1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Client\Directives; |
3
|
|
|
|
4
|
|
|
use PDO; |
5
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\SQLException; |
6
|
|
|
use vipnytt\RobotsTxtParser\SQL\SQLInterface; |
7
|
|
|
use vipnytt\UserAgentParser; |
8
|
|
|
|
9
|
|
|
/** |
10
|
|
|
* Class DelayHandlerClient |
11
|
|
|
* |
12
|
|
|
* @package vipnytt\RobotsTxtParser\Client\Directives |
13
|
|
|
*/ |
14
|
|
|
class DelayHandlerClient implements SQLInterface |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* Supported database drivers |
18
|
|
|
*/ |
19
|
|
|
const SUPPORTED_DRIVERS = [ |
20
|
|
|
self::DRIVER_MYSQL, |
21
|
|
|
]; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* Database connection |
25
|
|
|
* @var PDO |
26
|
|
|
*/ |
27
|
|
|
private $pdo; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* PDO driver |
31
|
|
|
* @var string |
32
|
|
|
*/ |
33
|
|
|
private $driver; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* Base UriClient |
37
|
|
|
* @var string |
38
|
|
|
*/ |
39
|
|
|
private $base; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* User-agent |
43
|
|
|
* @var string |
44
|
|
|
*/ |
45
|
|
|
private $userAgent; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Delay |
49
|
|
|
* @var float|int |
50
|
|
|
*/ |
51
|
|
|
private $delay; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* DelayClient constructor. |
55
|
|
|
* |
56
|
|
|
* @param PDO $pdo |
57
|
|
|
* @param string $baseUri |
58
|
|
|
* @param string $userAgent |
59
|
|
|
* @param float|int $delay |
60
|
|
|
* @throws SQLException |
61
|
|
|
*/ |
62
|
|
|
public function __construct(PDO $pdo, $baseUri, $userAgent, $delay) |
63
|
|
|
{ |
64
|
|
|
$this->pdo = $this->pdoInitialize($pdo); |
65
|
|
|
$this->base = $baseUri; |
66
|
|
|
$uaStringParser = new UserAgentParser($userAgent); |
67
|
|
|
$this->userAgent = $uaStringParser->stripVersion(); |
68
|
|
|
$this->delay = round($delay, 6, PHP_ROUND_HALF_UP); |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* Initialize PDO connection |
73
|
|
|
* |
74
|
|
|
* @param PDO $pdo |
75
|
|
|
* @return PDO |
76
|
|
|
* @throws SQLException |
77
|
|
|
*/ |
78
|
|
|
private function pdoInitialize(PDO $pdo) |
79
|
|
|
{ |
80
|
|
|
if ($pdo->getAttribute(PDO::ATTR_ERRMODE) === PDO::ERRMODE_SILENT) { |
81
|
|
|
$pdo->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING); |
82
|
|
|
} |
83
|
|
|
$pdo->setAttribute(PDO::ATTR_CASE, PDO::CASE_NATURAL); |
84
|
|
|
$pdo->exec('SET NAMES ' . self::SQL_ENCODING); |
85
|
|
|
$this->driver = $pdo->getAttribute(PDO::ATTR_DRIVER_NAME); |
86
|
|
|
if (!in_array($this->driver, self::SUPPORTED_DRIVERS)) { |
87
|
|
|
throw new SQLException('Unsupported database. ' . self::README_SQL_DELAY); |
88
|
|
|
} |
89
|
|
|
try { |
90
|
|
|
$pdo->query("SELECT 1 FROM robotstxt__delay0 LIMIT 1;"); |
91
|
|
|
} catch (\Exception $exception1) { |
92
|
|
|
try { |
93
|
|
|
$pdo->query(file_get_contents(__DIR__ . '/../../SQL/delay.sql')); |
94
|
|
|
} catch (\Exception $exception2) { |
95
|
|
|
throw new SQLException('Missing table `' . self::TABLE_DELAY . '`. Setup instructions: ' . self::README_SQL_DELAY); |
96
|
|
|
} |
97
|
|
|
} |
98
|
|
|
return $pdo; |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* Queue |
103
|
|
|
* |
104
|
|
|
* @return float|int |
105
|
|
|
*/ |
106
|
|
|
public function getQueue() |
107
|
|
|
{ |
108
|
|
|
if ($this->delay == 0) { |
109
|
|
|
return 0; |
110
|
|
|
} |
111
|
|
|
$query = $this->pdo->prepare(<<<SQL |
112
|
|
|
SELECT GREATEST(0, (delayUntil / 1000000) - UNIX_TIMESTAMP(CURTIME(6))) AS sec |
113
|
|
|
FROM robotstxt__delay0 |
114
|
|
|
WHERE base = :base AND userAgent = :userAgent; |
115
|
|
|
SQL |
116
|
|
|
); |
117
|
|
|
$query->bindParam(':base', $this->base, PDO::PARAM_STR); |
118
|
|
|
$query->bindParam(':userAgent', $this->userAgent, PDO::PARAM_STR); |
119
|
|
|
$query->execute(); |
120
|
|
|
if ($query->rowCount() > 0) { |
121
|
|
|
$row = $query->fetch(PDO::FETCH_ASSOC); |
122
|
|
|
return $row['sec']; |
123
|
|
|
} |
124
|
|
|
return 0; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Reset queue |
129
|
|
|
* |
130
|
|
|
* @return bool |
131
|
|
|
*/ |
132
|
|
|
public function reset() |
133
|
|
|
{ |
134
|
|
|
$query = $this->pdo->prepare(<<<SQL |
135
|
|
|
DELETE FROM robotstxt__delay0 |
136
|
|
|
WHERE base = :base AND userAgent = :useragent; |
137
|
|
|
SQL |
138
|
|
|
); |
139
|
|
|
$query->bindParam(':base', $this->base, PDO::PARAM_INT); |
140
|
|
|
$query->bindParam(':useragent', $this->userAgent, PDO::PARAM_INT); |
141
|
|
|
return $query->execute(); |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
/** |
145
|
|
|
* Sleep |
146
|
|
|
* |
147
|
|
|
* @return float|int |
148
|
|
|
*/ |
149
|
|
|
public function sleep() |
150
|
|
|
{ |
151
|
|
|
$start = microtime(true); |
152
|
|
|
$until = $this->getTimeSleepUntil(); |
153
|
|
|
if (microtime(true) > $until) { |
154
|
|
|
return 0; |
155
|
|
|
} |
156
|
|
|
try { |
157
|
|
|
time_sleep_until($until); |
158
|
|
|
} catch (\Exception $warning) { |
159
|
|
|
// Timestamp already in the past |
160
|
|
|
} |
161
|
|
|
return microtime(true) - $start; |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
/** |
165
|
|
|
* Timestamp with milliseconds |
166
|
|
|
* |
167
|
|
|
* @return float|int |
168
|
|
|
* @throws SQLException |
169
|
|
|
*/ |
170
|
|
|
public function getTimeSleepUntil() |
171
|
|
|
{ |
172
|
|
|
if ($this->delay == 0) { |
173
|
|
|
return 0; |
174
|
|
|
} |
175
|
|
|
$query = $this->pdo->prepare(<<<SQL |
176
|
|
|
SELECT |
177
|
|
|
delayUntil, |
178
|
|
|
UNIX_TIMESTAMP() |
179
|
|
|
FROM robotstxt__delay0 |
180
|
|
|
WHERE base = :base AND userAgent = :userAgent; |
181
|
|
|
SQL |
182
|
|
|
); |
183
|
|
|
$query->bindParam(':base', $this->base, PDO::PARAM_STR); |
184
|
|
|
$query->bindParam(':userAgent', $this->userAgent, PDO::PARAM_STR); |
185
|
|
|
$query->execute(); |
186
|
|
|
$this->increment(); |
187
|
|
|
if ($query->rowCount() > 0) { |
188
|
|
|
$row = $query->fetch(PDO::FETCH_ASSOC); |
189
|
|
|
if (abs(time() - $row['UNIX_TIMESTAMP()']) > 10) { |
190
|
|
|
throw new SQLException('`PHP server` and `SQL server` timestamps are out of sync. Please fix!'); |
191
|
|
|
} |
192
|
|
|
return $row['delayUntil'] / 1000000; |
193
|
|
|
} |
194
|
|
|
return 0; |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* Set new delayUntil timestamp |
199
|
|
|
* |
200
|
|
|
* @return bool |
201
|
|
|
*/ |
202
|
|
|
private function increment() |
203
|
|
|
{ |
204
|
|
|
$query = $this->pdo->prepare(<<<SQL |
205
|
|
|
INSERT INTO robotstxt__delay0 (base, userAgent, delayUntil, lastDelay) |
206
|
|
|
VALUES (:base, :userAgent, (UNIX_TIMESTAMP(CURTIME(6)) + :delay) * 1000000, :delay * 1000000) |
207
|
|
|
ON DUPLICATE KEY UPDATE |
208
|
|
|
delayUntil = GREATEST((UNIX_TIMESTAMP(CURTIME(6)) + :delay) * 1000000, delayUntil + (:delay * 1000000)), |
209
|
|
|
lastDelay = :delay * 1000000; |
210
|
|
|
SQL |
211
|
|
|
); |
212
|
|
|
$query->bindParam(':base', $this->base, PDO::PARAM_STR); |
213
|
|
|
$query->bindParam(':userAgent', $this->userAgent, PDO::PARAM_STR); |
214
|
|
|
$query->bindParam(':delay', $this->delay, is_int($this->delay) ? PDO::PARAM_INT : PDO::PARAM_STR); |
215
|
|
|
return $query->execute(); |
216
|
|
|
} |
217
|
|
|
} |
218
|
|
|
|