|
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
|
|
|
|
|
13
|
|
|
class CrawlController |
|
14
|
|
|
{ |
|
15
|
|
|
protected $crawler; |
|
16
|
|
|
protected $crawler_id; |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Get the latest crawler data from the database |
|
21
|
|
|
*/ |
|
22
|
|
|
protected function getCrawler() |
|
23
|
|
|
{ |
|
24
|
|
|
if(empty($this->crawler) || $this->crawler->id != $this->crawler_id) { |
|
25
|
|
|
$this->crawler = Crawlers::findOrFail($this->crawler_id); |
|
26
|
|
|
} else { |
|
27
|
|
|
$this->crawler = $this->crawler->fresh(); |
|
28
|
|
|
} |
|
29
|
|
|
} |
|
30
|
|
|
public function setCrawlerId($crawler_id) { |
|
31
|
|
|
$this->crawler_id = $crawler_id; |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Setup the crawler so it won't run twice at the same time |
|
36
|
|
|
* |
|
37
|
|
|
* @param $crawler_id |
|
38
|
|
|
*/ |
|
39
|
|
|
public function setupCrawler($crawler_id) |
|
40
|
|
|
{ |
|
41
|
|
|
$this->setCrawlerId($crawler_id); |
|
42
|
|
|
$times = config('laravel-job-handler.run_times', 10); |
|
43
|
|
|
|
|
44
|
|
|
for ($x = 0; $x <= $times; $x++) { |
|
45
|
|
|
//fetch the last data |
|
46
|
|
|
$this->getCrawler(); |
|
47
|
|
|
|
|
48
|
|
|
if (!$this->crawler->enabled) { |
|
49
|
|
|
throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - crawler isnt enabled in database'); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
|
|
53
|
|
|
$checkIfCrawlerCanBeRunned = $this->canCrawlerRunAfterPeriod(); |
|
54
|
|
|
|
|
55
|
|
|
if($checkIfCrawlerCanBeRunned['status']) { |
|
56
|
|
|
if (is_null($this->crawler->latest_status)) { |
|
57
|
|
|
//first time it runs... |
|
58
|
|
|
break; |
|
59
|
|
|
} |
|
60
|
|
|
if ($this->crawler->latest_status == 2) { |
|
61
|
|
|
//Done running... |
|
62
|
|
|
break; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
if ($this->crawler->latest_status == 3) { |
|
67
|
|
|
throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - last run had an error'); |
|
68
|
|
|
} |
|
69
|
|
|
} else { |
|
70
|
|
|
throw new CrawlerNotReachedTimeBetweenJobsException('Has to wait '.$checkIfCrawlerCanBeRunned['retry_in'].' more seconds to run'); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
if ($x == $times) { |
|
74
|
|
|
$this->failCrawler('Crawler (#'.$this->crawler_id.') - max execution time'); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
if ($this->crawler->status == 1) { |
|
78
|
|
|
if ($this->crawler->multiple_crawlers) { |
|
79
|
|
|
break; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
sleep(config('laravel-job-handler.retry_in_seconds', 3)); //retry in 3 seconds |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
$this->startCrawler(); |
|
87
|
|
|
} |
|
88
|
|
|
/** |
|
89
|
|
|
* Start the crawler and save it to the database |
|
90
|
|
|
* |
|
91
|
|
|
* @param string $output |
|
92
|
|
|
*/ |
|
93
|
|
|
public function startCrawler($output = '') |
|
94
|
|
|
{ |
|
95
|
|
|
return $this->addStatus(1, $output); //start running |
|
96
|
|
|
} |
|
97
|
|
|
/** |
|
98
|
|
|
* set the crawler as done so other scripts can run |
|
99
|
|
|
* |
|
100
|
|
|
* @param string $output |
|
101
|
|
|
*/ |
|
102
|
|
|
public function doneCrawler($output = '') |
|
103
|
|
|
{ |
|
104
|
|
|
return $this->addStatus(2, $output); //done running |
|
105
|
|
|
} |
|
106
|
|
|
public function finish($output = '') |
|
107
|
|
|
{ |
|
108
|
|
|
return $this->doneCrawler($output); |
|
109
|
|
|
} |
|
110
|
|
|
/** |
|
111
|
|
|
* crawler failed... |
|
112
|
|
|
* |
|
113
|
|
|
* @param string $output |
|
114
|
|
|
*/ |
|
115
|
|
|
public function failCrawler($output = '') |
|
116
|
|
|
{ |
|
117
|
|
|
$this->addStatus(3, $output); //failed |
|
118
|
|
|
|
|
119
|
|
|
throw new CrawlerException($output.' - status 3'); |
|
120
|
|
|
} |
|
121
|
|
|
/** |
|
122
|
|
|
* save the latest crawler status to the database |
|
123
|
|
|
* |
|
124
|
|
|
* @param $status |
|
125
|
|
|
* @param string $output |
|
126
|
|
|
* @return bool |
|
127
|
|
|
*/ |
|
128
|
|
|
protected function addStatus($status, $output = '') |
|
129
|
|
|
{ |
|
130
|
|
|
$crawlerstatus = new CrawlerStatus(); |
|
131
|
|
|
|
|
132
|
|
|
$crawlerstatus->crawler_id = $this->crawler_id; |
|
133
|
|
|
$crawlerstatus->status = $status; |
|
134
|
|
|
|
|
135
|
|
|
if ($crawlerstatus->save()) { |
|
136
|
|
|
$this->crawler->latest_status = $status; |
|
137
|
|
|
|
|
138
|
|
|
$this->crawler->save(); |
|
139
|
|
|
|
|
140
|
|
|
if (!empty($output)) { |
|
141
|
|
|
$crawlerstatuslog = new CrawlerStatusLogs(); |
|
142
|
|
|
|
|
143
|
|
|
$crawlerstatuslog->status_id = $crawlerstatus->id; |
|
|
|
|
|
|
144
|
|
|
$crawlerstatuslog->output = $output; |
|
|
|
|
|
|
145
|
|
|
|
|
146
|
|
|
$crawlerstatuslog->save(); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
return true; |
|
150
|
|
|
} else { |
|
151
|
|
|
throw new CrawlerSaveException('Cannot save crawlerstatus to database...'); |
|
152
|
|
|
} |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
/** |
|
156
|
|
|
* This will define when the job can be runned again |
|
157
|
|
|
* |
|
158
|
|
|
* @return array |
|
159
|
|
|
*/ |
|
160
|
|
|
public function canCrawlerRunAfterPeriod() { |
|
161
|
|
|
$this->getCrawler(); |
|
162
|
|
|
|
|
163
|
|
|
if(is_null($this->crawler->time_between)) { |
|
164
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
|
165
|
|
|
} else { |
|
166
|
|
|
$seconds = $this->crawler->time_between; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
if(!is_null($this->crawler->last_runned_at)) { |
|
170
|
|
|
if ($this->crawler->last_runned_at <= Carbon::now()->subSeconds($seconds)) { |
|
171
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(false, Carbon::parse($this->crawler->last_runned_at)->diffInSeconds(Carbon::now()->subSeconds($seconds))); |
|
175
|
|
|
} else { |
|
176
|
|
|
//crawler never runned, so it can run now |
|
177
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
|
178
|
|
|
} |
|
179
|
|
|
} |
|
180
|
|
|
public function canCrawlerRunAfterPeriodStatus($status, $retry_in = 0) { |
|
181
|
|
|
return [ |
|
182
|
|
|
'status' => $status, |
|
183
|
|
|
'retry_in' => $retry_in |
|
184
|
|
|
]; |
|
185
|
|
|
} |
|
186
|
|
|
} |
|
187
|
|
|
|
Checks if undeclared accessed properties appear in database migrations and if the creating migration is correct.