|
1
|
|
|
<?php |
|
2
|
|
|
namespace vipnytt\RobotsTxtParser\Modules; |
|
3
|
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\ParserException; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* Trait DirectiveTools |
|
8
|
|
|
* |
|
9
|
|
|
* @package vipnytt\RobotsTxtParser\Modules |
|
10
|
|
|
*/ |
|
11
|
|
|
trait Toolbox |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* Check basic rule |
|
15
|
|
|
* |
|
16
|
|
|
* @param string $path |
|
17
|
|
|
* @param array $paths |
|
18
|
|
|
* @return bool |
|
19
|
|
|
*/ |
|
20
|
|
|
protected function checkPath($path, $paths) |
|
21
|
|
|
{ |
|
22
|
|
|
// bug: https://github.com/t1gor/Robots.txt-Parser-Class/issues/62 |
|
23
|
|
|
foreach ($paths as $robotPath) { |
|
24
|
|
|
$escaped = strtr($robotPath, ["@" => '\@']); |
|
25
|
|
|
if (preg_match('@' . $escaped . '@', $path)) { |
|
26
|
|
|
if (mb_stripos($escaped, '$') !== false) { |
|
27
|
|
|
if (mb_strlen($escaped) - 1 == mb_strlen($path)) { |
|
28
|
|
|
return true; |
|
29
|
|
|
} |
|
30
|
|
|
} else { |
|
31
|
|
|
return true; |
|
32
|
|
|
} |
|
33
|
|
|
} |
|
34
|
|
|
} |
|
35
|
|
|
return false; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Generate directive/rule pair |
|
40
|
|
|
* |
|
41
|
|
|
* @param string $line |
|
42
|
|
|
* @param array $whiteList |
|
43
|
|
|
* @return array|false |
|
44
|
|
|
*/ |
|
45
|
|
|
protected function generateRulePair($line, $whiteList) |
|
46
|
|
|
{ |
|
47
|
|
|
$whiteList = array_map('mb_strtolower', $whiteList); |
|
48
|
|
|
// Split by directive and rule |
|
49
|
|
|
$pair = array_map('trim', mb_split(':', $line, 2)); |
|
50
|
|
|
// Check if the line contains a rule |
|
51
|
|
|
if ( |
|
52
|
|
|
empty($pair[1]) || |
|
53
|
|
|
empty($pair[0]) || |
|
54
|
|
|
!in_array(($pair[0] = mb_strtolower($pair[0])), $whiteList) |
|
55
|
|
|
) { |
|
56
|
|
|
// Line does not contain any supported directive |
|
57
|
|
|
return false; |
|
58
|
|
|
} |
|
59
|
|
|
return [ |
|
60
|
|
|
'directive' => $pair[0], |
|
61
|
|
|
'value' => $pair[1], |
|
62
|
|
|
]; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Validate directive |
|
67
|
|
|
* |
|
68
|
|
|
* @param $directive |
|
69
|
|
|
* @param $directives |
|
70
|
|
|
* @return string |
|
71
|
|
|
* @throws ParserException |
|
72
|
|
|
*/ |
|
73
|
|
|
protected function validateDirective($directive, $directives) |
|
74
|
|
|
{ |
|
75
|
|
|
if (!in_array($directive, $directives, true)) { |
|
76
|
|
|
throw new ParserException('Directive is not allowed here'); |
|
77
|
|
|
} |
|
78
|
|
|
return mb_strtolower($directive); |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
|
VIPnytt /
RobotsTxtParser
