CrawlController::setupCrawler() - Code Metrics - Inspection of "wip" - Famdirksen/LaravelJobHandler - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 534ea0...bf2ec8 )

by Robin

created 2018-08-07 09:17 UTC

CrawlController::setupCrawler() C

↳ Parent: CrawlController

Complexity

Conditions	13
Paths	20

Size

Total Lines	81
Code Lines	45

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	45
dl	0
loc	81
rs	6.6166
c	0
b	0
f	0
cc	13
nc	20
nop	1

How to fix Long Method Complexity

<?php

namespace Famdirksen\LaravelJobHandler\Http\Controllers;

use Carbon\Carbon;
use Famdirksen\LaravelJobHandler\Exceptions\CrawlerException;
use Famdirksen\LaravelJobHandler\Exceptions\CrawlerNotReachedTimeBetweenJobsException;
use Famdirksen\LaravelJobHandler\Exceptions\CrawlerSaveException;
use Famdirksen\LaravelJobHandler\Models\Crawlers;
use Famdirksen\LaravelJobHandler\Models\CrawlerStatus;
use Famdirksen\LaravelJobHandler\Models\CrawlerStatusLogs;
use Illuminate\Support\Facades\Log;

class CrawlController
{
    protected $crawler;
    protected $crawler_id;
    protected $override_fail_status = false;
    protected $logging = false;
    protected $logs = [];


    public function __construct()
    {
        $this->startLogging();
    }
    public function __destruct()
    {
        $this->stopLogging();
    }

    public function overrideFailStatus(bool $state) {
        $this->log('Setup overrideFailStatus to: '.$state);

        $this->override_fail_status = $state;
    }

    /**
     * Get the latest crawler data from the database
     */
    protected function getCrawler()
    {
        if (empty($this->crawler) || $this->crawler->id != $this->crawler_id) {
            $this->log('Loading new crawler data');
            $this->crawler = Crawlers::findOrFail($this->crawler_id);
            $this->log('Loaded new crawler data');
        } else {
            $this->log('Refreshing crawler data');
            $this->crawler = $this->crawler->fresh();
            $this->log('Refreshed crawler data');
        }
    }

    /**
     * Set the crawler id
     *
     * @param $crawler_id
     */
    public function setCrawlerId($crawler_id)
    {
        $this->log('Setting crawler_id');
        $this->crawler_id = $crawler_id;
        $this->log('Set crawler_id');
    }

    /**
     * Return the crawler id
     *
     * @return mixed
     */
    public function getCrawlerId()
    {
        $this->log('Getting crawler_id');

        return $this->crawler_id;
    }

    /**
     * Check if the controller is setup correctly
     *
     * @return bool
     */
    protected function controllerIsSetup() {
        $this->log('Check if controllerIsSetup');

        if(!is_null($this->crawler_id)) {
            return true;
        }

        return false;
    }

    /**
     * Setup the crawler so it won't run twice at the same time
     *
     * @param $crawler_id
     */
    public function setupCrawler($crawler_id = null)
    {
        $this->log('Setup crawler');

        if(!is_null($crawler_id)) {
            $this->log('Setup crawler, crawler_id is not set');
            $this->setCrawlerId($crawler_id);
        }

        if ($this->controllerIsSetup()) {
            $times = config('laravel-job-handler.run_times', 10);

            for ($x = 0; $x <= $times; $x++) {
                //fetch the last data
                $this->getCrawler();

                $this->log('Checking if crawler is enabled');
                if (!$this->crawler->enabled) {
                    $this->log('Crawler is not enabled');
                    throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - crawler isnt enabled in database');
                }

                $this->log('Checking if crawler can be runned');
                $checkIfCrawlerCanBeRunned = $this->canCrawlerRunAfterPeriod();

                if ($checkIfCrawlerCanBeRunned['status']) {
                    $this->log('Checked if crawler can runned');
                    if (is_null($this->crawler->latest_status)) {
                        $this->log('Crawler can be runned, it the first time');

                        //first time it runs...
                        break;
                    }
                    if ($this->crawler->latest_status == 2) {
                        $this->log('Crawler can be runned, last crawler runned successfully');

                        //Done running...
                        break;
                    }



                    if ($this->crawler->latest_status == 3) {
                        if($this->override_fail_status) {
                            $this->log('Last crawler failed, but it is forced to run');

                            //override the failed state, this will force to rerun...
                            break;
                        }

                        $this->log('Last crawler failed, force run is not enabled');
                        throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - last run had an error and override_fail_status is not enabled');
                    }
                } else {
                    $this->log('Crawler needs to wait ('.$checkIfCrawlerCanBeRunned['retry_in'].' seconds) before running again');
                    throw new CrawlerNotReachedTimeBetweenJobsException('Has to wait ' . $checkIfCrawlerCanBeRunned['retry_in'] . ' more seconds to run');
                }

                if ($x == $times) {
                    $this->log('Crawler exceeded the max execution time');
                    $this->failCrawler('Crawler (#' . $this->crawler_id . ') - max execution time');
                }

                if ($this->crawler->status == 1) {
                    if ($this->crawler->multiple_crawlers) {
                        $this->log('Crawler can run multiple crawlers at the same time');
                        break;
                    }

                    $wait = config('laravel-job-handler.retry_in_seconds', 3);

                    $this->log('Waiting for rechecking ('.$wait.' seconds) if crawler can be runned');

                    sleep($wait);
                }
            }

            $this->log('All setup, starting crawler');
            $this->startCrawler();
        } else {
            throw new CrawlerException('CrawlController is not setup correctly.');
        }
    }
    /**
     * Start the crawler and save it to the database
     *
     * @param string $output
     */
    public function startCrawler($output = '')
    {
        $this->log('Starting crawler');

        return $this->addStatus(1, $output); //start running
    }
    /**
     * set the crawler as done so other scripts can run
     *
     * @param string $output
     */
    public function doneCrawler($output = '')
    {
        $this->log('Crawler done');

        return $this->addStatus(2, $output); //done running
    }

    /**
     * Finishing the crawler
     *
     * @param string $output
     * @return bool
     */
    public function finish($output = '')
    {
        $this->log('Finishing crawler');

        return $this->doneCrawler($output);
    }
    /**
     * crawler failed...
     *
     * @param string $output
     */
    public function failCrawler($output = '')
    {
        $this->log('Crawler failed');

        $this->addStatus(3, $output); //failed

        throw new CrawlerException($output.' - status 3');
    }
    /**
     * Save the latest crawler status to the database
     *
     * @param $status
     * @param string $output
     * @return bool
     */
    protected function addStatus($status, $output = '')
    {
        $this->log('Registering status ('.$status.')');

        $crawlerstatus = new CrawlerStatus();

        $crawlerstatus->crawler_id = $this->crawler_id;
        $crawlerstatus->status = $status;

        if ($crawlerstatus->save()) {
            $this->log('Registered status ('.$status.')');
            $this->log('Setting crawler latest status ('.$status.') attribute');

            $this->crawler->latest_status = $status;

            $this->crawler->save();
            $this->log('Set crawler latest status ('.$status.') attribute');


            if (!empty($output)) {
                $formatted_logs[] = [

                    'status_id' => $crawlerstatus->id,
                    'output' => $output
                ];

                CrawlerStatusLogs::insert($formatted_logs);
            }

            if($status == 2) {
                $this->stopLogging($crawlerstatus->id);
            }

            $this->getCrawler();

            return true;
        } else {
            throw new CrawlerSaveException('Cannot save crawlerstatus to database...');
        }
    }
    protected function saveLog($crawlerstatus_id) {
        $formatted_logs = [];

        foreach($this->logs as $log) {
            $formatted_logs[] = [
                'status_id' => $crawlerstatus_id,
                'output' => $log
            ];
        }
        if(count($formatted_logs) > 0) {
            $this->log('Registering crawler logs');

            CrawlerStatusLogs::insert($formatted_logs);

            $this->log('Registered crawler logs (count: ' . count($formatted_logs) . ')');
        } else {
            $this->log('Log output is not set, skipping inserting');
        }
    }

    /**
     * This will define when the job can be runned again
     *
     * @return array
     */
    public function canCrawlerRunAfterPeriod()
    {
        $this->getCrawler();

        if (is_null($this->crawler->time_between)) {
            $this->log('Not time_between specified');

            return $this->canCrawlerRunAfterPeriodStatus(true);
        } else {
            $seconds = $this->crawler->time_between;
        }

        if (!is_null($this->crawler->last_runned_at)) {
            if ($this->crawler->last_runned_at <= Carbon::now()->subSeconds($seconds)) {
                return $this->canCrawlerRunAfterPeriodStatus(true);
            }

            return $this->canCrawlerRunAfterPeriodStatus(false, Carbon::parse($this->crawler->last_runned_at)->diffInSeconds(Carbon::now()->subSeconds($seconds)));
        } else {
            //crawler never runned, so it can run now
            return $this->canCrawlerRunAfterPeriodStatus(true);
        }
    }

    /**
     * Return the status for canCrawlerRunAfterPeriod method
     *
     * @param $status
     * @param int $retry_in
     * @return array
     */
    public function canCrawlerRunAfterPeriodStatus($status, $retry_in = 0)
    {
        return [
            'status' => $status,
            'retry_in' => $retry_in
        ];
    }



    protected function startLogging()
    {
        $this->logging = true;
        $this->log('Started logging');
    }
    protected function stopLogging($crawlerstatus_id = null)
    {
        $this->log('Stop logging');
        $this->logging = false;

        if(!is_null($crawlerstatus_id)) {
            $this->saveLog($crawlerstatus_id);
        }
    }
    protected function log($item = '')
    {
        if($this->logging)
        {
            $log = $item.' (crawler_id: '.$this->crawler_id.')';

            $this->logs[] = $log;
            Log::info($log);
        }
    }
}


1			<?php
2
3			namespace Famdirksen\LaravelJobHandler\Http\Controllers;
4
5			use Carbon\Carbon;
6			use Famdirksen\LaravelJobHandler\Exceptions\CrawlerException;
7			use Famdirksen\LaravelJobHandler\Exceptions\CrawlerNotReachedTimeBetweenJobsException;
8			use Famdirksen\LaravelJobHandler\Exceptions\CrawlerSaveException;
9			use Famdirksen\LaravelJobHandler\Models\Crawlers;
10			use Famdirksen\LaravelJobHandler\Models\CrawlerStatus;
11			use Famdirksen\LaravelJobHandler\Models\CrawlerStatusLogs;
12			use Illuminate\Support\Facades\Log;
13
14			class CrawlController
15			{
16			protected $crawler;
17			protected $crawler_id;
18			protected $override_fail_status = false;
19			protected $logging = false;
20			protected $logs = [];
21
22
23			public function __construct()
24			{
25			$this->startLogging();
26			}
27			public function __destruct()
28			{
29			$this->stopLogging();
30			}
31
32			public function overrideFailStatus(bool $state) {
33			$this->log('Setup overrideFailStatus to: '.$state);
34
35			$this->override_fail_status = $state;
36			}
37
38			/**
39			* Get the latest crawler data from the database
40			*/
41			protected function getCrawler()
42			{
43			if (empty($this->crawler) \|\| $this->crawler->id != $this->crawler_id) {
44			$this->log('Loading new crawler data');
45			$this->crawler = Crawlers::findOrFail($this->crawler_id);
46			$this->log('Loaded new crawler data');
47			} else {
48			$this->log('Refreshing crawler data');
49			$this->crawler = $this->crawler->fresh();
50			$this->log('Refreshed crawler data');
51			}
52			}
53
54			/**
55			* Set the crawler id
56			*
57			* @param $crawler_id
58			*/
59			public function setCrawlerId($crawler_id)
60			{
61			$this->log('Setting crawler_id');
62			$this->crawler_id = $crawler_id;
63			$this->log('Set crawler_id');
64			}
65
66			/**
67			* Return the crawler id
68			*
69			* @return mixed
70			*/
71			public function getCrawlerId()
72			{
73			$this->log('Getting crawler_id');
74
75			return $this->crawler_id;
76			}
77
78			/**
79			* Check if the controller is setup correctly
80			*
81			* @return bool
82			*/
83			protected function controllerIsSetup() {
84			$this->log('Check if controllerIsSetup');
85
86			if(!is_null($this->crawler_id)) {
87			return true;
88			}
89
90			return false;
91			}
92
93			/**
94			* Setup the crawler so it won't run twice at the same time
95			*
96			* @param $crawler_id
97			*/
98			public function setupCrawler($crawler_id = null)
99			{
100			$this->log('Setup crawler');
101
102			if(!is_null($crawler_id)) {
103			$this->log('Setup crawler, crawler_id is not set');
104			$this->setCrawlerId($crawler_id);
105			}
106
107			if ($this->controllerIsSetup()) {
108			$times = config('laravel-job-handler.run_times', 10);
109
110			for ($x = 0; $x <= $times; $x++) {
111			//fetch the last data
112			$this->getCrawler();
113
114			$this->log('Checking if crawler is enabled');
115			if (!$this->crawler->enabled) {
116			$this->log('Crawler is not enabled');
117			throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - crawler isnt enabled in database');
118			}
119
120			$this->log('Checking if crawler can be runned');
121			$checkIfCrawlerCanBeRunned = $this->canCrawlerRunAfterPeriod();
122
123			if ($checkIfCrawlerCanBeRunned['status']) {
124			$this->log('Checked if crawler can runned');
125			if (is_null($this->crawler->latest_status)) {
126			$this->log('Crawler can be runned, it the first time');
127
128			//first time it runs...
129			break;
130			}
131			if ($this->crawler->latest_status == 2) {
132			$this->log('Crawler can be runned, last crawler runned successfully');
133
134			//Done running...
135			break;
136			}
137
138
139
140			if ($this->crawler->latest_status == 3) {
141			if($this->override_fail_status) {
142			$this->log('Last crawler failed, but it is forced to run');
143
144			//override the failed state, this will force to rerun...
145			break;
146			}
147
148			$this->log('Last crawler failed, force run is not enabled');
149			throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - last run had an error and override_fail_status is not enabled');
150			}
151			} else {
152			$this->log('Crawler needs to wait ('.$checkIfCrawlerCanBeRunned['retry_in'].' seconds) before running again');
153			throw new CrawlerNotReachedTimeBetweenJobsException('Has to wait ' . $checkIfCrawlerCanBeRunned['retry_in'] . ' more seconds to run');
154			}
155
156			if ($x == $times) {
157			$this->log('Crawler exceeded the max execution time');
158			$this->failCrawler('Crawler (#' . $this->crawler_id . ') - max execution time');
159			}
160
161			if ($this->crawler->status == 1) {
162			if ($this->crawler->multiple_crawlers) {
163			$this->log('Crawler can run multiple crawlers at the same time');
164			break;
165			}
166
167			$wait = config('laravel-job-handler.retry_in_seconds', 3);
168
169			$this->log('Waiting for rechecking ('.$wait.' seconds) if crawler can be runned');
170
171			sleep($wait);
172			}
173			}
174
175			$this->log('All setup, starting crawler');
176			$this->startCrawler();
177			} else {
178			throw new CrawlerException('CrawlController is not setup correctly.');
179			}
180			}
181			/**
182			* Start the crawler and save it to the database
183			*
184			* @param string $output
185			*/
186			public function startCrawler($output = '')
187			{
188			$this->log('Starting crawler');
189
190			return $this->addStatus(1, $output); //start running
191			}
192			/**
193			* set the crawler as done so other scripts can run
194			*
195			* @param string $output
196			*/
197			public function doneCrawler($output = '')
198			{
199			$this->log('Crawler done');
200
201			return $this->addStatus(2, $output); //done running
202			}
203
204			/**
205			* Finishing the crawler
206			*
207			* @param string $output
208			* @return bool
209			*/
210			public function finish($output = '')
211			{
212			$this->log('Finishing crawler');
213
214			return $this->doneCrawler($output);
215			}
216			/**
217			* crawler failed...
218			*
219			* @param string $output
220			*/
221			public function failCrawler($output = '')
222			{
223			$this->log('Crawler failed');
224
225			$this->addStatus(3, $output); //failed
226
227			throw new CrawlerException($output.' - status 3');
228			}
229			/**
230			* Save the latest crawler status to the database
231			*
232			* @param $status
233			* @param string $output
234			* @return bool
235			*/
236			protected function addStatus($status, $output = '')
237			{
238			$this->log('Registering status ('.$status.')');
239
240			$crawlerstatus = new CrawlerStatus();
241
242			$crawlerstatus->crawler_id = $this->crawler_id;
243			$crawlerstatus->status = $status;
244
245			if ($crawlerstatus->save()) {
246			$this->log('Registered status ('.$status.')');
247			$this->log('Setting crawler latest status ('.$status.') attribute');
248
249			$this->crawler->latest_status = $status;
250
251			$this->crawler->save();
252			$this->log('Set crawler latest status ('.$status.') attribute');
253
254
255			if (!empty($output)) {
256			$formatted_logs[] = [
			0 ignored issues – show Comprehensibility Best Practice introduced 2018-08-07 09:20 UTC by Report Bug Copy Issue Report `$formatted_logs` was never initialized. Although not strictly required by PHP, it is generally a good practice to add `$formatted_logs = array();` before regardless. Loading history...
257			'status_id' => $crawlerstatus->id,
258			'output' => $output
259			];
260
261			CrawlerStatusLogs::insert($formatted_logs);
262			}
263
264			if($status == 2) {
265			$this->stopLogging($crawlerstatus->id);
266			}
267
268			$this->getCrawler();
269
270			return true;
271			} else {
272			throw new CrawlerSaveException('Cannot save crawlerstatus to database...');
273			}
274			}
275			protected function saveLog($crawlerstatus_id) {
276			$formatted_logs = [];
277
278			foreach($this->logs as $log) {
279			$formatted_logs[] = [
280			'status_id' => $crawlerstatus_id,
281			'output' => $log
282			];
283			}
284			if(count($formatted_logs) > 0) {
285			$this->log('Registering crawler logs');
286
287			CrawlerStatusLogs::insert($formatted_logs);
288
289			$this->log('Registered crawler logs (count: ' . count($formatted_logs) . ')');
290			} else {
291			$this->log('Log output is not set, skipping inserting');
292			}
293			}
294
295			/**
296			* This will define when the job can be runned again
297			*
298			* @return array
299			*/
300			public function canCrawlerRunAfterPeriod()
301			{
302			$this->getCrawler();
303
304			if (is_null($this->crawler->time_between)) {
305			$this->log('Not time_between specified');
306
307			return $this->canCrawlerRunAfterPeriodStatus(true);
308			} else {
309			$seconds = $this->crawler->time_between;
310			}
311
312			if (!is_null($this->crawler->last_runned_at)) {
313			if ($this->crawler->last_runned_at <= Carbon::now()->subSeconds($seconds)) {
314			return $this->canCrawlerRunAfterPeriodStatus(true);
315			}
316
317			return $this->canCrawlerRunAfterPeriodStatus(false, Carbon::parse($this->crawler->last_runned_at)->diffInSeconds(Carbon::now()->subSeconds($seconds)));
318			} else {
319			//crawler never runned, so it can run now
320			return $this->canCrawlerRunAfterPeriodStatus(true);
321			}
322			}
323
324			/**
325			* Return the status for canCrawlerRunAfterPeriod method
326			*
327			* @param $status
328			* @param int $retry_in
329			* @return array
330			*/
331			public function canCrawlerRunAfterPeriodStatus($status, $retry_in = 0)
332			{
333			return [
334			'status' => $status,
335			'retry_in' => $retry_in
336			];
337			}
338
339
340
341			protected function startLogging()
342			{
343			$this->logging = true;
344			$this->log('Started logging');
345			}
346			protected function stopLogging($crawlerstatus_id = null)
347			{
348			$this->log('Stop logging');
349			$this->logging = false;
350
351			if(!is_null($crawlerstatus_id)) {
352			$this->saveLog($crawlerstatus_id);
353			}
354			}
355			protected function log($item = '')
356			{
357			if($this->logging)
358			{
359			$log = $item.' (crawler_id: '.$this->crawler_id.')';
360
361			$this->logs[] = $log;
362			Log::info($log);
363			}
364			}
365			}
366

Famdirksen / LaravelJobHandler

Push — master ( 534ea0...bf2ec8 )

CrawlController::setupCrawler() C

Complexity

Size

Duplication

Importance

How to fix Long Method Complexity

Long Method

Duplication Side-by-Side

Filter issues like