Typo3PageIndexer::getPageDocument() - Code Metrics - Inspection of "Merge pull request #1086 from thomashohn/task/mast..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 48f77b...329461 )

by Timo

created 2017-02-25 09:49 UTC

Typo3PageIndexer::getPageDocument() A

↳ Parent: Typo3PageIndexer

Complexity

Conditions	1
Paths	1

Size

Total Lines	54
Code Lines	37

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	37
CRAP Score	1.0004

Importance

Changes

Metric	Value
dl	0
loc	54
ccs	37
cts	40
cp	0.925
rs	9.6716
c	0
b	0
f	0
cc	1
eloc	37
nc	1
nop	0
crap	1.0004

How to fix Long Method

<?php
namespace ApacheSolrForTypo3\Solr;

/***************************************************************
 *  Copyright notice
 *
 *  (c) 2009-2015 Ingo Renner <[email protected]>
 *  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 ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager;
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 <[email protected]>
 * @author Daniel Poetzinger <[email protected]>
 * @author Timo Schmidt <[email protected]>
 */
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;

    /**
     * @var \ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager
     */
    protected $logger = null;

    /**
     * Constructor
     *
     * @param TypoScriptFrontendController $page The page to index
     * @param IdBuilder $variantIdBuilder
     */
    public function __construct(TypoScriptFrontendController $page, IdBuilder $variantIdBuilder = null)
    {
        $this->logger = GeneralUtility::makeInstance(SolrLogManager::class, __CLASS__);

        $this->page = $page;
        $this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
        $this->configuration = Util::getSolrConfiguration();

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

            // TODO extract to a class "ExceptionLogger"
            if ($this->configuration->getLoggingExceptions()) {
                $this->logger->log(
                    SolrLogManager::ERROR,
                    'Exception while trying to index a page',
                    [
                        $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;
    }

    /**
     * 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) {

                $message = 'The document returned by ' . get_class($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 {
            $this->logger->log(
                SolrLogManager::INFO,
                'Adding ' . count($documents) . ' documents.',
                $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->logger->log(
                SolrLogManager::ERROR,
                $e->getMessage() . ' Error code: ' . $e->getCode()
            );

            if ($this->configuration->getLoggingExceptions()) {
                $this->logger->log(
                    SolrLogManager::ERROR,
                    'Exception while adding documents',
                    [
                        $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;
    }
}


1		<?php
2		namespace ApacheSolrForTypo3\Solr;
3
4		/***************************************************************
5		* Copyright notice
6		*
7		* (c) 2009-2015 Ingo Renner <[email protected]>
8		* All rights reserved
9		*
10		* This script is part of the TYPO3 project. The TYPO3 project is
11		* free software; you can redistribute it and/or modify
12		* it under the terms of the GNU General Public License as published by
13		* the Free Software Foundation; either version 2 of the License, or
14		* (at your option) any later version.
15		*
16		* The GNU General Public License can be found at
17		* http://www.gnu.org/copyleft/gpl.html.
18		*
19		* This script is distributed in the hope that it will be useful,
20		* but WITHOUT ANY WARRANTY; without even the implied warranty of
21		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22		* GNU General Public License for more details.
23		*
24		* This copyright notice MUST APPEAR in all copies of the script!
25		***************************************************************/
26
27		use Apache_Solr_Document;
28		use ApacheSolrForTypo3\Solr\Access\Rootline;
29		use ApacheSolrForTypo3\Solr\Domain\Variants\IdBuilder;
30		use ApacheSolrForTypo3\Solr\FieldProcessor\Service;
31		use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
32		use ApacheSolrForTypo3\Solr\IndexQueue\Item;
33		use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration;
34		use ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager;
35		use TYPO3\CMS\Core\Utility\GeneralUtility;
36		use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;
37
38		/**
39		* Page Indexer to index TYPO3 pages used by the Index Queue.
40		*
41		* @author Ingo Renner <[email protected]>
42		* @author Daniel Poetzinger <[email protected]>
43		* @author Timo Schmidt <[email protected]>
44		*/
45		class Typo3PageIndexer
46		{
47
48		/**
49		* ID of the current page's Solr document.
50		*
51		* @var string
52		*/
53		protected static $pageSolrDocumentId = '';
54		/**
55		* The Solr document generated for the current page.
56		*
57		* @var \Apache_Solr_Document
58		*/
59		protected static $pageSolrDocument = null;
60		/**
61		* The mount point parameter used in the Frontend controller.
62		*
63		* @var string
64		*/
65		protected $mountPointParameter;
66		/**
67		* Solr server connection.
68		*
69		* @var SolrService
70		*/
71		protected $solrConnection = null;
72		/**
73		* Frontend page object (TSFE).
74		*
75		* @var TypoScriptFrontendController
76		*/
77		protected $page = null;
78		/**
79		* Content extractor to extract content from TYPO3 pages
80		*
81		* @var Typo3PageContentExtractor
82		*/
83		protected $contentExtractor = null;
84		/**
85		* URL to be indexed as the page's URL
86		*
87		* @var string
88		*/
89		protected $pageUrl = '';
90		/**
91		* The page's access rootline
92		*
93		* @var Rootline
94		*/
95		protected $pageAccessRootline = null;
96		/**
97		* Documents that have been sent to Solr
98		*
99		* @var array
100		*/
101		protected $documentsSentToSolr = [];
102
103		/**
104		* @var TypoScriptConfiguration
105		*/
106		protected $configuration;
107
108		/**
109		* @var Item
110		*/
111		protected $indexQueueItem;
112
113		/**
114		* @var IdBuilder
115		*/
116		protected $variantIdBuilder;
117
118		/**
119		* @var \ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager
120		*/
121		protected $logger = null;
122
123		/**
124		* Constructor
125		*
126		* @param TypoScriptFrontendController $page The page to index
127		* @param IdBuilder $variantIdBuilder
128		*/
129	42	public function __construct(TypoScriptFrontendController $page, IdBuilder $variantIdBuilder = null)
130		{
131	42	$this->logger = GeneralUtility::makeInstance(SolrLogManager::class, __CLASS__);
132
133	42	$this->page = $page;
134	42	$this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
135	42	$this->configuration = Util::getSolrConfiguration();
136
137		try {
138	42	$this->initializeSolrConnection();
139		} catch (\Exception $e) {
140		$this->logger->log(
141		SolrLogManager::ERROR,
142		$e->getMessage() . ' Error code: ' . $e->getCode()
143		);
144
145		// TODO extract to a class "ExceptionLogger"
146		if ($this->configuration->getLoggingExceptions()) {
147		$this->logger->log(
148		SolrLogManager::ERROR,
149		'Exception while trying to index a page',
150		[
151		$e->__toString()
152		]
153		);
154		}
155		}
156
157	42	$this->contentExtractor = GeneralUtility::makeInstance(Typo3PageContentExtractor::class, $this->page->content);
158	42	$this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, '');
159	42	$this->variantIdBuilder = is_null($variantIdBuilder) ? GeneralUtility::makeInstance(IdBuilder::class) : $variantIdBuilder;
160	42	}
161
162		/**
163		* @param Item $indexQueueItem
164		*/
165	6	public function setIndexQueueItem($indexQueueItem)
166		{
167	6	$this->indexQueueItem = $indexQueueItem;
168	6	}
169
170
171		/**
172		* Initializes the Solr server connection.
173		*
174		* @throws \Exception when no Solr connection can be established.
175		*/
176	42	protected function initializeSolrConnection()
177		{
178	42	$solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, $this->page->sys_language_uid);
179
180		// do not continue if no server is available
181	42	if (!$solr->ping()) {
182		throw new \Exception(
183		'No Solr instance available while trying to index a page.',
184		1234790825
185		);
186		}
187
188	42	$this->solrConnection = $solr;
189	42	}
190
191		/**
192		* Gets the current page's Solr document ID.
193		*
194		* @return string\|NULL The page's Solr document ID or NULL in case no document was generated yet.
195		*/
196		public static function getPageSolrDocumentId()
197		{
198		return self::$pageSolrDocumentId;
199		}
200
201		/**
202		* Gets the Solr document generated for the current page.
203		*
204		* @return \Apache_Solr_Document\|NULL The page's Solr document or NULL if it has not been generated yet.
205		*/
206	6	public static function getPageSolrDocument()
207		{
208	6	return self::$pageSolrDocument;
209		}
210
211		/**
212		* Allows to provide a Solr server connection other than the one
213		* initialized by the constructor.
214		*
215		* @param SolrService $solrConnection Solr connection
216		* @throws \Exception if the Solr server cannot be reached
217		*/
218	6	public function setSolrConnection(SolrService $solrConnection)
219		{
220	6	if (!$solrConnection->ping()) {
221		throw new \Exception(
222		'Could not connect to Solr server.',
223		1323946472
224		);
225		}
226
227	6	$this->solrConnection = $solrConnection;
228	6	}
229
230		/**
231		* Indexes a page.
232		*
233		* @return bool TRUE after successfully indexing the page, FALSE on error
234		* @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
235		*/
236	42	public function indexPage()
237		{
238	42	$pageIndexed = false;
239	42	$documents = []; // this will become useful as soon as when starting to index individual records instead of whole pages
240
241	42	if (is_null($this->solrConnection)) {
242		// intended early return as it doesn't make sense to continue
243		// and waste processing time if the solr server isn't available
244		// anyways
245		// FIXME use an exception
246		return $pageIndexed;
247		}
248
249	42	$pageDocument = $this->getPageDocument();
250	42	$pageDocument = $this->substitutePageDocument($pageDocument);
251
252	42	$this->applyIndexPagePostProcessors($pageDocument);
253
254	42	self::$pageSolrDocument = $pageDocument;
255	42	$documents[] = $pageDocument;
256	42	$documents = $this->getAdditionalDocuments($pageDocument, $documents);
257	42	$this->processDocuments($documents);
258
259	42	$pageIndexed = $this->addDocumentsToSolrIndex($documents);
260	42	$this->documentsSentToSolr = $documents;
261
262	42	return $pageIndexed;
263		}
264
265		/**
266		* Applies the configured post processors (indexPagePostProcessPageDocument)
267		*
268		* @param \Apache_Solr_Document $pageDocument
269		*/
270	42	protected function applyIndexPagePostProcessors($pageDocument)
271		{
272	42	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
273	41	return;
274		}
275
276	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
277	1	$postProcessor = GeneralUtility::getUserObj($classReference);
278	1	if (!$postProcessor instanceof PageDocumentPostProcessor) {
279		throw new \UnexpectedValueException(get_class($pageDocument) . ' must implement interface ' . PageDocumentPostProcessor::class, 1397739154);
280		}
281
282	1	$postProcessor->postProcessPageDocument($pageDocument, $this->page);
283		}
284	1	}
285
286		/**
287		* Builds the Solr document for the current page.
288		*
289		* @return \Apache_Solr_Document A document representing the page
290		*/
291	42	protected function getPageDocument()
292		{
293	42	$document = GeneralUtility::makeInstance(Apache_Solr_Document::class);
294		/* @var $document \Apache_Solr_Document */
295	42	$site = Site::getSiteByPageId($this->page->id);
296	42	$pageRecord = $this->page->page;
297
298	42	self::$pageSolrDocumentId = $documentId = Util::getPageDocumentId(
299	42	$this->page->id,
300	42	$this->page->type,
301	42	$this->page->sys_language_uid,
302	42	$this->getDocumentIdGroups(),
303	42	$this->getMountPointParameter()
304		);
305	42	$document->setField('id', $documentId);
306	42	$document->setField('site', $site->getDomain());
307	42	$document->setField('siteHash', $site->getSiteHash());
308	42	$document->setField('appKey', 'EXT:solr');
309	42	$document->setField('type', 'pages');
310
311		// system fields
312	42	$document->setField('uid', $this->page->id);
313	42	$document->setField('pid', $pageRecord['pid']);
314
315		// variantId
316	42	$variantId = $this->variantIdBuilder->buildFromTypeAndUid('pages', $this->page->id);
317	42	$document->setField('variantId', $variantId);
318
319	42	$document->setField('typeNum', $this->page->type);
320	42	$document->setField('created', $pageRecord['crdate']);
321	42	$document->setField('changed', $pageRecord['SYS_LASTCHANGED']);
322
323	42	$rootline = $this->getRootLineFieldValue();
324	42	$document->setField('rootline', $rootline);
325
326		// access
327	42	$this->addAccessField($document);
328	42	$this->addEndtimeField($document, $pageRecord);
329
330		// content
331	42	$document->setField('title', $this->contentExtractor->getPageTitle());
332	42	$document->setField('subTitle', $pageRecord['subtitle']);
333	42	$document->setField('navTitle', $pageRecord['nav_title']);
334	42	$document->setField('author', $pageRecord['author']);
335	42	$document->setField('description', $pageRecord['description']);
336	42	$document->setField('abstract', $pageRecord['abstract']);
337	42	$document->setField('content', $this->contentExtractor->getIndexableContent());
338	42	$document->setField('url', $this->pageUrl);
339
340	42	$this->addKeywordsField($document, $pageRecord);
341	42	$this->addTagContentFields($document);
342
343	42	return $document;
344		}
345
346		/**
347		* Adds the access field to the document if needed.
348		*
349		* @param \Apache_Solr_Document $document
350		*/
351	42	protected function addAccessField(\Apache_Solr_Document $document)
352		{
353	42	$access = (string)$this->pageAccessRootline;
354	42	if (trim($access) !== '') {
355	9	$document->setField('access', $access);
356		}
357	42	}
358
359		/**
360		* @param $document
361		* @param $pageRecord
362		*/
363	42	protected function addEndtimeField(\Apache_Solr_Document $document, $pageRecord)
364		{
365	42	if ($this->page->page['endtime']) {
366		$document->setField('endtime', $pageRecord['endtime']);
367		}
368	42	}
369
370		/**
371		* Adds keywords, multi valued.
372		*
373		* @param \Apache_Solr_Document $document
374		* @param array $pageRecord
375		*/
376	42	protected function addKeywordsField(\Apache_Solr_Document $document, $pageRecord)
377		{
378	42	$keywords = array_unique(GeneralUtility::trimExplode(',', $pageRecord['keywords'], true));
379	42	foreach ($keywords as $keyword) {
380		$document->addField('keywords', $keyword);
381		}
382	42	}
383
384		/**
385		* Add content from several tags like headers, anchors, ...
386		*
387		* @param \Apache_Solr_Document $document
388		*/
389	42	protected function addTagContentFields(\Apache_Solr_Document $document)
390		{
391	42	$tagContent = $this->contentExtractor->getTagContent();
392	42	foreach ($tagContent as $fieldName => $fieldValue) {
393		$document->setField($fieldName, $fieldValue);
394		}
395	42	}
396
397		/**
398		* Builds the content for the rootline field.
399		*
400		* @return string
401		*/
402	42	protected function getRootLineFieldValue()
403		{
404	42	$rootline = $this->page->id;
405	42	$mountPointParameter = $this->getMountPointParameter();
406	42	if ($mountPointParameter !== '') {
407	36	$rootline .= ',' . $mountPointParameter;
408		}
409	42	return $rootline;
410		}
411
412		/**
413		* Gets a comma separated list of frontend user groups to use for the
414		* document ID.
415		*
416		* @return string A comma separated list of frontend user groups.
417		*/
418	42	protected function getDocumentIdGroups()
419		{
420	42	$groups = $this->pageAccessRootline->getGroups();
421	42	$groups = Rootline::cleanGroupArray($groups);
422
423	42	if (empty($groups)) {
424	35	$groups[] = 0;
425		}
426
427	42	$groups = implode(',', $groups);
428
429	42	return $groups;
430		}
431
432		// Logging
433		// TODO replace by a central logger
434
435		/**
436		* Gets the mount point parameter that is used in the Frontend controller.
437		*
438		* @return string
439		*/
440	42	public function getMountPointParameter()
441		{
442	42	return $this->mountPointParameter;
443		}
444
445		// Misc
446
447		/**
448		* Sets the mount point parameter that is used in the Frontend controller.
449		*
450		* @param string $mountPointParameter
451		*/
452	6	public function setMountPointParameter($mountPointParameter)
453		{
454	6	$this->mountPointParameter = (string)$mountPointParameter;
455	6	}
456
457		/**
458		* Allows third party extensions to replace or modify the page document
459		* created by this indexer.
460		*
461		* @param \Apache_Solr_Document $pageDocument The page document created by this indexer.
462		* @return \Apache_Solr_Document An Apache Solr document representing the currently indexed page
463		*/
464	42	protected function substitutePageDocument(\Apache_Solr_Document $pageDocument)
465		{
466	42	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
467	36	return $pageDocument;
468		}
469
470	6	$indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
471	6	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
472	6	$substituteIndexer = GeneralUtility::getUserObj($classReference);
473
474	6	if (!$substituteIndexer instanceof SubstitutePageIndexer) {
475		$message = get_class($substituteIndexer) . ' must implement interface ' . SubstitutePageIndexer::class;
476		throw new \UnexpectedValueException($message, 1310491001);
477		}
478
479	6	if ($substituteIndexer instanceof PageFieldMappingIndexer) {
480	6	$substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
481		}
482
483	6	$substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
484	6	if (!$substituteDocument instanceof Apache_Solr_Document) {
		0 ignored issues – show Bug introduced 2017-01-13 06:22 UTC by Report Bug Copy Issue Report 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? Loading history...
485		$message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Apache_Solr_Document document.';
486		throw new \UnexpectedValueException($message, 1310490952);
487		}
488	6	$pageDocument = $substituteDocument;
489		}
490
491	6	return $pageDocument;
492		}
493
494		/**
495		* Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
496		*
497		* @return string
498		*/
499	6	protected function getIndexConfigurationNameForCurrentPage()
500		{
501	6	return isset($this->indexQueueItem) ? $this->indexQueueItem->getIndexingConfigurationName() : 'pages';
502		}
503
504		/**
505		* Allows third party extensions to provide additional documents which
506		* should be indexed for the current page.
507		*
508		* @param \Apache_Solr_Document $pageDocument The main document representing this page.
509		* @param \Apache_Solr_Document[] $existingDocuments An array of documents already created for this page.
510		* @return array An array of additional \Apache_Solr_Document objects to index
511		*/
512	42	protected function getAdditionalDocuments(\Apache_Solr_Document $pageDocument, array $existingDocuments)
513		{
514	42	$documents = $existingDocuments;
515
516	42	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
517	41	return $documents;
518		}
519
520	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
521	1	$additionalIndexer = GeneralUtility::getUserObj($classReference);
522
523	1	if (!$additionalIndexer instanceof AdditionalPageIndexer) {
524		$message = get_class($additionalIndexer) . ' must implement interface ' . AdditionalPageIndexer::class;
525		throw new \UnexpectedValueException($message, 1310491024);
526		}
527
528	1	$additionalDocuments = $additionalIndexer->getAdditionalPageDocuments($pageDocument, $documents);
529	1	if (is_array($additionalDocuments)) {
530	1	$documents = array_merge($documents, $additionalDocuments);
531		}
532		}
533
534	1	return $documents;
535		}
536
537		/**
538		* Sends the given documents to the field processing service which takes
539		* care of manipulating fields as defined in the field's configuration.
540		*
541		* @param array $documents An array of documents to manipulate
542		*/
543	42	protected function processDocuments(array $documents)
544		{
545	42	$processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
546	42	if (count($processingInstructions) > 0) {
547	42	$service = GeneralUtility::makeInstance(Service::class);
548	42	$service->processDocuments($documents, $processingInstructions);
549		}
550	42	}
551
552		/**
553		* Adds the collected documents to the Solr index.
554		*
555		* @param array $documents An array of \Apache_Solr_Document objects.
556		* @return bool TRUE if documents were added successfully, FALSE otherwise
557		*/
558	42	protected function addDocumentsToSolrIndex(array $documents)
559		{
560	42	$documentsAdded = false;
561
562	42	if (!count($documents)) {
563		return $documentsAdded;
564		}
565
566		try {
567	42	$this->logger->log(
568	42	SolrLogManager::INFO,
569	42	'Adding ' . count($documents) . ' documents.',
570		$documents
571		);
572
573		// chunk adds by 20
574	42	$documentChunks = array_chunk($documents, 20);
575	42	foreach ($documentChunks as $documentChunk) {
576	42	$response = $this->solrConnection->addDocuments($documentChunk);
577
578	42	if ($response->getHttpStatus() != 200) {
579		$transportException = new \Apache_Solr_HttpTransportException($response);
580	42	throw new \RuntimeException('Solr Request failed.', 1331834983, $transportException);
581		}
582		}
583
584	42	$documentsAdded = true;
585		} catch (\Exception $e) {
586		$this->logger->log(
587		SolrLogManager::ERROR,
588		$e->getMessage() . ' Error code: ' . $e->getCode()
589		);
590
591		if ($this->configuration->getLoggingExceptions()) {
592		$this->logger->log(
593		SolrLogManager::ERROR,
594		'Exception while adding documents',
595		[
596		$e->__toString()
597		]
598		);
599		}
600		}
601
602	42	return $documentsAdded;
603		}
604
605		/**
606		* Gets the current page's URL.
607		*
608		* @return string URL of the current page.
609		*/
610		public function getPageUrl()
611		{
612		return $this->pageUrl;
613		}
614
615		/**
616		* Sets the URL to use for the page document.
617		*
618		* @param string $url The page's URL.
619		*/
620	6	public function setPageUrl($url)
621		{
622	6	$this->pageUrl = $url;
623	6	}
624
625		/**
626		* Gets the page's access rootline.
627		*
628		* @return Rootline The page's access rootline
629		*/
630		public function getPageAccessRootline()
631		{
632		return $this->pageAccessRootline;
633		}
634
635		/**
636		* Sets the page's access rootline.
637		*
638		* @param Rootline $accessRootline The page's access rootline
639		*/
640	41	public function setPageAccessRootline(Rootline $accessRootline)
641		{
642	41	$this->pageAccessRootline = $accessRootline;
643	41	}
644
645		/**
646		* Gets the documents that have been sent to Solr
647		*
648		* @return array An array of \Apache_Solr_Document objects
649		*/
650	6	public function getDocumentsSentToSolr()
651		{
652	6	return $this->documentsSentToSolr;
653		}
654		}
655

TYPO3-Solr / ext-solr

Push — master ( 48f77b...329461 )

Typo3PageIndexer::getPageDocument() A

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Long Method

Long Method

Duplication Side-by-Side

Filter issues like