|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace whm\Smoke\Rules\Seo; |
|
4
|
|
|
|
|
5
|
|
|
use phm\HttpWebdriverClient\Http\Response\UriAwareResponse; |
|
6
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
7
|
|
|
use whm\Smoke\Http\ClientAware; |
|
8
|
|
|
use whm\Smoke\Http\Response; |
|
9
|
|
|
use whm\Smoke\Rules\Rule; |
|
10
|
|
|
use whm\Smoke\Rules\ValidationFailedException; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* This rule checks if robots.txt has no entry "Disallow:/". |
|
14
|
|
|
*/ |
|
15
|
|
|
class RobotsDisallowAllRule implements Rule |
|
16
|
|
|
{ |
|
17
|
|
|
public function init() |
|
18
|
|
|
{ |
|
19
|
|
|
|
|
20
|
|
|
} |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @param ResponseInterface $response |
|
24
|
|
|
* @throws ValidationFailedException |
|
25
|
|
|
*/ |
|
26
|
|
|
public function validate(ResponseInterface $response) |
|
27
|
|
|
{ |
|
28
|
|
|
if ($response instanceof UriAwareResponse) { |
|
29
|
|
|
$url = $response->getUri()->getScheme() . '://' . $response->getUri()->getHost(); |
|
30
|
|
|
|
|
31
|
|
|
if (substr_count($url, '/') === 2) { |
|
32
|
|
|
$filename = $robotsUrl = $url . '/robots.txt'; |
|
33
|
|
|
} elseif (substr_count($url, '/') === 3) { |
|
34
|
|
|
$filename = $robotsUrl = $url . 'robots.txt'; |
|
35
|
|
|
} else { |
|
36
|
|
|
return; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
try { |
|
40
|
|
|
$content = @file_get_contents($filename); |
|
41
|
|
|
} catch (\Exception $e) { |
|
42
|
|
|
return; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
$normalizedContent = $this->normalizeContent($content); |
|
46
|
|
|
|
|
47
|
|
|
if (strpos($normalizedContent, 'user-agent:* disallow:/' . PHP_EOL) !== false) { |
|
48
|
|
|
throw new ValidationFailedException('The robots.txt contains disallow all (Disallow: /)'); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
if (strpos($normalizedContent, 'user-agent:* disallow:/') === strlen($normalizedContent) - 23) { |
|
52
|
|
|
throw new ValidationFailedException('The robots.txt contains disallow all (Disallow: /)'); |
|
53
|
|
|
} |
|
54
|
|
|
} |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
private function normalizeContent($content) |
|
58
|
|
|
{ |
|
59
|
|
|
$normalizedContent = strtolower($content); |
|
60
|
|
|
$normalizedContent = str_replace(' ', '', $normalizedContent); |
|
61
|
|
|
|
|
62
|
|
|
$normalizedContent = trim(preg_replace('/\s+/', ' ', $normalizedContent)); |
|
63
|
|
|
|
|
64
|
|
|
return $normalizedContent; |
|
65
|
|
|
} |
|
66
|
|
|
} |
|
67
|
|
|
|
phmLabs /
Smoke
