ApacheSolrForTypo3\Solr\Typo3PageIndexer

Complexity

Total Complexity

63

Size/Duplication

Total Lines	607
Duplicated Lines	0 %

Coupling/Cohesion

Components	1
Dependencies	14

Test Coverage

Coverage

64.48%

Importance

Changes

0

28 Methods

Rating	Name	Duplication	Size	Complexity
A	setIndexQueueItem()	0	4	1
A	initializeSolrConnection()	0	14	2
A	getPageSolrDocumentId()	0	4	1
A	getPageSolrDocument()	0	4	1
A	setSolrConnection()	0	11	2
B	indexPage()	0	28	2
A	addEndtimeField()	0	6	2
A	getMountPointParameter()	0	4	1
A	setMountPointParameter()	0	4	1
A	getIndexConfigurationNameForCurrentPage()	0	4	2
A	getPageUrl()	0	4	1
A	setPageUrl()	0	4	1
A	getPageAccessRootline()	0	4	1
A	setPageAccessRootline()	0	4	1
A	getDocumentsSentToSolr()	0	4	1
A	__construct()	0	21	4
B	log()	0	17	5
A	applyIndexPagePostProcessors()	0	15	4
A	getPageDocument()	0	54	1
A	addAccessField()	0	7	2
A	addKeywordsField()	0	7	2
A	addTagContentFields()	0	7	2
A	getRootLineFieldValue()	0	9	2
A	getDocumentIdGroups()	0	13	2
B	substitutePageDocument()	0	29	6
B	getAdditionalDocuments()	0	24	5
A	processDocuments()	0	8	2
B	addDocumentsToSolrIndex()	0	33	6

<?php
namespace ApacheSolrForTypo3\Solr;

/***************************************************************
 *  Copyright notice
 *
 *  (c) 2009-2015 Ingo Renner <ingo@typo3.org>
 *  All rights reserved
 *
 *  This script is part of the TYPO3 project. The TYPO3 project is
 *  free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License as published by
 *  the Free Software Foundation; either version 2 of the License, or
 *  (at your option) any later version.
 *
 *  The GNU General Public License can be found at
 *  http://www.gnu.org/copyleft/gpl.html.
 *
 *  This script is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU General Public License for more details.
 *
 *  This copyright notice MUST APPEAR in all copies of the script!
 ***************************************************************/

use Apache_Solr_Document;
use ApacheSolrForTypo3\Solr\Access\Rootline;
use ApacheSolrForTypo3\Solr\Domain\Variants\IdBuilder;
use ApacheSolrForTypo3\Solr\FieldProcessor\Service;
use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
use ApacheSolrForTypo3\Solr\IndexQueue\Item;
use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration;
use TYPO3\CMS\Core\Utility\GeneralUtility;
use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;

/**
 * Page Indexer to index TYPO3 pages used by the Index Queue.
 *
 * @author Ingo Renner <ingo@typo3.org>
 * @author Daniel Poetzinger <poetzinger@aoemedia.de>
 * @author Timo Schmidt <schmidt@aoemedia.de>
 */
class Typo3PageIndexer
{

    /**
     * ID of the current page's Solr document.
     *
     * @var string
     */
    protected static $pageSolrDocumentId = '';
    /**
     * The Solr document generated for the current page.
     *
     * @var \Apache_Solr_Document
     */
    protected static $pageSolrDocument = null;
    /**
     * The mount point parameter used in the Frontend controller.
     *
     * @var string
     */
    protected $mountPointParameter;
    /**
     * Solr server connection.
     *
     * @var SolrService
     */
    protected $solrConnection = null;
    /**
     * Frontend page object (TSFE).
     *
     * @var TypoScriptFrontendController
     */
    protected $page = null;
    /**
     * Content extractor to extract content from TYPO3 pages
     *
     * @var Typo3PageContentExtractor
     */
    protected $contentExtractor = null;
    /**
     * URL to be indexed as the page's URL
     *
     * @var string
     */
    protected $pageUrl = '';
    /**
     * The page's access rootline
     *
     * @var Rootline
     */
    protected $pageAccessRootline = null;
    /**
     * Documents that have been sent to Solr
     *
     * @var array
     */
    protected $documentsSentToSolr = [];

    /**
     * @var TypoScriptConfiguration
     */
    protected $configuration;

    /**
     * @var Item
     */
    protected $indexQueueItem;

    /**
     * @var IdBuilder
     */
    protected $variantIdBuilder;

    /**
     * Constructor
     *
     * @param TypoScriptFrontendController $page The page to index
     * @param IdBuilder $variantIdBuilder
     */
    public function __construct(TypoScriptFrontendController $page, IdBuilder $variantIdBuilder = null)
    {
        $this->page = $page;
        $this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
        $this->configuration = Util::getSolrConfiguration();

        try {
            $this->initializeSolrConnection();
        } catch (\Exception $e) {
            $this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 3);

            // TODO extract to a class "ExceptionLogger"
            if ($this->configuration->getLoggingExceptions()) {
                GeneralUtility::devLog('Exception while trying to index a page', 'solr', 3, [$e->__toString()]);
            }
        }

        $this->contentExtractor = GeneralUtility::makeInstance(Typo3PageContentExtractor::class, $this->page->content);
        $this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, '');
        $this->variantIdBuilder = is_null($variantIdBuilder) ? GeneralUtility::makeInstance(IdBuilder::class) : $variantIdBuilder;
    }

    /**
     * @param Item $indexQueueItem
     */
    public function setIndexQueueItem($indexQueueItem)
    {
        $this->indexQueueItem = $indexQueueItem;
    }


    /**
     * Initializes the Solr server connection.
     *
     * @throws    \Exception when no Solr connection can be established.
     */
    protected function initializeSolrConnection()
    {
        $solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, $this->page->sys_language_uid);

        // do not continue if no server is available
        if (!$solr->ping()) {
            throw new \Exception(
                'No Solr instance available while trying to index a page.',
                1234790825
            );
        }

        $this->solrConnection = $solr;
    }

    /**
     * Logs messages to devlog and TS log (admin panel)
     *
     * @param string $message Message to set
     * @param int $errorNum Error number
     * @param array $data Additional data to log
     * @return void
     */
    protected function log($message, $errorNum = 0, array $data = [])
    {
        if (is_object($GLOBALS['TT'])) {
            $GLOBALS['TT']->setTSlogMessage('tx_solr: ' . $message, $errorNum);
        }

        if ($this->configuration->getLoggingIndexing()) {
            $logData = [];
            if (!empty($data)) {
                foreach ($data as $value) {
                    $logData[] = (array)$value;
                }
            }

            GeneralUtility::devLog($message, 'solr', $errorNum, $logData);
        }
    }

    /**
     * Gets the current page's Solr document ID.
     *
     * @return string|NULL The page's Solr document ID or NULL in case no document was generated yet.
     */
    public static function getPageSolrDocumentId()
    {
        return self::$pageSolrDocumentId;
    }

    /**
     * Gets the Solr document generated for the current page.
     *
     * @return \Apache_Solr_Document|NULL The page's Solr document or NULL if it has not been generated yet.
     */
    public static function getPageSolrDocument()
    {
        return self::$pageSolrDocument;
    }

    /**
     * Allows to provide a Solr server connection other than the one
     * initialized by the constructor.
     *
     * @param SolrService $solrConnection Solr connection
     * @throws \Exception if the Solr server cannot be reached
     */
    public function setSolrConnection(SolrService $solrConnection)
    {
        if (!$solrConnection->ping()) {
            throw new \Exception(
                'Could not connect to Solr server.',
                1323946472
            );
        }

        $this->solrConnection = $solrConnection;
    }

    /**
     * Indexes a page.
     *
     * @return bool TRUE after successfully indexing the page, FALSE on error
     * @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
     */
    public function indexPage()
    {
        $pageIndexed = false;
        $documents = []; // this will become useful as soon as when starting to index individual records instead of whole pages

        if (is_null($this->solrConnection)) {
            // intended early return as it doesn't make sense to continue
            // and waste processing time if the solr server isn't available
            // anyways
            // FIXME use an exception
            return $pageIndexed;
        }

        $pageDocument = $this->getPageDocument();
        $pageDocument = $this->substitutePageDocument($pageDocument);

        $this->applyIndexPagePostProcessors($pageDocument);

        self::$pageSolrDocument = $pageDocument;
        $documents[] = $pageDocument;
        $documents = $this->getAdditionalDocuments($pageDocument, $documents);
        $this->processDocuments($documents);

        $pageIndexed = $this->addDocumentsToSolrIndex($documents);
        $this->documentsSentToSolr = $documents;

        return $pageIndexed;
    }

    /**
     * Applies the configured post processors (indexPagePostProcessPageDocument)
     *
     * @param \Apache_Solr_Document $pageDocument
     */
    protected function applyIndexPagePostProcessors($pageDocument)
    {
        if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
            return;
        }

        foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
            $postProcessor = GeneralUtility::getUserObj($classReference);
            if (!$postProcessor instanceof PageDocumentPostProcessor) {
                throw new \UnexpectedValueException(get_class($pageDocument) . ' must implement interface ' . PageDocumentPostProcessor::class, 1397739154);
            }

            $postProcessor->postProcessPageDocument($pageDocument, $this->page);
        }
    }

    /**
     * Builds the Solr document for the current page.
     *
     * @return \Apache_Solr_Document A document representing the page
     */
    protected function getPageDocument()
    {
        $document = GeneralUtility::makeInstance(Apache_Solr_Document::class);
        /* @var $document \Apache_Solr_Document */
        $site = Site::getSiteByPageId($this->page->id);
        $pageRecord = $this->page->page;

        self::$pageSolrDocumentId = $documentId = Util::getPageDocumentId(
            $this->page->id,
            $this->page->type,
            $this->page->sys_language_uid,
            $this->getDocumentIdGroups(),
            $this->getMountPointParameter()
        );
        $document->setField('id', $documentId);
        $document->setField('site', $site->getDomain());
        $document->setField('siteHash', $site->getSiteHash());
        $document->setField('appKey', 'EXT:solr');
        $document->setField('type', 'pages');

        // system fields
        $document->setField('uid', $this->page->id);
        $document->setField('pid', $pageRecord['pid']);

        // variantId
        $variantId = $this->variantIdBuilder->buildFromTypeAndUid('pages', $this->page->id);
        $document->setField('variantId', $variantId);

        $document->setField('typeNum', $this->page->type);
        $document->setField('created', $pageRecord['crdate']);
        $document->setField('changed', $pageRecord['SYS_LASTCHANGED']);

        $rootline = $this->getRootLineFieldValue();
        $document->setField('rootline', $rootline);

        // access
        $this->addAccessField($document);
        $this->addEndtimeField($document, $pageRecord);

        // content
        $document->setField('title', $this->contentExtractor->getPageTitle());
        $document->setField('subTitle', $pageRecord['subtitle']);
        $document->setField('navTitle', $pageRecord['nav_title']);
        $document->setField('author', $pageRecord['author']);
        $document->setField('description', $pageRecord['description']);
        $document->setField('abstract', $pageRecord['abstract']);
        $document->setField('content', $this->contentExtractor->getIndexableContent());
        $document->setField('url', $this->pageUrl);

        $this->addKeywordsField($document, $pageRecord);
        $this->addTagContentFields($document);

        return $document;
    }

    /**
     * Adds the access field to the document if needed.
     *
     * @param \Apache_Solr_Document $document
     */
    protected function addAccessField(\Apache_Solr_Document $document)
    {
        $access = (string)$this->pageAccessRootline;
        if (trim($access) !== '') {
            $document->setField('access', $access);
        }
    }

    /**
     * @param $document
     * @param $pageRecord
     */
    protected function addEndtimeField(\Apache_Solr_Document  $document, $pageRecord)
    {
        if ($this->page->page['endtime']) {
            $document->setField('endtime', $pageRecord['endtime']);
        }
    }

    /**
     * Adds keywords, multi valued.
     *
     * @param \Apache_Solr_Document $document
     * @param array $pageRecord
     */
    protected function addKeywordsField(\Apache_Solr_Document $document, $pageRecord)
    {
        $keywords = array_unique(GeneralUtility::trimExplode(',', $pageRecord['keywords'], true));
        foreach ($keywords as $keyword) {
            $document->addField('keywords', $keyword);
        }
    }

    /**
     * Add content from several tags like headers, anchors, ...
     *
     * @param \Apache_Solr_Document $document
     */
    protected function addTagContentFields(\Apache_Solr_Document  $document)
    {
        $tagContent = $this->contentExtractor->getTagContent();
        foreach ($tagContent as $fieldName => $fieldValue) {
            $document->setField($fieldName, $fieldValue);
        }
    }

    /**
     * Builds the content for the rootline field.
     *
     * @return string
     */
    protected function getRootLineFieldValue()
    {
        $rootline = $this->page->id;
        $mountPointParameter = $this->getMountPointParameter();
        if ($mountPointParameter !== '') {
            $rootline .= ',' . $mountPointParameter;
        }
        return $rootline;
    }

    /**
     * Gets a comma separated list of frontend user groups to use for the
     * document ID.
     *
     * @return string A comma separated list of frontend user groups.
     */
    protected function getDocumentIdGroups()
    {
        $groups = $this->pageAccessRootline->getGroups();
        $groups = Rootline::cleanGroupArray($groups);

        if (empty($groups)) {
            $groups[] = 0;
        }

        $groups = implode(',', $groups);

        return $groups;
    }

    // Logging
    // TODO replace by a central logger

    /**
     * Gets the mount point parameter that is used in the Frontend controller.
     *
     * @return string
     */
    public function getMountPointParameter()
    {
        return $this->mountPointParameter;
    }

    // Misc

    /**
     * Sets the mount point parameter that is used in the Frontend controller.
     *
     * @param string $mountPointParameter
     */
    public function setMountPointParameter($mountPointParameter)
    {
        $this->mountPointParameter = (string)$mountPointParameter;
    }

    /**
     * Allows third party extensions to replace or modify the page document
     * created by this indexer.
     *
     * @param \Apache_Solr_Document $pageDocument The page document created by this indexer.
     * @return \Apache_Solr_Document An Apache Solr document representing the currently indexed page
     */
    protected function substitutePageDocument(\Apache_Solr_Document $pageDocument)
    {
        if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
            return $pageDocument;
        }

        $indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
        foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
            $substituteIndexer = GeneralUtility::getUserObj($classReference);

            if (!$substituteIndexer instanceof SubstitutePageIndexer) {
                $message = get_class($substituteIndexer) . ' must implement interface ' . SubstitutePageIndexer::class;
                throw new \UnexpectedValueException($message, 1310491001);
            }

            if ($substituteIndexer instanceof PageFieldMappingIndexer) {
                $substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
            }

            $substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
            if (!$substituteDocument instanceof \Apache_Solr_Document) {

The class Apache_Solr_Document does not exist. Is this class maybe located in a folder that is not analyzed, or in a newer version of your dependencies than listed in your composer.lock/composer.json?

                $message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Apache_Solr_Document document.';
                throw new \UnexpectedValueException($message, 1310490952);
            }
            $pageDocument = $substituteDocument;
        }

        return $pageDocument;
    }

    /**
     * Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
     *
     * @return string
     */
    protected function getIndexConfigurationNameForCurrentPage()
    {
        return isset($this->indexQueueItem) ? $this->indexQueueItem->getIndexingConfigurationName() : 'pages';
    }

    /**
     * Allows third party extensions to provide additional documents which
     * should be indexed for the current page.
     *
     * @param \Apache_Solr_Document $pageDocument The main document representing this page.
     * @param \Apache_Solr_Document[] $existingDocuments An array of documents already created for this page.
     * @return array An array of additional \Apache_Solr_Document objects to index
     */
    protected function getAdditionalDocuments(\Apache_Solr_Document $pageDocument, array $existingDocuments)
    {
        $documents = $existingDocuments;

        if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
            return $documents;
        }

        foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
            $additionalIndexer = GeneralUtility::getUserObj($classReference);

            if (!$additionalIndexer instanceof AdditionalPageIndexer) {
                $message = get_class($additionalIndexer) . ' must implement interface ' . AdditionalPageIndexer::class;
                throw new \UnexpectedValueException($message, 1310491024);
            }

            $additionalDocuments = $additionalIndexer->getAdditionalPageDocuments($pageDocument, $documents);
            if (is_array($additionalDocuments)) {
                $documents = array_merge($documents, $additionalDocuments);
            }
        }

        return $documents;
    }

    /**
     * Sends the given documents to the field processing service which takes
     * care of manipulating fields as defined in the field's configuration.
     *
     * @param array $documents An array of documents to manipulate
     */
    protected function processDocuments(array $documents)
    {
        $processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
        if (count($processingInstructions) > 0) {
            $service = GeneralUtility::makeInstance(Service::class);
            $service->processDocuments($documents, $processingInstructions);
        }
    }

    /**
     * Adds the collected documents to the Solr index.
     *
     * @param array $documents An array of \Apache_Solr_Document objects.
     * @return bool TRUE if documents were added successfully, FALSE otherwise
     */
    protected function addDocumentsToSolrIndex(array $documents)
    {
        $documentsAdded = false;

        if (!count($documents)) {
            return $documentsAdded;
        }

        try {
            $this->log('Adding ' . count($documents) . ' documents.', 0, $documents);

            // chunk adds by 20
            $documentChunks = array_chunk($documents, 20);
            foreach ($documentChunks as $documentChunk) {
                $response = $this->solrConnection->addDocuments($documentChunk);

                if ($response->getHttpStatus() != 200) {
                    $transportException = new \Apache_Solr_HttpTransportException($response);
                    throw new \RuntimeException('Solr Request failed.', 1331834983, $transportException);
                }
            }

            $documentsAdded = true;
        } catch (\Exception $e) {
            $this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 2);

            if ($this->configuration->getLoggingExceptions()) {
                GeneralUtility::devLog('Exception while adding documents', 'solr', 3, [$e->__toString()]);
            }
        }

        return $documentsAdded;
    }

    /**
     * Gets the current page's URL.
     *
     * @return string URL of the current page.
     */
    public function getPageUrl()
    {
        return $this->pageUrl;
    }

    /**
     * Sets the URL to use for the page document.
     *
     * @param string $url The page's URL.
     */
    public function setPageUrl($url)
    {
        $this->pageUrl = $url;
    }

    /**
     * Gets the page's access rootline.
     *
     * @return Rootline The page's access rootline
     */
    public function getPageAccessRootline()
    {
        return $this->pageAccessRootline;
    }

    /**
     * Sets the page's access rootline.
     *
     * @param Rootline $accessRootline The page's access rootline
     */
    public function setPageAccessRootline(Rootline $accessRootline)
    {
        $this->pageAccessRootline = $accessRootline;
    }

    /**
     * Gets the documents that have been sent to Solr
     *
     * @return array An array of \Apache_Solr_Document objects
     */
    public function getDocumentsSentToSolr()
    {
        return $this->documentsSentToSolr;
    }
}