CrawlerApi::countUnprocessedItems() - Code Metrics - Inspection of "[FEATURE] Deprecates EventDispatcher + Adds Signal..." - AOEpeople/crawler - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — issue/252 ( b288df...34f654 )

by Tomas Norre

created 2019-04-29 17:05 UTC

CrawlerApi::countUnprocessedItems() A

↳ Parent: CrawlerApi

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	2

Importance

Changes

Metric	Value
cc	1
nc	1
nop	0
dl	0
loc	9
ccs	0
cts	8
cp	0
crap	2
rs	9.9666
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\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;
    }

    /**
     * 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 $time timestamp
     */
    public function addPageToQueueTimed($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 $timestamp
     *
     * @return bool
     */
    public function isPageInQueue($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 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);
        $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 $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.
     *
     * @return array
     */
    public function getUnprocessedItems()
    {
        $query = '*';
        $where = 'exec_time = 0';
        $rows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows($query, 'tx_crawler_queue', $where, '', '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
     */
    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 $show_unprocessed only respect unprocessed pages
     *
     * @return boolean
     */
    public function isPageInQueueTimed($uid, $show_unprocessed = true)
    {
        $uid = intval($uid);

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

AOEpeople / crawler

Push — issue/252 ( b288df...34f654 )

CrawlerApi::countUnprocessedItems() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like