Typo3PageIndexer::getPageDocument() - Code Metrics - Inspection of "Merge pull request #914 from timohund/task/master/..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Branch — master (b9fc31)

by Timo

created 2017-01-23 12:02 UTC

Typo3PageIndexer::getPageDocument() A

↳ Parent: Typo3PageIndexer

Complexity

Conditions	1
Paths	1

Size

Total Lines	53
Code Lines	36

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	37
CRAP Score	1.0001

Importance

Changes

Metric	Value
dl	0
loc	53
ccs	37
cts	39
cp	0.9487
rs	9.5797
c	0
b	0
f	0
cc	1
eloc	36
nc	1
nop	0
crap	1.0001

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\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;

    /**
     * Constructor
     *
     * @param TypoScriptFrontendController $page The page to index
     */
    public function __construct(TypoScriptFrontendController $page)
    {
        $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, '');
    }

    /**
     * @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
        $document->setField('variantId', 'pages/' . $this->page->id);

        $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\FieldProcessor\Service;
30		use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
31		use ApacheSolrForTypo3\Solr\IndexQueue\Item;
32		use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration;
33		use TYPO3\CMS\Core\Utility\GeneralUtility;
34		use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;
35
36		/**
37		* Page Indexer to index TYPO3 pages used by the Index Queue.
38		*
39		* @author Ingo Renner <[email protected]>
40		* @author Daniel Poetzinger <[email protected]>
41		* @author Timo Schmidt <[email protected]>
42		*/
43		class Typo3PageIndexer
44		{
45
46		/**
47		* ID of the current page's Solr document.
48		*
49		* @var string
50		*/
51		protected static $pageSolrDocumentId = '';
52		/**
53		* The Solr document generated for the current page.
54		*
55		* @var \Apache_Solr_Document
56		*/
57		protected static $pageSolrDocument = null;
58		/**
59		* The mount point parameter used in the Frontend controller.
60		*
61		* @var string
62		*/
63		protected $mountPointParameter;
64		/**
65		* Solr server connection.
66		*
67		* @var SolrService
68		*/
69		protected $solrConnection = null;
70		/**
71		* Frontend page object (TSFE).
72		*
73		* @var TypoScriptFrontendController
74		*/
75		protected $page = null;
76		/**
77		* Content extractor to extract content from TYPO3 pages
78		*
79		* @var Typo3PageContentExtractor
80		*/
81		protected $contentExtractor = null;
82		/**
83		* URL to be indexed as the page's URL
84		*
85		* @var string
86		*/
87		protected $pageUrl = '';
88		/**
89		* The page's access rootline
90		*
91		* @var Rootline
92		*/
93		protected $pageAccessRootline = null;
94		/**
95		* Documents that have been sent to Solr
96		*
97		* @var array
98		*/
99		protected $documentsSentToSolr = [];
100
101		/**
102		* @var TypoScriptConfiguration
103		*/
104		protected $configuration;
105
106		/**
107		* @var Item
108		*/
109		protected $indexQueueItem;
110
111		/**
112		* Constructor
113		*
114		* @param TypoScriptFrontendController $page The page to index
115		*/
116	35	public function __construct(TypoScriptFrontendController $page)
117		{
118	35	$this->page = $page;
119	35	$this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
120	35	$this->configuration = Util::getSolrConfiguration();
121
122		try {
123	35	$this->initializeSolrConnection();
124	35	} catch (\Exception $e) {
125		$this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 3);
126
127		// TODO extract to a class "ExceptionLogger"
128		if ($this->configuration->getLoggingExceptions()) {
129		GeneralUtility::devLog('Exception while trying to index a page', 'solr', 3, [$e->__toString()]);
130		}
131		}
132
133	35	$this->contentExtractor = GeneralUtility::makeInstance(Typo3PageContentExtractor::class, $this->page->content);
134	35	$this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, '');
135	35	}
136
137		/**
138		* @param Item $indexQueueItem
139		*/
140	5	public function setIndexQueueItem($indexQueueItem)
141		{
142	5	$this->indexQueueItem = $indexQueueItem;
143	5	}
144
145
146		/**
147		* Initializes the Solr server connection.
148		*
149		* @throws \Exception when no Solr connection can be established.
150		*/
151	35	protected function initializeSolrConnection()
152		{
153	35	$solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, $this->page->sys_language_uid);
154
155		// do not continue if no server is available
156	35	if (!$solr->ping()) {
157		throw new \Exception(
158		'No Solr instance available while trying to index a page.',
159		1234790825
160		);
161		}
162
163	35	$this->solrConnection = $solr;
164	35	}
165
166		/**
167		* Logs messages to devlog and TS log (admin panel)
168		*
169		* @param string $message Message to set
170		* @param int $errorNum Error number
171		* @param array $data Additional data to log
172		* @return void
173		*/
174	35	protected function log($message, $errorNum = 0, array $data = [])
175		{
176	35	if (is_object($GLOBALS['TT'])) {
177	35	$GLOBALS['TT']->setTSlogMessage('tx_solr: ' . $message, $errorNum);
178	35	}
179
180	35	if ($this->configuration->getLoggingIndexing()) {
181		$logData = [];
182		if (!empty($data)) {
183		foreach ($data as $value) {
184	8	$logData[] = (array)$value;
185		}
186		}
187
188		GeneralUtility::devLog($message, 'solr', $errorNum, $logData);
189		}
190	35	}
191
192		/**
193		* Gets the current page's Solr document ID.
194		*
195		* @return string\|NULL The page's Solr document ID or NULL in case no document was generated yet.
196		*/
197		public static function getPageSolrDocumentId()
198		{
199		return self::$pageSolrDocumentId;
200		}
201
202		/**
203		* Gets the Solr document generated for the current page.
204		*
205		* @return \Apache_Solr_Document\|NULL The page's Solr document or NULL if it has not been generated yet.
206		*/
207	5	public static function getPageSolrDocument()
208		{
209	5	return self::$pageSolrDocument;
210		}
211
212		/**
213		* Allows to provide a Solr server connection other than the one
214		* initialized by the constructor.
215		*
216		* @param SolrService $solrConnection Solr connection
217		* @throws \Exception if the Solr server cannot be reached
218		*/
219	5	public function setSolrConnection(SolrService $solrConnection)
220		{
221	5	if (!$solrConnection->ping()) {
222		throw new \Exception(
223		'Could not connect to Solr server.',
224		1323946472
225		);
226		}
227
228	5	$this->solrConnection = $solrConnection;
229	5	}
230
231		/**
232		* Indexes a page.
233		*
234		* @return bool TRUE after successfully indexing the page, FALSE on error
235		* @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
236		*/
237	35	public function indexPage()
238		{
239	35	$pageIndexed = false;
240	35	$documents = []; // this will become useful as soon as when starting to index individual records instead of whole pages
241
242	35	if (is_null($this->solrConnection)) {
243		// intended early return as it doesn't make sense to continue
244		// and waste processing time if the solr server isn't available
245		// anyways
246		// FIXME use an exception
247		return $pageIndexed;
248		}
249
250	35	$pageDocument = $this->getPageDocument();
251	35	$pageDocument = $this->substitutePageDocument($pageDocument);
252
253	35	$this->applyIndexPagePostProcessors($pageDocument);
254
255	35	self::$pageSolrDocument = $pageDocument;
256	35	$documents[] = $pageDocument;
257	35	$documents = $this->getAdditionalDocuments($pageDocument, $documents);
258	35	$this->processDocuments($documents);
259
260	35	$pageIndexed = $this->addDocumentsToSolrIndex($documents);
261	35	$this->documentsSentToSolr = $documents;
262
263	35	return $pageIndexed;
264		}
265
266		/**
267		* Applies the configured post processors (indexPagePostProcessPageDocument)
268		*
269		* @param \Apache_Solr_Document $pageDocument
270		*/
271	35	protected function applyIndexPagePostProcessors($pageDocument)
272		{
273	35	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
274	34	return;
275		}
276
277	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
278	1	$postProcessor = GeneralUtility::getUserObj($classReference);
279	1	if (!$postProcessor instanceof PageDocumentPostProcessor) {
280		throw new \UnexpectedValueException(get_class($pageDocument) . ' must implement interface ' . PageDocumentPostProcessor::class, 1397739154);
281		}
282
283	1	$postProcessor->postProcessPageDocument($pageDocument, $this->page);
284	1	}
285	1	}
286
287		/**
288		* Builds the Solr document for the current page.
289		*
290		* @return \Apache_Solr_Document A document representing the page
291		*/
292	35	protected function getPageDocument()
293		{
294	35	$document = GeneralUtility::makeInstance(Apache_Solr_Document::class);
295		/* @var $document \Apache_Solr_Document */
296	35	$site = Site::getSiteByPageId($this->page->id);
297	35	$pageRecord = $this->page->page;
298
299	35	self::$pageSolrDocumentId = $documentId = Util::getPageDocumentId(
300	35	$this->page->id,
301	35	$this->page->type,
302	35	$this->page->sys_language_uid,
303	35	$this->getDocumentIdGroups(),
304	35	$this->getMountPointParameter()
305	35	);
306	35	$document->setField('id', $documentId);
307	35	$document->setField('site', $site->getDomain());
308	35	$document->setField('siteHash', $site->getSiteHash());
309	35	$document->setField('appKey', 'EXT:solr');
310	35	$document->setField('type', 'pages');
311
312		// system fields
313	35	$document->setField('uid', $this->page->id);
314	35	$document->setField('pid', $pageRecord['pid']);
315
316		// variantId
317	35	$document->setField('variantId', 'pages/' . $this->page->id);
318
319	35	$document->setField('typeNum', $this->page->type);
320	35	$document->setField('created', $pageRecord['crdate']);
321	35	$document->setField('changed', $pageRecord['SYS_LASTCHANGED']);
322
323	35	$rootline = $this->getRootLineFieldValue();
324	35	$document->setField('rootline', $rootline);
325
326		// access
327	35	$this->addAccessField($document);
328	35	$this->addEndtimeField($document, $pageRecord);
329
330		// content
331	35	$document->setField('title', $this->contentExtractor->getPageTitle());
332	35	$document->setField('subTitle', $pageRecord['subtitle']);
333	35	$document->setField('navTitle', $pageRecord['nav_title']);
334	35	$document->setField('author', $pageRecord['author']);
335	35	$document->setField('description', $pageRecord['description']);
336	35	$document->setField('abstract', $pageRecord['abstract']);
337	35	$document->setField('content', $this->contentExtractor->getIndexableContent());
338	35	$document->setField('url', $this->pageUrl);
339
340	35	$this->addKeywordsField($document, $pageRecord);
341	35	$this->addTagContentFields($document);
342
343	35	return $document;
344		}
345
346		/**
347		* Adds the access field to the document if needed.
348		*
349		* @param \Apache_Solr_Document $document
350		*/
351	35	protected function addAccessField(\Apache_Solr_Document $document)
352		{
353	35	$access = (string)$this->pageAccessRootline;
354	35	if (trim($access) !== '') {
355	8	$document->setField('access', $access);
356	8	}
357	35	}
358
359		/**
360		* @param $document
361		* @param $pageRecord
362		*/
363	35	protected function addEndtimeField(\Apache_Solr_Document $document, $pageRecord)
364		{
365	35	if ($this->page->page['endtime']) {
366		$document->setField('endtime', $pageRecord['endtime']);
367		}
368	35	}
369
370		/**
371		* Adds keywords, multi valued.
372		*
373		* @param \Apache_Solr_Document $document
374		* @param array $pageRecord
375		*/
376	35	protected function addKeywordsField(\Apache_Solr_Document $document, $pageRecord)
377		{
378	35	$keywords = array_unique(GeneralUtility::trimExplode(',', $pageRecord['keywords'], true));
379	35	foreach ($keywords as $keyword) {
380		$document->addField('keywords', $keyword);
381	35	}
382	35	}
383
384		/**
385		* Add content from several tags like headers, anchors, ...
386		*
387		* @param \Apache_Solr_Document $document
388		*/
389	35	protected function addTagContentFields(\Apache_Solr_Document $document)
390		{
391	35	$tagContent = $this->contentExtractor->getTagContent();
392	35	foreach ($tagContent as $fieldName => $fieldValue) {
393		$document->setField($fieldName, $fieldValue);
394	35	}
395	35	}
396
397		/**
398		* Builds the content for the rootline field.
399		*
400		* @return string
401		*/
402	35	protected function getRootLineFieldValue()
403		{
404	35	$rootline = $this->page->id;
405	35	$mountPointParameter = $this->getMountPointParameter();
406	35	if ($mountPointParameter !== '') {
407	30	$rootline .= ',' . $mountPointParameter;
408	30	}
409	35	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	35	protected function getDocumentIdGroups()
419		{
420	35	$groups = $this->pageAccessRootline->getGroups();
421	35	$groups = Rootline::cleanGroupArray($groups);
422
423	35	if (empty($groups)) {
424	29	$groups[] = 0;
425	29	}
426
427	35	$groups = implode(',', $groups);
428
429	35	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	35	public function getMountPointParameter()
441		{
442	35	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 $mountPointParameter
451		*/
452	5	public function setMountPointParameter($mountPointParameter)
453		{
454	5	$this->mountPointParameter = (string)$mountPointParameter;
455	5	}
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	35	protected function substitutePageDocument(\Apache_Solr_Document $pageDocument)
465		{
466	35	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
467	30	return $pageDocument;
468		}
469
470	5	$indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
471	5	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
472	5	$substituteIndexer = GeneralUtility::getUserObj($classReference);
473
474	5	if (!$substituteIndexer instanceof SubstitutePageIndexer) {
475		$message = get_class($substituteIndexer) . ' must implement interface ' . SubstitutePageIndexer::class;
476		throw new \UnexpectedValueException($message, 1310491001);
477		}
478
479	5	if ($substituteIndexer instanceof PageFieldMappingIndexer) {
480	5	$substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
481	5	}
482
483	5	$substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
484	5	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	5	$pageDocument = $substituteDocument;
489	5	}
490
491	5	return $pageDocument;
492		}
493
494		/**
495		* Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
496		*
497		* @return string
498		*/
499	5	protected function getIndexConfigurationNameForCurrentPage()
500		{
501	5	return isset($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	35	protected function getAdditionalDocuments(\Apache_Solr_Document $pageDocument, array $existingDocuments)
513		{
514	35	$documents = $existingDocuments;
515
516	35	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
517	34	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	1	}
532	1	}
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	35	protected function processDocuments(array $documents)
544		{
545	35	$processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
546	35	if (count($processingInstructions) > 0) {
547	35	$service = GeneralUtility::makeInstance(Service::class);
548	35	$service->processDocuments($documents, $processingInstructions);
549	35	}
550	35	}
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	35	protected function addDocumentsToSolrIndex(array $documents)
559		{
560	35	$documentsAdded = false;
561
562	35	if (!count($documents)) {
563		return $documentsAdded;
564		}
565
566		try {
567	35	$this->log('Adding ' . count($documents) . ' documents.', 0, $documents);
568
569		// chunk adds by 20
570	35	$documentChunks = array_chunk($documents, 20);
571	35	foreach ($documentChunks as $documentChunk) {
572	35	$response = $this->solrConnection->addDocuments($documentChunk);
573
574	35	if ($response->getHttpStatus() != 200) {
575		$transportException = new \Apache_Solr_HttpTransportException($response);
576		throw new \RuntimeException('Solr Request failed.', 1331834983, $transportException);
577		}
578	35	}
579
580	35	$documentsAdded = true;
581	35	} catch (\Exception $e) {
582		$this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 2);
583
584		if ($this->configuration->getLoggingExceptions()) {
585		GeneralUtility::devLog('Exception while adding documents', 'solr', 3, [$e->__toString()]);
586		}
587		}
588
589	35	return $documentsAdded;
590		}
591
592		/**
593		* Gets the current page's URL.
594		*
595		* @return string URL of the current page.
596		*/
597		public function getPageUrl()
598		{
599		return $this->pageUrl;
600		}
601
602		/**
603		* Sets the URL to use for the page document.
604		*
605		* @param string $url The page's URL.
606		*/
607	5	public function setPageUrl($url)
608		{
609	5	$this->pageUrl = $url;
610	5	}
611
612		/**
613		* Gets the page's access rootline.
614		*
615		* @return Rootline The page's access rootline
616		*/
617		public function getPageAccessRootline()
618		{
619		return $this->pageAccessRootline;
620		}
621
622		/**
623		* Sets the page's access rootline.
624		*
625		* @param Rootline $accessRootline The page's access rootline
626		*/
627	34	public function setPageAccessRootline(Rootline $accessRootline)
628		{
629	34	$this->pageAccessRootline = $accessRootline;
630	34	}
631
632		/**
633		* Gets the documents that have been sent to Solr
634		*
635		* @return array An array of \Apache_Solr_Document objects
636		*/
637	5	public function getDocumentsSentToSolr()
638		{
639	5	return $this->documentsSentToSolr;
640		}
641		}
642

TYPO3-Solr / ext-solr

Branch — master (b9fc31)

Typo3PageIndexer::getPageDocument() A

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Long Method

Long Method

Duplication Side-by-Side

Filter issues like