1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Parser; |
3
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Parser\Directives\DirectiveParserCommons; |
5
|
|
|
use vipnytt\RobotsTxtParser\Parser\Directives\RootDirectiveHandler; |
6
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* Class RobotsTxtParser |
10
|
|
|
* |
11
|
|
|
* @package vipnytt\RobotsTxtParser\Parser |
12
|
|
|
*/ |
13
|
|
|
class RobotsTxtParser implements RobotsTxtInterface |
14
|
|
|
{ |
15
|
|
|
use DirectiveParserCommons; |
16
|
|
|
use UrlParser; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Directive white list |
20
|
|
|
*/ |
21
|
|
|
const TOP_LEVEL_DIRECTIVES = [ |
22
|
|
|
self::DIRECTIVE_CLEAN_PARAM => 'cleanParam', |
23
|
|
|
self::DIRECTIVE_HOST => 'host', |
24
|
|
|
self::DIRECTIVE_SITEMAP => 'sitemap', |
25
|
|
|
]; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* Root directive handler |
29
|
|
|
* @var RootDirectiveHandler |
30
|
|
|
*/ |
31
|
|
|
protected $handler; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* TxtClient constructor. |
35
|
|
|
* |
36
|
|
|
* @param string $baseUri |
37
|
|
|
* @param string $content |
38
|
|
|
*/ |
39
|
|
|
public function __construct($baseUri, $content) |
40
|
|
|
{ |
41
|
|
|
mb_internal_encoding(self::ENCODING); |
42
|
|
|
$this->handler = new RootDirectiveHandler($this->urlBase($this->urlEncode($baseUri))); |
43
|
|
|
$this->parseTxt($content); |
44
|
|
|
} |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* Client robots.txt |
48
|
|
|
* |
49
|
|
|
* @param string $txt |
50
|
|
|
* @return bool |
51
|
|
|
*/ |
52
|
|
|
private function parseTxt($txt) |
53
|
|
|
{ |
54
|
|
|
$result = []; |
55
|
|
|
$lines = array_filter(array_map('trim', mb_split('\r\n|\n|\r', $txt))); |
56
|
|
|
// Client each line individually |
57
|
|
|
foreach ($lines as $line) { |
58
|
|
|
// Limit rule length |
59
|
|
|
$line = mb_substr($line, 0, self::MAX_LENGTH_RULE); |
60
|
|
|
// Remove comments |
61
|
|
|
$line = mb_split('#', $line, 2)[0]; |
62
|
|
|
// Parse line |
63
|
|
|
$result[] = $this->parseLine($line); |
64
|
|
|
} |
65
|
|
|
return in_array(true, $result, true); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Add line |
70
|
|
|
* |
71
|
|
|
* @param string $line |
72
|
|
|
* @return bool |
73
|
|
|
*/ |
74
|
|
|
private function parseLine($line) |
75
|
|
|
{ |
76
|
|
|
if (($pair = $this->generateRulePair($line, array_keys(self::TOP_LEVEL_DIRECTIVES))) !== false) { |
77
|
|
|
return $this->handler->{self::TOP_LEVEL_DIRECTIVES[$pair['directive']]}()->add($pair['value']); |
78
|
|
|
} |
79
|
|
|
return $this->handler->userAgent()->add($line); |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Render |
84
|
|
|
* |
85
|
|
|
* @param string $lineSeparator |
86
|
|
|
* @return string |
87
|
|
|
*/ |
88
|
|
|
public function render($lineSeparator = "\n") |
89
|
|
|
{ |
90
|
|
|
return implode($lineSeparator, array_merge( |
91
|
|
|
$this->handler->host()->render(), |
92
|
|
|
$this->handler->cleanParam()->render(), |
93
|
|
|
$this->handler->sitemap()->render(), |
94
|
|
|
$this->handler->userAgent()->render() |
95
|
|
|
)); |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
/** |
99
|
|
|
* Rule array |
100
|
|
|
* |
101
|
|
|
* @return array |
102
|
|
|
*/ |
103
|
|
|
public function export() |
104
|
|
|
{ |
105
|
|
|
return [ |
106
|
|
|
self::DIRECTIVE_HOST => $this->handler->host()->client()->export(), |
107
|
|
|
self::DIRECTIVE_CLEAN_PARAM => $this->handler->cleanParam()->client()->export(), |
108
|
|
|
self::DIRECTIVE_SITEMAP => $this->handler->sitemap()->client()->export(), |
109
|
|
|
self::DIRECTIVE_USER_AGENT => $this->handler->userAgent()->export(), |
110
|
|
|
]; |
111
|
|
|
} |
112
|
|
|
} |
113
|
|
|
|