Typo3PageIndexer::getPageUrl() - Code Metrics - Inspection of "Merge pull request #844 from timohund/task/master/..." - TYPO3-Solr/ext-solr - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( b6b629...72d15a )

by Timo

created 2016-12-20 08:58 UTC

Typo3PageIndexer::getPageUrl() A

↳ Parent: Typo3PageIndexer

Complexity

Conditions	1
Paths	1

Size

Total Lines	4
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	2

Importance

Changes

Metric	Value
dl	0
loc	4
ccs	0
cts	4
cp	0
rs	10
c	0
b	0
f	0
cc	1
eloc	2
nc	1
nop	0
crap	2

<?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 ApacheSolrForTypo3\Solr\Access\Rootline;
use ApacheSolrForTypo3\Solr\ConnectionManager;
use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
use ApacheSolrForTypo3\Solr\IndexQueue\Item;
use ApacheSolrForTypo3\Solr\IndexQueue\Queue;
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 = array();

    /**
     * @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, array(
                        $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 $data = array())
    {
        if (is_object($GLOBALS['TT'])) {
            $GLOBALS['TT']->setTSlogMessage('tx_solr: ' . $message, $errorNum);
        }

        if ($this->configuration->getLoggingIndexing()) {
            $logData = array();
            if (!empty($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 = array(); // 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);

        if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
            foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
                $postProcessor = GeneralUtility::getUserObj($classReference);

                if ($postProcessor instanceof PageDocumentPostProcessor) {
                    $postProcessor->postProcessPageDocument($pageDocument, $this->page);
                } else {
                    throw new \UnexpectedValueException(
                        get_class($pageDocument) . ' must implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor',
                        1397739154
                    );
                }
            }
        }

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

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

        return $pageIndexed;
    }

    /**
     * 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');
        /* @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->page->id;
        $mountPointParameter = $this->getMountPointParameter();
        if ($mountPointParameter !== '') {
            $rootline .= ',' . $mountPointParameter;
        }
        $document->setField('rootline', $rootline);

        // access
        $access = (string)$this->pageAccessRootline;
        if (trim($access) !== '') {
            $document->setField('access', $access);
        }
        if ($this->page->page['endtime']) {
            $document->setField('endtime', $pageRecord['endtime']);
        }

        // 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);

        // keywords, multi valued
        $keywords = array_unique(GeneralUtility::trimExplode(
            ',',
            $pageRecord['keywords'],
            true
        ));
        foreach ($keywords as $keyword) {
            $document->addField('keywords', $keyword);
        }

        // content from several tags like headers, anchors, ...
        $tagContent = $this->contentExtractor->getTagContent();
        foreach ($tagContent as $fieldName => $fieldValue) {
            $document->setField($fieldName, $fieldValue);
        }

        return $document;
    }

    /**
     * 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 ApacheSolrForTypo3\Solr\SubstitutePageIndexer';
                throw new \UnexpectedValueException($message, 1310491001);
            }

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

            $substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
            if (!$substituteDocument instanceof \Apache_Solr_Document) {
if ($x instanceof DoesNotExist) {
    // Do something.
}
                $message = 'The document returned by ' . get_class($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 array $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($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
            foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
                $additionalIndexer = GeneralUtility::getUserObj($classReference);

                if ($additionalIndexer instanceof AdditionalPageIndexer) {
                    $additionalDocuments = $additionalIndexer->getAdditionalPageDocuments($pageDocument,
                        $documents);

                    if (is_array($additionalDocuments)) {
                        $documents = array_merge($documents,
                            $additionalDocuments);
                    }
                } else {
                    throw new \UnexpectedValueException(
                        get_class($additionalIndexer) . ' must implement interface ApacheSolrForTypo3\Solr\AdditionalPageIndexer',
                        1310491024
                    );
                }
            }
        }

        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('ApacheSolrForTypo3\\Solr\\FieldProcessor\\Service');
            $service->processDocuments($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, array(
                        $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;
    }
}


Push — master ( b6b629...72d15a )

Typo3PageIndexer::getPageUrl() A

Complexity

Size

Duplication

Code Coverage

Importance

1. Missing dependencies

2. Missing use statement

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 ApacheSolrForTypo3\Solr\Access\Rootline;
28		use ApacheSolrForTypo3\Solr\ConnectionManager;
29		use ApacheSolrForTypo3\Solr\IndexQueue\FrontendHelper\PageFieldMappingIndexer;
30		use ApacheSolrForTypo3\Solr\IndexQueue\Item;
31		use ApacheSolrForTypo3\Solr\IndexQueue\Queue;
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 = array();
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	33	public function __construct(TypoScriptFrontendController $page)
117		{
118	33	$this->page = $page;
119	33	$this->pageUrl = GeneralUtility::getIndpEnv('TYPO3_REQUEST_URL');
120	33	$this->configuration = Util::getSolrConfiguration();
121
122		try {
123	33	$this->initializeSolrConnection();
124		} 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',
130		'solr', 3, array(
131		$e->__toString()
132		));
133		}
134		}
135
136	33	$this->contentExtractor = GeneralUtility::makeInstance(Typo3PageContentExtractor::class, $this->page->content);
137	33	$this->pageAccessRootline = GeneralUtility::makeInstance(Rootline::class, '');
138	33	}
139
140		/**
141		* @param Item $indexQueueItem
142		*/
143	3	public function setIndexQueueItem($indexQueueItem)
144		{
145	3	$this->indexQueueItem = $indexQueueItem;
146	3	}
147
148
149		/**
150		* Initializes the Solr server connection.
151		*
152		* @throws \Exception when no Solr connection can be established.
153		*/
154	33	protected function initializeSolrConnection()
155		{
156	33	$solr = GeneralUtility::makeInstance(ConnectionManager::class)->getConnectionByPageId($this->page->id, $this->page->sys_language_uid);
157
158		// do not continue if no server is available
159	33	if (!$solr->ping()) {
160		throw new \Exception(
161		'No Solr instance available while trying to index a page.',
162		1234790825
163		);
164		}
165
166	33	$this->solrConnection = $solr;
167	33	}
168
169		/**
170		* Logs messages to devlog and TS log (admin panel)
171		*
172		* @param string $message Message to set
173		* @param int $errorNum Error number
174		* @param array $data Additional data to log
175		* @return void
176		*/
177	33	protected function log($message, $errorNum = 0, array $data = array())
178		{
179	33	if (is_object($GLOBALS['TT'])) {
180	33	$GLOBALS['TT']->setTSlogMessage('tx_solr: ' . $message, $errorNum);
181		}
182
183	33	if ($this->configuration->getLoggingIndexing()) {
184		$logData = array();
185		if (!empty($data)) {
186		foreach ($data as $value) {
187		$logData[] = (array)$value;
188		}
189		}
190
191		GeneralUtility::devLog($message, 'solr', $errorNum, $logData);
192		}
193	33	}
194
195		/**
196		* Gets the current page's Solr document ID.
197		*
198		* @return string\|NULL The page's Solr document ID or NULL in case no document was generated yet.
199		*/
200		public static function getPageSolrDocumentId()
201		{
202		return self::$pageSolrDocumentId;
203		}
204
205		/**
206		* Gets the Solr document generated for the current page.
207		*
208		* @return \Apache_Solr_Document\|NULL The page's Solr document or NULL if it has not been generated yet.
209		*/
210	3	public static function getPageSolrDocument()
211		{
212	3	return self::$pageSolrDocument;
213		}
214
215		/**
216		* Allows to provide a Solr server connection other than the one
217		* initialized by the constructor.
218		*
219		* @param SolrService $solrConnection Solr connection
220		* @throws \Exception if the Solr server cannot be reached
221		*/
222	3	public function setSolrConnection(SolrService $solrConnection)
223		{
224	3	if (!$solrConnection->ping()) {
225		throw new \Exception(
226		'Could not connect to Solr server.',
227		1323946472
228		);
229		}
230
231	3	$this->solrConnection = $solrConnection;
232	3	}
233
234		/**
235		* Indexes a page.
236		*
237		* @return bool TRUE after successfully indexing the page, FALSE on error
238		* @throws \UnexpectedValueException if a page document post processor fails to implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor
239		*/
240	33	public function indexPage()
241		{
242	33	$pageIndexed = false;
243	33	$documents = array(); // this will become useful as soon as when starting to index individual records instead of whole pages
244
245	33	if (is_null($this->solrConnection)) {
246		// intended early return as it doesn't make sense to continue
247		// and waste processing time if the solr server isn't available
248		// anyways
249		// FIXME use an exception
250		return $pageIndexed;
251		}
252
253	33	$pageDocument = $this->getPageDocument();
254	33	$pageDocument = $this->substitutePageDocument($pageDocument);
255
256	33	if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'])) {
257		foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPagePostProcessPageDocument'] as $classReference) {
258		$postProcessor = GeneralUtility::getUserObj($classReference);
259
260		if ($postProcessor instanceof PageDocumentPostProcessor) {
261		$postProcessor->postProcessPageDocument($pageDocument, $this->page);
262		} else {
263		throw new \UnexpectedValueException(
264		get_class($pageDocument) . ' must implement interface ApacheSolrForTypo3\Solr\PageDocumentPostProcessor',
265		1397739154
266		);
267		}
268		}
269		}
270
271	33	self::$pageSolrDocument = $pageDocument;
272	33	$documents[] = $pageDocument;
273	33	$documents = $this->getAdditionalDocuments($pageDocument, $documents);
274	33	$this->processDocuments($documents);
275
276	33	$pageIndexed = $this->addDocumentsToSolrIndex($documents);
277	33	$this->documentsSentToSolr = $documents;
278
279	33	return $pageIndexed;
280		}
281
282		/**
283		* Builds the Solr document for the current page.
284		*
285		* @return \Apache_Solr_Document A document representing the page
286		*/
287	33	protected function getPageDocument()
288		{
289	33	$document = GeneralUtility::makeInstance('Apache_Solr_Document');
290		/* @var $document \Apache_Solr_Document */
291	33	$site = Site::getSiteByPageId($this->page->id);
292	33	$pageRecord = $this->page->page;
293
294	33	self::$pageSolrDocumentId = $documentId = Util::getPageDocumentId(
295	33	$this->page->id,
296	33	$this->page->type,
297	33	$this->page->sys_language_uid,
298	33	$this->getDocumentIdGroups(),
299	33	$this->getMountPointParameter()
300		);
301	33	$document->setField('id', $documentId);
302	33	$document->setField('site', $site->getDomain());
303	33	$document->setField('siteHash', $site->getSiteHash());
304	33	$document->setField('appKey', 'EXT:solr');
305	33	$document->setField('type', 'pages');
306
307		// system fields
308	33	$document->setField('uid', $this->page->id);
309	33	$document->setField('pid', $pageRecord['pid']);
310
311		// variantId
312	33	$document->setField('variantId', 'pages/' . $this->page->id);
313
314	33	$document->setField('typeNum', $this->page->type);
315	33	$document->setField('created', $pageRecord['crdate']);
316	33	$document->setField('changed', $pageRecord['SYS_LASTCHANGED']);
317
318	33	$rootline = $this->page->id;
319	33	$mountPointParameter = $this->getMountPointParameter();
320	33	if ($mountPointParameter !== '') {
321	30	$rootline .= ',' . $mountPointParameter;
322		}
323	33	$document->setField('rootline', $rootline);
324
325		// access
326	33	$access = (string)$this->pageAccessRootline;
327	33	if (trim($access) !== '') {
328	6	$document->setField('access', $access);
329		}
330	33	if ($this->page->page['endtime']) {
331		$document->setField('endtime', $pageRecord['endtime']);
332		}
333
334		// content
335	33	$document->setField('title', $this->contentExtractor->getPageTitle());
336	33	$document->setField('subTitle', $pageRecord['subtitle']);
337	33	$document->setField('navTitle', $pageRecord['nav_title']);
338	33	$document->setField('author', $pageRecord['author']);
339	33	$document->setField('description', $pageRecord['description']);
340	33	$document->setField('abstract', $pageRecord['abstract']);
341	33	$document->setField('content',
342	33	$this->contentExtractor->getIndexableContent());
343	33	$document->setField('url', $this->pageUrl);
344
345		// keywords, multi valued
346	33	$keywords = array_unique(GeneralUtility::trimExplode(
347	33	',',
348	33	$pageRecord['keywords'],
349	33	true
350		));
351	33	foreach ($keywords as $keyword) {
352		$document->addField('keywords', $keyword);
353		}
354
355		// content from several tags like headers, anchors, ...
356	33	$tagContent = $this->contentExtractor->getTagContent();
357	33	foreach ($tagContent as $fieldName => $fieldValue) {
358		$document->setField($fieldName, $fieldValue);
359		}
360
361	33	return $document;
362		}
363
364		/**
365		* Gets a comma separated list of frontend user groups to use for the
366		* document ID.
367		*
368		* @return string A comma separated list of frontend user groups.
369		*/
370	33	protected function getDocumentIdGroups()
371		{
372	33	$groups = $this->pageAccessRootline->getGroups();
373	33	$groups = Rootline::cleanGroupArray($groups);
374
375	33	if (empty($groups)) {
376	29	$groups[] = 0;
377		}
378
379	33	$groups = implode(',', $groups);
380
381	33	return $groups;
382		}
383
384		// Logging
385		// TODO replace by a central logger
386
387		/**
388		* Gets the mount point parameter that is used in the Frontend controller.
389		*
390		* @return string
391		*/
392	33	public function getMountPointParameter()
393		{
394	33	return $this->mountPointParameter;
395		}
396
397		// Misc
398
399		/**
400		* Sets the mount point parameter that is used in the Frontend controller.
401		*
402		* @param string $mountPointParameter
403		*/
404	3	public function setMountPointParameter($mountPointParameter)
405		{
406	3	$this->mountPointParameter = (string)$mountPointParameter;
407	3	}
408
409		/**
410		* Allows third party extensions to replace or modify the page document
411		* created by this indexer.
412		*
413		* @param \Apache_Solr_Document $pageDocument The page document created by this indexer.
414		* @return \Apache_Solr_Document An Apache Solr document representing the currently indexed page
415		*/
416	33	protected function substitutePageDocument(\Apache_Solr_Document $pageDocument)
417		{
418	33	if (!is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'])) {
419	30	return $pageDocument;
420		}
421
422	3	$indexConfigurationName = $this->getIndexConfigurationNameForCurrentPage();
423	3	foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageSubstitutePageDocument'] as $classReference) {
424	3	$substituteIndexer = GeneralUtility::getUserObj($classReference);
425
426	3	if (!$substituteIndexer instanceof SubstitutePageIndexer) {
427		$message = get_class($substituteIndexer) . ' must implement interface ApacheSolrForTypo3\Solr\SubstitutePageIndexer';
428		throw new \UnexpectedValueException($message, 1310491001);
429		}
430
431	3	if ($substituteIndexer instanceof PageFieldMappingIndexer) {
432	3	$substituteIndexer->setPageIndexingConfigurationName($indexConfigurationName);
433		}
434
435	3	$substituteDocument = $substituteIndexer->getPageDocument($pageDocument);
436	3	if (!$substituteDocument instanceof \Apache_Solr_Document) {
		0 ignored issues – show Bug introduced 2016-08-18 11:24 UTC by Report Bug Copy Issue Report The class `Apache_Solr_Document` does not exist. Did you forget a USE statement, or did you not list all dependencies? This error could be the result of: 1. Missing dependencies PHP Analyzer uses your `composer.json` file (if available) to determine the dependencies of your project and to determine all the available classes and functions. It expects the `composer.json` to be in the root folder of your repository. Are you sure this class is defined by one of your dependencies, or did you maybe not list a dependency in either the `require` or `require-dev` section? 2. Missing use statement PHP does not complain about undefined classes in `ìnstanceof` checks. For example, the following PHP code will work perfectly fine: if ($x instanceof DoesNotExist) { // Do something. } If you have not tested against this specific condition, such errors might go unnoticed. Loading history...
437		$message = 'The document returned by ' . get_class($substituteIndexer) . ' is not a valid Apache_Solr_Document document.';
438		throw new \UnexpectedValueException($message, 1310490952);
439		}
440	3	$pageDocument = $substituteDocument;
441		}
442
443	3	return $pageDocument;
444		}
445
446		/**
447		* Retrieves the indexConfigurationName from the related queueItem, or falls back to pages when no queue item set.
448		*
449		* @return string
450		*/
451	3	protected function getIndexConfigurationNameForCurrentPage()
452		{
453	3	return isset($this->indexQueueItem) ? $this->indexQueueItem->getIndexingConfigurationName() : 'pages';
454		}
455
456		/**
457		* Allows third party extensions to provide additional documents which
458		* should be indexed for the current page.
459		*
460		* @param \Apache_Solr_Document $pageDocument The main document representing this page.
461		* @param array $existingDocuments An array of documents already created for this page.
462		* @return array An array of additional \Apache_Solr_Document objects to index
463		*/
464	33	protected function getAdditionalDocuments(
465		\Apache_Solr_Document $pageDocument,
466		array $existingDocuments
467		) {
468	33	$documents = $existingDocuments;
469
470	33	if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'])) {
471		foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['solr']['Indexer']['indexPageAddDocuments'] as $classReference) {
472		$additionalIndexer = GeneralUtility::getUserObj($classReference);
473
474		if ($additionalIndexer instanceof AdditionalPageIndexer) {
475		$additionalDocuments = $additionalIndexer->getAdditionalPageDocuments($pageDocument,
476		$documents);
477
478		if (is_array($additionalDocuments)) {
479		$documents = array_merge($documents,
480		$additionalDocuments);
481		}
482		} else {
483		throw new \UnexpectedValueException(
484		get_class($additionalIndexer) . ' must implement interface ApacheSolrForTypo3\Solr\AdditionalPageIndexer',
485		1310491024
486		);
487		}
488		}
489		}
490
491	33	return $documents;
492		}
493
494		/**
495		* Sends the given documents to the field processing service which takes
496		* care of manipulating fields as defined in the field's configuration.
497		*
498		* @param array $documents An array of documents to manipulate
499		*/
500	33	protected function processDocuments(array $documents)
501		{
502	33	$processingInstructions = $this->configuration->getIndexFieldProcessingInstructionsConfiguration();
503	33	if (count($processingInstructions) > 0) {
504	33	$service = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Solr\\FieldProcessor\\Service');
505	33	$service->processDocuments($documents, $processingInstructions);
506		}
507	33	}
508
509		/**
510		* Adds the collected documents to the Solr index.
511		*
512		* @param array $documents An array of \Apache_Solr_Document objects.
513		* @return bool TRUE if documents were added successfully, FALSE otherwise
514		*/
515	33	protected function addDocumentsToSolrIndex(array $documents)
516		{
517	33	$documentsAdded = false;
518
519	33	if (!count($documents)) {
520		return $documentsAdded;
521		}
522
523		try {
524	33	$this->log('Adding ' . count($documents) . ' documents.', 0,
525		$documents);
526
527		// chunk adds by 20
528	33	$documentChunks = array_chunk($documents, 20);
529	33	foreach ($documentChunks as $documentChunk) {
530	33	$response = $this->solrConnection->addDocuments($documentChunk);
531
532	33	if ($response->getHttpStatus() != 200) {
533		$transportException = new \Apache_Solr_HttpTransportException($response);
534		throw new \RuntimeException('Solr Request failed.',
535	33	1331834983, $transportException);
536		}
537		}
538
539	33	$documentsAdded = true;
540		} catch (\Exception $e) {
541		$this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 2);
542
543		if ($this->configuration->getLoggingExceptions()) {
544		GeneralUtility::devLog('Exception while adding documents',
545		'solr', 3, array(
546		$e->__toString()
547		));
548		}
549		}
550
551	33	return $documentsAdded;
552		}
553
554		/**
555		* Gets the current page's URL.
556		*
557		* @return string URL of the current page.
558		*/
559		public function getPageUrl()
560		{
561		return $this->pageUrl;
562		}
563
564		/**
565		* Sets the URL to use for the page document.
566		*
567		* @param string $url The page's URL.
568		*/
569	3	public function setPageUrl($url)
570		{
571	3	$this->pageUrl = $url;
572	3	}
573
574		/**
575		* Gets the page's access rootline.
576		*
577		* @return Rootline The page's access rootline
578		*/
579		public function getPageAccessRootline()
580		{
581		return $this->pageAccessRootline;
582		}
583
584		/**
585		* Sets the page's access rootline.
586		*
587		* @param Rootline $accessRootline The page's access rootline
588		*/
589	32	public function setPageAccessRootline(Rootline $accessRootline)
590		{
591	32	$this->pageAccessRootline = $accessRootline;
592	32	}
593
594		/**
595		* Gets the documents that have been sent to Solr
596		*
597		* @return array An array of \Apache_Solr_Document objects
598		*/
599	3	public function getDocumentsSentToSolr()
600		{
601	3	return $this->documentsSentToSolr;
602		}
603		}
604

TYPO3-Solr / ext-solr

Push — master ( b6b629...72d15a )

Typo3PageIndexer::getPageUrl() A

Complexity

Size

Duplication

Code Coverage

Importance

1. Missing dependencies

2. Missing use statement

Duplication Side-by-Side

Filter issues like