1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Bee4\RobotsTxt; |
4
|
|
|
|
5
|
|
|
use Bee4\RobotsTxt\Exception\InvalidContentException; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Class Parser |
9
|
|
|
* Take the content of a robots.txt file and transform it to rules |
10
|
|
|
* |
11
|
|
|
* @copyright Bee4 2015 |
12
|
|
|
* @author Stephane HULARD <[email protected]> |
13
|
|
|
*/ |
14
|
|
|
class Parser |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* Parse the current content |
18
|
|
|
* @return Rules |
19
|
|
|
*/ |
20
|
|
|
public function analyze($content) |
21
|
|
|
{ |
22
|
1 |
|
$content = $this->handleContent($content); |
23
|
|
|
|
24
|
1 |
|
$rules = new Rules(); |
25
|
1 |
|
$current = []; |
26
|
1 |
|
$ua = false; |
27
|
|
|
|
28
|
1 |
|
foreach ($content as $line) { |
29
|
1 |
|
if (0 === strpos($line, '#')) { |
30
|
1 |
|
continue; |
31
|
|
|
} |
32
|
|
|
|
33
|
1 |
|
if (preg_match('/^\s*User-Agent\: (.*)$/i', $line, $matches)) { |
34
|
1 |
|
if ($ua !== true) { |
35
|
1 |
|
$this->populateRules($rules, $current); |
36
|
1 |
|
$current = []; |
37
|
1 |
|
} |
38
|
1 |
|
$current[] = new Rule($matches[1]); |
39
|
1 |
|
$ua = true; |
40
|
1 |
|
} else { |
41
|
1 |
|
$ua = false; |
42
|
1 |
|
$this->parseLine($current, $line, $rules); |
43
|
|
|
} |
44
|
1 |
|
} |
45
|
1 |
|
$this->populateRules($rules, $current); |
46
|
1 |
|
return $rules; |
47
|
|
|
} |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* Handle content to build a valid instance |
51
|
|
|
* @param string|Content $content |
52
|
|
|
* @return Content |
53
|
|
|
*/ |
54
|
|
|
private function handleContent($content) |
55
|
|
|
{ |
56
|
1 |
|
if (is_string($content)) { |
57
|
1 |
|
$content = new Content($content); |
58
|
1 |
|
} |
59
|
1 |
|
if (!($content instanceof Content)) { |
60
|
1 |
|
throw (new InvalidContentException( |
61
|
|
|
'Content must be a `string` or a `Content` instance' |
62
|
1 |
|
))->setContent($content); |
63
|
1 |
|
} |
64
|
|
|
|
65
|
1 |
|
return $content; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Transform file content to structured Rules |
70
|
|
|
* @param string|Content $content |
71
|
|
|
* @return Rules |
72
|
|
|
*/ |
73
|
|
|
public static function parse($content) |
74
|
|
|
{ |
75
|
1 |
|
$parser = new self(); |
76
|
1 |
|
return $parser->analyze($content); |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
/** |
80
|
|
|
* Parse a line of data |
81
|
|
|
* @param array &$current |
82
|
|
|
* @param string $line |
83
|
|
|
*/ |
84
|
|
|
private function parseLine(array &$current, $line, Rules $rules) |
85
|
|
|
{ |
86
|
1 |
|
if (preg_match('/^\s*Allow: (.*)$/i', $line, $matches)) { |
87
|
1 |
|
$this->apply($current, 'allow', $matches[1]); |
88
|
1 |
|
} elseif (preg_match('/^\s*Disallow: (.*)$/i', $line, $matches)) { |
89
|
1 |
|
$this->apply($current, 'disallow', $matches[1]); |
90
|
1 |
|
} elseif (preg_match('/^\s*Sitemap: (.*)$/i', $line, $matches)) { |
91
|
1 |
|
$rules->addSitemap($matches[1]); |
92
|
1 |
|
} |
93
|
1 |
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Apply a method on all element of a given array |
97
|
|
|
* @param array $data |
98
|
|
|
* @param string $method |
99
|
|
|
* @param string $param |
100
|
|
|
*/ |
101
|
|
|
private function apply(array $data, $method, $param) |
102
|
|
|
{ |
103
|
1 |
|
array_walk($data, function ($item) use ($method, $param) { |
104
|
1 |
|
$item->$method($param); |
105
|
1 |
|
}); |
106
|
1 |
|
} |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* Populate rules property with build Rule instance |
110
|
|
|
* @param Rules $rules |
111
|
|
|
* @param array $current Collection of Rule objects |
112
|
|
|
*/ |
113
|
|
|
private function populateRules(Rules $rules, array $current) |
114
|
|
|
{ |
115
|
1 |
|
foreach ($current as $item) { |
116
|
1 |
|
$rules->add($item); |
117
|
1 |
|
} |
118
|
1 |
|
} |
119
|
|
|
} |
120
|
|
|
|