Typo3PageIndexer::indexPage() - Code Metrics - Inspection of "Merge pull request #1317 from dkd-kaehm/TASK/1315-..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( a95893...a4d2df )

by Timo

created 2017-06-07 13:04 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	15
CRAP Score	2.0185

Importance

Changes

Metric	Value
dl	0
loc	28
ccs	15
cts	18
cp	0.8333
rs	8.8571
c	0
b	0
f	0
cc	2
eloc	15
nc	2
nop	0
crap	2.0185

<?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\Search\ApacheSolrDocument\Builder;
use ApacheSolrForTypo3\Solr\Domain\Variants\IdBuilder;
use ApacheSolrForTypo3\Solr\FieldProcessor\Service;
use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
use ApacheSolrForTypo3\Solr\IndexQueue\Item;
use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration;
use ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager;
use TYPO3\CMS\Core\Utility\GeneralUtility;
use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;

/**
 * Page Indexer to index TYPO3 pages used by the Index Queue.
 *
 * @author Ingo Renner <[email protected]>
 * @author Daniel Poetzinger <[email protected]>
 * @author Timo Schmidt <[email protected]>
 */
class Typo3PageIndexer
{

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

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

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

    /**
     * @var \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, __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, '');
    }

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

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

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

        $this->solrConnection = $solr;
    }

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

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

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

        $this->solrConnection = $solrConnection;
    }

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

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

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

        $this->applyIndexPagePostProcessors($pageDocument);

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

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

        return $pageIndexed;
    }

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

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

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

    /**
     * Builds the Solr document for the current page.
     *
     * @return \Apache_Solr_Document A document representing the page
     */
    protected function getPageDocument()
    {
        $documentBuilder = GeneralUtility::makeInstance(Builder::class);
        $document = $documentBuilder->fromPage($this->page, $this->pageUrl, $this->pageAccessRootline, $this->mountPointParameter);
        $idField = $document->getField('id');

        self::$pageSolrDocumentId = $idField['value'];

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

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

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

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

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

        return $pageDocument;
    }

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

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

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

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

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

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

        return $documents;
    }

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

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

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

        try {
            $this->logger->log(
                SolrLogManager::INFO,
                'Adding ' . count($documents) . ' documents.',
                $documents
            );

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

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

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

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

        return $documentsAdded;
    }

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

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

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

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

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


1		<?php
2		namespace ApacheSolrForTypo3\Solr;
3
4		/***************************************************************
5		* Copyright notice
6		*
7		* (c) 2009-2015 Ingo Renner <[email protected]>
8		* All rights reserved
9		*
10		* This script is part of the TYPO3 project. The TYPO3 project is
11		* free software; you can redistribute it and/or modify
12		* it under the terms of the GNU General Public License as published by
13		* the Free Software Foundation; either version 2 of the License, or
14		* (at your option) any later version.
15		*
16		* The GNU General Public License can be found at
17		* http://www.gnu.org/copyleft/gpl.html.
18		*
19		* This script is distributed in the hope that it will be useful,
20		* but WITHOUT ANY WARRANTY; without even the implied warranty of
21		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22		* GNU General Public License for more details.
23		*
24		* This copyright notice MUST APPEAR in all copies of the script!
25		***************************************************************/
26
27		use Apache_Solr_Document;
28		use ApacheSolrForTypo3\Solr\Access\Rootline;
29		use ApacheSolrForTypo3\Solr\Domain\Search\ApacheSolrDocument\Builder;
30		use ApacheSolrForTypo3\Solr\Domain\Variants\IdBuilder;
31		use ApacheSolrForTypo3\Solr\FieldProcessor\Service;
32		use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
33		use ApacheSolrForTypo3\Solr\IndexQueue\Item;
34		use ApacheSolrForTypo3\Solr\System\Configuration\TypoScriptConfiguration;
35		use ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager;
36		use TYPO3\CMS\Core\Utility\GeneralUtility;
37		use TYPO3\CMS\Frontend\Controller\TypoScriptFrontendController;
38
39		/**
40		* Page Indexer to index TYPO3 pages used by the Index Queue.
41		*
42		* @author Ingo Renner <[email protected]>
43		* @author Daniel Poetzinger <[email protected]>
44		* @author Timo Schmidt <[email protected]>
45		*/
46		class Typo3PageIndexer
47		{
48
49		/**
50		* ID of the current page's Solr document.
51		*
52		* @var string
53		*/
54		protected static $pageSolrDocumentId = '';
55		/**
56		* The Solr document generated for the current page.
57		*
58		* @var \Apache_Solr_Document
59		*/
60		protected static $pageSolrDocument = null;
61		/**
62		* The mount point parameter used in the Frontend controller.
63		*
64		* @var string
65		*/
66		protected $mountPointParameter;
67		/**
68		* Solr server connection.
69		*
70		* @var SolrService
71		*/
72		protected $solrConnection = null;
73		/**
74		* Frontend page object (TSFE).
75		*
76		* @var TypoScriptFrontendController
77		*/
78		protected $page = null;
79		/**
80		* Content extractor to extract content from TYPO3 pages
81		*
82		* @var Typo3PageContentExtractor
83		*/
84		protected $contentExtractor = null;
85		/**
86		* URL to be indexed as the page's URL
87		*
88		* @var string
89		*/
90		protected $pageUrl = '';
91		/**
92		* The page's access rootline
93		*
94		* @var Rootline
95		*/
96		protected $pageAccessRootline = null;
97		/**
98		* Documents that have been sent to Solr
99		*
100		* @var array
101		*/
102		protected $documentsSentToSolr = [];
103
104		/**
105		* @var TypoScriptConfiguration
106		*/
107		protected $configuration;
108
109		/**
110		* @var Item
111		*/
112		protected $indexQueueItem;
113
114		/**
115		* @var \ApacheSolrForTypo3\Solr\System\Logging\SolrLogManager
116		*/
117		protected $logger = null;
118
119		/**
120		* Constructor
121		*
122		* @param TypoScriptFrontendController $page The page to index
123		*/
124	48	public function __construct(TypoScriptFrontendController $page)
125		{
126	48	$this->logger = GeneralUtility::makeInstance(SolrLogManager::class, __CLASS__);
127
128	48	$this->page = $page;
129	48	$this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
130	48	$this->configuration = Util::getSolrConfiguration();
131
132		try {
133	48	$this->initializeSolrConnection();
134	3	} catch (\Exception $e) {
135	3	$this->logger->log(
136	3	SolrLogManager::ERROR,
137	3	$e->getMessage() . ' Error code: ' . $e->getCode()
138		);
139
140		// TODO extract to a class "ExceptionLogger"
141	3	if ($this->configuration->getLoggingExceptions()) {
142	3	$this->logger->log(
143	3	SolrLogManager::ERROR,
144	3	'Exception while trying to index a page',
145		[
146	3	$e->__toString()
147		]
148		);
149		}
150		}
151
152	48	$this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, '');
153	48	}
154
155		/**
156		* @param Item $indexQueueItem
157		*/
158	8	public function setIndexQueueItem($indexQueueItem)
159		{
160	8	$this->indexQueueItem = $indexQueueItem;
161	8	}
162
163		/**
164		* Initializes the Solr server connection.
165		*
166		* @throws \Exception when no Solr connection can be established.
167		*/
168	48	protected function initializeSolrConnection()
169		{
170	48	$solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, $this->page->sys_language_uid);
171
172		// do not continue if no server is available
173	46	if (!$solr->ping()) {
174	1	throw new \Exception(
175	1	'No Solr instance available while trying to index a page.',
176	1	1234790825
177		);
178		}
179
180	45	$this->solrConnection = $solr;
181	45	}
182
183		/**
184		* Gets the current page's Solr document ID.
185		*
186		* @return string\|NULL The page's Solr document ID or NULL in case no document was generated yet.
187		*/
188		public static function getPageSolrDocumentId()
189		{
190		return self::$pageSolrDocumentId;
191		}
192
193		/**
194		* Gets the Solr document generated for the current page.
195		*
196		* @return \Apache_Solr_Document\|NULL The page's Solr document or NULL if it has not been generated yet.
197		*/
198	8	public static function getPageSolrDocument()
199		{
200	8	return self::$pageSolrDocument;
201		}
202
203		/**
204		* Allows to provide a Solr server connection other than the one
205		* initialized by the constructor.
206		*
207		* @param SolrService $solrConnection Solr connection
208		* @throws \Exception if the Solr server cannot be reached
209		*/
210	8	public function setSolrConnection(SolrService $solrConnection)
211		{
212	8	if (!$solrConnection->ping()) {
213		throw new \Exception(
214		'Could not connect to Solr server.',
215		1323946472
216		);
217		}
218
219	8	$this->solrConnection = $solrConnection;
220	8	}
221
222		/**
223		* Indexes a page.
224		*
225		* @return bool TRUE after successfully indexing the page, FALSE on error
226		* @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
227		*/
228	48	public function indexPage()
229		{
230	48	$pageIndexed = false;
231	48	$documents = []; // this will become useful as soon as when starting to index individual records instead of whole pages
232
233	48	if (is_null($this->solrConnection)) {
234		// intended early return as it doesn't make sense to continue
235		// and waste processing time if the solr server isn't available
236		// anyways
237		// FIXME use an exception
238	1	return $pageIndexed;
239		}
240
241	47	$pageDocument = $this->getPageDocument();
242	47	$pageDocument = $this->substitutePageDocument($pageDocument);
243
244	47	$this->applyIndexPagePostProcessors($pageDocument);
245
246	47	self::$pageSolrDocument = $pageDocument;
247	47	$documents[] = $pageDocument;
248	47	$documents = $this->getAdditionalDocuments($pageDocument, $documents);
249	47	$this->processDocuments($documents);
250
251	47	$pageIndexed = $this->addDocumentsToSolrIndex($documents);
252	47	$this->documentsSentToSolr = $documents;
253
254	47	return $pageIndexed;
255		}
256
257		/**
258		* Applies the configured post processors (indexPagePostProcessPageDocument)
259		*
260		* @param \Apache_Solr_Document $pageDocument
261		*/
262	47	protected function applyIndexPagePostProcessors($pageDocument)
263		{
264	47	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
265	46	return;
266		}
267
268	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
269	1	$postProcessor = GeneralUtility::getUserObj($classReference);
270	1	if (!$postProcessor instanceof PageDocumentPostProcessor) {
271		throw new \UnexpectedValueException(get_class($pageDocument) . ' must implement interface ' . PageDocumentPostProcessor::class, 1397739154);
272		}
273
274	1	$postProcessor->postProcessPageDocument($pageDocument, $this->page);
275		}
276	1	}
277
278		/**
279		* Builds the Solr document for the current page.
280		*
281		* @return \Apache_Solr_Document A document representing the page
282		*/
283	47	protected function getPageDocument()
284		{
285	47	$documentBuilder = GeneralUtility::makeInstance(Builder::class);
286	47	$document = $documentBuilder->fromPage($this->page, $this->pageUrl, $this->pageAccessRootline, $this->mountPointParameter);
287	47	$idField = $document->getField('id');
288
289	47	self::$pageSolrDocumentId = $idField['value'];
290
291	47	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 $mountPointParameter
314		*/
315	8	public function setMountPointParameter($mountPointParameter)
316		{
317	8	$this->mountPointParameter = (string)$mountPointParameter;
318	8	}
319
320		/**
321		* Allows third party extensions to replace or modify the page document
322		* created by this indexer.
323		*
324		* @param \Apache_Solr_Document $pageDocument The page document created by this indexer.
325		* @return \Apache_Solr_Document An Apache Solr document representing the currently indexed page
326		*/
327	47	protected function substitutePageDocument(\Apache_Solr_Document $pageDocument)
328		{
329	47	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
330	39	return $pageDocument;
331		}
332
333	8	$indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
334	8	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
335	8	$substituteIndexer = GeneralUtility::getUserObj($classReference);
336
337	8	if (!$substituteIndexer instanceof SubstitutePageIndexer) {
338		$message = get_class($substituteIndexer) . ' must implement interface ' . SubstitutePageIndexer::class;
339		throw new \UnexpectedValueException($message, 1310491001);
340		}
341
342	8	if ($substituteIndexer instanceof PageFieldMappingIndexer) {
343	8	$substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
344		}
345
346	8	$substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
347	8	if (!$substituteDocument instanceof Apache_Solr_Document) {
348		$message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Apache_Solr_Document document.';
349		throw new \UnexpectedValueException($message, 1310490952);
350		}
351	8	$pageDocument = $substituteDocument;
352		}
353
354	8	return $pageDocument;
355		}
356
357		/**
358		* Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
359		*
360		* @return string
361		*/
362	8	protected function getIndexConfigurationNameForCurrentPage()
363		{
364	8	return isset($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 \Apache_Solr_Document $pageDocument The main document representing this page.
372		* @param \Apache_Solr_Document[] $existingDocuments An array of documents already created for this page.
373		* @return array An array of additional \Apache_Solr_Document objects to index
374		*/
375	47	protected function getAdditionalDocuments(\Apache_Solr_Document $pageDocument, array $existingDocuments)
376		{
377	47	$documents = $existingDocuments;
378
379	47	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
380	46	return $documents;
381		}
382
383	1	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
384	1	$additionalIndexer = GeneralUtility::getUserObj($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	47	protected function processDocuments(array $documents)
407		{
408	47	$processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
409	47	if (count($processingInstructions) > 0) {
410	47	$service = GeneralUtility::makeInstance(Service::class);
411	47	$service->processDocuments($documents, $processingInstructions);
412		}
413	47	}
414
415		/**
416		* Adds the collected documents to the Solr index.
417		*
418		* @param array $documents An array of \Apache_Solr_Document objects.
419		* @return bool TRUE if documents were added successfully, FALSE otherwise
420		*/
421	47	protected function addDocumentsToSolrIndex(array $documents)
422		{
423	47	$documentsAdded = false;
424
425	47	if (!count($documents)) {
426		return $documentsAdded;
427		}
428
429		try {
430	47	$this->logger->log(
431	47	SolrLogManager::INFO,
432	47	'Adding ' . count($documents) . ' documents.',
433	47	$documents
434		);
435
436		// chunk adds by 20
437	47	$documentChunks = array_chunk($documents, 20);
438	47	foreach ($documentChunks as $documentChunk) {
439	47	$response = $this->solrConnection->addDocuments($documentChunk);
440
441	47	if ($response->getHttpStatus() != 200) {
442		$transportException = new \Apache_Solr_HttpTransportException($response);
443		throw new \RuntimeException('Solr Request failed.', 1331834983, $transportException);
444		}
445		}
446
447	47	$documentsAdded = true;
448		} catch (\Exception $e) {
449		$this->logger->log(
450		SolrLogManager::ERROR,
451		$e->getMessage() . ' Error code: ' . $e->getCode()
452		);
453
454		if ($this->configuration->getLoggingExceptions()) {
455		$this->logger->log(
456		SolrLogManager::ERROR,
457		'Exception while adding documents',
458		[
459		$e->__toString()
460		]
461		);
462		}
463		}
464
465	47	return $documentsAdded;
466		}
467
468		/**
469		* Gets the current page's URL.
470		*
471		* @return string URL of the current page.
472		*/
473		public function getPageUrl()
474		{
475		return $this->pageUrl;
476		}
477
478		/**
479		* Sets the URL to use for the page document.
480		*
481		* @param string $url The page's URL.
482		*/
483	8	public function setPageUrl($url)
484		{
485	8	$this->pageUrl = $url;
486	8	}
487
488		/**
489		* Gets the page's access rootline.
490		*
491		* @return Rootline The page's access rootline
492		*/
493		public function getPageAccessRootline()
494		{
495		return $this->pageAccessRootline;
496		}
497
498		/**
499		* Sets the page's access rootline.
500		*
501		* @param Rootline $accessRootline The page's access rootline
502		*/
503	21	public function setPageAccessRootline(Rootline $accessRootline)
504		{
505	21	$this->pageAccessRootline = $accessRootline;
506	21	}
507
508		/**
509		* Gets the documents that have been sent to Solr
510		*
511		* @return array An array of \Apache_Solr_Document objects
512		*/
513	8	public function getDocumentsSentToSolr()
514		{
515	8	return $this->documentsSentToSolr;
516		}
517		}
518

TYPO3-Solr / ext-solr

Push — master ( a95893...a4d2df )

Typo3PageIndexer::indexPage() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like