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
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Set the crawler id |
33
|
|
|
* |
34
|
|
|
* @param $crawler_id |
35
|
|
|
*/ |
36
|
|
|
public function setCrawlerId($crawler_id) |
37
|
|
|
{ |
38
|
|
|
$this->crawler_id = $crawler_id; |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Return the crawler id |
43
|
|
|
* |
44
|
|
|
* @return mixed |
45
|
|
|
*/ |
46
|
|
|
public function getCrawlerId() { |
47
|
|
|
return $this->crawler_id; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* Check if the controller is setup correctly |
52
|
|
|
* |
53
|
|
|
* @return bool |
54
|
|
|
*/ |
55
|
|
|
protected function controllerIsSetup() { |
56
|
|
|
if(!is_null($this->crawler_id)) { |
57
|
|
|
return true; |
58
|
|
|
} |
59
|
|
|
|
60
|
|
|
return false; |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Setup the crawler so it won't run twice at the same time |
65
|
|
|
* |
66
|
|
|
* @param $crawler_id |
67
|
|
|
*/ |
68
|
|
|
public function setupCrawler($crawler_id = null) |
69
|
|
|
{ |
70
|
|
|
if(!is_null($crawler_id)) { |
71
|
|
|
$this->setCrawlerId($crawler_id); |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
if($this->controllerIsSetup()) { |
75
|
|
|
$times = config('laravel-job-handler.run_times', 10); |
76
|
|
|
|
77
|
|
|
for ($x = 0; $x <= $times; $x++) { |
78
|
|
|
//fetch the last data |
79
|
|
|
$this->getCrawler(); |
80
|
|
|
|
81
|
|
|
if (!$this->crawler->enabled) { |
82
|
|
|
throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - crawler isnt enabled in database'); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
$checkIfCrawlerCanBeRunned = $this->canCrawlerRunAfterPeriod(); |
86
|
|
|
|
87
|
|
|
if ($checkIfCrawlerCanBeRunned['status']) { |
88
|
|
|
if (is_null($this->crawler->latest_status)) { |
89
|
|
|
//first time it runs... |
90
|
|
|
break; |
91
|
|
|
} |
92
|
|
|
if ($this->crawler->latest_status == 2) { |
93
|
|
|
//Done running... |
94
|
|
|
break; |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
|
98
|
|
|
if ($this->crawler->latest_status == 3) { |
99
|
|
|
throw new CrawlerException('Crawler (#' . $this->crawler_id . ') - last run had an error'); |
100
|
|
|
} |
101
|
|
|
} else { |
102
|
|
|
throw new CrawlerNotReachedTimeBetweenJobsException('Has to wait ' . $checkIfCrawlerCanBeRunned['retry_in'] . ' more seconds to run'); |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
if ($x == $times) { |
106
|
|
|
$this->failCrawler('Crawler (#' . $this->crawler_id . ') - max execution time'); |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
if ($this->crawler->status == 1) { |
110
|
|
|
if ($this->crawler->multiple_crawlers) { |
111
|
|
|
break; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
sleep(config('laravel-job-handler.retry_in_seconds', 3)); //retry in 3 seconds |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
$this->startCrawler(); |
119
|
|
|
} else { |
120
|
|
|
throw new CrawlerException('CrawlController is not setup correctly.'); |
121
|
|
|
} |
122
|
|
|
} |
123
|
|
|
/** |
124
|
|
|
* Start the crawler and save it to the database |
125
|
|
|
* |
126
|
|
|
* @param string $output |
127
|
|
|
*/ |
128
|
|
|
public function startCrawler($output = '') |
129
|
|
|
{ |
130
|
|
|
return $this->addStatus(1, $output); //start running |
131
|
|
|
} |
132
|
|
|
/** |
133
|
|
|
* set the crawler as done so other scripts can run |
134
|
|
|
* |
135
|
|
|
* @param string $output |
136
|
|
|
*/ |
137
|
|
|
public function doneCrawler($output = '') |
138
|
|
|
{ |
139
|
|
|
return $this->addStatus(2, $output); //done running |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
/** |
143
|
|
|
* Finishing the crawler |
144
|
|
|
* |
145
|
|
|
* @param string $output |
146
|
|
|
* @return bool |
147
|
|
|
*/ |
148
|
|
|
public function finish($output = '') |
149
|
|
|
{ |
150
|
|
|
return $this->doneCrawler($output); |
151
|
|
|
} |
152
|
|
|
/** |
153
|
|
|
* crawler failed... |
154
|
|
|
* |
155
|
|
|
* @param string $output |
156
|
|
|
*/ |
157
|
|
|
public function failCrawler($output = '') |
158
|
|
|
{ |
159
|
|
|
$this->addStatus(3, $output); //failed |
160
|
|
|
|
161
|
|
|
throw new CrawlerException($output.' - status 3'); |
162
|
|
|
} |
163
|
|
|
/** |
164
|
|
|
* Save the latest crawler status to the database |
165
|
|
|
* |
166
|
|
|
* @param $status |
167
|
|
|
* @param string $output |
168
|
|
|
* @return bool |
169
|
|
|
*/ |
170
|
|
|
protected function addStatus($status, $output = '') |
171
|
|
|
{ |
172
|
|
|
$crawlerstatus = new CrawlerStatus(); |
173
|
|
|
|
174
|
|
|
$crawlerstatus->crawler_id = $this->crawler_id; |
175
|
|
|
$crawlerstatus->status = $status; |
176
|
|
|
|
177
|
|
|
if ($crawlerstatus->save()) { |
178
|
|
|
$this->crawler->latest_status = $status; |
179
|
|
|
|
180
|
|
|
$this->crawler->save(); |
181
|
|
|
|
182
|
|
|
if (!empty($output)) { |
183
|
|
|
$crawlerstatuslog = new CrawlerStatusLogs(); |
184
|
|
|
|
185
|
|
|
$crawlerstatuslog->status_id = $crawlerstatus->id; |
|
|
|
|
186
|
|
|
$crawlerstatuslog->output = $output; |
|
|
|
|
187
|
|
|
|
188
|
|
|
$crawlerstatuslog->save(); |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
return true; |
192
|
|
|
} else { |
193
|
|
|
throw new CrawlerSaveException('Cannot save crawlerstatus to database...'); |
194
|
|
|
} |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* This will define when the job can be runned again |
199
|
|
|
* |
200
|
|
|
* @return array |
201
|
|
|
*/ |
202
|
|
|
public function canCrawlerRunAfterPeriod() |
203
|
|
|
{ |
204
|
|
|
$this->getCrawler(); |
205
|
|
|
|
206
|
|
|
if (is_null($this->crawler->time_between)) { |
207
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
208
|
|
|
} else { |
209
|
|
|
$seconds = $this->crawler->time_between; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
if (!is_null($this->crawler->last_runned_at)) { |
213
|
|
|
if ($this->crawler->last_runned_at <= Carbon::now()->subSeconds($seconds)) { |
214
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(false, Carbon::parse($this->crawler->last_runned_at)->diffInSeconds(Carbon::now()->subSeconds($seconds))); |
218
|
|
|
} else { |
219
|
|
|
//crawler never runned, so it can run now |
220
|
|
|
return $this->canCrawlerRunAfterPeriodStatus(true); |
221
|
|
|
} |
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Return the status for canCrawlerRunAfterPeriod method |
226
|
|
|
* |
227
|
|
|
* @param $status |
228
|
|
|
* @param int $retry_in |
229
|
|
|
* @return array |
230
|
|
|
*/ |
231
|
|
|
public function canCrawlerRunAfterPeriodStatus($status, $retry_in = 0) |
232
|
|
|
{ |
233
|
|
|
return [ |
234
|
|
|
'status' => $status, |
235
|
|
|
'retry_in' => $retry_in |
236
|
|
|
]; |
237
|
|
|
} |
238
|
|
|
} |
239
|
|
|
|
Checks if undeclared accessed properties appear in database migrations and if the creating migration is correct.