CrawlerApi::__construct() - Code Metrics - Inspection of "[TASK] Deprecated functions (#387)" - AOEpeople/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 6a3171...e461ad )

by Tomas Norre

created 2019-07-19 12:36 UTC

CrawlerApi::__construct() A

↳ Parent: CrawlerApi

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	1.008

Importance

Changes

Metric	Value
cc	1
nc	1
nop	0
dl	0
loc	5
ccs	4
cts	5
cp	0.8
crap	1.008
rs	10
c	0
b	0
f	0

<?php
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 TYPO3\CMS\Core\Utility\GeneralUtility;
use TYPO3\CMS\Core\Utility\MathUtility;
use TYPO3\CMS\Extbase\Object\ObjectManager;
use TYPO3\CMS\Frontend\Page\PageRepository;

/**
 * Class CrawlerApi
 *
 * @package AOE\Crawler\Api
 */
class CrawlerApi
{
    /**
     * @var CrawlerController
     */
    private $crawlerController;

    /**
     * @var QueueRepository
     */
    protected $queueRepository;

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

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

    /**
     * This method is used to limit the configuration selection to
     * a set of configurations.
     *
     * @param array $allowedConfigurations
     */
    public function setAllowedConfigurations(array $allowedConfigurations)
    {
        $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;
    }

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

    /**
     * Method to get an instance of the internal crawler singleton
     *
     * @return CrawlerController Instance of the crawler lib
     *
     * @throws \Exception
     */
    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;
        } else {
            throw new \Exception('no crawler object', 1512659759);
        }
    }

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

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

        return $configurations;
    }

    /**
     * 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)
    {
        $uid = intval($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);
            }
        } else {
            //no configuration found
        }
    }

    /**
     * Counts all entries in the database which are scheduled for a given page id and a schedule timestamp.
     *
     * @param int $page_uid
     * @param int $schedule_timestamp
     *
     * @return int
     *
     * @deprecated since crawler v6.2.0, will be removed in crawler v7.0.0.
     */
    protected function countEntriesInQueueForPageByScheduleTime($page_uid, $schedule_timestamp)
    {
        $page_uid = intval($page_uid);
        $schedule_timestamp = intval($schedule_timestamp);

        //if the same page is scheduled for the same time and has not be executed?
        if ($schedule_timestamp == 0) {
            //un-timed elements need an exec_time with 0 because they can occur multiple times
            $where = 'page_id=' . $page_uid . ' AND exec_time = 0 AND scheduled=' . $schedule_timestamp;
        } else {
            //timed elements have got a fixed schedule time, if a record with this time
            //exists it is maybe queued for the future, or is has been queue for the past and therefore
            //also been processed.
            $where = 'page_id=' . $page_uid . ' AND scheduled=' . $schedule_timestamp;
        }

        $row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($GLOBALS['TYPO3_DB']->exec_SELECTquery(
            'count(*) as cnt',
            'tx_crawler_queue',
            $where
        ));

        return intval($row['cnt']);
    }

    /**
     * Determines if a page is queued
     *
     * @param $uid
     * @param bool $unprocessed_only
     * @param bool $timed_only
     * @param bool $timestamp
     *
     * @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
     *
     * @return bool
     */
    public function isPageInQueue($uid, $unprocessed_only = true, $timed_only = false, $timestamp = false)
    {
        if (!MathUtility::canBeInterpretedAsInteger($uid)) {
            throw new \InvalidArgumentException('Invalid parameter type', 1468931945);
        }

        $isPageInQueue = false;

        $whereClause = 'page_id = ' . (integer)$uid;

        if (false !== $unprocessed_only) {
            $whereClause .= ' AND exec_time = 0';
        }

        if (false !== $timed_only) {
            $whereClause .= ' AND scheduled != 0';
        }

        if (false !== $timestamp) {
            $whereClause .= ' AND scheduled = ' . (integer)$timestamp;
        }

        $count = $GLOBALS['TYPO3_DB']->exec_SELECTcountRows(
            '*',
            'tx_crawler_queue',
            $whereClause
        );

        if (false !== $count && $count > 0) {
            $isPageInQueue = true;
        }

        return $isPageInQueue;
    }

    /**
     * Method to return the latest Crawler 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);
        $query = 'max(scheduled) as latest';
        $where = ' page_id = ' . $uid;

        if ($future_crawldates_only) {
            $where .= ' AND scheduled > ' . time();
        }

        if ($unprocessed_only) {
            $where .= ' AND exec_time = 0';
        }

        $rs = $GLOBALS['TYPO3_DB']->exec_SELECTquery($query, 'tx_crawler_queue', $where);
        if ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($rs)) {
            $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.
     *
     * @param int $uid uid of the page
     * @param bool $limit
     *
     * @return array array with the crawl-history of a page => 0 : scheduled time , 1 : executed_time, 2 : set_id
     */
    public function getCrawlHistoryForPage($uid, $limit = 0)
    {
        $uid = intval($uid);
        $limit = intval($limit);

        $query = 'scheduled, exec_time, set_id';
        $where = ' page_id = ' . $uid;

        $limit_query = ($limit) ? $limit : null;

        $rows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows($query, 'tx_crawler_queue', $where, null, null, $limit_query);
        return $rows;
    }

    /**
     * Method to determine unprocessed Items in the crawler queue.
     *
     * @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
     *
     * @return array
     */
    public function getUnprocessedItems()
    {
        $rows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
            '*',
            'tx_crawler_queue',
            'exec_time = 0',
            '',
            'page_id, scheduled'
        );

        return $rows;
    }

    /**
     * Method to get the number of unprocessed items in the crawler
     *
     * @param int number of unprocessed items in the queue
     * @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
     *
     */
    public function countUnprocessedItems()
    {
        $query = 'count(page_id) as anz';
        $where = 'exec_time = 0';
        $rs = $GLOBALS['TYPO3_DB']->exec_SELECTquery($query, 'tx_crawler_queue', $where);
        $row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($rs);

        return $row['anz'];
    }

    /**
     * Method to check if a page is in the queue which is timed for a
     * date when it should be crawled
     *
     * @param int $uid uid of the page
     * @param boolean $show_unprocessed only respect unprocessed pages
     *
     * @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
     *
     * @return boolean
     */
    public function isPageInQueueTimed($uid, $show_unprocessed = true)
    {
        $uid = intval($uid);
        return $this->queueRepository->isPageInQueue($uid, $show_unprocessed);
    }

    /**
     * Reads the registered processingInstructions of the crawler
     *
     * @return array
     */
    private function getCrawlerProcInstructions()
    {
        if (isset($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'])) {
            return $GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'];
        }

        return [];
    }

    /**
     * Removes an queue entry with a given queue id
     *
     * @param int $qid
     * @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0. Please use the QueueRepository instead
     */
    public function removeQueueEntrie($qid)
    {
        $qid = intval($qid);
        $table = 'tx_crawler_queue';
        $where = ' qid=' . $qid;
        $GLOBALS['TYPO3_DB']->exec_DELETEquery($table, $where);
    }

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

    /**
     * Get queue repository
     *
     * @return QueueRepository
     */
    protected function getQueueRepository()
    {
        if (!$this->queueRepository instanceof QueueRepository) {
            $this->queueRepository = new QueueRepository();
        }

        return $this->queueRepository;
    }

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

        $setIds = $this->getQueueRepository()->getSetIdWithUnprocessedEntries();

        $totals = $this->getQueueRepository()->getTotalQueueEntriesByConfiguration($setIds);

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

        return $statistics;
    }

    /**
     * Get active processes count
     *
     * @param void
     *
     * @return int
     */
    public function getActiveProcessesCount()
    {
        $processRepository = new ProcessRepository();

        return $processRepository->countActive();
    }

    /**
     * Get last processed entries
     *
     * @param int $limit
     *
     * @return array
     */
    public function getLastProcessedQueueEntries($limit)
    {
        return $this->getQueueRepository()->getLastProcessedEntries('*', $limit);
    }

    /**
     * Get current crawling speed
     *
     * @param float|false page speed in pages per minute
     *
     * @return int
     */
    public function getCurrentCrawlingSpeed()
    {
        $lastProcessedEntries = $this->getQueueRepository()->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 $key => $timestamp) {
            if ($compareValue - $timestamp > $tooOldDelta) {
                break;
            }
            $compareValue = $timestamp;
            $pages++;
        }

        if ($pages < 10) {
            // not enough information
            return false;
        }
        $oldestTimestampThatIsNotTooOld = $compareValue;
        $time = $startTime - $oldestTimestampThatIsNotTooOld;
        $speed = $pages / ($time / 60);

        return $speed;
    }

    /**
     * Get some performance data
     *
     * @param integer $start
     * @param integer $end
     * @param integer $resolution
     *
     * @return array data
     *
     * @throws \Exception
     */
    public function getPerformanceData($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 \Exception('End timestamp must be after start timestamp', 1512659945);
        }

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

            $slotUrlCount = 0;
            foreach ($slotData as $processId => &$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;
    }
}


1		<?php
2		namespace AOE\Crawler\Api;
3
4		/***************************************************************
5		* Copyright notice
6		*
7		* (c) 2018 AOE GmbH <[email protected]>
8		*
9		* All rights reserved
10		*
11		* This script is part of the TYPO3 project. The TYPO3 project is
12		* free software; you can redistribute it and/or modify
13		* it under the terms of the GNU General Public License as published by
14		* the Free Software Foundation; either version 3 of the License, or
15		* (at your option) any later version.
16		*
17		* The GNU General Public License can be found at
18		* http://www.gnu.org/copyleft/gpl.html.
19		*
20		* This script is distributed in the hope that it will be useful,
21		* but WITHOUT ANY WARRANTY; without even the implied warranty of
22		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23		* GNU General Public License for more details.
24		*
25		* This copyright notice MUST APPEAR in all copies of the script!
26		***************************************************************/
27
28		use AOE\Crawler\Controller\CrawlerController;
29		use AOE\Crawler\Domain\Repository\ProcessRepository;
30		use AOE\Crawler\Domain\Repository\QueueRepository;
31		use TYPO3\CMS\Core\Utility\GeneralUtility;
32		use TYPO3\CMS\Core\Utility\MathUtility;
33		use TYPO3\CMS\Extbase\Object\ObjectManager;
34		use TYPO3\CMS\Frontend\Page\PageRepository;
35
36		/**
37		* Class CrawlerApi
38		*
39		* @package AOE\Crawler\Api
40		*/
41		class CrawlerApi
42		{
43		/**
44		* @var CrawlerController
45		*/
46		private $crawlerController;
47
48		/**
49		* @var QueueRepository
50		*/
51		protected $queueRepository;
52
53		/**
54		* @var $allowedConfigurations array
55		*/
56		protected $allowedConfigurations = [];
57
58		/**
59		* Each crawler run has a setid, this facade method delegates
60		* the it to the crawler object
61		*
62		* @param int
63		*/
64	1	public function overwriteSetId($id)
65		{
66	1	$this->findCrawler()->setID = intval($id);
67	1	}
68
69		/**
70		* This method is used to limit the configuration selection to
71		* a set of configurations.
72		*
73		* @param array $allowedConfigurations
74		*/
75	1	public function setAllowedConfigurations(array $allowedConfigurations)
76		{
77	1	$this->allowedConfigurations = $allowedConfigurations;
78	1	}
79
80		/**
81		* @return array
82		*/
83	1	public function getAllowedConfigurations()
84		{
85	1	return $this->allowedConfigurations;
86		}
87
88		/**
89		* Returns the setID of the crawler
90		*
91		* @return int
92		*/
93	1	public function getSetId()
94		{
95	1	return $this->findCrawler()->setID;
96		}
97
98	19	public function __construct()
99		{
100	19	$objectManager = GeneralUtility::makeInstance(ObjectManager::class);
101	19	$this->queueRepository = $objectManager->get(QueueRepository::class);
102	19	}
103
104		/**
105		* Method to get an instance of the internal crawler singleton
106		*
107		* @return CrawlerController Instance of the crawler lib
108		*
109		* @throws \Exception
110		*/
111	2	protected function findCrawler()
112		{
113	2	if (!is_object($this->crawlerController)) {
114	2	$this->crawlerController = GeneralUtility::makeInstance(CrawlerController::class);
115	2	$this->crawlerController->setID = GeneralUtility::md5int(microtime());
116		}
117
118	2	if (is_object($this->crawlerController)) {
119	2	return $this->crawlerController;
120		} else {
121		throw new \Exception('no crawler object', 1512659759);
122		}
123		}
124
125		/**
126		* Adds a page to the crawlerqueue by uid
127		*
128		* @param int $uid uid
129		*/
130		public function addPageToQueue($uid)
131		{
132		$uid = intval($uid);
133		//non timed elements will be added with timestamp 0
134		$this->addPageToQueueTimed($uid, 0);
135		}
136
137		/**
138		* This method is used to limit the processing instructions to the processing instructions
139		* that are allowed.
140		*
141		* @return array
142		*/
143	4	protected function filterUnallowedConfigurations($configurations)
144		{
145	4	if (count($this->allowedConfigurations) > 0) {
146		// remove configuration that does not match the current selection
147		foreach ($configurations as $confKey => $confArray) {
148		if (!in_array($confKey, $this->allowedConfigurations)) {
149		unset($configurations[$confKey]);
150		}
151		}
152		}
153
154	4	return $configurations;
155		}
156
157		/**
158		* Adds a page to the crawlerqueue by uid and sets a
159		* timestamp when the page should be crawled.
160		*
161		* @param int $uid pageid
162		* @param int $time timestamp
163		*
164		* @throws \Exception
165		*/
166	4	public function addPageToQueueTimed($uid, $time)
167		{
168	4	$uid = intval($uid);
169	4	$time = intval($time);
170
171	4	$crawler = $this->findCrawler();
172	4	$pageData = GeneralUtility::makeInstance(PageRepository::class)->getPage($uid);
173	4	$configurations = $crawler->getUrlsForPageRow($pageData);
174	4	$configurations = $this->filterUnallowedConfigurations($configurations);
175	4	$downloadUrls = [];
176	4	$duplicateTrack = [];
177
178	4	if (is_array($configurations)) {
179	4	foreach ($configurations as $cv) {
180		//enable inserting of entries
181	4	$crawler->registerQueueEntriesInternallyOnly = false;
182	4	$crawler->urlListFromUrlArray(
183	4	$cv,
184	4	$pageData,
185	4	$time,
186	4	300,
187	4	true,
188	4	false,
189	4	$duplicateTrack,
190	4	$downloadUrls,
191	4	array_keys($this->getCrawlerProcInstructions())
192		);
193
194		//reset the queue because the entries have been written to the db
195	4	unset($crawler->queueEntries);
196		}
197		} else {
198		//no configuration found
199		}
200	4	}
201
202		/**
203		* Counts all entries in the database which are scheduled for a given page id and a schedule timestamp.
204		*
205		* @param int $page_uid
206		* @param int $schedule_timestamp
207		*
208		* @return int
209		*
210		* @deprecated since crawler v6.2.0, will be removed in crawler v7.0.0.
211		*/
212	1	protected function countEntriesInQueueForPageByScheduleTime($page_uid, $schedule_timestamp)
213		{
214	1	$page_uid = intval($page_uid);
215	1	$schedule_timestamp = intval($schedule_timestamp);
216
217		//if the same page is scheduled for the same time and has not be executed?
218	1	if ($schedule_timestamp == 0) {
219		//un-timed elements need an exec_time with 0 because they can occur multiple times
220	1	$where = 'page_id=' . $page_uid . ' AND exec_time = 0 AND scheduled=' . $schedule_timestamp;
221		} else {
222		//timed elements have got a fixed schedule time, if a record with this time
223		//exists it is maybe queued for the future, or is has been queue for the past and therefore
224		//also been processed.
225	1	$where = 'page_id=' . $page_uid . ' AND scheduled=' . $schedule_timestamp;
226		}
227
228	1	$row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($GLOBALS['TYPO3_DB']->exec_SELECTquery(
229	1	'count(*) as cnt',
230	1	'tx_crawler_queue',
231	1	$where
232		));
233
234	1	return intval($row['cnt']);
235		}
236
237		/**
238		* Determines if a page is queued
239		*
240		* @param $uid
241		* @param bool $unprocessed_only
242		* @param bool $timed_only
243		* @param bool $timestamp
244		*
245		* @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
246		*
247		* @return bool
248		*/
249	5	public function isPageInQueue($uid, $unprocessed_only = true, $timed_only = false, $timestamp = false)
250		{
251	5	if (!MathUtility::canBeInterpretedAsInteger($uid)) {
252	1	throw new \InvalidArgumentException('Invalid parameter type', 1468931945);
253		}
254
255	4	$isPageInQueue = false;
256
257	4	$whereClause = 'page_id = ' . (integer)$uid;
258
259	4	if (false !== $unprocessed_only) {
260	1	$whereClause .= ' AND exec_time = 0';
261		}
262
263	4	if (false !== $timed_only) {
264	1	$whereClause .= ' AND scheduled != 0';
265		}
266
267	4	if (false !== $timestamp) {
268	1	$whereClause .= ' AND scheduled = ' . (integer)$timestamp;
269		}
270
271	4	$count = $GLOBALS['TYPO3_DB']->exec_SELECTcountRows(
272	4	'*',
273	4	'tx_crawler_queue',
274	4	$whereClause
275		);
276
277	4	if (false !== $count && $count > 0) {
278	3	$isPageInQueue = true;
279		}
280
281	4	return $isPageInQueue;
282		}
283
284		/**
285		* Method to return the latest Crawler Timestamp for a page.
286		*
287		* @param int $uid uid id of the page
288		* @param bool $future_crawldates_only
289		* @param bool $unprocessed_only
290		*
291		* @return int
292		*/
293	1	public function getLatestCrawlTimestampForPage($uid, $future_crawldates_only = false, $unprocessed_only = false)
294		{
295	1	$uid = intval($uid);
296	1	$query = 'max(scheduled) as latest';
297	1	$where = ' page_id = ' . $uid;
298
299	1	if ($future_crawldates_only) {
300		$where .= ' AND scheduled > ' . time();
301		}
302
303	1	if ($unprocessed_only) {
304		$where .= ' AND exec_time = 0';
305		}
306
307	1	$rs = $GLOBALS['TYPO3_DB']->exec_SELECTquery($query, 'tx_crawler_queue', $where);
308	1	if ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($rs)) {
309	1	$res = $row['latest'];
310		} else {
311		$res = 0;
312		}
313
314	1	return intval($res);
315		}
316
317		/**
318		* Returns an array with timestamps when the page has been scheduled for crawling and
319		* at what time the scheduled crawl has been executed. The array also contains items that are
320		* scheduled but have note been crawled yet.
321		*
322		* @param int $uid uid of the page
323		* @param bool $limit
324		*
325		* @return array array with the crawl-history of a page => 0 : scheduled time , 1 : executed_time, 2 : set_id
326		*/
327	1	public function getCrawlHistoryForPage($uid, $limit = 0)
328		{
329	1	$uid = intval($uid);
330	1	$limit = intval($limit);
331
332	1	$query = 'scheduled, exec_time, set_id';
333	1	$where = ' page_id = ' . $uid;
334
335	1	$limit_query = ($limit) ? $limit : null;
336
337	1	$rows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows($query, 'tx_crawler_queue', $where, null, null, $limit_query);
338	1	return $rows;
339		}
340
341		/**
342		* Method to determine unprocessed Items in the crawler queue.
343		*
344		* @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
345		*
346		* @return array
347		*/
348	1	public function getUnprocessedItems()
349		{
350	1	$rows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows(
351	1	'*',
352	1	'tx_crawler_queue',
353	1	'exec_time = 0',
354	1	'',
355	1	'page_id, scheduled'
356		);
357
358	1	return $rows;
359		}
360
361		/**
362		* Method to get the number of unprocessed items in the crawler
363		*
364		* @param int number of unprocessed items in the queue
365		* @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
366		*
367		*/
368	3	public function countUnprocessedItems()
369		{
370	3	$query = 'count(page_id) as anz';
371	3	$where = 'exec_time = 0';
372	3	$rs = $GLOBALS['TYPO3_DB']->exec_SELECTquery($query, 'tx_crawler_queue', $where);
373	3	$row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($rs);
374
375	3	return $row['anz'];
376		}
377
378		/**
379		* Method to check if a page is in the queue which is timed for a
380		* date when it should be crawled
381		*
382		* @param int $uid uid of the page
383		* @param boolean $show_unprocessed only respect unprocessed pages
384		*
385		* @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0.
386		*
387		* @return boolean
388		*/
389	1	public function isPageInQueueTimed($uid, $show_unprocessed = true)
390		{
391	1	$uid = intval($uid);
392	1	return $this->queueRepository->isPageInQueue($uid, $show_unprocessed);
393		}
394
395		/**
396		* Reads the registered processingInstructions of the crawler
397		*
398		* @return array
399		*/
400	4	private function getCrawlerProcInstructions()
401		{
402	4	if (isset($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'])) {
403		return $GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['procInstructions'];
404		}
405
406	4	return [];
407		}
408
409		/**
410		* Removes an queue entry with a given queue id
411		*
412		* @param int $qid
413		* @deprecated since crawler v6.5.1, will be removed in crawler v9.0.0. Please use the QueueRepository instead
414		*/
415		public function removeQueueEntrie($qid)
416		{
417		$qid = intval($qid);
418		$table = 'tx_crawler_queue';
419		$where = ' qid=' . $qid;
420		$GLOBALS['TYPO3_DB']->exec_DELETEquery($table, $where);
421		}
422
423		/**
424		* Get queue statistics
425		*
426		* @param void
427		*
428		* @return array array('assignedButUnprocessed' => <>, 'unprocessed' => <>);
429		*/
430	1	public function getQueueStatistics()
431		{
432		return [
433	1	'assignedButUnprocessed' => $this->getQueueRepository()->countAllAssignedPendingItems(),
434	1	'unprocessed' => $this->getQueueRepository()->countAllPendingItems()
435		];
436		}
437
438		/**
439		* Get queue repository
440		*
441		* @return QueueRepository
442		*/
443	2	protected function getQueueRepository()
444		{
445	2	if (!$this->queueRepository instanceof QueueRepository) {
446		$this->queueRepository = new QueueRepository();
447		}
448
449	2	return $this->queueRepository;
450		}
451
452		/**
453		* Get queue statistics by configuration
454		*
455		* @return array array of array('configuration' => <>, 'assignedButUnprocessed' => <>, 'unprocessed' => <>)
456		*/
457		public function getQueueStatisticsByConfiguration()
458		{
459		$statistics = $this->getQueueRepository()->countPendingItemsGroupedByConfigurationKey();
460
461		$setIds = $this->getQueueRepository()->getSetIdWithUnprocessedEntries();
462
463		$totals = $this->getQueueRepository()->getTotalQueueEntriesByConfiguration($setIds);
464
465		// "merge" arrays
466		foreach ($statistics as $key => &$value) {
467		$value['total'] = $totals[$value['configuration']];
468		}
469
470		return $statistics;
471		}
472
473		/**
474		* Get active processes count
475		*
476		* @param void
477		*
478		* @return int
479		*/
480		public function getActiveProcessesCount()
481		{
482		$processRepository = new ProcessRepository();
483
484		return $processRepository->countActive();
485		}
486
487		/**
488		* Get last processed entries
489		*
490		* @param int $limit
491		*
492		* @return array
493		*/
494		public function getLastProcessedQueueEntries($limit)
495		{
496		return $this->getQueueRepository()->getLastProcessedEntries('*', $limit);
497		}
498
499		/**
500		* Get current crawling speed
501		*
502		* @param float\|false page speed in pages per minute
503		*
504		* @return int
505		*/
506		public function getCurrentCrawlingSpeed()
507		{
508		$lastProcessedEntries = $this->getQueueRepository()->getLastProcessedEntriesTimestamps();
509
510		if (count($lastProcessedEntries) < 10) {
511		// not enough information
512		return false;
513		}
514
515		$tooOldDelta = 60; // time between two entries is "too old"
516
517		$compareValue = time();
518		$startTime = $lastProcessedEntries[0];
519
520		$pages = 0;
521
522		reset($lastProcessedEntries);
523		foreach ($lastProcessedEntries as $key => $timestamp) {
524		if ($compareValue - $timestamp > $tooOldDelta) {
525		break;
526		}
527		$compareValue = $timestamp;
528		$pages++;
529		}
530
531		if ($pages < 10) {
532		// not enough information
533		return false;
534		}
535		$oldestTimestampThatIsNotTooOld = $compareValue;
536		$time = $startTime - $oldestTimestampThatIsNotTooOld;
537		$speed = $pages / ($time / 60);
538
539		return $speed;
540		}
541
542		/**
543		* Get some performance data
544		*
545		* @param integer $start
546		* @param integer $end
547		* @param integer $resolution
548		*
549		* @return array data
550		*
551		* @throws \Exception
552		*/
553		public function getPerformanceData($start, $end, $resolution)
554		{
555		$data = [];
556
557		$data['urlcount'] = 0;
558		$data['start'] = $start;
559		$data['end'] = $end;
560		$data['duration'] = $data['end'] - $data['start'];
561
562		if ($data['duration'] < 1) {
563		throw new \Exception('End timestamp must be after start timestamp', 1512659945);
564		}
565
566		for ($slotStart = $start; $slotStart < $end; $slotStart += $resolution) {
567		$slotEnd = min($slotStart + $resolution - 1, $end);
568		$slotData = $this->getQueueRepository()->getPerformanceData($slotStart, $slotEnd);
569
570		$slotUrlCount = 0;
571		foreach ($slotData as $processId => &$processData) {
572		$duration = $processData['end'] - $processData['start'];
573		if ($processData['urlcount'] > 5 && $duration > 0) {
574		$processData['speed'] = 60 * 1 / ($duration / $processData['urlcount']);
575		}
576		$slotUrlCount += $processData['urlcount'];
577		}
578
579		$data['urlcount'] += $slotUrlCount;
580
581		$data['slots'][$slotEnd] = [
582		'amountProcesses' => count($slotData),
583		'urlcount' => $slotUrlCount,
584		'processes' => $slotData,
585		];
586
587		if ($slotUrlCount > 5) {
588		$data['slots'][$slotEnd]['speed'] = 60 * 1 / ($slotEnd - $slotStart / $slotUrlCount);
589		} else {
590		$data['slots'][$slotEnd]['speed'] = 0;
591		}
592		}
593
594		if ($data['urlcount'] > 5) {
595		$data['speed'] = 60 * 1 / ($data['duration'] / $data['urlcount']);
596		} else {
597		$data['speed'] = 0;
598		}
599
600		return $data;
601		}
602		}
603

AOEpeople / crawler

Push — master ( 6a3171...e461ad )

CrawlerApi::__construct() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like