1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Cache\SQL; |
3
|
|
|
|
4
|
|
|
use PDO; |
5
|
|
|
use vipnytt\RobotsTxtParser\Client; |
6
|
|
|
use vipnytt\RobotsTxtParser\Core\RobotsTxtInterface; |
7
|
|
|
use vipnytt\RobotsTxtParser\Core\UrlParser; |
8
|
|
|
use vipnytt\RobotsTxtParser\Request; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Class Cache |
12
|
|
|
* |
13
|
|
|
* @package vipnytt\RobotsTxtParser\Cache |
14
|
|
|
*/ |
15
|
|
|
class Cache implements RobotsTxtInterface |
16
|
|
|
{ |
17
|
|
|
use UrlParser; |
18
|
|
|
|
19
|
|
|
/** |
20
|
|
|
* Database connection |
21
|
|
|
* @var PDO |
22
|
|
|
*/ |
23
|
|
|
protected $pdo; |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* GuzzleHTTP config |
27
|
|
|
* @var array |
28
|
|
|
*/ |
29
|
|
|
protected $guzzleConfig = []; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Byte limit |
33
|
|
|
* @var int |
34
|
|
|
*/ |
35
|
|
|
protected $byteLimit = self::BYTE_LIMIT; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* Client nextUpdate margin in seconds |
39
|
|
|
* @var int |
40
|
|
|
*/ |
41
|
|
|
protected $clientNextUpdateMargin = 300; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* Cache constructor. |
45
|
|
|
* |
46
|
|
|
* @param PDO $pdo |
47
|
|
|
* @param array $guzzleConfig |
48
|
|
|
* @param int $byteLimit |
49
|
|
|
*/ |
50
|
|
|
public function __construct(PDO $pdo, array $guzzleConfig = [], $byteLimit = self::BYTE_LIMIT) |
51
|
|
|
{ |
52
|
|
|
$this->pdo = $pdo; |
53
|
|
|
$this->guzzleConfig = $guzzleConfig; |
54
|
|
|
$this->byteLimit = $byteLimit; |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Process the update queue |
59
|
|
|
* |
60
|
|
|
* @param int|null $workerID |
61
|
|
|
* @return bool |
62
|
|
|
*/ |
63
|
|
|
public function cron($workerID = null) |
64
|
|
|
{ |
65
|
|
|
$worker = $this->setWorkerID($workerID); |
66
|
|
|
$result = true; |
67
|
|
|
while ($result) { |
68
|
|
|
$query = $this->pdo->prepare(<<<SQL |
69
|
|
|
UPDATE robotstxt__cache0 |
70
|
|
|
SET worker = :workerID |
71
|
|
|
WHERE worker IS NULL AND nextUpdate <= UNIX_TIMESTAMP() |
72
|
|
|
ORDER BY nextUpdate ASC |
73
|
|
|
LIMIT 1; |
74
|
|
|
SELECT |
75
|
|
|
base, |
76
|
|
|
validUntil |
77
|
|
|
FROM robotstxt__cache0 |
78
|
|
|
WHERE worker = :worker; |
79
|
|
|
SQL |
80
|
|
|
); |
81
|
|
|
$query->bindParam(':workerID', $worker, PDO::PARAM_INT); |
82
|
|
|
$query->execute(); |
83
|
|
|
if ($query->rowCount() > 0) { |
84
|
|
|
while ($row = $query->fetch(PDO::FETCH_ASSOC)) { |
85
|
|
|
$result = $this->push(new Request($row['base'], $this->guzzleConfig, $this->byteLimit), $row['validUntil']); |
86
|
|
|
} |
87
|
|
|
continue; |
88
|
|
|
} |
89
|
|
|
return true; |
90
|
|
|
} |
91
|
|
|
return false; |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* Set WorkerID |
96
|
|
|
* |
97
|
|
|
* @param int|null $workerID |
98
|
|
|
* @return int |
99
|
|
|
*/ |
100
|
|
|
protected function setWorkerID($workerID = null) |
101
|
|
|
{ |
102
|
|
|
if ( |
103
|
|
|
is_int($workerID) && |
104
|
|
|
$workerID <= 255 && |
105
|
|
|
$workerID >= 1 |
106
|
|
|
) { |
107
|
|
|
return $workerID; |
108
|
|
|
} elseif ($workerID !== null) { |
109
|
|
|
trigger_error('WorkerID out of range (1-255)', E_USER_WARNING); |
110
|
|
|
} |
111
|
|
|
return rand(1, 255); |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* Update an robots.txt in the database |
116
|
|
|
* |
117
|
|
|
* @param Request $request |
118
|
|
|
* @param int $existingValidUntil |
119
|
|
|
* @return bool |
120
|
|
|
*/ |
121
|
|
|
public function push(Request $request, $existingValidUntil = 0) |
122
|
|
|
{ |
123
|
|
|
$time = time(); |
124
|
|
|
$base = $request->getBaseUri(); |
125
|
|
|
$statusCode = $request->getStatusCode(); |
126
|
|
|
$nextUpdate = $request->nextUpdate(); |
127
|
|
|
if ( |
128
|
|
|
$existingValidUntil > $time && |
129
|
|
|
$statusCode >= 500 && |
130
|
|
|
$statusCode < 600 && |
131
|
|
|
mb_strpos(parse_url($base, PHP_URL_SCHEME), 'http') === 0 |
132
|
|
|
) { |
133
|
|
|
$nextUpdate = min($existingValidUntil, $nextUpdate); |
134
|
|
|
$query = $this->pdo->prepare(<<<SQL |
135
|
|
|
UPDATE robotstxt__cache0 |
136
|
|
|
SET nextUpdate = :nextUpdate, worker = NULL |
137
|
|
|
WHERE base = :base; |
138
|
|
|
SQL |
139
|
|
|
); |
140
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
141
|
|
|
$query->bindParam(':nextUpdate', $nextUpdate, PDO::PARAM_INT); |
142
|
|
|
return $query->execute(); |
143
|
|
|
} |
144
|
|
|
$validUntil = $request->validUntil(); |
145
|
|
|
$content = $request->getContents(); |
146
|
|
|
$query = $this->pdo->prepare(<<<SQL |
147
|
|
|
INSERT INTO robotstxt__cache0 (base, content, statusCode, validUntil, nextUpdate) |
148
|
|
|
VALUES (:base, :content, :statusCode, :validUntil, :nextUpdate) |
149
|
|
|
ON DUPLICATE KEY UPDATE content = :content, statusCode = :statusCode, validUntil = :validUntil, nextUpdate = :nextUpdate, worker = 0; |
150
|
|
|
SQL |
151
|
|
|
); |
152
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
153
|
|
|
$query->bindParam(':content', $content, PDO::PARAM_STR); |
154
|
|
|
$query->bindParam(':statusCode', $statusCode, PDO::PARAM_INT); |
155
|
|
|
$query->bindParam(':validUntil', $validUntil, PDO::PARAM_INT); |
156
|
|
|
$query->bindParam(':nextUpdate', $nextUpdate, PDO::PARAM_INT); |
157
|
|
|
return $query->execute(); |
158
|
|
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* Parser client |
162
|
|
|
* |
163
|
|
|
* @param string $baseUri |
164
|
|
|
* @return Client|Request |
165
|
|
|
*/ |
166
|
|
|
public function client($baseUri) |
167
|
|
|
{ |
168
|
|
|
$base = $this->urlBase($this->urlEncode($baseUri)); |
169
|
|
|
$query = $this->pdo->prepare(<<<SQL |
170
|
|
|
SELECT content,statusCode,nextUpdate,worker |
171
|
|
|
FROM robotstxt__cache0 |
172
|
|
|
WHERE base = :base; |
173
|
|
|
SQL |
174
|
|
|
); |
175
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
176
|
|
|
$query->execute(); |
177
|
|
|
if ($query->rowCount() > 0) { |
178
|
|
|
$row = $query->fetch(PDO::FETCH_ASSOC); |
179
|
|
|
if ($row['nextUpdate'] >= (time() - $this->clientNextUpdateMargin)) { |
180
|
|
|
$this->markAsActive($base, $row['worker']); |
181
|
|
|
return new Client($base, $row['code'], $row['content'], self::ENCODING, $this->byteLimit); |
182
|
|
|
} |
183
|
|
|
} |
184
|
|
|
$request = new Request($base, $this->guzzleConfig, $this->byteLimit); |
185
|
|
|
$this->push($request); |
186
|
|
|
$this->markAsActive($base); |
187
|
|
|
return $request; |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
/** |
191
|
|
|
* Mark robots.txt as active |
192
|
|
|
* |
193
|
|
|
* @param string $base |
194
|
|
|
* @param int|null $workerID |
195
|
|
|
* @return bool |
196
|
|
|
*/ |
197
|
|
|
protected function markAsActive($base, $workerID = 0) |
198
|
|
|
{ |
199
|
|
|
if ($workerID == 0) { |
200
|
|
|
$query = $this->pdo->prepare(<<<SQL |
201
|
|
|
UPDATE robotstxt__cache0 |
202
|
|
|
SET worker = NULL |
203
|
|
|
WHERE base = :base; |
204
|
|
|
SQL |
205
|
|
|
); |
206
|
|
|
$query->bindParam(':base', $base, PDO::PARAM_STR); |
207
|
|
|
return $query->execute(); |
208
|
|
|
} |
209
|
|
|
return true; |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
/** |
213
|
|
|
* Delay |
214
|
|
|
* |
215
|
|
|
* @param float|int $delay |
216
|
|
|
* @param string $baseUri |
217
|
|
|
* @param string $userAgent |
218
|
|
|
* @return Delay |
219
|
|
|
*/ |
220
|
|
|
public function delay($delay, $baseUri, $userAgent) |
221
|
|
|
{ |
222
|
|
|
return new Delay($this->pdo, $delay, $baseUri, $userAgent); |
223
|
|
|
} |
224
|
|
|
} |
225
|
|
|
|