Typo3PageIndexer::indexPage() - Code Metrics - Inspection of "Merge pull request #1088 from timohund/bugfix/mast..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 4904bb...48f77b )

by Timo

created 2017-02-24 17:40 UTC

Typo3PageIndexer::indexPage() B

↳ Parent: Typo3PageIndexer

Complexity

Conditions	2
Paths	2

Size

Total Lines	28
Code Lines	15

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	14
CRAP Score	2.0438

Importance

Changes

Metric	Value
dl	0
loc	28
ccs	14
cts	18
cp	0.7778
rs	8.8571
c	0
b	0
f	0
cc	2
eloc	15
nc	2
nop	0
crap	2.0438

<?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 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;

    /**
     * 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) {

                $message = 'The document returned by ' . get_class($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;
    }
}


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

TYPO3-Solr / ext-solr

Push — master ( 4904bb...48f77b )

Typo3PageIndexer::indexPage() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like