Typo3PageIndexer::getPageSolrDocument() - Code Metrics - Inspection of "[FIX] typoscript in Tsfe::initializeTsfe() parsed..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Failed Conditions

Push — task/2976_TYPO3.11_compatibili... ( 38a128...0b41c4 )

by Rafael

created 2021-12-17 19:26 UTC

Typo3PageIndexer::getPageSolrDocument() A

↳ Parent: Typo3PageIndexer

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	3
ccs	2
cts	2
cp	1
rs	10
c	0
b	0
f	0
cc	1
eloc	1
nc	1
nop	0
crap	1

<?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 3 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 ApacheSolrForTypo3\Solr\Access\Rootline;
use ApacheSolrForTypo3\Solr\Domain\Search\ApacheSolrDocument\Builder;
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 ApacheSolrForTypo3\Solr\System\Solr\Document\Document;
use ApacheSolrForTypo3\Solr\System\Solr\SolrConnection;
use Exception;
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 Document
     */
    protected static $pageSolrDocument = null;
    /**
     * The mount point parameter used in the Frontend controller.
     *
     * @var string
     */
    protected $mountPointParameter;
    /**
     * Solr server connection.
     *
     * @var SolrConnection
     */
    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 \ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager
     */
    protected $logger = null;

    /**
     * Constructor
     *
     * @param TypoScriptFrontendController $page The page to index
     */
    public function __construct(TypoScriptFrontendController $page)
    {
        $this->logger = GeneralUtility::makeInstance(SolrLogManager::class, /** @scrutinizer ignore-type */ __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->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, /** @scrutinizer ignore-type */ '');
    }

    /**
     * @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, Util::getLanguageUid());

        // do not continue if no server is available
        if (!$solr->getWriteService()->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 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 SolrConnection $solrConnection Solr connection
     * @throws Exception if the Solr server cannot be reached
     */
    public function setSolrConnection(SolrConnection $solrConnection)
    {
        if (!$solrConnection->getWriteService()->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 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) {
            $postProcessor = GeneralUtility::makeInstance($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 Document A document representing the page
     */
    protected function getPageDocument()
    {
        $documentBuilder = GeneralUtility::makeInstance(Builder::class);
        $document = $documentBuilder->fromPage($this->page, $this->pageUrl, $this->pageAccessRootline, (string)$this->mountPointParameter);

        self::$pageSolrDocumentId = $document['id'];

        return $document;
    }


    // 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 Document $pageDocument The page document created by this indexer.
     * @return Document An Apache Solr document representing the currently indexed page
     */
    protected function substitutePageDocument(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) {
            $substituteIndexer = GeneralUtility::makeInstance($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 Document) {
                $message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Document object.';
                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 Document $pageDocument The main document representing this page.
     * @param Document[] $existingDocuments An array of documents already created for this page.
     * @return array An array of additional Document objects to index
     */
    protected function getAdditionalDocuments(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) {
            $additionalIndexer = GeneralUtility::makeInstance($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 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 = $this->solrConnection->getWriteService()->addDocuments($documentChunk);
                if ($response->getHttpStatus() != 200) {
                    throw new \RuntimeException('Solr Request failed.', 1331834983);
                }
            }

            $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 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 3 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 ApacheSolrForTypo3\Solr\Access\Rootline;
28		use ApacheSolrForTypo3\Solr\Domain\Search\ApacheSolrDocument\Builder;
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 ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager;
34		use ApacheSolrForTypo3\Solr\System\Solr\Document\Document;
35		use ApacheSolrForTypo3\Solr\System\Solr\SolrConnection;
36		use Exception;
37		use TYPO3\CMS\Core\Utility\GeneralUtility;
38		use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;
39
40		/**
41		* Page Indexer to index TYPO3 pages used by the Index Queue.
42		*
43		* @author Ingo Renner <[email protected]>
44		* @author Daniel Poetzinger <[email protected]>
45		* @author Timo Schmidt <[email protected]>
46		*/
47		class Typo3PageIndexer
48		{
49
50		/**
51		* ID of the current page's Solr document.
52		*
53		* @var string
54		*/
55		protected static $pageSolrDocumentId = '';
56		/**
57		* The Solr document generated for the current page.
58		*
59		* @var Document
60		*/
61		protected static $pageSolrDocument = null;
62		/**
63		* The mount point parameter used in the Frontend controller.
64		*
65		* @var string
66		*/
67		protected $mountPointParameter;
68		/**
69		* Solr server connection.
70		*
71		* @var SolrConnection
72		*/
73		protected $solrConnection = null;
74		/**
75		* Frontend page object (TSFE).
76		*
77		* @var TypoScriptFrontendController
78		*/
79		protected $page = null;
80		/**
81		* Content extractor to extract content from TYPO3 pages
82		*
83		* @var Typo3PageContentExtractor
84		*/
85		protected $contentExtractor = null;
86		/**
87		* URL to be indexed as the page's URL
88		*
89		* @var string
90		*/
91		protected $pageUrl = '';
92		/**
93		* The page's access rootline
94		*
95		* @var Rootline
96		*/
97		protected $pageAccessRootline = null;
98		/**
99		* Documents that have been sent to Solr
100		*
101		* @var array
102		*/
103		protected $documentsSentToSolr = [];
104
105		/**
106		* @var TypoScriptConfiguration
107		*/
108		protected $configuration;
109
110		/**
111		* @var Item
112		*/
113		protected $indexQueueItem;
114
115		/**
116		* @var \ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager
117		*/
118		protected $logger = null;
119
120		/**
121		* Constructor
122		*
123		* @param TypoScriptFrontendController $page The page to index
124		*/
125	31	public function __construct(TypoScriptFrontendController $page)
126		{
127	31	$this->logger = GeneralUtility::makeInstance(SolrLogManager::class, /** @scrutinizer ignore-type */ __CLASS__);
128
129	31	$this->page = $page;
130	31	$this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
131	31	$this->configuration = Util::getSolrConfiguration();
132
133		try {
134	31	$this->initializeSolrConnection();
135		} catch (Exception $e) {
136		$this->logger->log(
137		SolrLogManager::ERROR,
138		$e->getMessage() . ' Error code: ' . $e->getCode()
139		);
140
141		// TODO extract to a class "ExceptionLogger"
142		if ($this->configuration->getLoggingExceptions()) {
143		$this->logger->log(
144		SolrLogManager::ERROR,
145		'Exception while trying to index a page',
146		[
147		$e->__toString()
148		]
149		);
150		}
151		}
152
153	31	$this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, /** @scrutinizer ignore-type */ '');
154	31	}
155
156		/**
157		* @param Item $indexQueueItem
158		*/
159	10	public function setIndexQueueItem($indexQueueItem)
160		{
161	10	$this->indexQueueItem = $indexQueueItem;
162	10	}
163
164		/**
165		* Initializes the Solr server connection.
166		*
167		* @throws Exception when no Solr connection can be established.
168		*/
169	31	protected function initializeSolrConnection()
170		{
171	31	$solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, Util::getLanguageUid());
172
173		// do not continue if no server is available
174	31	if (!$solr->getWriteService()->ping()) {
175		throw new Exception(
176		'No Solr instance available while trying to index a page.',
177		1234790825
178		);
179		}
180
181	31	$this->solrConnection = $solr;
182	31	}
183
184		/**
185		* Gets the current page's Solr document ID.
186		*
187		* @return string\|NULL The page's Solr document ID or NULL in case no document was generated yet.
188		*/
189		public static function getPageSolrDocumentId()
190		{
191		return self::$pageSolrDocumentId;
192		}
193
194		/**
195		* Gets the Solr document generated for the current page.
196		*
197		* @return Document\|NULL The page's Solr document or NULL if it has not been generated yet.
198		*/
199	10	public static function getPageSolrDocument()
200		{
201	10	return self::$pageSolrDocument;
202		}
203
204		/**
205		* Allows to provide a Solr server connection other than the one
206		* initialized by the constructor.
207		*
208		* @param SolrConnection $solrConnection Solr connection
209		* @throws Exception if the Solr server cannot be reached
210		*/
211	10	public function setSolrConnection(SolrConnection $solrConnection)
212		{
213	10	if (!$solrConnection->getWriteService()->ping()) {
214		throw new Exception(
215		'Could not connect to Solr server.',
216		1323946472
217		);
218		}
219
220	10	$this->solrConnection = $solrConnection;
221	10	}
222
223		/**
224		* Indexes a page.
225		*
226		* @return bool TRUE after successfully indexing the page, FALSE on error
227		* @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
228		*/
229	31	public function indexPage()
230		{
231	31	$pageIndexed = false;
232	31	$documents = []; // this will become useful as soon as when starting to index individual records instead of whole pages
233
234	31	if (is_null($this->solrConnection)) {
235		// intended early return as it doesn't make sense to continue
236		// and waste processing time if the solr server isn't available
237		// anyways
238		// FIXME use an exception
239		return $pageIndexed;
240		}
241
242	31	$pageDocument = $this->getPageDocument();
243	31	$pageDocument = $this->substitutePageDocument($pageDocument);
244
245	31	$this->applyIndexPagePostProcessors($pageDocument);
246
247	31	self::$pageSolrDocument = $pageDocument;
248	31	$documents[] = $pageDocument;
249	31	$documents = $this->getAdditionalDocuments($pageDocument, $documents);
250	31	$this->processDocuments($documents);
251
252	31	$pageIndexed = $this->addDocumentsToSolrIndex($documents);
253	31	$this->documentsSentToSolr = $documents;
254
255	31	return $pageIndexed;
256		}
257
258		/**
259		* Applies the configured post processors (indexPagePostProcessPageDocument)
260		*
261		* @param Document $pageDocument
262		*/
263	31	protected function applyIndexPagePostProcessors($pageDocument)
264		{
265	31	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
266	30	return;
267		}
268
269	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
270	1	$postProcessor = GeneralUtility::makeInstance($classReference);
271	1	if (!$postProcessor instanceof PageDocumentPostProcessor) {
272		throw new \UnexpectedValueException(get_class($pageDocument) . ' must implement interface ' . PageDocumentPostProcessor::class, 1397739154);
273		}
274
275	1	$postProcessor->postProcessPageDocument($pageDocument, $this->page);
276		}
277	1	}
278
279		/**
280		* Builds the Solr document for the current page.
281		*
282		* @return Document A document representing the page
283		*/
284	31	protected function getPageDocument()
285		{
286	31	$documentBuilder = GeneralUtility::makeInstance(Builder::class);
287	31	$document = $documentBuilder->fromPage($this->page, $this->pageUrl, $this->pageAccessRootline, (string)$this->mountPointParameter);
288
289	31	self::$pageSolrDocumentId = $document['id'];
290
291	31	return $document;
292		}
293
294
295		// Logging
296		// TODO replace by a central logger
297
298		/**
299		* Gets the mount point parameter that is used in the Frontend controller.
300		*
301		* @return string
302		*/
303		public function getMountPointParameter()
304		{
305		return $this->mountPointParameter;
306		}
307
308		// Misc
309
310		/**
311		* Sets the mount point parameter that is used in the Frontend controller.
312		*
313		* @param string $mountPointParameter
314		*/
315	10	public function setMountPointParameter($mountPointParameter)
316		{
317	10	$this->mountPointParameter = (string)$mountPointParameter;
318	10	}
319
320		/**
321		* Allows third party extensions to replace or modify the page document
322		* created by this indexer.
323		*
324		* @param Document $pageDocument The page document created by this indexer.
325		* @return Document An Apache Solr document representing the currently indexed page
326		*/
327	31	protected function substitutePageDocument(Document $pageDocument)
328		{
329	31	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
330		return $pageDocument;
331		}
332
333	31	$indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
334	31	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
335	31	$substituteIndexer = GeneralUtility::makeInstance($classReference);
336
337	31	if (!$substituteIndexer instanceof SubstitutePageIndexer) {
338		$message = get_class($substituteIndexer) . ' must implement interface ' . SubstitutePageIndexer::class;
339		throw new \UnexpectedValueException($message, 1310491001);
340		}
341
342	31	if ($substituteIndexer instanceof PageFieldMappingIndexer) {
343	10	$substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
344		}
345
346	31	$substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
347	31	if (!$substituteDocument instanceof Document) {
348		$message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Document object.';
349		throw new \UnexpectedValueException($message, 1310490952);
350		}
351	31	$pageDocument = $substituteDocument;
352		}
353
354	31	return $pageDocument;
355		}
356
357		/**
358		* Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
359		*
360		* @return string
361		*/
362	31	protected function getIndexConfigurationNameForCurrentPage()
363		{
364	31	return isset($this->indexQueueItem) ? $this->indexQueueItem->getIndexingConfigurationName() : 'pages';
365		}
366
367		/**
368		* Allows third party extensions to provide additional documents which
369		* should be indexed for the current page.
370		*
371		* @param Document $pageDocument The main document representing this page.
372		* @param Document[] $existingDocuments An array of documents already created for this page.
373		* @return array An array of additional Document objects to index
374		*/
375	31	protected function getAdditionalDocuments(Document $pageDocument, array $existingDocuments)
376		{
377	31	$documents = $existingDocuments;
378
379	31	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
380	30	return $documents;
381		}
382
383	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
384	1	$additionalIndexer = GeneralUtility::makeInstance($classReference);
385
386	1	if (!$additionalIndexer instanceof AdditionalPageIndexer) {
387		$message = get_class($additionalIndexer) . ' must implement interface ' . AdditionalPageIndexer::class;
388		throw new \UnexpectedValueException($message, 1310491024);
389		}
390
391	1	$additionalDocuments = $additionalIndexer->getAdditionalPageDocuments($pageDocument, $documents);
392	1	if (is_array($additionalDocuments)) {
393	1	$documents = array_merge($documents, $additionalDocuments);
394		}
395		}
396
397	1	return $documents;
398		}
399
400		/**
401		* Sends the given documents to the field processing service which takes
402		* care of manipulating fields as defined in the field's configuration.
403		*
404		* @param array $documents An array of documents to manipulate
405		*/
406	31	protected function processDocuments(array $documents)
407		{
408	31	$processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
409	31	if (count($processingInstructions) > 0) {
410	31	$service = GeneralUtility::makeInstance(Service::class);
411	31	$service->processDocuments($documents, $processingInstructions);
412		}
413	31	}
414
415		/**
416		* Adds the collected documents to the Solr index.
417		*
418		* @param array $documents An array of Document objects.
419		* @return bool TRUE if documents were added successfully, FALSE otherwise
420		*/
421	31	protected function addDocumentsToSolrIndex(array $documents)
422		{
423	31	$documentsAdded = false;
424
425	31	if (!count($documents)) {
426		return $documentsAdded;
427		}
428
429		try {
430	31	$this->logger->log(SolrLogManager::INFO, 'Adding ' . count($documents) . ' documents.', $documents);
431
432		// chunk adds by 20
433	31	$documentChunks = array_chunk($documents, 20);
434	31	foreach ($documentChunks as $documentChunk) {
435	31	$response = $this->solrConnection->getWriteService()->addDocuments($documentChunk);
436	31	if ($response->getHttpStatus() != 200) {
437		throw new \RuntimeException('Solr Request failed.', 1331834983);
438		}
439		}
440
441	31	$documentsAdded = true;
442		} catch (Exception $e) {
443		$this->logger->log(SolrLogManager::ERROR, $e->getMessage() . ' Error code: ' . $e->getCode());
444
445		if ($this->configuration->getLoggingExceptions()) {
446		$this->logger->log(SolrLogManager::ERROR, 'Exception while adding documents', [$e->__toString()]);
447		}
448		}
449
450	31	return $documentsAdded;
451		}
452
453		/**
454		* Gets the current page's URL.
455		*
456		* @return string URL of the current page.
457		*/
458		public function getPageUrl()
459		{
460		return $this->pageUrl;
461		}
462
463		/**
464		* Sets the URL to use for the page document.
465		*
466		* @param string $url The page's URL.
467		*/
468	10	public function setPageUrl($url)
469		{
470	10	$this->pageUrl = $url;
471	10	}
472
473		/**
474		* Gets the page's access rootline.
475		*
476		* @return Rootline The page's access rootline
477		*/
478		public function getPageAccessRootline()
479		{
480		return $this->pageAccessRootline;
481		}
482
483		/**
484		* Sets the page's access rootline.
485		*
486		* @param Rootline $accessRootline The page's access rootline
487		*/
488	24	public function setPageAccessRootline(Rootline $accessRootline)
489		{
490	24	$this->pageAccessRootline = $accessRootline;
491	24	}
492
493		/**
494		* Gets the documents that have been sent to Solr
495		*
496		* @return array An array of Document objects
497		*/
498	10	public function getDocumentsSentToSolr()
499		{
500	10	return $this->documentsSentToSolr;
501		}
502		}
503

TYPO3-Solr / ext-solr

Push — task/2976_TYPO3.11_compatibili... ( 38a128...0b41c4 )

Typo3PageIndexer::getPageSolrDocument() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like