|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Parser\Directives; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions; |
|
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\RobotsTxtInterface; |
|
6
|
|
|
use vipnytt\RobotsTxtParser\Parser\Toolbox; |
|
7
|
|
|
use vipnytt\RobotsTxtParser\Parser\UrlParser; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Class DisAllow |
|
11
|
|
|
* |
|
12
|
|
|
* @package vipnytt\RobotsTxtParser\Parser\Directives |
|
13
|
|
|
*/ |
|
14
|
|
|
class DisAllow implements DirectiveInterface, RobotsTxtInterface |
|
15
|
|
|
{ |
|
16
|
|
|
use Toolbox; |
|
17
|
|
|
use UrlParser; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Directive alternatives |
|
21
|
|
|
*/ |
|
22
|
|
|
const DIRECTIVE = [ |
|
23
|
|
|
self::DIRECTIVE_ALLOW, |
|
24
|
|
|
self::DIRECTIVE_DISALLOW, |
|
25
|
|
|
]; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Sub directives white list |
|
29
|
|
|
*/ |
|
30
|
|
|
const SUB_DIRECTIVES = [ |
|
31
|
|
|
self::DIRECTIVE_CLEAN_PARAM, |
|
32
|
|
|
self::DIRECTIVE_HOST, |
|
33
|
|
|
]; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Directive |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $directive; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Rule array |
|
43
|
|
|
* @var array |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $array = []; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Sub-directive Clean-param |
|
49
|
|
|
* @var CleanParam |
|
50
|
|
|
*/ |
|
51
|
|
|
protected $cleanParam; |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* Sub-directive Host |
|
55
|
|
|
* @var Host |
|
56
|
|
|
*/ |
|
57
|
|
|
protected $host; |
|
58
|
|
|
|
|
59
|
|
|
/** |
|
60
|
|
|
* DisAllow constructor |
|
61
|
|
|
* |
|
62
|
|
|
* @param string $directive |
|
63
|
|
|
*/ |
|
64
|
|
|
public function __construct($directive) |
|
65
|
|
|
{ |
|
66
|
|
|
$this->directive = $this->validateDirective($directive, self::DIRECTIVE); |
|
67
|
|
|
$this->cleanParam = new CleanParam(); |
|
68
|
|
|
$this->host = new Host(); |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Add |
|
73
|
|
|
* |
|
74
|
|
|
* @param string $line |
|
75
|
|
|
* @return bool |
|
76
|
|
|
*/ |
|
77
|
|
|
public function add($line) |
|
78
|
|
|
{ |
|
79
|
|
|
$pair = $this->generateRulePair($line, self::SUB_DIRECTIVES); |
|
80
|
|
|
switch ($pair['directive']) { |
|
81
|
|
|
case self::DIRECTIVE_CLEAN_PARAM: |
|
82
|
|
|
return $this->cleanParam->add($pair['value']); |
|
83
|
|
|
case self::DIRECTIVE_HOST: |
|
84
|
|
|
return $this->host->add($pair['value']); |
|
85
|
|
|
} |
|
86
|
|
|
return $this->addPath($line); |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Add plain path to allow/disallow |
|
91
|
|
|
* |
|
92
|
|
|
* @param string $rule |
|
93
|
|
|
* @return bool |
|
94
|
|
|
*/ |
|
95
|
|
|
protected function addPath($rule) |
|
96
|
|
|
{ |
|
97
|
|
|
if (isset($this->array['path']) && in_array($rule, $this->array['path'])) { |
|
98
|
|
|
return false; |
|
99
|
|
|
} |
|
100
|
|
|
$this->array['path'][] = $rule; |
|
101
|
|
|
return true; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* Check |
|
106
|
|
|
* |
|
107
|
|
|
* @param string $url |
|
108
|
|
|
* @return bool |
|
109
|
|
|
*/ |
|
110
|
|
|
public function check($url) |
|
111
|
|
|
{ |
|
112
|
|
|
$path = $this->getPath($url); |
|
113
|
|
|
return ($path === false) ? false : ( |
|
114
|
|
|
$this->checkPath($path, isset($this->array['path']) ? $this->array['path'] : []) || |
|
115
|
|
|
$this->cleanParam->check($path) || |
|
116
|
|
|
$this->host->check($url) |
|
117
|
|
|
); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Get path and query |
|
122
|
|
|
* |
|
123
|
|
|
* @param string $url |
|
124
|
|
|
* @return string |
|
125
|
|
|
* @throws Exceptions\ClientException |
|
126
|
|
|
*/ |
|
127
|
|
|
protected function getPath($url) |
|
128
|
|
|
{ |
|
129
|
|
|
// Encode |
|
130
|
|
|
$url = $this->urlEncode($url); |
|
131
|
|
|
if (mb_stripos($url, '/') === 0) { |
|
132
|
|
|
// Strip fragments |
|
133
|
|
|
$url = mb_split('#', $url)[0]; |
|
134
|
|
|
return $url; |
|
135
|
|
|
} |
|
136
|
|
|
if (!$this->urlValidate($url)) { |
|
137
|
|
|
throw new Exceptions\ClientException('Invalid URL'); |
|
138
|
|
|
} |
|
139
|
|
|
$path = (($path = parse_url($url, PHP_URL_PATH)) === null) ? '/' : $path; |
|
140
|
|
|
$query = (($query = parse_url($url, PHP_URL_QUERY)) === null) ? '' : '?' . $query; |
|
141
|
|
|
return $path . $query; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* Export rules |
|
146
|
|
|
* |
|
147
|
|
|
* @return array |
|
148
|
|
|
*/ |
|
149
|
|
|
public function export() |
|
150
|
|
|
{ |
|
151
|
|
|
$result = array_merge( |
|
152
|
|
|
$this->array, |
|
153
|
|
|
$this->cleanParam->export(), |
|
154
|
|
|
$this->host->export() |
|
155
|
|
|
); |
|
156
|
|
|
return empty($result) ? [] : [$this->directive => $result]; |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
/** |
|
160
|
|
|
* Render |
|
161
|
|
|
* |
|
162
|
|
|
* @return string[] |
|
163
|
|
|
*/ |
|
164
|
|
|
public function render() |
|
165
|
|
|
{ |
|
166
|
|
|
$result = []; |
|
167
|
|
|
$render = array_merge( |
|
168
|
|
|
$this->array, |
|
169
|
|
|
$this->cleanParam->render(), |
|
170
|
|
|
$this->host->render() |
|
171
|
|
|
); |
|
172
|
|
|
foreach ($render as $value) { |
|
173
|
|
|
if (is_array($value)) { |
|
174
|
|
|
foreach ($value as $path) { |
|
175
|
|
|
$result[] = $this->directive . ':' . $path; |
|
176
|
|
|
} |
|
177
|
|
|
continue; |
|
178
|
|
|
} |
|
179
|
|
|
$result[] = $this->directive . ': ' . $value; |
|
180
|
|
|
} |
|
181
|
|
|
return $result; |
|
182
|
|
|
} |
|
183
|
|
|
} |
|
184
|
|
|
|
VIPnytt /
RobotsTxtParser
