|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\Robots; |
|
4
|
|
|
|
|
5
|
|
|
use InvalidArgumentException; |
|
6
|
|
|
|
|
7
|
|
|
class RobotsHeaders |
|
8
|
|
|
{ |
|
9
|
|
|
protected $robotHeadersProperties = []; |
|
10
|
|
|
|
|
11
|
|
|
public static function readFrom(string $source): self |
|
12
|
|
|
{ |
|
13
|
|
|
$content = @file_get_contents($source); |
|
14
|
|
|
|
|
15
|
|
|
if ($content === false) { |
|
16
|
|
|
throw new InvalidArgumentException("Could not read from source `{$source}`"); |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
return new self($http_response_header ?? []); |
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
public static function create(array $headers): self |
|
23
|
|
|
{ |
|
24
|
|
|
return new self($headers); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
public function __construct(array $headers) |
|
28
|
|
|
{ |
|
29
|
|
|
$this->robotHeadersProperties = $this->parseHeaders($headers); |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
public function mayIndex(string $userAgent = '*'): bool |
|
33
|
|
|
{ |
|
34
|
|
|
return ! $this->noindex($userAgent); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function mayFollow(string $userAgent = '*'): bool |
|
38
|
|
|
{ |
|
39
|
|
|
return ! $this->nofollow($userAgent); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
|
|
43
|
|
|
public function noindex(string $userAgent = '*'): bool |
|
44
|
|
|
{ |
|
45
|
|
|
$userAgent = strtolower($userAgent); |
|
46
|
|
|
|
|
47
|
|
|
return |
|
48
|
|
|
$this->robotHeadersProperties[$userAgent]['noindex'] |
|
49
|
|
|
?? $this->robotHeadersProperties[$this->getWildCardUserAgent($userAgent)]['noindex'] |
|
50
|
|
|
?? $this->robotHeadersProperties['*']['noindex'] |
|
51
|
|
|
?? false; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
public function nofollow(string $userAgent = '*'): bool |
|
55
|
|
|
{ |
|
56
|
|
|
$userAgent = strtolower($userAgent); |
|
57
|
|
|
|
|
58
|
|
|
return |
|
59
|
|
|
$this->robotHeadersProperties[$userAgent]['nofollow'] |
|
60
|
|
|
?? $this->robotHeadersProperties[$this->getWildCardUserAgent($userAgent)]['nofollow'] |
|
61
|
|
|
?? $this->robotHeadersProperties['*']['nofollow'] |
|
62
|
|
|
?? false; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
protected function getWildCardUserAgent(string $userAgent): ?string |
|
67
|
|
|
{ |
|
68
|
|
|
if ($userAgent !== '*') { |
|
69
|
|
|
for ($i = 1; $i <= strlen($userAgent); $i++) { |
|
70
|
|
|
$wildCardUserAgent = substr($userAgent, 0, $i).'*'; |
|
71
|
|
|
if (isset($this->robotHeadersProperties[$wildCardUserAgent])) { |
|
72
|
|
|
return $wildCardUserAgent; |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
return null; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
protected function parseHeaders(array $headers): array |
|
81
|
|
|
{ |
|
82
|
|
|
$robotHeaders = $this->filterRobotHeaders($headers); |
|
83
|
|
|
|
|
84
|
|
|
return array_reduce($robotHeaders, function (array $parsedHeaders, $header) { |
|
85
|
|
|
$header = $this->normalizeHeaders($header); |
|
86
|
|
|
|
|
87
|
|
|
$headerParts = explode(':', $header); |
|
88
|
|
|
|
|
89
|
|
|
$userAgent = count($headerParts) === 3 |
|
90
|
|
|
? trim($headerParts[1]) |
|
91
|
|
|
: '*'; |
|
92
|
|
|
|
|
93
|
|
|
$userAgent = strtolower($userAgent); |
|
94
|
|
|
$options = end($headerParts); |
|
95
|
|
|
|
|
96
|
|
|
$parsedHeaders[$userAgent] = []; |
|
97
|
|
|
|
|
98
|
|
|
if (strpos(strtolower($options), 'index')) { |
|
99
|
|
|
$parsedHeaders[$userAgent]['noindex'] = strpos(strtolower($options), 'noindex') !== false; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
if (strpos(strtolower($options), 'follow')) { |
|
103
|
|
|
$parsedHeaders[$userAgent]['nofollow'] = strpos(strtolower($options), 'nofollow') !== false; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
return $parsedHeaders; |
|
107
|
|
|
}, []); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
protected function filterRobotHeaders(array $headers): array |
|
111
|
|
|
{ |
|
112
|
|
|
return array_filter($headers, function ($header) use ($headers) { |
|
113
|
|
|
$headerContent = $this->normalizeHeaders($headers[$header] ?? []); |
|
114
|
|
|
|
|
115
|
|
|
return strpos(strtolower($header), 'x-robots-tag') === 0 |
|
116
|
|
|
|| strpos(strtolower($headerContent), 'x-robots-tag') === 0; |
|
117
|
|
|
}, ARRAY_FILTER_USE_KEY); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
protected function normalizeHeaders($headers): string |
|
121
|
|
|
{ |
|
122
|
|
|
return implode(',', (array) $headers); |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
|
spatie /
robots-txt
