|
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\Directives; |
|
10
|
|
|
|
|
11
|
|
|
use DateTime; |
|
12
|
|
|
use DateTimeZone; |
|
13
|
|
|
use vipnytt\RobotsTxtParser\Handler\ErrorHandler; |
|
14
|
|
|
use vipnytt\RobotsTxtParser\Parser\UriParser; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Trait DirectiveClientTrait |
|
18
|
|
|
* |
|
19
|
|
|
* @package vipnytt\RobotsTxtParser\Client\Directives |
|
20
|
|
|
*/ |
|
21
|
|
|
trait DirectiveClientTrait |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* Get path and query |
|
25
|
|
|
* |
|
26
|
|
|
* @param string $uri |
|
27
|
|
|
* @return string |
|
28
|
|
|
* @throws \InvalidArgumentException |
|
29
|
|
|
*/ |
|
30
|
|
|
private function getPathFromUri($uri) |
|
31
|
|
|
{ |
|
32
|
|
|
$uriParser = new UriParser($uri); |
|
33
|
|
|
// Prepare uri |
|
34
|
|
|
$uriParser->encode(); |
|
35
|
|
|
$uri = $uriParser->stripFragment(); |
|
36
|
|
|
if (strpos($uri, '/') === 0) { |
|
37
|
|
|
// URI is already an path |
|
38
|
|
|
return $uri; |
|
39
|
|
|
} |
|
40
|
|
|
if (!$uriParser->validate()) { |
|
41
|
|
|
throw new \InvalidArgumentException('Invalid URI'); |
|
42
|
|
|
} |
|
43
|
|
|
$path = (($path = parse_url($uri, PHP_URL_PATH)) === null) ? '/' : $path; |
|
44
|
|
|
$query = (($query = parse_url($uri, PHP_URL_QUERY)) === null) ? '' : '?' . $query; |
|
45
|
|
|
return $path . $query; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Is time between |
|
50
|
|
|
* |
|
51
|
|
|
* @param int $timestamp |
|
52
|
|
|
* @param string $fromTime |
|
53
|
|
|
* @param string $toTime |
|
54
|
|
|
* @param string $format |
|
55
|
|
|
* @return bool |
|
56
|
|
|
*/ |
|
57
|
|
|
private function isBetween($timestamp, $fromTime, $toTime, $format = 'Hi') |
|
58
|
|
|
{ |
|
59
|
|
|
$dateTime = new DateTime(); |
|
60
|
|
|
$timezone = new DateTimeZone('UTC'); |
|
61
|
|
|
$dtRef = $dateTime->createFromFormat('U', $timestamp, $timezone); |
|
62
|
|
|
$dtFrom = $dateTime->createFromFormat($format, $fromTime, $timezone); |
|
63
|
|
|
$dtTo = $dateTime->createFromFormat($format, $toTime, $timezone); |
|
64
|
|
|
if ($dtFrom > $dtTo) { |
|
65
|
|
|
$dtTo->modify('+1 day'); |
|
66
|
|
|
} |
|
67
|
|
|
return ( |
|
68
|
|
|
$dtFrom <= $dtRef && |
|
69
|
|
|
$dtRef <= $dtTo |
|
70
|
|
|
) || ( |
|
71
|
|
|
$dtFrom <= $dtRef->modify('+1 day') && |
|
72
|
|
|
$dtRef <= $dtTo |
|
73
|
|
|
); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Check path rule |
|
78
|
|
|
* |
|
79
|
|
|
* @param string $path |
|
80
|
|
|
* @param string[] $paths |
|
81
|
|
|
* @return int|false |
|
82
|
|
|
*/ |
|
83
|
|
|
private function checkPaths($path, array $paths) |
|
84
|
|
|
{ |
|
85
|
|
|
$pairs = [ |
|
86
|
|
|
'?' => '\?', |
|
87
|
|
|
'.' => '\.', |
|
88
|
|
|
'*' => '.*', |
|
89
|
|
|
]; |
|
90
|
|
|
$errorHandler = new ErrorHandler(); |
|
91
|
|
|
set_error_handler([$errorHandler, 'callback'], E_NOTICE | E_WARNING); |
|
92
|
|
|
foreach ($paths as $rule) { |
|
93
|
|
|
$escaped = $rule; |
|
94
|
|
|
foreach ($pairs as $search => $replace) { |
|
95
|
|
|
$escaped = str_replace($search, $replace, $escaped); |
|
96
|
|
|
} |
|
97
|
|
|
if (preg_match('#^' . $escaped . '#', $path) === 1) { |
|
98
|
|
|
restore_error_handler(); |
|
99
|
|
|
return mb_strlen($rule); |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
restore_error_handler(); |
|
103
|
|
|
return false; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
|
|
|