|
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\Client; |
|
10
|
|
|
|
|
11
|
|
|
use vipnytt\RobotsTxtParser\Handler\Directives\RootDirectiveHandler; |
|
12
|
|
|
use vipnytt\RobotsTxtParser\Handler\RenderHandler; |
|
13
|
|
|
use vipnytt\RobotsTxtParser\RobotsTxtInterface; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Class RenderClient |
|
17
|
|
|
* |
|
18
|
|
|
* @package vipnytt\RobotsTxtParser\Client |
|
19
|
|
|
*/ |
|
20
|
|
|
class RenderClient implements RobotsTxtInterface |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* Handler |
|
24
|
|
|
* @var RootDirectiveHandler |
|
25
|
|
|
*/ |
|
26
|
|
|
private $root; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* RenderClient constructor. |
|
30
|
|
|
* |
|
31
|
|
|
* @param RootDirectiveHandler $handler |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct(RootDirectiveHandler $handler) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->root = $handler; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Compatibility optimized robots.txt |
|
40
|
|
|
* |
|
41
|
|
|
* Differences compared to Normal: |
|
42
|
|
|
* - Each User-agent is listed with it's own separate rule set (even if it's equal to others) |
|
43
|
|
|
* - Byte consuming, may be multiple times larger (depending on the number of user-agents) |
|
44
|
|
|
* - Maximum compatibility, optimized for badly written 3rd party parsers with limited specification support |
|
45
|
|
|
* |
|
46
|
|
|
* @param string $eol |
|
47
|
|
|
* @return string |
|
48
|
|
|
*/ |
|
49
|
|
|
public function compatibility($eol = "\r\n") |
|
50
|
|
|
{ |
|
51
|
|
|
return $this->generate(2, $eol) . $eol; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* Generate an rule string array |
|
56
|
|
|
* |
|
57
|
|
|
* @param int $level |
|
58
|
|
|
* @param string $eol |
|
59
|
|
|
* @return string |
|
60
|
|
|
*/ |
|
61
|
|
|
private function generate($level, $eol) |
|
62
|
|
|
{ |
|
63
|
|
|
$handler = new RenderHandler($level, $eol); |
|
64
|
|
|
if ($level == 2) { |
|
65
|
|
|
$this->root->userAgent->render($handler); |
|
66
|
|
|
} |
|
67
|
|
|
$this->root->host->render($handler); |
|
68
|
|
|
$this->root->cleanParam->render($handler); |
|
69
|
|
|
$this->root->sitemap->render($handler); |
|
70
|
|
|
if ($level != 2) { |
|
71
|
|
|
$this->root->userAgent->render($handler); |
|
72
|
|
|
} |
|
73
|
|
|
return $handler->generate(); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Normal looking robots.txt |
|
78
|
|
|
* |
|
79
|
|
|
* Differences compared to Compressed: |
|
80
|
|
|
* - The directives first character is uppercase |
|
81
|
|
|
* - Whitespace between the directive and it's value |
|
82
|
|
|
* - Easy to edit |
|
83
|
|
|
* - User-agent groups are separated with blank lines |
|
84
|
|
|
* |
|
85
|
|
|
* @param string $eol |
|
86
|
|
|
* @return string |
|
87
|
|
|
*/ |
|
88
|
|
|
public function normal($eol = "\r\n") |
|
89
|
|
|
{ |
|
90
|
|
|
return $this->generate(1, $eol); |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Compressed robots.txt |
|
95
|
|
|
* |
|
96
|
|
|
* Generates an robots.txt compressed to a absolute minimum. |
|
97
|
|
|
* Best suited for parsing purposes and storage in databases. |
|
98
|
|
|
* |
|
99
|
|
|
* @param string $eol |
|
100
|
|
|
* @return string |
|
101
|
|
|
*/ |
|
102
|
|
|
public function compressed($eol = PHP_EOL) |
|
103
|
|
|
{ |
|
104
|
|
|
return $this->generate(0, $eol); |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
VIPnytt /
RobotsTxtParser
