/where it is linked on a page
* @param Asm_Solr_Model_Indexqueue_File $file The file to index
* @return Apache_Solr_Document
*/
protected function buildFileDocument($storeId, Asm_Solr_Model_Indexqueue_File $file)
{
$helper = Mage::helper('solr');
$baseUrl = Mage::getBaseUrl(Mage_Core_Model_Store::URL_TYPE_WEB);
$host = parse_url($baseUrl, PHP_URL_HOST);
$document = new Apache_Solr_Document();
$document->setField('appKey', 'Asm_Solr');
$document->setField('type', 'solr/indexqueue_file');
$document->setField('id', $helper->getFileDocumentId($file->getId()));
$document->setField('site', $host);
$document->setField('siteHash', $helper->getSiteHashForDomain($host));
$document->setField('storeId', $storeId);
$document->setField('changed', $helper->dateToIso($file->getFileLastChangedTime()));
$document->setField('productId', 0);
$document->setField('sku', 'solr/indexqueue_file');
$document->setField('title', $file->getName());
$document->setField('content', $file->getContent());
$document->setField('url', $file->getUrl());
return $document;
}
/**
* Adds fields to the document as defined in $indexingConfiguration
*
* @param Apache_Solr_Document $document base document to add fields to
* @param array $indexingConfiguration Indexing configuration / mapping
* @param array $data Record data
* @return Apache_Solr_Document Modified document with added fields
*/
protected function addDocumentFieldsFromTyposcript(Apache_Solr_Document $document, array $indexingConfiguration, array $data)
{
// mapping of record fields => solr document fields, resolving cObj
foreach ($indexingConfiguration as $solrFieldName => $recordFieldName) {
if (is_array($recordFieldName)) {
// configuration for a content object, skipping
continue;
}
$fieldValue = $this->resolveFieldValue($indexingConfiguration, $solrFieldName, $data);
if (is_array($fieldValue)) {
// multi value
foreach ($fieldValue as $multiValue) {
$document->addField($solrFieldName, $multiValue);
}
} else {
$document->setField($solrFieldName, $fieldValue);
}
}
return $document;
}
public function solrStore($contact) {
$doc = new Apache_Solr_Document();
$doc->setField('id', $contact->id);
$doc->setField('timecreate', $contact->timecreate->format('Y-m-d\TH:i:s\Z'));
$doc->setField('timeupdate', $contact->timeupdate->format('Y-m-d\TH:i:s\Z'));
$doc->setField('usercreate', $contact->usercreate);
$doc->setField('userupdate', $contact->userupdate);
$doc->setField('datasource', $contact->datasource_id);
$doc->setField('domain', $GLOBALS['obm']['domain_id']);
$doc->setField('in', $contact->addressbook);
$doc->setField('addressbookId', $contact->addressbook_id);
$doc->setField('company', $contact->company);
$doc->setField('companyId', $contact->company_id);
$doc->setField('commonname', $contact->commonname);
$doc->setField('lastname', $contact->lastname);
$doc->setField('firstname', $contact->firstname);
$doc->setField('middlename', $contact->mname);
$doc->setField('sortable', $contact->lastname." ".$contact->firstname);
$doc->setField('suffix', $contact->suffix);
$doc->setField('aka', $contact->aka);
$doc->setField('kind', $contact->kind);
//$doc->setField('kind', $db->f('kind_header'));
$doc->setField('manager', $contact->manager);
$doc->setField('assistant', $contact->assistant);
$doc->setField('spouse', $contact->spouse);
$doc->setField('birthdayId', $contact->birthday_event);
$doc->setField('anniversaryId', $contact->anniversary_event);
if($contact->birthday) $doc->setField('birthday', $contact->birthday->format('Y-m-d\TH:i:s\Z'));
if($contact->anniversary) $doc->setField('anniversary', $contact->anniversary->format('Y-m-d\TH:i:s\Z'));
$doc->setField('category', $contact->category);
foreach($contact->categories as $category) {
foreach($category as $c) {
$doc->setMultiValue('categoryId', $c['id']);
}
}
$doc->setField('service', $contact->service);
$doc->setField('function', $contact->function);
$doc->setField('title', $contact->title);
if ($contact->archive) {
$doc->setField('is', 'archive');
}
if ($contact->collected) {
$doc->setField('is', 'collected');
}
if ($contact->mailok) {
$doc->setField('is', 'mailing');
}
if ($contact->newsletter) {
$doc->setField('is', 'newsletter');
}
if($contact->date) $doc->setField('date', $contact->date->format('Y-m-d\TH:i:s\Z'));
$doc->setField('comment', $contact->comment);
$doc->setField('comment2', $contact->comment2);
$doc->setField('comment3', $contact->comment3);
$doc->setField('from', $contact->origin);
foreach($contact->email as $email) {
$doc->setMultiValue('email', $email['address']);
}
foreach($contact->phone as $phone) {
$doc->setMultiValue('phone', $phone['number']);
}
foreach($contact->im as $im) {
$doc->setMultiValue('jabber', $im['address']);
}
foreach($contact->address as $address) {
$doc->setMultiValue('street', $address['street']);
$doc->setMultiValue('zipcode', $address['zipcode']);
$doc->setMultiValue('expresspostal', $address['expresspostal']);
$doc->setMultiValue('town', $address['town']);
$doc->setMultiValue('country', $address['country']);
}
if($contact->hasACalendarUrl()){
$doc->setField('hasACalendar', "true");
}
else {
$doc->setField('hasACalendar', "false");
}
OBM_IndexingService::store('contact', array($doc));
}
/**
* modifies a document according to the given configuration
*
* @param \Apache_Solr_Document $document
* @param array $processingConfiguration
*/
public function processDocument(\Apache_Solr_Document $document, array $processingConfiguration)
{
foreach ($processingConfiguration as $fieldName => $instruction) {
$fieldInformation = $document->getField($fieldName);
$isSingleValueField = false;
if ($fieldInformation !== false) {
$fieldValue = $fieldInformation['value'];
if (!is_array($fieldValue)) {
// turn single value field into multi value field
$fieldValue = array($fieldValue);
$isSingleValueField = true;
}
switch ($instruction) {
case 'timestampToUtcIsoDate':
$processor = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\TimestampToUtcIsoDate');
$fieldValue = $processor->process($fieldValue);
break;
case 'timestampToIsoDate':
$processor = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\TimestampToIsoDate');
$fieldValue = $processor->process($fieldValue);
break;
case 'pathToHierarchy':
$processor = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\PathToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'pageUidToHierarchy':
$processor = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\PageUidToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'categoryUidToHierarchy':
$processor = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\CategoryUidToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'uppercase':
$fieldValue = array_map('strtoupper', $fieldValue);
break;
}
if ($isSingleValueField) {
// turn multi value field back into single value field
$fieldValue = $fieldValue[0];
}
$document->setField($fieldName, $fieldValue);
}
}
}
/**
* Build a Solr document for a given page
*
* @param integer $storeId Store ID
* @param Mage_Cms_Model_Page $page Page instance
* @return Apache_Solr_Document
*/
protected function buildPageDocument($storeId, $page)
{
$helper = Mage::helper('solr');
$baseUrl = Mage::getBaseUrl(Mage_Core_Model_Store::URL_TYPE_WEB);
$host = parse_url($baseUrl, PHP_URL_HOST);
$document = new Apache_Solr_Document();
$document->setField('appKey', 'Asm_Solr');
$document->setField('type', 'cms/page');
$document->setField('id', $helper->getPageDocumentId($page->getId()));
$document->setField('site', $host);
$document->setField('siteHash', $helper->getSiteHashForDomain($host));
$document->setField('storeId', $storeId);
$document->setField('created', $helper->dateToIso($page->getCreationTime()));
$document->setField('changed', $helper->dateToIso($page->getUpdateTime()));
$document->setField('sku', 'cms/page');
$document->setField('productId', 0);
$document->setField('pageId', $page->getId());
$document->setField('title', $page->getTitle());
$document->setField('content', Mage::helper('solr/contentExtractor')->getIndexableContent($page->getContent()));
$document->setField('keywords', $helper->trimExplode(',', $page->getMetaKeywords(), true));
$document->setField('url', Mage::helper('cms/page')->getPageUrl($page->getId()));
return $document;
}
/**
* Processes a physical unit for the Solr index
*
* @access protected
*
* @param tx_dlf_document &$doc: The METS document
* @param integer $page: The page number
* @param array $physicalUnit: Array of the physical unit to process
*
* @return integer 0 on success or 1 on failure
*/
protected static function processPhysical(tx_dlf_document &$doc, $page, array $physicalUnit)
{
$errors = 0;
// Read extension configuration.
$extConf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]);
if (!empty($physicalUnit['files'][$extConf['fileGrpFulltext']])) {
$file = $doc->getFileLocation($physicalUnit['files'][$extConf['fileGrpFulltext']]);
// Load XML file.
if (\TYPO3\CMS\Core\Utility\GeneralUtility::isValidUrl($file) || version_compare(phpversion(), '5.3.3', '<')) {
// Set user-agent to identify self when fetching XML data.
if (!empty($extConf['useragent'])) {
@ini_set('user_agent', $extConf['useragent']);
}
// Turn off libxml's error logging.
$libxmlErrors = libxml_use_internal_errors(TRUE);
// disable entity loading
$previousValueOfEntityLoader = libxml_disable_entity_loader(TRUE);
// Load XML from file.
$xml = simplexml_load_string(file_get_contents($file));
// reset entity loader setting
libxml_disable_entity_loader($previousValueOfEntityLoader);
// Reset libxml's error logging.
libxml_use_internal_errors($libxmlErrors);
if ($xml === FALSE) {
return 1;
}
} else {
return 1;
}
// Load class.
if (!class_exists('Apache_Solr_Document')) {
require_once \TYPO3\CMS\Core\Utility\GeneralUtility::getFileAbsFileName('EXT:' . self::$extKey . '/lib/SolrPhpClient/Apache/Solr/Document.php');
}
// Create new Solr document.
$solrDoc = new Apache_Solr_Document();
// Create unique identifier from document's UID and unit's XML ID.
$solrDoc->setField('id', $doc->uid . $physicalUnit['id']);
$solrDoc->setField('uid', $doc->uid);
$solrDoc->setField('pid', $doc->pid);
$solrDoc->setField('page', $page);
if (!empty($physicalUnit['files'][$extConf['fileGrpThumbs']])) {
$solrDoc->setField('thumbnail', $doc->getFileLocation($physicalUnit['files'][$extConf['fileGrpThumbs']]));
}
$solrDoc->setField('partof', $doc->parentId);
$solrDoc->setField('root', $doc->rootId);
$solrDoc->setField('sid', $physicalUnit['id']);
$solrDoc->setField('toplevel', FALSE);
$solrDoc->setField('type', $physicalUnit['type'], self::$fields['fieldboost']['type']);
$solrDoc->setField('fulltext', tx_dlf_alto::getRawText($xml));
try {
self::$solr->service->addDocument($solrDoc);
} catch (Exception $e) {
if (!defined('TYPO3_cliMode')) {
$message = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('TYPO3\\CMS\\Core\\Messaging\\FlashMessage', tx_dlf_helper::getLL('flash.solrException', TRUE) . '<br />' . htmlspecialchars($e->getMessage()), tx_dlf_helper::getLL('flash.error', TRUE), \TYPO3\CMS\Core\Messaging\FlashMessage::ERROR, TRUE);
tx_dlf_helper::addMessage($message);
}
return 1;
}
}
return $errors;
}
/**
*
* @param Apache_Solr_Document or Array $parts
*
* @return SP_Controller_Action_Helper_Solr
*/
public function pushDocuments($parts)
{
$this->_setSolrService();
if ($parts instanceof Apache_Solr_Document) {
$this->documents[] = $parts;
} else {
if (is_array($parts)) {
foreach ($parts as $item => $fields) {
if ($fields instanceof Apache_Solr_Document) {
$this->documents[] = $fields;
} else {
$part = new Apache_Solr_Document();
foreach ($fields as $key => $value) {
if (is_array($value)) {
foreach ($value as $datum) {
$part->setMultiValue($key, $datum);
}
} else {
$part->setField($key, $value);
}
}
$this->documents[] = $part;
}
}
} else {
trigger_error("the paramter \$part must be an object of Apache_Solr_Document or an array");
}
}
return $this;
}
<?php
require dirname(__FILE__) . '/../../bootstrap/unit.php';
$t = new limeade_test(2, limeade_output::get());
$t->diag('document format a document');
$document = new Apache_Solr_Document();
$document->setBoost(10);
$document->setField('sfl_guid', 'GUID_1234');
$document->setField('name', 'Thomas Rabaix', 1);
$document->setMultiValue('skills', 'php');
$document->setMultiValue('skills', 'symfony');
$document->addField('skills', 'objective-c');
$expected = array('name' => 'skills', 'value' => array(0 => 'php', 1 => 'symfony', 2 => 'objective-c'), 'boost' => false);
$t->cmp_ok($document->getField('skills'), '==', $expected, '::getField test multivalue setter');
$expected = array('name' => 'name', 'value' => 'Thomas Rabaix', 'boost' => 1);
$t->cmp_ok($document->getField('name'), '==', $expected, '::getField test setter');
/**
* Processes a logical unit (and its children) for the Solr index
*
* @access protected
*
* @param tx_dlf_document &$doc: The METS document
* @param array $logicalUnit: Array of the logical unit to process
*
* @return integer 0 on success or 1 on failure
*/
protected static function process(tx_dlf_document &$doc, array $logicalUnit)
{
$errors = 0;
// Get metadata for logical unit.
$metadata = $doc->metadataArray[$logicalUnit['id']];
if (!empty($metadata)) {
// Load class.
if (!class_exists('Apache_Solr_Document')) {
require_once t3lib_div::getFileAbsFileName('EXT:' . self::$extKey . '/lib/SolrPhpClient/Apache/Solr/Document.php');
}
// Create new Solr document.
$solrDoc = new Apache_Solr_Document();
// Create unique identifier from document's UID and unit's XML ID.
$solrDoc->setField('id', $doc->uid . $logicalUnit['id']);
$solrDoc->setField('uid', $doc->uid);
$solrDoc->setField('pid', $doc->pid);
if (tx_dlf_helper::testInt($logicalUnit['points'])) {
$solrDoc->setField('page', $logicalUnit['points']);
}
if ($logicalUnit['id'] == $doc->toplevelId) {
$solrDoc->setField('thumbnail', $doc->thumbnail);
} elseif (!empty($logicalUnit['thumbnailId'])) {
$solrDoc->setField('thumbnail', $doc->getFileLocation($logicalUnit['thumbnailId']));
}
$solrDoc->setField('partof', $doc->parentId);
$solrDoc->setField('sid', $logicalUnit['id']);
$solrDoc->setField('toplevel', in_array($logicalUnit['type'], self::$toplevel));
$solrDoc->setField('type', $logicalUnit['type'], self::$fields['fieldboost']['type']);
$solrDoc->setField('title', $metadata['title'][0], self::$fields['fieldboost']['title']);
$solrDoc->setField('volume', $metadata['volume'][0], self::$fields['fieldboost']['volume']);
$autocomplete = array();
foreach ($metadata as $index_name => $data) {
if (!empty($data) && substr($index_name, -8) !== '_sorting') {
$solrDoc->setField(self::getIndexFieldName($index_name, $doc->pid), $data, self::$fields['fieldboost'][$index_name]);
if (in_array($index_name, self::$fields['sortables'])) {
// Add sortable fields to index.
$solrDoc->setField($index_name . '_sorting', $metadata[$index_name . '_sorting'][0]);
}
if (in_array($index_name, self::$fields['facets'])) {
// Add facets to index.
$solrDoc->setField($index_name . '_faceting', $data);
}
if (in_array($index_name, self::$fields['autocompleted'])) {
$autocomplete = array_merge($autocomplete, $data);
}
}
}
// Add autocomplete values to index.
if (!empty($autocomplete)) {
$solrDoc->setField('autocomplete', $autocomplete);
}
try {
self::$solr->service->addDocument($solrDoc);
} catch (Exception $e) {
if (!defined('TYPO3_cliMode')) {
$message = t3lib_div::makeInstance('t3lib_FlashMessage', tx_dlf_helper::getLL('flash.solrException', TRUE) . '<br />' . htmlspecialchars($e->getMessage()), tx_dlf_helper::getLL('flash.error', TRUE), t3lib_FlashMessage::ERROR, TRUE);
t3lib_FlashMessageQueue::addMessage($message);
}
return 1;
}
}
// Check for child elements...
if (!empty($logicalUnit['children'])) {
foreach ($logicalUnit['children'] as $child) {
if (!$errors) {
// ...and process them, too.
$errors = self::process($doc, $child);
} else {
break;
}
}
}
return $errors;
}
public function convertObjectToDocument($dataObject, $stage = null, $fieldBoost = array())
{
$document = new Apache_Solr_Document();
$fieldsToIndex = array();
$object = null;
// whether the original item is an object or array.
// determines how we treat the object later when checking all the fields
$sourceObject = true;
$id = 0;
if (is_object($dataObject)) {
if ($dataObject->hasMethod('hasField') && $dataObject->hasField('ShowInSearch') && !$dataObject->ShowInSearch) {
return;
}
$fieldsToIndex = $this->getSearchableFieldsFor($dataObject);
// $dataObject->searchableFields();
$object = $this->objectToFields($dataObject);
$id = $dataObject->ID;
} else {
$object = $dataObject;
$id = isset($dataObject['ID']) ? $dataObject['ID'] : 0;
$fieldsToIndex = isset($object['index_fields']) ? $object['index_fields'] : array_flip(array_keys($object));
$sourceObject = false;
}
$fieldsToIndex['SS_URL'] = true;
$fieldsToIndex['SS_ID'] = true;
$fieldsToIndex['LastEdited'] = true;
$fieldsToIndex['Created'] = true;
$fieldsToIndex['ClassName'] = true;
$fieldsToIndex['ClassNameHierarchy'] = true;
$fieldsToIndex['ParentsHierarchy'] = true;
// the stage we're on when we write this doc to the index.
// this is used for versioned AND non-versioned objects; we just cheat and
// set it BOTH stages if it's non-versioned object
$fieldsToIndex['SS_Stage'] = true;
// if it's a versioned object, just save ONE stage value.
if ($stage) {
$object['SS_Stage'] = array('Type' => 'Enum', 'Value' => $stage);
$id = $id . '_' . $stage;
} else {
$object['SS_Stage'] = array('Type' => 'Enum', 'Value' => array('Stage', 'Live'));
}
if (!$id) {
return false;
}
// specially handle the subsite module - this has serious implications for our search
// @TODO we want to genercise this later for other modules to hook into it!
if (ClassInfo::exists('Subsite')) {
$fieldsToIndex['SubsiteID'] = true;
if (is_object($dataObject)) {
$object['SubsiteID'] = array('Type' => 'Int', 'Value' => $dataObject->SubsiteID);
}
}
$classType = isset($object['ClassName']) ? $object['ClassName']['Value'] : null;
// we're not indexing the ID field because it conflicts with Solr's internal ID
unset($object['ID']);
// a special type hierarchy
if ($classType) {
$classes = array_values(ClassInfo::ancestry($classType));
$object['ClassNameHierarchy'] = array('Type' => 'MultiValueField', 'Value' => $classes);
$object['ParentsHierarchy'] = $this->getParentsHierarchyField($dataObject);
}
foreach ($object as $field => $valueDesc) {
if (!$valueDesc) {
continue;
}
if (!is_array($valueDesc) || !isset($valueDesc['Type'])) {
// if we're indexing an object and there's no valueDesc, just skip this field
if ($sourceObject) {
continue;
}
$valueDesc = array('Value' => $valueDesc, 'Type' => $this->mapper->mapValueToType($field, $valueDesc));
}
$type = $valueDesc['Type'];
$value = $valueDesc['Value'];
// this should have already been taken care of, but just in case...
if ($type == 'MultiValueField' && $value instanceof MultiValueField) {
$value = $value->getValues();
}
if (!isset($fieldsToIndex[$field])) {
continue;
}
$fieldName = $this->mapper->mapFieldNameFromType($field, $type, $fieldsToIndex[$field]);
if (!$fieldName) {
continue;
}
$value = $this->mapper->convertValue($value, $type);
if (is_array($value)) {
foreach ($value as $v) {
$document->addField($fieldName, $v);
}
} else {
$boost = false;
if (isset($fieldBoost["{$fieldName}:{$value}"])) {
$boost = $fieldBoost["{$fieldName}:{$value}"];
}
$document->setField($fieldName, $value, $boost);
$document->{$fieldName} = $value;
}
}
$document->id = $classType ? $classType . '_' . $id : SolrSearchService::RAW_DATA_KEY . $id;
return $document;
}
/**
* This takes an Omeka_Record instance and returns a populated
* Apache_Solr_Document.
*
* @param Omeka_Record $item The record to index.
*
* @return Apache_Solr_Document
* @author Eric Rochester <*****@*****.**>
**/
public static function itemToDocument($item)
{
$fields = get_db()->getTable('SolrSearchField');
$doc = new Apache_Solr_Document();
$doc->setField('id', "Item_{$item->id}");
$doc->setField('resulttype', 'Item');
$doc->setField('model', 'Item');
$doc->setField('modelid', $item->id);
$doc->setField('owner_id', $item->owner_id);
// extend $doc to to include and items public / private status
$doc->setField('public', $item->public);
// Title:
$title = metadata($item, array('Dublin Core', 'Title'));
$doc->setField('title', $title);
// Elements
self::indexItem($fields, $item, $doc);
// Tags:
foreach ($item->getTags() as $tag) {
$doc->setMultiValue('tag_t', $tag->name);
$doc->setMultiValue('tag_s', $tag->name);
}
// Collection:
if ($collection = $item->getCollection()) {
$doc->collection = metadata($collection, array('Dublin Core', 'Title'));
}
// Item type:
if ($itemType = $item->getItemType()) {
$doc->itemtype = $itemType->name;
}
$doc->featured = (bool) $item->featured;
// File metadata (this is weird)
// foreach ($item->getFiles() as $file) {
// self::indexItem($fields, $file, $doc);
// }
###############################################################################
//ADDITION FOR VISUALIZATIONS, FACETS AND PRIVACY --> move to indexitem
//VISUALIZATION:
if ($itemType = $item->getItemType()) {
$doc->setField('itemtype_id', $item->item_type_id);
}
if ($collection = $item->getCollection()) {
$doc->setField('collection_id', $item->collection_id);
}
// Start and end date(s):
$date = metadata($item, array('Dublin Core', 'Date'));
if (self::date_validate($date)) {
$date_span = explode(' ', $date, 2);
if (count($date_span) == 2) {
$doc->setField('date_start', self::date_supplement($date_span[0]));
$doc->setField('date_end', self::date_supplement($date_span[1]));
} elseif (count($date_span) == 1) {
$doc->setField('date_start', self::date_supplement($date_span[0]));
$doc->setField('date_end', self::date_supplement($date_span[0]));
}
$doc->setField('decennium_group', self::date_decennium($date_span[0]));
}
// FACETS
//text size and text size group
if ($text = metadata($item, array('Item Type Metadata', 'Text'))) {
$main_word_count = substr_count($text, ' ');
$doc->setField('94_t', $main_word_count);
$doc->setField('95_t', self::classify_length($main_word_count));
$doc->setField('94_s', $main_word_count);
$doc->setField('95_s', self::classify_length($main_word_count));
}
//Locations
$db = get_db();
$locations = $db->getTable('Location')->findLocationByItem($item, false);
if ($locations) {
if (array_key_exists('narration_location', $locations)) {
$location = $locations['narration_location'];
$doc->setField("latitude", $location->latitude);
$doc->setField("longitude", $location->longitude);
$doc->setField("zoom_level", $location->zoom_level);
$doc->setField("map_type_t", $location->map_type);
$doc->setField("address_t", $location->address);
$doc->setField("route_t", $location->route);
$doc->setField("street_number_t", $location->street_number);
$doc->setField("postal_code_t", $location->postal_code);
$doc->setField("postal_code_prefix_t", $location->postal_code_prefix);
$doc->setField("sublocality_t", $location->sublocality);
$doc->setField("locality_t", $location->locality);
$doc->setField("natural_feature_t", $location->natural_feature);
$doc->setField("establishment_t", $location->establishment);
$doc->setField("point_of_interest_t", $location->point_of_interest);
$doc->setField("administrative_area_level_3_t", $location->administrative_area_level_3);
$doc->setField("administrative_area_level_2_t", $location->administrative_area_level_2);
$doc->setField("administrative_area_level_1_t", $location->administrative_area_level_1);
$doc->setField("country_t", $location->country);
$doc->setField("continent_t", $location->continent);
$doc->setField("planetary_body_t", $location->planetary_body);
$doc->setField("map_type_s", $location->map_type);
$doc->setField("address_s", $location->address);
$doc->setField("route_s", $location->route);
$doc->setField("street_number_s", $location->street_number);
$doc->setField("postal_code_s", $location->postal_code);
$doc->setField("postal_code_prefix_s", $location->postal_code_prefix);
$doc->setField("sublocality_s", $location->sublocality);
$doc->setField("locality_s", $location->locality);
$doc->setField("natural_feature_s", $location->natural_feature);
$doc->setField("establishment_s", $location->establishment);
$doc->setField("point_of_interest_s", $location->point_of_interest);
$doc->setField("administrative_area_level_3_s", $location->administrative_area_level_3);
$doc->setField("administrative_area_level_2_s", $location->administrative_area_level_2);
$doc->setField("administrative_area_level_1_s", $location->administrative_area_level_1);
$doc->setField("country_s", $location->country);
$doc->setField("continent_s", $location->continent);
$doc->setField("planetary_body_s", $location->planetary_body);
}
if (array_key_exists('action_location', $locations)) {
$location = $locations['action_location'];
$doc->setField("action_latitude", $location->latitude);
$doc->setField("action_longitude", $location->longitude);
$doc->setField("action_zoom_level", $location->zoom_level);
$doc->setField("action_map_type_t", $location->map_type);
$doc->setField("action_address_t", $location->address);
$doc->setField("action_route_t", $location->route);
$doc->setField("action_street_number_t", $location->street_number);
$doc->setField("action_postal_code_t", $location->postal_code);
$doc->setField("action_postal_code_prefix_t", $location->postal_code_prefix);
$doc->setField("action_sublocality_t", $location->sublocality);
$doc->setField("action_locality_t", $location->locality);
$doc->setField("action_natural_feature_t", $location->natural_feature);
$doc->setField("action_establishment_t", $location->establishment);
$doc->setField("action_point_of_interest_t", $location->point_of_interest);
$doc->setField("action_administrative_area_level_3_t", $location->administrative_area_level_3);
$doc->setField("action_administrative_area_level_2_t", $location->administrative_area_level_2);
$doc->setField("action_administrative_area_level_1_t", $location->administrative_area_level_1);
$doc->setField("action_country_t", $location->country);
$doc->setField("action_continent_t", $location->continent);
$doc->setField("action_planetary_body_t", $location->planetary_body);
$doc->setField("action_map_type_s", $location->map_type);
$doc->setField("action_address_s", $location->address);
$doc->setField("action_route_s", $location->route);
$doc->setField("action_street_number_s", $location->street_number);
$doc->setField("action_postal_code_s", $location->postal_code);
$doc->setField("action_postal_code_prefix_s", $location->postal_code_prefix);
$doc->setField("action_sublocality_s", $location->sublocality);
$doc->setField("action_locality_s", $location->locality);
$doc->setField("action_natural_feature_s", $location->natural_feature);
$doc->setField("action_establishment_s", $location->establishment);
$doc->setField("action_point_of_interest_s", $location->point_of_interest);
$doc->setField("action_administrative_area_level_3_s", $location->administrative_area_level_3);
$doc->setField("action_administrative_area_level_2_s", $location->administrative_area_level_2);
$doc->setField("action_administrative_area_level_1_s", $location->administrative_area_level_1);
$doc->setField("action_country_s", $location->country);
$doc->setField("action_continent_s", $location->continent);
$doc->setField("action_planetary_body_s", $location->planetary_body);
}
}
###############################################################################
return $doc;
}
function s4w_build_document($post_info, $domain = NULL, $path = NULL)
{
global $blog_id;
global $current_blog;
$doc = NULL;
$plugin_s4w_settings = s4w_get_option();
$exclude_ids = $plugin_s4w_settings['s4w_exclude_pages'];
$categoy_as_taxonomy = $plugin_s4w_settings['s4w_cat_as_taxo'];
$index_comments = $plugin_s4w_settings['s4w_index_comments'];
$index_custom_fields = $plugin_s4w_settings['s4w_index_custom_fields'];
if ($post_info) {
# check if we need to exclude this document
if (is_multisite() && in_array($current_blog->domain . $post_info->ID, (array) $exclude_ids)) {
return NULL;
} else {
if (!is_multisite() && in_array($post_info->ID, (array) $exclude_ids)) {
return NULL;
}
}
$doc = new Apache_Solr_Document();
$auth_info = get_userdata($post_info->post_author);
# wpmu specific info
if (is_multisite()) {
// if we get here we expect that we've "switched" what blog we're running
// as
if ($domain == NULL) {
$domain = $current_blog->domain;
}
if ($path == NULL) {
$path = $current_blog->path;
}
$blogid = get_blog_id_from_url($domain, $path);
$doc->setField('id', $domain . $path . $post_info->ID);
$doc->setField('permalink', get_blog_permalink($blogid, $post_info->ID));
$doc->setField('blogid', $blogid);
$doc->setField('blogdomain', $domain);
$doc->setField('blogpath', $path);
$doc->setField('wp', 'multisite');
} else {
$doc->setField('id', $post_info->ID);
$doc->setField('permalink', get_permalink($post_info->ID));
$doc->setField('wp', 'wp');
}
$numcomments = 0;
if ($index_comments) {
$comments = get_comments("status=approve&post_id={$post_info->ID}");
foreach ($comments as $comment) {
$doc->addField('comments', $comment->comment_content);
$numcomments += 1;
}
}
$doc->setField('title', $post_info->post_title);
$doc->setField('content', strip_tags($post_info->post_content));
// rawcontent strips out characters lower than 0x20
$doc->setField('rawcontent', strip_tags(preg_replace('/[^(\\x20-\\x7F)\\x0A]*/', '', $post_info->post_content)));
// contentnoshortcodes also strips characters below 0x20 but also strips shortcodes
// used in WP to add images or other content, useful if you're pulling this data
// into another system
//
// For example
// [caption id="attachment_92495" align="alignright" width="160" caption="Duane Sand"][/caption] FARGO - Republican U.S. Senate...
//
// Will become
// FARGO - Republican U.S. Senate...
$doc->setField('contentnoshortcodes', strip_tags(preg_replace('/[^(\\x20-\\x7F)\\x0A]*/', '', strip_tags(strip_shortcodes($post_info->post_content)))));
$doc->setField('numcomments', $numcomments);
$doc->setField('author', $auth_info->display_name);
$doc->setField('author_s', get_author_posts_url($auth_info->ID, $auth_info->user_nicename));
$doc->setField('type', $post_info->post_type);
$doc->setField('date', s4w_format_date($post_info->post_date_gmt));
$doc->setField('tdate', s4w_format_date($post_info->post_date_gmt));
$doc->setField('modified', s4w_format_date($post_info->post_modified_gmt));
$doc->setField('displaydate', $post_info->post_date);
$doc->setField('displaymodified', $post_info->post_modified);
$categories = get_the_category($post_info->ID);
if (!$categories == NULL) {
foreach ($categories as $category) {
if ($categoy_as_taxonomy) {
$doc->addField('categories', get_category_parents($category->cat_ID, FALSE, '^^'));
} else {
$doc->addField('categories', $category->cat_name);
}
}
}
//get all the taxonomy names used by wp
$taxonomies = (array) get_taxonomies(array('_builtin' => FALSE), 'names');
foreach ($taxonomies as $parent) {
$terms = get_the_terms($post_info->ID, $parent);
if ((array) $terms === $terms) {
//we are creating *_taxonomy as dynamic fields using our schema
//so lets set up all our taxonomies in that format
$parent = $parent . "_taxonomy";
foreach ($terms as $term) {
$doc->addField($parent, $term->name);
}
}
}
$tags = get_the_tags($post_info->ID);
if (!$tags == NULL) {
foreach ($tags as $tag) {
$doc->addField('tags', $tag->name);
}
}
if (count($index_custom_fields) > 0 && count($custom_fields = get_post_custom($post_info->ID))) {
foreach ((array) $index_custom_fields as $field_name) {
$field = (array) $custom_fields[$field_name];
foreach ($field as $key => $value) {
$doc->addField($field_name . '_str', $value);
$doc->addField($field_name . '_srch', $value);
}
}
}
// add full json respone
$json_response = new JSON_API_Response();
$json_post = new JSON_API_Post_full(get_post($post_info->ID));
$response = (object) array('post' => $json_post);
$doc->setField('json', $json_response->get_json($response));
// add tone: /nieuws/beste van het web/weblog/column/
$tone = 'Nieuws';
if ($blog_id > 1) {
$nmt_blog_type = stripslashes(get_blog_option($blog_id, 'nmt_blog_type'));
$nmt_blog_types = array(1 => 'Weblog', 2 => 'Column', 3 => 'Nieuws', 4 => 'Fotoserie');
$tone = $nmt_blog_types[$nmt_blog_type];
} else {
$category_slugs = array();
if (!$categories == NULL) {
foreach ($categories as $category) {
$category_slugs[] = $category->slug;
}
}
if (in_array('beste-van-het-web', $category_slugs)) {
$tone = 'Beste van het web';
}
}
$doc->setField('tone', $tone);
} else {
// this will fire during blog sign up on multisite, not sure why
_e('Post Information is NULL', 'solr4wp');
}
syslog(LOG_ERR, "built document for {$blog_id} - {$domain}{$path} with title " . $post_info->post_title . " and status of " . $post_info->post_status);
return $doc;
}
protected function buildProductDocument($storeId, $productId, $indexableAttributes)
{
$helper = Mage::helper('solr');
$indexableAttributes = Mage::helper('solr/attribute')->getNamedProductAttributes($indexableAttributes);
$product = Mage::getModel('catalog/product')->setStoreId($storeId)->load($productId);
/** @var Mage_Catalog_Model_Product $product */
$baseUrl = Mage::getBaseUrl(Mage_Core_Model_Store::URL_TYPE_WEB);
$host = parse_url($baseUrl, PHP_URL_HOST);
$document = new Apache_Solr_Document();
$document->setField('appKey', 'Asm_Solr');
$document->setField('type', 'catalog/product');
$document->setField('id', $helper->getProductDocumentId($product->getEntityId()));
$document->setField('site', $host);
$document->setField('siteHash', $helper->getSiteHashForDomain($host));
$document->setField('storeId', $storeId);
$document->setField('created', $helper->dateToIso($product->getCreatedAt()));
$document->setField('changed', $helper->dateToIso($product->getUpdatedAt()));
$document->setField('sku', $product->getSku());
$document->setField('productId', $product->getEntityId());
$categoryIds = $product->getCategoryIds();
foreach ($categoryIds as $categoryId) {
$document->addField('categoryId', $categoryId);
}
$document->setField('isSalable', $product->isSalable());
$document->setField('inStock', $product->isInStock());
$document->setField('isVisible', $product->getStatus());
$document->setField('isVisibleInCatalog', $product->isVisibleInCatalog());
$document->setField('title', $product->getName());
$document->setField('content', $product->getDescription());
$document->setField('keywords', $helper->trimExplode(',', $product->getMetaKeyword(), true));
$document->setField('url', $product->getProductUrl());
$document->setField('price', $product->getPrice());
if ($product->getManufacturer()) {
$document->setField('manufacturer', $product->getAttributeText('manufacturer'));
}
$document->setField('image_stringS', $product->getImage());
$document->setField('small_image_stringS', $product->getSmallImage());
$document->setField('thumbnail_stringS', $product->getThumbnail());
$productType = $product->getTypeId();
$document->setField('type_id_stringS', $productType);
if ($productType == 'configurable') {
$childProductAttributes = $this->getConfigurableProductChildProductAttributes($storeId, $product);
$indexableAttributes = array_merge($indexableAttributes, $childProductAttributes);
}
$fieldProcessorFactory = Mage::getResourceModel('solr/indexer_fieldprocessor_factory');
// add other searchable attributes as dynamic fields
foreach ($indexableAttributes as $attributeCode => $attributeValue) {
if (empty($attributeValue) || in_array($attributeCode, $this->fixedSchemaFieldAttributes)) {
// don't add fixed schema fields twice
continue;
}
$fieldProcessor = $fieldProcessorFactory->getFieldProcessor($attributeCode, $attributeValue);
$document->setField($fieldProcessor->getFieldName(), $fieldProcessor->getFieldValue());
}
return $document;
}
protected function _addAs($object, $base, $options)
{
$includeSubs = $options['include_children'];
$doc = new Apache_Solr_Document();
// Always present fields
$doc->setField('_documentid', $this->getDocumentID($object, $base, $includeSubs));
$doc->setField('ID', $object->ID);
$doc->setField('ClassName', $object->ClassName);
foreach (SearchIntrospection::hierarchy(get_class($object), false) as $class) {
$doc->addField('ClassHierarchy', $class);
}
// Add the user-specified fields
foreach ($this->getFieldsIterator() as $name => $field) {
if ($field['base'] == $base) {
$this->_addField($doc, $object, $field);
}
}
try {
$this->getService()->addDocument($doc);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
return $doc;
}
/**
* modifies a document according to the given configuration
*
* @param Apache_Solr_Document $document
* @param array $processingConfiguration
*/
public function processDocument(Apache_Solr_Document $document, array $processingConfiguration)
{
foreach ($processingConfiguration as $fieldName => $instruction) {
$fieldInformation = $document->getField($fieldName);
$isSingleValueField = FALSE;
if ($fieldInformation !== FALSE) {
$fieldValue = $fieldInformation['value'];
if (!is_array($fieldValue)) {
// turn single value field into multi value field
$fieldValue = array($fieldValue);
$isSingleValueField = TRUE;
}
switch ($instruction) {
case 'timestampToIsoDate':
$processor = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('Tx_Solr_FieldProcessor_TimestampToIsoDate');
$fieldValue = $processor->process($fieldValue);
break;
case 'pathToHierarchy':
$processor = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('Tx_Solr_FieldProcessor_PathToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'pageUidToHierarchy':
$processor = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('Tx_Solr_FieldProcessor_PageUidToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'categoryUidToHierarchy':
$processor = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('Tx_Solr_FieldProcessor_CategoryUidToHierarchy');
$fieldValue = $processor->process($fieldValue);
break;
case 'uppercase':
$fieldValue = array_map('strtoupper', $fieldValue);
break;
}
if ($isSingleValueField) {
// turn multi value field back into single value field
$fieldValue = $fieldValue[0];
}
$document->setField($fieldName, $fieldValue);
}
}
}
