1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client\SQL\Cache; |
3
|
|
|
|
4
|
|
|
use PDO; |
5
|
|
|
use vipnytt\RobotsTxtParser\Client\SQL\SQLInterface; |
6
|
|
|
use vipnytt\RobotsTxtParser\Client\SQL\SQLTrait; |
7
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
8
|
|
|
use vipnytt\RobotsTxtParser\URI; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Class CacheCoreSQL |
12
|
|
|
* |
13
|
|
|
* @package vipnytt\RobotsTxtParser\Client\SQL\Cache |
14
|
|
|
*/ |
15
|
|
|
abstract class CacheCoreSQL implements RobotsTxtInterface, SQLInterface |
16
|
|
|
{ |
17
|
|
|
use SQLTrait; |
18
|
|
|
|
19
|
|
|
/** |
20
|
|
|
* Database connection |
21
|
|
|
* @var PDO |
22
|
|
|
*/ |
23
|
|
|
protected $pdo; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* PDO driver |
27
|
|
|
* @var string |
28
|
|
|
*/ |
29
|
|
|
protected $driver; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* GuzzleHTTP config |
33
|
|
|
* @var array |
34
|
|
|
*/ |
35
|
|
|
protected $guzzleConfig = []; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Byte limit |
39
|
|
|
* @var int |
40
|
|
|
*/ |
41
|
|
|
protected $byteLimit = self::BYTE_LIMIT; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* CacheCoreSQL constructor. |
45
|
|
|
* |
46
|
|
|
* @param PDO $pdo |
47
|
|
|
* @param array $guzzleConfig |
48
|
|
|
* @param int|null $byteLimit |
49
|
|
|
*/ |
50
|
|
|
public function __construct(PDO $pdo, array $guzzleConfig = [], $byteLimit = self::BYTE_LIMIT) |
51
|
|
|
{ |
52
|
|
|
$this->pdo = $this->pdoInitialize($pdo); |
53
|
|
|
$this->driver = $this->pdo->getAttribute(PDO::ATTR_DRIVER_NAME); |
54
|
|
|
if ($this->driver != 'mysql') { |
55
|
|
|
trigger_error('Unsupported database. Currently supports MySQL only. ' . self::README_SQL_CACHE, E_USER_WARNING); |
56
|
|
|
} |
57
|
|
|
$this->guzzleConfig = $guzzleConfig; |
58
|
|
|
$this->byteLimit = $byteLimit; |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* Process the update queue |
63
|
|
|
* |
64
|
|
|
* @param int|null $workerID |
65
|
|
|
* @return bool |
66
|
|
|
*/ |
67
|
|
|
public function cron($workerID = null) |
68
|
|
|
{ |
69
|
|
|
$worker = $this->setWorkerID($workerID); |
70
|
|
|
$result = true; |
71
|
|
|
while ($result) { |
72
|
|
|
$query = $this->pdo->prepare(<<<SQL |
73
|
|
|
UPDATE robotstxt__cache0 |
74
|
|
|
SET worker = :workerID |
75
|
|
|
WHERE worker IS NULL AND nextUpdate <= UNIX_TIMESTAMP() |
76
|
|
|
ORDER BY nextUpdate ASC |
77
|
|
|
LIMIT 1; |
78
|
|
|
SELECT base |
79
|
|
|
FROM robotstxt__cache0 |
80
|
|
|
WHERE worker = :worker; |
81
|
|
|
SQL |
82
|
|
|
); |
83
|
|
|
$query->bindParam(':workerID', $worker, PDO::PARAM_INT); |
84
|
|
|
$query->execute(); |
85
|
|
|
if ($query->rowCount() > 0) { |
86
|
|
|
while ($row = $query->fetch(PDO::FETCH_ASSOC)) { |
87
|
|
|
$result = $this->push(new URI($row['base'], $this->guzzleConfig, $this->byteLimit)); |
88
|
|
|
} |
89
|
|
|
continue; |
90
|
|
|
} |
91
|
|
|
return true; |
92
|
|
|
} |
93
|
|
|
return false; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* Set WorkerID |
98
|
|
|
* |
99
|
|
|
* @param int|null $workerID |
100
|
|
|
* @return int |
101
|
|
|
*/ |
102
|
|
|
protected function setWorkerID($workerID = null) |
103
|
|
|
{ |
104
|
|
|
if ( |
105
|
|
|
is_int($workerID) && |
106
|
|
|
$workerID <= 255 && |
107
|
|
|
$workerID >= 1 |
108
|
|
|
) { |
109
|
|
|
return $workerID; |
110
|
|
|
} elseif ($workerID !== null) { |
111
|
|
|
trigger_error('WorkerID out of range (1-255)', E_USER_WARNING); |
112
|
|
|
} |
113
|
|
|
return rand(1, 255); |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* Update an robots.txt in the database |
118
|
|
|
* |
119
|
|
|
* @param URI $request |
120
|
|
|
* @return bool |
121
|
|
|
*/ |
122
|
|
|
public function push(URI $request) |
123
|
|
|
{ |
124
|
|
|
$base = $request->getBaseUri(); |
125
|
|
|
$statusCode = $request->getStatusCode(); |
126
|
|
|
$nextUpdate = $request->nextUpdate(); |
127
|
|
|
if ( |
128
|
|
|
$statusCode >= 500 && |
129
|
|
|
$statusCode < 600 && |
130
|
|
|
mb_strpos($base, 'http') === 0 |
131
|
|
|
) { |
132
|
|
|
$query = $this->pdo->prepare(<<<SQL |
133
|
|
|
SELECT validUntil |
134
|
|
|
FROM robotstxt__cache0 |
135
|
|
|
WHERE base = :base; |
136
|
|
|
SQL |
137
|
|
|
); |
138
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
139
|
|
|
$query->execute(); |
140
|
|
|
if ( |
141
|
|
|
$query->rowCount() > 0 && |
142
|
|
|
($existingValidUntil = $query->fetch(PDO::FETCH_ASSOC)['validUntil']) > time() |
143
|
|
|
) { |
144
|
|
|
$nextUpdate = min($existingValidUntil, $nextUpdate); |
145
|
|
|
$query = $this->pdo->prepare(<<<SQL |
146
|
|
|
UPDATE robotstxt__cache0 |
147
|
|
|
SET nextUpdate = :nextUpdate, worker = NULL |
148
|
|
|
WHERE base = :base; |
149
|
|
|
SQL |
150
|
|
|
); |
151
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
152
|
|
|
$query->bindParam(':nextUpdate', $nextUpdate, PDO::PARAM_INT); |
153
|
|
|
return $query->execute(); |
154
|
|
|
} |
155
|
|
|
} |
156
|
|
|
$validUntil = $request->validUntil(); |
157
|
|
|
$content = $request->render("\r\n"); |
158
|
|
|
$query = $this->pdo->prepare(<<<SQL |
159
|
|
|
INSERT INTO robotstxt__cache0 (base, content, statusCode, validUntil, nextUpdate) |
160
|
|
|
VALUES (:base, :content, :statusCode, :validUntil, :nextUpdate) |
161
|
|
|
ON DUPLICATE KEY UPDATE content = :content, statusCode = :statusCode, validUntil = :validUntil, |
162
|
|
|
nextUpdate = :nextUpdate, worker = 0; |
163
|
|
|
SQL |
164
|
|
|
); |
165
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
166
|
|
|
$query->bindParam(':content', $content, PDO::PARAM_STR); |
167
|
|
|
$query->bindParam(':statusCode', $statusCode, PDO::PARAM_INT); |
168
|
|
|
$query->bindParam(':validUntil', $validUntil, PDO::PARAM_INT); |
169
|
|
|
$query->bindParam(':nextUpdate', $nextUpdate, PDO::PARAM_INT); |
170
|
|
|
return $query->execute(); |
171
|
|
|
} |
172
|
|
|
} |
173
|
|
|
|