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\CleanParamClient; |
12
|
|
|
use vipnytt\RobotsTxtParser\Handler\RenderHandler; |
13
|
|
|
use vipnytt\RobotsTxtParser\Parser\UriParser; |
14
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* Class CleanParamParser |
18
|
|
|
* |
19
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
20
|
|
|
*/ |
21
|
|
|
class CleanParamParser implements ParserInterface, RobotsTxtInterface |
22
|
|
|
{ |
23
|
|
|
/** |
24
|
|
|
* Clean-param array |
25
|
|
|
* @var string[][] |
26
|
|
|
*/ |
27
|
|
|
private $cleanParam = []; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* Sort result |
31
|
|
|
* @var bool |
32
|
|
|
*/ |
33
|
|
|
private $sort = false; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* CleanParamParser constructor. |
37
|
|
|
*/ |
38
|
|
|
public function __construct() |
39
|
|
|
{ |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Client |
44
|
|
|
* |
45
|
|
|
* @return CleanParamClient |
46
|
|
|
*/ |
47
|
|
|
public function client() |
48
|
|
|
{ |
49
|
|
|
$this->sort(); |
50
|
|
|
return new CleanParamClient($this->cleanParam); |
51
|
|
|
} |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* Sort by length |
55
|
|
|
* |
56
|
|
|
* @return bool |
57
|
|
|
*/ |
58
|
|
|
private function sort() |
59
|
|
|
{ |
60
|
|
|
if ($this->sort) { |
61
|
|
|
return $this->sort; |
62
|
|
|
}; |
63
|
|
|
return $this->sort = array_multisort($this->cleanParam) && ksort($this->cleanParam); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* Add |
68
|
|
|
* |
69
|
|
|
* @param string $line |
70
|
|
|
* @return bool |
71
|
|
|
*/ |
72
|
|
|
public function add($line) |
73
|
|
|
{ |
74
|
|
|
$path = '/'; |
75
|
|
|
// split into parameter and path |
76
|
|
|
$array = array_map('trim', mb_split('\s+', $line, 2)); |
77
|
|
|
if (isset($array[1])) { |
78
|
|
|
// strip any invalid characters from path prefix |
79
|
|
|
$uriParser = new UriParser(preg_replace('/[^A-Za-z0-9\.-\/\*\_]/', '', $array[1])); |
80
|
|
|
$path = rtrim($uriParser->encode(), '*'); |
81
|
|
|
// make sure path is valid |
82
|
|
|
$path = in_array(substr($path, 0, 1), [ |
83
|
|
|
'/', |
84
|
|
|
'*', |
85
|
|
|
]) ? $path : '/'; |
86
|
|
|
} |
87
|
|
|
$param = array_map('trim', explode('&', $array[0])); |
88
|
|
|
foreach ($param as $key) { |
89
|
|
|
$this->cleanParam[$key][] = $path; |
90
|
|
|
} |
91
|
|
|
return true; |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* Render |
96
|
|
|
* |
97
|
|
|
* @param RenderHandler $handler |
98
|
|
|
* @return bool |
99
|
|
|
*/ |
100
|
|
|
public function render(RenderHandler $handler) |
101
|
|
|
{ |
102
|
|
|
$this->sort(); |
103
|
|
|
return $handler->getLevel() == 2 ? $this->renderExtensive($handler) : $this->renderCompressed($handler); |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* Render extensive |
108
|
|
|
* |
109
|
|
|
* @param RenderHandler $handler |
110
|
|
|
* @return bool |
111
|
|
|
*/ |
112
|
|
|
private function renderExtensive(RenderHandler $handler) |
113
|
|
|
{ |
114
|
|
|
foreach ($this->cleanParam as $param => $paths) { |
115
|
|
|
foreach ($paths as $path) { |
116
|
|
|
$handler->add(self::DIRECTIVE_CLEAN_PARAM, $param . ' ' . $path); |
117
|
|
|
} |
118
|
|
|
} |
119
|
|
|
return true; |
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
/** |
123
|
|
|
* Render compressed |
124
|
|
|
* |
125
|
|
|
* @param RenderHandler $handler |
126
|
|
|
* @return bool |
127
|
|
|
*/ |
128
|
|
|
private function renderCompressed(RenderHandler $handler) |
129
|
|
|
{ |
130
|
|
|
$pair = $this->cleanParam; |
131
|
|
|
while (($currentPair = current($pair)) !== false) { |
132
|
|
|
$equalParams = array_keys($pair, $currentPair); |
133
|
|
|
foreach ($currentPair as $path) { |
134
|
|
|
$handler->add(self::DIRECTIVE_CLEAN_PARAM, implode('&', $equalParams) . ' ' . $path); |
135
|
|
|
} |
136
|
|
|
foreach ($equalParams as $param) { |
137
|
|
|
unset($pair[$param]); |
138
|
|
|
} |
139
|
|
|
} |
140
|
|
|
return true; |
141
|
|
|
} |
142
|
|
|
} |
143
|
|
|
|