|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of the Veslo project <https://github.com/symfony-doge/veslo>. |
|
5
|
|
|
* |
|
6
|
|
|
* (C) 2019 Pavel Petrov <[email protected]>. |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
9
|
|
|
* file that was distributed with this source code. |
|
10
|
|
|
* |
|
11
|
|
|
* @license https://opensource.org/licenses/GPL-3.0 GPL-3.0 |
|
12
|
|
|
*/ |
|
13
|
|
|
|
|
14
|
|
|
declare(strict_types=1); |
|
15
|
|
|
|
|
16
|
|
|
namespace Veslo\AnthillBundle\Vacancy\Digger; |
|
17
|
|
|
|
|
18
|
|
|
use Psr\Cache\InvalidArgumentException; |
|
19
|
|
|
use Psr\Log\LoggerInterface; |
|
20
|
|
|
use Symfony\Component\Cache\Adapter\AdapterInterface; |
|
21
|
|
|
use Symfony\Component\Cache\CacheItem; |
|
22
|
|
|
use Veslo\AnthillBundle\Vacancy\DiggerInterface; |
|
23
|
|
|
use Veslo\AnthillBundle\Vacancy\Roadmap\ConveyorAwareRoadmap; |
|
24
|
|
|
use Veslo\AppBundle\Workflow\Vacancy\WorkerInterface; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Cranky dung beetle temporarily refuses to dig if it had too much unsuccessful attempts in a row |
|
28
|
|
|
* So stressful, be nice with him. |
|
29
|
|
|
* |
|
30
|
|
|
* @see DungBeetle |
|
31
|
|
|
*/ |
|
32
|
|
|
class CrankyDungBeetle implements WorkerInterface, DiggerInterface |
|
33
|
|
|
{ |
|
34
|
|
|
/** |
|
35
|
|
|
* Cache for handling pause on too much unsuccessful search attempts and other values |
|
36
|
|
|
* |
|
37
|
|
|
* @const string |
|
38
|
|
|
*/ |
|
39
|
|
|
private const CACHE_NAMESPACE = 'veslo.anthill.vacancy.cranky_dung_beetle'; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Cache instance for managing pause state of roadmaps |
|
43
|
|
|
* |
|
44
|
|
|
* @var AdapterInterface |
|
45
|
|
|
*/ |
|
46
|
|
|
private $cache; |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Count of unsuccessful vacancy search attempts after which dung beetle will temporarily stops digging |
|
50
|
|
|
* |
|
51
|
|
|
* This prevents external website DDoSing and possible rate limiting |
|
52
|
|
|
* |
|
53
|
|
|
* @var int |
|
54
|
|
|
*/ |
|
55
|
|
|
private $attemptsUntilPause; |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* Digging pause duration (in seconds) |
|
59
|
|
|
* |
|
60
|
|
|
* Dung beetle will refuse dig command if it doesn't get |
|
61
|
|
|
* any vacancies by roadmap {$attemptsUntilPause} times in a row |
|
62
|
|
|
* |
|
63
|
|
|
* @var int |
|
64
|
|
|
*/ |
|
65
|
|
|
private $pauseDuration; |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Provides base implementation of digging logic |
|
69
|
|
|
* |
|
70
|
|
|
* @var DungBeetle |
|
71
|
|
|
*/ |
|
72
|
|
|
private $_dungBeetle; |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* CrankyDungBeetle constructor. |
|
76
|
|
|
* |
|
77
|
|
|
* @param DungBeetle $dungBeetle Provides base implementation of digging logic |
|
78
|
|
|
* @param AdapterInterface $cache Cache instance for managing pause state of roadmaps |
|
79
|
|
|
* @param int $attemptsUntilPause Count of unsuccessful vacancy search attempts for pausing |
|
80
|
|
|
* @param int $pauseDuration Digging pause duration (in seconds) |
|
81
|
|
|
*/ |
|
82
|
|
|
public function __construct( |
|
83
|
|
|
DungBeetle $dungBeetle, |
|
84
|
|
|
AdapterInterface $cache, |
|
85
|
|
|
int $attemptsUntilPause, |
|
86
|
|
|
int $pauseDuration |
|
87
|
|
|
) { |
|
88
|
|
|
$this->_dungBeetle = $dungBeetle; |
|
89
|
|
|
$this->cache = $cache; |
|
90
|
|
|
$this->attemptsUntilPause = $attemptsUntilPause; |
|
91
|
|
|
$this->pauseDuration = $pauseDuration; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* Digs dung (vacancies) from internet by specified roadmap and attempts count |
|
96
|
|
|
* Temporarily stops digging if too much unsuccessful attempts occurred |
|
97
|
|
|
* |
|
98
|
|
|
* {@inheritdoc} |
|
99
|
|
|
*/ |
|
100
|
|
|
public function dig(ConveyorAwareRoadmap $roadmap, int $iterations = 1): int |
|
101
|
|
|
{ |
|
102
|
|
|
if ($this->isPaused($roadmap)) { |
|
103
|
|
|
$logger = $this->getLogger(); |
|
104
|
|
|
|
|
105
|
|
|
if ($logger instanceof LoggerInterface) { |
|
|
|
|
|
|
106
|
|
|
$roadmapName = $roadmap->getName(); |
|
107
|
|
|
$logger->debug('Roadmap is on pause.', ['roadmap' => $roadmapName]); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
return 0; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
$successfulIterations = $this->_dungBeetle->dig($roadmap, $iterations); |
|
114
|
|
|
|
|
115
|
|
|
if (empty($successfulIterations)) { |
|
116
|
|
|
$this->pause($roadmap); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
return $successfulIterations; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* {@inheritdoc} |
|
124
|
|
|
*/ |
|
125
|
|
|
public function getLogger(): ?LoggerInterface |
|
126
|
|
|
{ |
|
127
|
|
|
return $this->_dungBeetle->getLogger(); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* Returns positive if roadmap is paused |
|
132
|
|
|
* |
|
133
|
|
|
* @param ConveyorAwareRoadmap $roadmap Provides URL of vacancies |
|
134
|
|
|
* |
|
135
|
|
|
* @return bool |
|
136
|
|
|
* |
|
137
|
|
|
* @throws InvalidArgumentException |
|
138
|
|
|
*/ |
|
139
|
|
|
private function isPaused(ConveyorAwareRoadmap $roadmap): bool |
|
140
|
|
|
{ |
|
141
|
|
|
$pause = $this->getPause($roadmap->getName()); |
|
142
|
|
|
|
|
143
|
|
|
if (!$pause->isHit()) { |
|
144
|
|
|
return false; |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
$unsuccessfulAttempts = (int) $pause->get(); |
|
148
|
|
|
|
|
149
|
|
|
return $unsuccessfulAttempts >= $this->attemptsUntilPause; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* Pausing roadmap that makes it unusable by dung beetle for {$pauseDuration} seconds |
|
154
|
|
|
* |
|
155
|
|
|
* @param ConveyorAwareRoadmap $roadmap Provides URL of vacancies |
|
156
|
|
|
* |
|
157
|
|
|
* @return void |
|
158
|
|
|
* |
|
159
|
|
|
* @throws InvalidArgumentException |
|
160
|
|
|
*/ |
|
161
|
|
|
private function pause(ConveyorAwareRoadmap $roadmap): void |
|
162
|
|
|
{ |
|
163
|
|
|
$pause = $this->getPause($roadmap->getName()); |
|
164
|
|
|
$pause->expiresAfter($this->pauseDuration); |
|
165
|
|
|
|
|
166
|
|
|
$unsuccessfulAttempts = $pause->isHit() ? (int) $pause->get() : 0; |
|
167
|
|
|
|
|
168
|
|
|
if ($unsuccessfulAttempts >= $this->attemptsUntilPause) { |
|
169
|
|
|
return; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
$pause->set($unsuccessfulAttempts + 1); |
|
173
|
|
|
$this->cache->save($pause); |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
/** |
|
177
|
|
|
* Returns cache item that represents pause state with unsuccessful attempts counter |
|
178
|
|
|
* |
|
179
|
|
|
* @param string $roadmapName Name of vacancy URL provider |
|
180
|
|
|
* |
|
181
|
|
|
* @return CacheItem |
|
182
|
|
|
* |
|
183
|
|
|
* @throws InvalidArgumentException |
|
184
|
|
|
*/ |
|
185
|
|
|
private function getPause(string $roadmapName): CacheItem |
|
186
|
|
|
{ |
|
187
|
|
|
$cacheKey = implode('.', [static::CACHE_NAMESPACE, 'dig', $roadmapName]); |
|
188
|
|
|
|
|
189
|
|
|
return $this->cache->getItem($cacheKey); |
|
190
|
|
|
} |
|
191
|
|
|
} |
|
192
|
|
|
|