CrawlerApi::__construct() - Code Metrics - Inspection of "[WIP] Documentation" - AOEpeople/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — features/addFlushedPagesToCraw... ( 44f5f4...d700c8 )

by Tomas Norre

created 2020-07-13 20:19 UTC

CrawlerApi::__construct() A

↳ Parent: CrawlerApi

Complexity

Conditions	1
Paths	1

Size

Total Lines	5
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	3
CRAP Score	1.0156

Importance

Changes

Metric	Value
cc	1
eloc	2
c	0
b	0
f	0
nc	1
nop	0
dl	0
loc	5
ccs	3
cts	4
cp	0.75
crap	1.0156
rs	10

<?php

declare(strict_types=1);

namespace AOE\Crawler\Api;

/***************************************************************
 *  Copyright notice
 *
 *  (c) 2018 AOE GmbH <[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 AOE\Crawler\Controller\CrawlerController;
use AOE\Crawler\Domain\Repository\ProcessRepository;
use AOE\Crawler\Domain\Repository\QueueRepository;
use AOE\Crawler\Exception\CrawlerObjectException;
use AOE\Crawler\Exception\TimeStampException;
use TYPO3\CMS\Core\Database\ConnectionPool;
use TYPO3\CMS\Core\Database\Query\QueryBuilder;
use TYPO3\CMS\Core\Utility\GeneralUtility;
use TYPO3\CMS\Extbase\Object\ObjectManager;
use TYPO3\CMS\Frontend\Page\PageRepository;

/**
 * Class CrawlerApi
 *
 * @package AOE\Crawler\Api
 */
class CrawlerApi
{
    /**
     * @var QueueRepository
     */
    protected $queueRepository;

    /**
     * @var array
     */
    protected $allowedConfigurations = [];

    /**
     * @var QueryBuilder
     */
    protected $queryBuilder;

    /**
     * @var string
     */
    protected $tableName = 'tx_crawler_queue';

    /**
     * @var CrawlerController
     */
    protected $crawlerController;

    public function __construct()
    {
        /** @var ObjectManager $objectManager */
        $objectManager = GeneralUtility::makeInstance(ObjectManager::class);
        $this->queueRepository = $objectManager->get(QueueRepository::class);
    }

    /**
     * Each crawler run has a setid, this facade method delegates
     * the it to the crawler object
     *
     * @throws \Exception
     */
    public function overwriteSetId(int $id): void
    {
        $this->findCrawler()->setID = $id;
    }

    /**
     * This method is used to limit the configuration selection to
     * a set of configurations.
     */
    public function setAllowedConfigurations(array $allowedConfigurations): void
    {
        $this->allowedConfigurations = $allowedConfigurations;
    }

    /**
     * @return array
     */
    public function getAllowedConfigurations()
    {
        return $this->allowedConfigurations;
    }

    /**
     * Returns the setID of the crawler
     *
     * @return int
     */
    public function getSetId()
    {
        return $this->findCrawler()->setID;
    }

    /**
     * Adds a page to the crawlerqueue by uid
     *
     * @param int $uid uid
     */
    public function addPageToQueue($uid): void
    {
        $uid = intval($uid);
        //non timed elements will be added with timestamp 0
        $this->addPageToQueueTimed($uid, 0);
    }

    /**
     * Adds a page to the crawlerqueue by uid and sets a
     * timestamp when the page should be crawled.
     *
     * @param int $uid pageid
     * @param int $time timestamp
     *
     * @throws \Exception
     */
    public function addPageToQueueTimed($uid, $time): void
    {
        $uid = intval($uid);
        $time = intval($time);

        $crawler = $this->findCrawler();
        $pageData = GeneralUtility::makeInstance(PageRepository::class)->getPage($uid);
        $configurations = $crawler->getUrlsForPageRow($pageData);
        $configurations = $this->filterUnallowedConfigurations($configurations);
        $downloadUrls = [];
        $duplicateTrack = [];

        if (is_array($configurations)) {

            foreach ($configurations as $cv) {
                //enable inserting of entries
                $crawler->registerQueueEntriesInternallyOnly = false;
                $crawler->urlListFromUrlArray(
                    $cv,
                    $pageData,
                    $time,
                    300,
                    true,
                    false,
                    $duplicateTrack,
                    $downloadUrls,
                    array_keys($this->getCrawlerProcInstructions())
                );

                //reset the queue because the entries have been written to the db
                unset($crawler->queueEntries);
            }
        }
    }

    /**
     * Method to return the latest Crawle Timestamp for a page.
     *
     * @param int $uid uid id of the page
     * @param bool $future_crawldates_only
     * @param bool $unprocessed_only
     *
     * @return int
     */
    public function getLatestCrawlTimestampForPage($uid, $future_crawldates_only = false, $unprocessed_only = false)
    {
        $uid = intval($uid);

        $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName);
        $query = $queryBuilder
            ->from($this->tableName)
            ->selectLiteral('max(scheduled) as latest')
            ->where(
                $queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid))
            );

        if ($future_crawldates_only) {
            $query->andWhere(
                $queryBuilder->expr()->gt('scheduled', time())
            );
        }

        if ($unprocessed_only) {
            $query->andWhere(
                $queryBuilder->expr()->eq('exec_time', 0)
            );
        }

        $row = $query->execute()->fetch(0);
        if ($row['latest']) {
            $res = $row['latest'];
        } else {
            $res = 0;
        }

        return intval($res);
    }

    /**
     * Returns an array with timestamps when the page has been scheduled for crawling and
     * at what time the scheduled crawl has been executed. The array also contains items that are
     * scheduled but have note been crawled yet.
     *
     * @return array array with the crawl-history of a page => 0 : scheduled time , 1 : executed_time, 2 : set_id
     */
    public function getCrawlHistoryForPage(int $uid, int $limit = 0)
    {
        $uid = intval($uid);
        $limit = intval($limit);

        $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName);
        $statement = $queryBuilder
            ->from($this->tableName)
            ->select('scheduled', 'exec_time', 'set_id')
            ->where(
                $queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid, \PDO::PARAM_INT))
            );
        if ($limit) {
            $statement->setMaxResults($limit);
        }

        return $statement->execute()->fetchAll();
    }

    /**
     * Get queue statistics
     */
    public function getQueueStatistics(): array
    {
        return [
            'assignedButUnprocessed' => $this->queueRepository->countAllAssignedPendingItems(),
            'unprocessed' => $this->queueRepository->countAllPendingItems(),
        ];
    }

    /**
     * Get queue statistics by configuration
     *
     * @return array array of array('configuration' => <>, 'assignedButUnprocessed' => <>, 'unprocessed' => <>)
     */
    public function getQueueStatisticsByConfiguration()
    {
        $statistics = $this->queueRepository->countPendingItemsGroupedByConfigurationKey();
        $setIds = $this->queueRepository->getSetIdWithUnprocessedEntries();
        $totals = $this->queueRepository->getTotalQueueEntriesByConfiguration($setIds);

        // "merge" arrays
        foreach ($statistics as &$value) {
            $value['total'] = $totals[$value['configuration']];
        }

        return $statistics;
    }

    /**
     * Get active processes count
     */
    public function getActiveProcessesCount(): int
    {
        $processRepository = new ProcessRepository();

        return $processRepository->countActive();
    }

    public function getLastProcessedQueueEntries(int $limit): array
    {
        return $this->queueRepository->getLastProcessedEntries($limit);
    }

    /**
     * Get current crawling speed
     *
     * @return int|float|bool
     */
    public function getCurrentCrawlingSpeed()
    {
        $lastProcessedEntries = $this->queueRepository->getLastProcessedEntriesTimestamps();

        if (count($lastProcessedEntries) < 10) {
            // not enough information
            return false;
        }

        $tooOldDelta = 60; // time between two entries is "too old"

        $compareValue = time();
        $startTime = $lastProcessedEntries[0];

        $pages = 0;

        reset($lastProcessedEntries);
        foreach ($lastProcessedEntries as $timestamp) {
            if ($compareValue - $timestamp > $tooOldDelta) {
                break;
            }
            $compareValue = $timestamp;
            $pages++;
        }

        if ($pages < 10) {
            // not enough information
            return false;
        }
        $oldestTimestampThatIsNotTooOld = $compareValue;
        $time = $startTime - $oldestTimestampThatIsNotTooOld;

        return $pages / ($time / 60);
    }

    /**
     * Get some performance data
     *
     * @param integer $start
     * @param integer $end
     * @param integer $resolution
     *
     * @return array data
     *
     * @throws TimeStampException
     */
    public function getPerformanceData($start, $end, $resolution)
    {
        $data = [];

        $data['urlcount'] = 0;
        $data['start'] = $start;
        $data['end'] = $end;
        $data['duration'] = $data['end'] - $data['start'];

        if ($data['duration'] < 1) {
            throw new TimeStampException('End timestamp must be after start timestamp', 1512659945);
        }

        for ($slotStart = $start; $slotStart < $end; $slotStart += $resolution) {
            $slotEnd = min($slotStart + $resolution - 1, $end);
            $slotData = $this->queueRepository->getPerformanceData($slotStart, $slotEnd);

            $slotUrlCount = 0;
            foreach ($slotData as &$processData) {
                $duration = $processData['end'] - $processData['start'];
                if ($processData['urlcount'] > 5 && $duration > 0) {
                    $processData['speed'] = 60 * 1 / ($duration / $processData['urlcount']);
                }
                $slotUrlCount += $processData['urlcount'];
            }

            $data['urlcount'] += $slotUrlCount;

            $data['slots'][$slotEnd] = [
                'amountProcesses' => count($slotData),
                'urlcount' => $slotUrlCount,
                'processes' => $slotData,
            ];

            if ($slotUrlCount > 5) {
                $data['slots'][$slotEnd]['speed'] = 60 * 1 / ($slotEnd - $slotStart / $slotUrlCount);
            } else {
                $data['slots'][$slotEnd]['speed'] = 0;
            }
        }

        if ($data['urlcount'] > 5) {
            $data['speed'] = 60 * 1 / ($data['duration'] / $data['urlcount']);
        } else {
            $data['speed'] = 0;
        }

        return $data;
    }

    /**
     * Method to get an instance of the internal crawler singleton
     *
     * @return CrawlerController Instance of the crawler lib
     *
     * @throws CrawlerObjectException
     */
    protected function findCrawler()
    {
        if (! is_object($this->crawlerController)) {
            $this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class);
            $this->crawlerController->setID = GeneralUtility::md5int(microtime());
        }

        if (is_object($this->crawlerController)) {
            return $this->crawlerController;
        }
        throw new CrawlerObjectException('no crawler object', 1512659759);
    }

    /**
     * This method is used to limit the processing instructions to the processing instructions
     * that are allowed.
     */
    protected function filterUnallowedConfigurations(array $configurations): array
    {
        if (count($this->allowedConfigurations) > 0) {
            // 	remove configuration that does not match the current selection
            foreach ($configurations as $confKey => $confArray) {
                if (! in_array($confKey, $this->allowedConfigurations, true)) {
                    unset($configurations[$confKey]);
                }
            }
        }

        return $configurations;
    }

    /**
     * Reads the registered processingInstructions of the crawler
     */
    private function getCrawlerProcInstructions(): array
    {
        $crawlerProcInstructions = [];
        if (! empty($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'])) {
            foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'] as $configuration) {
                $crawlerProcInstructions[$configuration['key']] = $configuration['value'];
            }
        }

        return $crawlerProcInstructions;
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		namespace AOE\Crawler\Api;
6
7		/***************************************************************
8		* Copyright notice
9		*
10		* (c) 2018 AOE GmbH <[email protected]>
11		*
12		* All rights reserved
13		*
14		* This script is part of the TYPO3 project. The TYPO3 project is
15		* free software; you can redistribute it and/or modify
16		* it under the terms of the GNU General Public License as published by
17		* the Free Software Foundation; either version 3 of the License, or
18		* (at your option) any later version.
19		*
20		* The GNU General Public License can be found at
21		* http://www.gnu.org/copyleft/gpl.html.
22		*
23		* This script is distributed in the hope that it will be useful,
24		* but WITHOUT ANY WARRANTY; without even the implied warranty of
25		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
26		* GNU General Public License for more details.
27		*
28		* This copyright notice MUST APPEAR in all copies of the script!
29		***************************************************************/
30
31		use AOE\Crawler\Controller\CrawlerController;
32		use AOE\Crawler\Domain\Repository\ProcessRepository;
33		use AOE\Crawler\Domain\Repository\QueueRepository;
34		use AOE\Crawler\Exception\CrawlerObjectException;
35		use AOE\Crawler\Exception\TimeStampException;
36		use TYPO3\CMS\Core\Database\ConnectionPool;
37		use TYPO3\CMS\Core\Database\Query\QueryBuilder;
38		use TYPO3\CMS\Core\Utility\GeneralUtility;
39		use TYPO3\CMS\Extbase\Object\ObjectManager;
40		use TYPO3\CMS\Frontend\Page\PageRepository;
41
42		/**
43		* Class CrawlerApi
44		*
45		* @package AOE\Crawler\Api
46		*/
47		class CrawlerApi
48		{
49		/**
50		* @var QueueRepository
51		*/
52		protected $queueRepository;
53
54		/**
55		* @var array
56		*/
57		protected $allowedConfigurations = [];
58
59		/**
60		* @var QueryBuilder
61		*/
62		protected $queryBuilder;
63
64		/**
65		* @var string
66		*/
67		protected $tableName = 'tx_crawler_queue';
68
69		/**
70		* @var CrawlerController
71		*/
72		protected $crawlerController;
73
74	10	public function __construct()
75		{
76		/** @var ObjectManager $objectManager */
77	10	$objectManager = GeneralUtility::makeInstance(ObjectManager::class);
78	10	$this->queueRepository = $objectManager->get(QueueRepository::class);
79	10	}
80
81		/**
82		* Each crawler run has a setid, this facade method delegates
83		* the it to the crawler object
84		*
85		* @throws \Exception
86		*/
87	1	public function overwriteSetId(int $id): void
88		{
89	1	$this->findCrawler()->setID = $id;
90	1	}
91
92		/**
93		* This method is used to limit the configuration selection to
94		* a set of configurations.
95		*/
96	1	public function setAllowedConfigurations(array $allowedConfigurations): void
97		{
98	1	$this->allowedConfigurations = $allowedConfigurations;
99	1	}
100
101		/**
102		* @return array
103		*/
104	1	public function getAllowedConfigurations()
105		{
106	1	return $this->allowedConfigurations;
107		}
108
109		/**
110		* Returns the setID of the crawler
111		*
112		* @return int
113		*/
114	1	public function getSetId()
115		{
116	1	return $this->findCrawler()->setID;
117		}
118
119		/**
120		* Adds a page to the crawlerqueue by uid
121		*
122		* @param int $uid uid
123		*/
124		public function addPageToQueue($uid): void
125		{
126		$uid = intval($uid);
127		//non timed elements will be added with timestamp 0
128		$this->addPageToQueueTimed($uid, 0);
129		}
130
131		/**
132		* Adds a page to the crawlerqueue by uid and sets a
133		* timestamp when the page should be crawled.
134		*
135		* @param int $uid pageid
136		* @param int $time timestamp
137		*
138		* @throws \Exception
139		*/
140	4	public function addPageToQueueTimed($uid, $time): void
141		{
142	4	$uid = intval($uid);
143	4	$time = intval($time);
144
145	4	$crawler = $this->findCrawler();
146	4	$pageData = GeneralUtility::makeInstance(PageRepository::class)->getPage($uid);
147	4	$configurations = $crawler->getUrlsForPageRow($pageData);
148	4	$configurations = $this->filterUnallowedConfigurations($configurations);
149	4	$downloadUrls = [];
150	4	$duplicateTrack = [];
151
152	4	if (is_array($configurations)) {
		0 ignored issues – show introduced 2019-12-30 17:56 UTC by Report Bug Copy Issue Report The condition `is_array($configurations)` is always `true`. Loading history...
153	4	foreach ($configurations as $cv) {
154		//enable inserting of entries
155	4	$crawler->registerQueueEntriesInternallyOnly = false;
156	4	$crawler->urlListFromUrlArray(
157	4	$cv,
158	4	$pageData,
159	4	$time,
160	4	300,
161	4	true,
162	4	false,
163	4	$duplicateTrack,
164	4	$downloadUrls,
165	4	array_keys($this->getCrawlerProcInstructions())
166		);
167
168		//reset the queue because the entries have been written to the db
169	4	unset($crawler->queueEntries);
170		}
171		}
172	4	}
173
174		/**
175		* Method to return the latest Crawle Timestamp for a page.
176		*
177		* @param int $uid uid id of the page
178		* @param bool $future_crawldates_only
179		* @param bool $unprocessed_only
180		*
181		* @return int
182		*/
183	1	public function getLatestCrawlTimestampForPage($uid, $future_crawldates_only = false, $unprocessed_only = false)
184		{
185	1	$uid = intval($uid);
186
187	1	$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName);
188		$query = $queryBuilder
189	1	->from($this->tableName)
190	1	->selectLiteral('max(scheduled) as latest')
191	1	->where(
192	1	$queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid))
193		);
194
195	1	if ($future_crawldates_only) {
196		$query->andWhere(
197		$queryBuilder->expr()->gt('scheduled', time())
198		);
199		}
200
201	1	if ($unprocessed_only) {
202		$query->andWhere(
203		$queryBuilder->expr()->eq('exec_time', 0)
204		);
205		}
206
207	1	$row = $query->execute()->fetch(0);
208	1	if ($row['latest']) {
209	1	$res = $row['latest'];
210		} else {
211		$res = 0;
212		}
213
214	1	return intval($res);
215		}
216
217		/**
218		* Returns an array with timestamps when the page has been scheduled for crawling and
219		* at what time the scheduled crawl has been executed. The array also contains items that are
220		* scheduled but have note been crawled yet.
221		*
222		* @return array array with the crawl-history of a page => 0 : scheduled time , 1 : executed_time, 2 : set_id
223		*/
224	1	public function getCrawlHistoryForPage(int $uid, int $limit = 0)
225		{
226	1	$uid = intval($uid);
227	1	$limit = intval($limit);
228
229	1	$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)->getQueryBuilderForTable($this->tableName);
230		$statement = $queryBuilder
231	1	->from($this->tableName)
232	1	->select('scheduled', 'exec_time', 'set_id')
233	1	->where(
234	1	$queryBuilder->expr()->eq('page_id', $queryBuilder->createNamedParameter($uid, \PDO::PARAM_INT))
235		);
236	1	if ($limit) {
237	1	$statement->setMaxResults($limit);
238		}
239
240	1	return $statement->execute()->fetchAll();
241		}
242
243		/**
244		* Get queue statistics
245		*/
246	1	public function getQueueStatistics(): array
247		{
248		return [
249	1	'assignedButUnprocessed' => $this->queueRepository->countAllAssignedPendingItems(),
250	1	'unprocessed' => $this->queueRepository->countAllPendingItems(),
251		];
252		}
253
254		/**
255		* Get queue statistics by configuration
256		*
257		* @return array array of array('configuration' => <>, 'assignedButUnprocessed' => <>, 'unprocessed' => <>)
258		*/
259		public function getQueueStatisticsByConfiguration()
260		{
261		$statistics = $this->queueRepository->countPendingItemsGroupedByConfigurationKey();
262		$setIds = $this->queueRepository->getSetIdWithUnprocessedEntries();
263		$totals = $this->queueRepository->getTotalQueueEntriesByConfiguration($setIds);
264
265		// "merge" arrays
266		foreach ($statistics as &$value) {
267		$value['total'] = $totals[$value['configuration']];
268		}
269
270		return $statistics;
271		}
272
273		/**
274		* Get active processes count
275		*/
276		public function getActiveProcessesCount(): int
277		{
278		$processRepository = new ProcessRepository();
279
280		return $processRepository->countActive();
281		}
282
283		public function getLastProcessedQueueEntries(int $limit): array
284		{
285		return $this->queueRepository->getLastProcessedEntries($limit);
286		}
287
288		/**
289		* Get current crawling speed
290		*
291		* @return int\|float\|bool
292		*/
293		public function getCurrentCrawlingSpeed()
294		{
295		$lastProcessedEntries = $this->queueRepository->getLastProcessedEntriesTimestamps();
296
297		if (count($lastProcessedEntries) < 10) {
298		// not enough information
299		return false;
300		}
301
302		$tooOldDelta = 60; // time between two entries is "too old"
303
304		$compareValue = time();
305		$startTime = $lastProcessedEntries[0];
306
307		$pages = 0;
308
309		reset($lastProcessedEntries);
310		foreach ($lastProcessedEntries as $timestamp) {
311		if ($compareValue - $timestamp > $tooOldDelta) {
312		break;
313		}
314		$compareValue = $timestamp;
315		$pages++;
316		}
317
318		if ($pages < 10) {
319		// not enough information
320		return false;
321		}
322		$oldestTimestampThatIsNotTooOld = $compareValue;
323		$time = $startTime - $oldestTimestampThatIsNotTooOld;
324
325		return $pages / ($time / 60);
326		}
327
328		/**
329		* Get some performance data
330		*
331		* @param integer $start
332		* @param integer $end
333		* @param integer $resolution
334		*
335		* @return array data
336		*
337		* @throws TimeStampException
338		*/
339		public function getPerformanceData($start, $end, $resolution)
340		{
341		$data = [];
342
343		$data['urlcount'] = 0;
344		$data['start'] = $start;
345		$data['end'] = $end;
346		$data['duration'] = $data['end'] - $data['start'];
347
348		if ($data['duration'] < 1) {
349		throw new TimeStampException('End timestamp must be after start timestamp', 1512659945);
350		}
351
352		for ($slotStart = $start; $slotStart < $end; $slotStart += $resolution) {
353		$slotEnd = min($slotStart + $resolution - 1, $end);
354		$slotData = $this->queueRepository->getPerformanceData($slotStart, $slotEnd);
355
356		$slotUrlCount = 0;
357		foreach ($slotData as &$processData) {
358		$duration = $processData['end'] - $processData['start'];
359		if ($processData['urlcount'] > 5 && $duration > 0) {
360		$processData['speed'] = 60 * 1 / ($duration / $processData['urlcount']);
361		}
362		$slotUrlCount += $processData['urlcount'];
363		}
364
365		$data['urlcount'] += $slotUrlCount;
366
367		$data['slots'][$slotEnd] = [
368		'amountProcesses' => count($slotData),
369		'urlcount' => $slotUrlCount,
370		'processes' => $slotData,
371		];
372
373		if ($slotUrlCount > 5) {
374		$data['slots'][$slotEnd]['speed'] = 60 * 1 / ($slotEnd - $slotStart / $slotUrlCount);
375		} else {
376		$data['slots'][$slotEnd]['speed'] = 0;
377		}
378		}
379
380		if ($data['urlcount'] > 5) {
381		$data['speed'] = 60 * 1 / ($data['duration'] / $data['urlcount']);
382		} else {
383		$data['speed'] = 0;
384		}
385
386		return $data;
387		}
388
389		/**
390		* Method to get an instance of the internal crawler singleton
391		*
392		* @return CrawlerController Instance of the crawler lib
393		*
394		* @throws CrawlerObjectException
395		*/
396	2	protected function findCrawler()
397		{
398	2	if (! is_object($this->crawlerController)) {
399	2	$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class);
400	2	$this->crawlerController->setID = GeneralUtility::md5int(microtime());
401		}
402
403	2	if (is_object($this->crawlerController)) {
404	2	return $this->crawlerController;
405		}
406		throw new CrawlerObjectException('no crawler object', 1512659759);
407		}
408
409		/**
410		* This method is used to limit the processing instructions to the processing instructions
411		* that are allowed.
412		*/
413	4	protected function filterUnallowedConfigurations(array $configurations): array
414		{
415	4	if (count($this->allowedConfigurations) > 0) {
416		// remove configuration that does not match the current selection
417		foreach ($configurations as $confKey => $confArray) {
418		if (! in_array($confKey, $this->allowedConfigurations, true)) {
419		unset($configurations[$confKey]);
420		}
421		}
422		}
423
424	4	return $configurations;
425		}
426
427		/**
428		* Reads the registered processingInstructions of the crawler
429		*/
430	4	private function getCrawlerProcInstructions(): array
431		{
432	4	$crawlerProcInstructions = [];
433	4	if (! empty($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'])) {
434		foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'] as $configuration) {
435		$crawlerProcInstructions[$configuration['key']] = $configuration['value'];
436		}
437		}
438
439	4	return $crawlerProcInstructions;
440		}
441		}
442

AOEpeople / crawler

Push — features/addFlushedPagesToCraw... ( 44f5f4...d700c8 )

CrawlerApi::__construct() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like