SolrIndexTask::runChildProcess() - Code Metrics - Inspection of "Improved background processing" - Firesphere/silverstripe-solr-search - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — bufferfix ( 9ef1b7...96911a )

by Simon

created 2019-11-24 02:45 UTC

SolrIndexTask::runChildProcess() A

↳ Parent: Project

Complexity

Conditions	6
Paths	9

Size

Total Lines	21
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	42

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	6
eloc	12
c	1
b	0
f	0
nc	9
nop	5
dl	0
loc	21
ccs	0
cts	0
cp	0
crap	42
rs	9.2222

<?php


namespace Firesphere\SolrSearch\Tasks;

use Exception;
use Firesphere\SolrSearch\Factories\DocumentFactory;
use Firesphere\SolrSearch\Helpers\SolrLogger;
use Firesphere\SolrSearch\Indexes\BaseIndex;
use Firesphere\SolrSearch\Services\SolrCoreService;
use Firesphere\SolrSearch\States\SiteState;
use Firesphere\SolrSearch\Traits\LoggerTrait;
use GuzzleHttp\Exception\GuzzleException;
use Psr\Log\LoggerInterface;
use ReflectionException;
use SilverStripe\Control\Director;
use SilverStripe\Control\HTTPRequest;
use SilverStripe\Core\Injector\Injector;
use SilverStripe\Dev\BuildTask;
use SilverStripe\ORM\ArrayList;
use SilverStripe\ORM\DataList;
use SilverStripe\ORM\DataObject;
use SilverStripe\ORM\DB;
use SilverStripe\ORM\ValidationException;
use SilverStripe\Versioned\Versioned;

/**
 * Class SolrIndexTask
 *
 * @description Index items to Solr through a tasks
 * @package Firesphere\SolrSearch\Tasks
 */
class SolrIndexTask extends BuildTask
{
    use LoggerTrait;
    /**
     * URLSegment of this task
     *
     * @var string
     */
    private static $segment = 'SolrIndexTask';
    /**
     * Store the current states for all instances of SiteState
     *
     * @var array
     */
    public $currentStates;
    /**
     * My name
     *
     * @var string
     */
    protected $title = 'Solr Index update';
    /**
     * What do I do?
     *
     * @var string
     */
    protected $description = 'Add or update documents to an existing Solr core.';
    /**
     * Debug mode enabled, default false
     *
     * @var bool
     */
    protected $debug = false;
    /**
     * Singleton of {@link SolrCoreService}
     *
     * @var SolrCoreService
     */
    protected $service;

    /**
     * SolrIndexTask constructor. Sets up the document factory
     *
     * @throws ReflectionException
     */
    public function __construct()
    {
        parent::__construct();
        // Only index live items.
        // The old FTS module also indexed Draft items. This is unnecessary
        Versioned::set_reading_mode(Versioned::DEFAULT_MODE);
        // If versioned is needed, a separate Versioned Search module is required
        $this->setService(Injector::inst()->get(SolrCoreService::class));
        $this->setLogger(Injector::inst()->get(LoggerInterface::class));
        $this->setDebug(Director::isDev() || Director::is_cli());
        $currentStates = SiteState::currentStates();
        SiteState::setDefaultStates($currentStates);
    }

    /**
     * Set the {@link SolrCoreService}
     *
     * @param SolrCoreService $service
     * @return SolrIndexTask
     */
    public function setService(SolrCoreService $service): SolrIndexTask
    {
        $this->service = $service;

        return $this;
    }

    /**
     * Set the debug mode
     *
     * @param bool $debug
     * @return SolrIndexTask
     */
    public function setDebug(bool $debug): SolrIndexTask
    {
        $this->debug = $debug;

        return $this;
    }

    /**
     * Implement this method in the task subclass to
     * execute via the TaskRunner
     *
     * @param HTTPRequest $request
     * @return int|bool
     * @throws Exception
     * @throws GuzzleException
     * @todo defer to background because it may run out of memory
     */
    public function run($request)
    {
        $startTime = time();
        $this->getLogger()->info(date('Y-m-d H:i:s'));
        list($vars, $group, $isGroup) = $this->taskSetup($request);
        $groups = 0;
        $indexes = $this->service->getValidIndexes($request->getVar('index'));

        foreach ($indexes as $indexName) {
            /** @var BaseIndex $index */
            $index = Injector::inst()->get($indexName, false);

            $indexClasses = $index->getClasses();
            $classes = $this->getClasses($vars, $indexClasses);
            if (!count($classes)) {
                continue;
            }

            $this->clearIndex($vars, $index);

            $groups = $this->indexClassForIndex($classes, $isGroup, $index, $group);
        }

        $this->getLogger()->info(sprintf('Finished in %s', date('H:i:s', time() - $startTime)));

        return $groups;
    }

    /**
     * Set up the requirements for this task
     *
     * @param HTTPRequest $request
     * @return array
     */
    protected function taskSetup($request): array
    {
        $vars = $request->getVars();
        $this->debug = $this->debug || isset($vars['debug']);
        $group = $vars['group'] ?? 0;
        $start = $vars['start'] ?? 0;
        $group = ($start > $group) ? $start : $group;
        $isGroup = isset($vars['group']);

        return [$vars, $group, $isGroup];
    }

    /**
     * get the classes to run for this task execution
     *
     * @param $vars
     * @param array $classes
     * @return bool|array
     */
    protected function getClasses($vars, array $classes): array
    {
        if (isset($vars['class'])) {
            return array_intersect($classes, [$vars['class']]);
        }

        return $classes;
    }

    /**
     * Clear the given index if a full re-index is needed
     *
     * @param $vars
     * @param BaseIndex $index
     * @throws Exception
     */
    public function clearIndex($vars, BaseIndex $index)
    {
        if (!empty($vars['clear'])) {
            $this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName()));
            $this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index);
        }
    }

    /**
     * Index the classes for a specific index
     *
     * @param $classes
     * @param $isGroup
     * @param BaseIndex $index
     * @param $group
     * @return int
     * @throws Exception
     * @throws GuzzleException
     */
    protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int
    {
        $groups = 0;
        foreach ($classes as $class) {
            $groups = $this->indexClass($isGroup, $class, $index, $group);
        }

        return $groups;
    }

    /**
     * Index a single class for a given index. {@link static::indexClassForIndex()}
     *
     * @param bool $isGroup
     * @param string $class
     * @param BaseIndex $index
     * @param int $group
     * @return int
     * @throws GuzzleException
     * @throws ValidationException
     */
    private function indexClass($isGroup, $class, BaseIndex $index, int $group): int
    {
        $this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()), []);

        $batchLength = DocumentFactory::config()->get('batchLength');
        $groups = (int)ceil($class::get()->count() / $batchLength);
        $groups = $isGroup ? $group : $groups;
        $this->getLogger()->info(sprintf('Total groups %s', $groups));
        do { // Run from oldest to newest
            try {
                // Temporary workaround for CircleCI
                if (function_exists('pcntl_fork')) {
                    $cores = SolrCoreService::config()->get('cores');
                    $pids = [];
                    // for each core, start a grouped indexing
                    for ($i = 0; $i < $cores; $i++) {
                        $group += $i;
                        $pid = pcntl_fork();
                        // PID needs to be pushed before anything else, for some reason
                        $pids[$i] = $pid;
                        if (!$pid && $group < $groups) {
                            $this->doReindex($group, $class, $batchLength, $index);
                        }
                    }
                    // Wait for each child to finish
                    foreach ($pids as $pid) {
                        if ($pid) {
                            pcntl_waitpid($pid, $status);
                        }
                    }
                } else {
                    $this->doReindex($group, $class, $batchLength, $index);
                }
            } catch (Exception $error) {
                $this->logException($index->getIndexName(), $group, $error);
                $group++;
                continue;
            }
            $group++;
        } while ($group <= $groups);

        return $groups;
    }

    /**
     * Reindex the given group, for each state
     *
     * @param int $group
     * @param string $class
     * @param int $batchLength
     * @param BaseIndex $index
     * @throws Exception
     */
    private function doReindex($group, $class, $batchLength, BaseIndex $index): int
    {
        $config = DB::getConfig();
        DB::connect($config);
        foreach (SiteState::getStates() as $state) {
            if ($state !== 'default' && !empty($state)) {
                SiteState::withState($state);
            }
            $this->stateReindex($group, $class, $batchLength, $index);
        }

        SiteState::withState(SiteState::DEFAULT_STATE);
        $this->getLogger()->info(sprintf('Indexed group %s', $group ));

        exit(0);
interface ReturnsInt {
    public function returnsIntHinted(): int;
}

class MyClass implements ReturnsInt {
    public function returnsIntHinted(): int
    {
        if (foo()) {
            return 123;
        }
        // here: null is implicitly returned
    }
}
    }

    /**
     * Index a group of a class for a specific state and index
     *
     * @param $group
     * @param $class
     * @param $batchLength
     * @param BaseIndex $index
     * @throws Exception
     */
    private function stateReindex($group, $class, $batchLength, BaseIndex $index): void
    {
        // Generate filtered list of local records
        $baseClass = DataObject::getSchema()->baseDataClass($class);
        /** @var DataList|DataObject[] $items */
        $items = DataObject::get($baseClass)
            ->sort('ID ASC')
            ->limit($batchLength, ($group * $batchLength));
        if ($items->count()) {
            $this->updateIndex($index, $items);
        }
    }

    /**
     * Execute the update on the client
     *
     * @param BaseIndex $index
     * @param $items
     * @throws Exception
     */
    private function updateIndex(BaseIndex $index, $items): void
    {
        $client = $index->getClient();
        $update = $client->createUpdate();
        $this->service->setInDebugMode($this->debug);
        $this->service->updateIndex($index, $items, $update);
        $update->addCommit();
        $client->update($update);
    }

    /**
     * Log an exception if it happens. Most are catched, these logs are for the developers
     * to identify problems and fix them.
     *
     * @param string $index
     * @param int $group
     * @param Exception $exception
     * @throws GuzzleException
     * @throws ValidationException
     */
    private function logException($index, int $group, Exception $exception): void
    {
        $this->getLogger()->error($exception->getMessage());
        $msg = sprintf(
            'Error indexing core %s on group %s,' . PHP_EOL .
            'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL,
            $index,
            $group
        );
        SolrLogger::logMessage('ERROR', $msg, $index);
    }
}


1		<?php
2
3
4		namespace Firesphere\SolrSearch\Tasks;
5
6		use Exception;
7		use Firesphere\SolrSearch\Factories\DocumentFactory;
8		use Firesphere\SolrSearch\Helpers\SolrLogger;
9		use Firesphere\SolrSearch\Indexes\BaseIndex;
10		use Firesphere\SolrSearch\Services\SolrCoreService;
11		use Firesphere\SolrSearch\States\SiteState;
12		use Firesphere\SolrSearch\Traits\LoggerTrait;
13		use GuzzleHttp\Exception\GuzzleException;
14		use Psr\Log\LoggerInterface;
15		use ReflectionException;
16		use SilverStripe\Control\Director;
17		use SilverStripe\Control\HTTPRequest;
18		use SilverStripe\Core\Injector\Injector;
19		use SilverStripe\Dev\BuildTask;
20		use SilverStripe\ORM\ArrayList;
21		use SilverStripe\ORM\DataList;
22		use SilverStripe\ORM\DataObject;
23		use SilverStripe\ORM\DB;
24		use SilverStripe\ORM\ValidationException;
25		use SilverStripe\Versioned\Versioned;
26
27		/**
28		* Class SolrIndexTask
29		*
30		* @description Index items to Solr through a tasks
31		* @package Firesphere\SolrSearch\Tasks
32		*/
33		class SolrIndexTask extends BuildTask
34		{
35		use LoggerTrait;
36		/**
37		* URLSegment of this task
38		*
39		* @var string
40		*/
41		private static $segment = 'SolrIndexTask';
42		/**
43		* Store the current states for all instances of SiteState
44		*
45		* @var array
46		*/
47		public $currentStates;
48		/**
49		* My name
50		*
51		* @var string
52		*/
53		protected $title = 'Solr Index update';
54		/**
55		* What do I do?
56		*
57		* @var string
58		*/
59		protected $description = 'Add or update documents to an existing Solr core.';
60		/**
61		* Debug mode enabled, default false
62		*
63		* @var bool
64		*/
65		protected $debug = false;
66		/**
67		* Singleton of {@link SolrCoreService}
68		*
69		* @var SolrCoreService
70		*/
71		protected $service;
72
73		/**
74		* SolrIndexTask constructor. Sets up the document factory
75		*
76		* @throws ReflectionException
77	14	*/
78		public function __construct()
79	14	{
80		parent::__construct();
81		// Only index live items.
82	14	// The old FTS module also indexed Draft items. This is unnecessary
83	14	Versioned::set_reading_mode(Versioned::DEFAULT_MODE);
84	14	// If versioned is needed, a separate Versioned Search module is required
85	14	$this->setService(Injector::inst()->get(SolrCoreService::class));
86	14	$this->setLogger(Injector::inst()->get(LoggerInterface::class));
87	14	$this->setDebug(Director::isDev() \|\| Director::is_cli());
88	14	$currentStates = SiteState::currentStates();
89		SiteState::setDefaultStates($currentStates);
90		}
91
92		/**
93		* Set the {@link SolrCoreService}
94		*
95		* @param SolrCoreService $service
96	14	* @return SolrIndexTask
97		*/
98	14	public function setService(SolrCoreService $service): SolrIndexTask
99		{
100	14	$this->service = $service;
101
102		return $this;
103		}
104
105		/**
106		* Set the debug mode
107		*
108		* @param bool $debug
109	14	* @return SolrIndexTask
110		*/
111	14	public function setDebug(bool $debug): SolrIndexTask
112		{
113	14	$this->debug = $debug;
114
115		return $this;
116		}
117
118		/**
119		* Implement this method in the task subclass to
120		* execute via the TaskRunner
121		*
122		* @param HTTPRequest $request
123		* @return int\|bool
124		* @throws Exception
125		* @throws GuzzleException
126	13	* @todo defer to background because it may run out of memory
127		*/
128	13	public function run($request)
129	13	{
130	13	$startTime = time();
131	13	$this->getLogger()->info(date('Y-m-d H:i:s'));
132		list($vars, $group, $isGroup) = $this->taskSetup($request);
133	13	$groups = 0;
134		$indexes = $this->service->getValidIndexes($request->getVar('index'));
135	13
136		foreach ($indexes as $indexName) {
137	13	/** @var BaseIndex $index */
138	13	$index = Injector::inst()->get($indexName, false);
139	13
140	10	$indexClasses = $index->getClasses();
141		$classes = $this->getClasses($vars, $indexClasses);
142		if (!count($classes)) {
143	13	continue;
144		}
145	13
146		$this->clearIndex($vars, $index);
147	13
148	13	$groups = $this->indexClassForIndex($classes, $isGroup, $index, $group);
149		}
150
151	13	$this->getLogger()->info(sprintf('Finished in %s', date('H:i:s', time() - $startTime)));
152	13
153		return $groups;
154	13	}
155
156		/**
157		* Set up the requirements for this task
158		*
159		* @param HTTPRequest $request
160		* @return array
161		*/
162		protected function taskSetup($request): array
163	13	{
164		$vars = $request->getVars();
165	13	$this->debug = $this->debug \|\| isset($vars['debug']);
166	13	$group = $vars['group'] ?? 0;
167	13	$start = $vars['start'] ?? 0;
168	13	$group = ($start > $group) ? $start : $group;
169	13	$isGroup = isset($vars['group']);
170
171	13	return [$vars, $group, $isGroup];
172		}
173
174		/**
175		* get the classes to run for this task execution
176		*
177		* @param $vars
178		* @param array $classes
179		* @return bool\|array
180		*/
181	13	protected function getClasses($vars, array $classes): array
182		{
183	13	if (isset($vars['class'])) {
184	1	return array_intersect($classes, [$vars['class']]);
185		}
186
187	12	return $classes;
188		}
189
190		/**
191		* Clear the given index if a full re-index is needed
192		*
193		* @param $vars
194		* @param BaseIndex $index
195		* @throws Exception
196		*/
197	13	public function clearIndex($vars, BaseIndex $index)
198		{
199	13	if (!empty($vars['clear'])) {
200	1	$this->getLogger()->info(sprintf('Clearing index %s', $index->getIndexName()));
201	1	$this->service->doManipulate(ArrayList::create([]), SolrCoreService::DELETE_TYPE_ALL, $index);
202		}
203	13	}
204
205		/**
206		* Index the classes for a specific index
207		*
208		* @param $classes
209		* @param $isGroup
210		* @param BaseIndex $index
211		* @param $group
212		* @return int
213		* @throws Exception
214		* @throws GuzzleException
215	13	*/
216		protected function indexClassForIndex($classes, $isGroup, BaseIndex $index, $group): int
217	13	{
218	13	$groups = 0;
219	13	foreach ($classes as $class) {
220		$groups = $this->indexClass($isGroup, $class, $index, $group);
221		}
222	13
223		return $groups;
224		}
225
226		/**
227		* Index a single class for a given index. {@link static::indexClassForIndex()}
228		*
229		* @param bool $isGroup
230		* @param string $class
231		* @param BaseIndex $index
232		* @param int $group
233		* @return int
234		* @throws GuzzleException
235	13	* @throws ValidationException
236		*/
237	13	private function indexClass($isGroup, $class, BaseIndex $index, int $group): int
238		{
239		$this->getLogger()->info(sprintf('Indexing %s for %s', $class, $index->getIndexName()), []);
240	13
241		$batchLength = DocumentFactory::config()->get('batchLength');
242		$groups = (int)ceil($class::get()->count() / $batchLength);
243		$groups = $isGroup ? $group : $groups;
244		$this->getLogger()->info(sprintf('Total groups %s', $groups));
245	13	do { // Run from oldest to newest
246		try {
247		// Temporary workaround for CircleCI
248		if (function_exists('pcntl_fork')) {
249		$cores = SolrCoreService::config()->get('cores');
250		$pids = [];
251		// for each core, start a grouped indexing
252		for ($i = 0; $i < $cores; $i++) {
253		$group += $i;
254		$pid = pcntl_fork();
255		// PID needs to be pushed before anything else, for some reason
256		$pids[$i] = $pid;
257	13	if (!$pid && $group < $groups) {
258		$this->doReindex($group, $class, $batchLength, $index);
259	13	}
260	13	}
261		// Wait for each child to finish
262		foreach ($pids as $pid) {
263	13	if ($pid) {
264		pcntl_waitpid($pid, $status);
265		}
266	13	}
267	13	} else {
268		$this->doReindex($group, $class, $batchLength, $index);
269		}
270		} catch (Exception $error) {
271		$this->logException($index->getIndexName(), $group, $error);
272		$group++;
273		continue;
274		}
275		$group++;
276		} while ($group <= $groups);
277	13
278		return $groups;
279		}
280	13
281		/**
282	13	* Reindex the given group, for each state
283	13	*
284	13	* @param int $group
285	13	* @param string $class
286	13	* @param int $batchLength
287		* @param BaseIndex $index
288	13	* @throws Exception
289	1	*/
290		private function doReindex($group, $class, $batchLength, BaseIndex $index): int
291	13	{
292		$config = DB::getConfig();
293		DB::connect($config);
294		foreach (SiteState::getStates() as $state) {
295		if ($state !== 'default' && !empty($state)) {
296		SiteState::withState($state);
297		}
298		$this->stateReindex($group, $class, $batchLength, $index);
299		}
300	1
301		SiteState::withState(SiteState::DEFAULT_STATE);
302	1	$this->getLogger()->info(sprintf('Indexed group %s', $group ));
303	1
304	1	exit(0);
		0 ignored issues – show Bug Best Practice introduced 2019-11-24 02:53 UTC by Report Bug Copy Issue Report In this branch, the function will implicitly return `null` which is incompatible with the type-hinted return `integer`. Consider adding a return statement or allowing `null` as return value. For hinted functions/methods where all return statements with the correct type are only reachable via conditions, ?null? gets implicitly returned which may be incompatible with the hinted type. Let?s take a look at an example: interface ReturnsInt { public function returnsIntHinted(): int; } class MyClass implements ReturnsInt { public function returnsIntHinted(): int { if (foo()) { return 123; } // here: null is implicitly returned } } Loading history... Best Practice introduced 2019-11-24 02:53 UTC by Report Bug Copy Issue Report Using `exit` here is not recommended. In general, usage of exit should be done with care and only when running in a scripting context like a CLI script. Loading history...
305		}
306
307		/**
308		* Index a group of a class for a specific state and index
309		*
310		* @param $group
311		* @param $class
312		* @param $batchLength
313		* @param BaseIndex $index
314		* @throws Exception
315		*/
316		private function stateReindex($group, $class, $batchLength, BaseIndex $index): void
317		{
318		// Generate filtered list of local records
319		$baseClass = DataObject::getSchema()->baseDataClass($class);
320		/** @var DataList\|DataObject[] $items */
321		$items = DataObject::get($baseClass)
322		->sort('ID ASC')
323		->limit($batchLength, ($group * $batchLength));
324		if ($items->count()) {
325		$this->updateIndex($index, $items);
326		}
327		}
328
329		/**
330		* Execute the update on the client
331		*
332		* @param BaseIndex $index
333		* @param $items
334		* @throws Exception
335		*/
336		private function updateIndex(BaseIndex $index, $items): void
337		{
338		$client = $index->getClient();
339		$update = $client->createUpdate();
340		$this->service->setInDebugMode($this->debug);
341		$this->service->updateIndex($index, $items, $update);
342		$update->addCommit();
343		$client->update($update);
344		}
345
346		/**
347		* Log an exception if it happens. Most are catched, these logs are for the developers
348		* to identify problems and fix them.
349		*
350		* @param string $index
351		* @param int $group
352		* @param Exception $exception
353		* @throws GuzzleException
354		* @throws ValidationException
355		*/
356		private function logException($index, int $group, Exception $exception): void
357		{
358		$this->getLogger()->error($exception->getMessage());
359		$msg = sprintf(
360		'Error indexing core %s on group %s,' . PHP_EOL .
361		'Please log in to the CMS to find out more about Indexing errors' . PHP_EOL,
362		$index,
363		$group
364		);
365		SolrLogger::logMessage('ERROR', $msg, $index);
366		}
367		}
368

Firesphere / silverstripe-solr-search

Push — bufferfix ( 9ef1b7...96911a )

SolrIndexTask::runChildProcess() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like