|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* vipnytt/RobotsTxtParser |
|
4
|
|
|
* |
|
5
|
|
|
* @link https://github.com/VIPnytt/RobotsTxtParser |
|
6
|
|
|
* @license https://github.com/VIPnytt/RobotsTxtParser/blob/master/LICENSE The MIT License (MIT) |
|
7
|
|
|
*/ |
|
8
|
|
|
|
|
9
|
|
|
namespace vipnytt\RobotsTxtParser\Parser\Directives; |
|
10
|
|
|
|
|
11
|
|
|
use vipnytt\RobotsTxtParser\Client\Directives\RequestRateClient; |
|
12
|
|
|
use vipnytt\RobotsTxtParser\Handler\RenderHandler; |
|
13
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Class RequestRateParser |
|
17
|
|
|
* |
|
18
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
|
19
|
|
|
*/ |
|
20
|
|
|
class RequestRateParser implements ParserInterface, RobotsTxtInterface |
|
21
|
|
|
{ |
|
22
|
|
|
use DirectiveParserTrait; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Base uri |
|
26
|
|
|
* @var string |
|
27
|
|
|
*/ |
|
28
|
|
|
private $base; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* RequestRate array |
|
32
|
|
|
* @var array |
|
33
|
|
|
*/ |
|
34
|
|
|
private $requestRates = []; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* RequestRate constructor. |
|
38
|
|
|
* |
|
39
|
|
|
* @param string $base |
|
40
|
|
|
*/ |
|
41
|
|
|
public function __construct($base) |
|
42
|
|
|
{ |
|
43
|
|
|
$this->base = $base; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Add |
|
48
|
|
|
* |
|
49
|
|
|
* @param string $line |
|
50
|
|
|
* @return bool |
|
51
|
|
|
*/ |
|
52
|
|
|
public function add($line) |
|
53
|
|
|
{ |
|
54
|
|
|
$array = preg_split('/\s+/', $line, 2); |
|
55
|
|
|
$result = [ |
|
56
|
|
|
'rate' => $this->draftParseRate($array[0]), |
|
57
|
|
|
'from' => null, |
|
58
|
|
|
'to' => null, |
|
59
|
|
|
]; |
|
60
|
|
|
if ($result['rate'] === false) { |
|
61
|
|
|
return false; |
|
62
|
|
|
} elseif (!empty($array[1]) && |
|
63
|
|
|
($times = $this->draftParseTime($array[1])) !== false |
|
64
|
|
|
) { |
|
65
|
|
|
$result = array_merge($result, $times); |
|
66
|
|
|
} |
|
67
|
|
|
$this->requestRates[] = $result; |
|
68
|
|
|
return true; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Client rate as specified in the `Robot exclusion standard` version 2.0 draft |
|
73
|
|
|
* rate = numDocuments / timeUnit |
|
74
|
|
|
* @link http://www.conman.org/people/spc/robots2.html#format.directives.request-rate |
|
75
|
|
|
* |
|
76
|
|
|
* @param string $string |
|
77
|
|
|
* @return float|int|false |
|
78
|
|
|
*/ |
|
79
|
|
|
private function draftParseRate($string) |
|
80
|
|
|
{ |
|
81
|
|
|
$parts = array_map('trim', explode('/', $string)); |
|
82
|
|
|
if (count($parts) != 2) { |
|
83
|
|
|
return false; |
|
84
|
|
|
} |
|
85
|
|
|
$multiplier = 1; |
|
86
|
|
|
switch (strtolower(substr(preg_replace('/[^A-Za-z]/', '', $parts[1]), 0, 1))) { |
|
87
|
|
|
case 'm': |
|
88
|
|
|
$multiplier = 60; |
|
89
|
|
|
break; |
|
90
|
|
|
case 'h': |
|
91
|
|
|
$multiplier = 3600; |
|
92
|
|
|
break; |
|
93
|
|
|
case 'd': |
|
94
|
|
|
$multiplier = 86400; |
|
95
|
|
|
break; |
|
96
|
|
|
} |
|
97
|
|
|
$num = floatval(preg_replace('/[^0-9]/', '', $parts[0])); |
|
98
|
|
|
$sec = floatval(preg_replace('/[^0-9.]/', '', $parts[1])) * $multiplier; |
|
99
|
|
|
$rate = $sec / $num; |
|
100
|
|
|
return $rate > 0 ? $rate : false; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* Client |
|
105
|
|
|
* |
|
106
|
|
|
* @param string $userAgent |
|
107
|
|
|
* @param float|int $fallbackValue |
|
108
|
|
|
* @return RequestRateClient |
|
109
|
|
|
*/ |
|
110
|
|
|
public function client($userAgent = self::USER_AGENT, $fallbackValue = 0) |
|
111
|
|
|
{ |
|
112
|
|
|
$this->sort(); |
|
113
|
|
|
return new RequestRateClient($this->base, $userAgent, $this->requestRates, $fallbackValue); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
/** |
|
117
|
|
|
* Sort |
|
118
|
|
|
* |
|
119
|
|
|
* @return void |
|
120
|
|
|
*/ |
|
121
|
|
|
private function sort() |
|
122
|
|
|
{ |
|
123
|
|
|
usort($this->requestRates, function ($requestRateA, $requestRateB) { |
|
124
|
|
|
// PHP 7: Switch to the <=> "Spaceship" operator |
|
125
|
|
|
return $requestRateB['rate'] > $requestRateA['rate']; |
|
126
|
|
|
}); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* Render |
|
131
|
|
|
* |
|
132
|
|
|
* @param RenderHandler $handler |
|
133
|
|
|
* @return bool |
|
134
|
|
|
*/ |
|
135
|
|
|
public function render(RenderHandler $handler) |
|
136
|
|
|
{ |
|
137
|
|
|
$this->sort(); |
|
138
|
|
|
foreach ($this->requestRates as $array) { |
|
139
|
|
|
$suffix = 's'; |
|
140
|
|
|
if (isset($array['from']) && |
|
141
|
|
|
isset($array['to']) |
|
142
|
|
|
) { |
|
143
|
|
|
$suffix .= ' ' . $array['from'] . '-' . $array['to']; |
|
144
|
|
|
} |
|
145
|
|
|
$handler->add(self::DIRECTIVE_REQUEST_RATE, '1/' . $array['rate'] . $suffix); |
|
146
|
|
|
} |
|
147
|
|
|
return true; |
|
148
|
|
|
} |
|
149
|
|
|
} |
|
150
|
|
|
|
VIPnytt /
RobotsTxtParser
