1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Modules; |
3
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\ClientException; |
5
|
|
|
|
6
|
|
|
/** |
7
|
|
|
* Trait UrlTools |
8
|
|
|
* |
9
|
|
|
* @package vipnytt\RobotsTxtParser\Modules |
10
|
|
|
*/ |
11
|
|
|
trait UrlTools |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* Convert relative to full URL |
15
|
|
|
* |
16
|
|
|
* @param string $url |
17
|
|
|
* @param string $base |
18
|
|
|
* @return string |
19
|
|
|
* @throws ClientException |
20
|
|
|
*/ |
21
|
|
|
protected function urlConvertToFull($url, $base) |
22
|
|
|
{ |
23
|
|
|
$url = $this->urlEncode($url); |
24
|
|
|
if ($this->urlValidate($url)) { |
25
|
|
|
return $url; |
26
|
|
|
} elseif (mb_stripos($url, '/') === 0) { |
27
|
|
|
return $this->urlBase($base) . $url; |
28
|
|
|
} |
29
|
|
|
throw new ClientException('Invalid URL'); |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* URL encoder according to RFC 3986 |
34
|
|
|
* Returns a string containing the encoded URL with disallowed characters converted to their percentage encodings. |
35
|
|
|
* @link http://publicmind.in/blog/url-encoding/ |
36
|
|
|
* |
37
|
|
|
* @param string $url |
38
|
|
|
* @return string |
39
|
|
|
*/ |
40
|
|
|
protected function urlEncode($url) |
41
|
|
|
{ |
42
|
|
|
$reserved = [ |
43
|
|
|
":" => '!%3A!ui', |
44
|
|
|
"/" => '!%2F!ui', |
45
|
|
|
"?" => '!%3F!ui', |
46
|
|
|
"#" => '!%23!ui', |
47
|
|
|
"[" => '!%5B!ui', |
48
|
|
|
"]" => '!%5D!ui', |
49
|
|
|
"@" => '!%40!ui', |
50
|
|
|
"!" => '!%21!ui', |
51
|
|
|
"$" => '!%24!ui', |
52
|
|
|
"&" => '!%26!ui', |
53
|
|
|
"'" => '!%27!ui', |
54
|
|
|
"(" => '!%28!ui', |
55
|
|
|
")" => '!%29!ui', |
56
|
|
|
"*" => '!%2A!ui', |
57
|
|
|
"+" => '!%2B!ui', |
58
|
|
|
"," => '!%2C!ui', |
59
|
|
|
";" => '!%3B!ui', |
60
|
|
|
"=" => '!%3D!ui', |
61
|
|
|
"%" => '!%25!ui' |
62
|
|
|
]; |
63
|
|
|
foreach ($reserved as $replace => $pattern) { |
64
|
|
|
$url = mb_ereg_replace($pattern, $replace, $url); |
65
|
|
|
} |
66
|
|
|
return $url; |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* Validate URL |
71
|
|
|
* |
72
|
|
|
* @param string $url |
73
|
|
|
* @return bool |
74
|
|
|
*/ |
75
|
|
|
protected function urlValidate($url) |
76
|
|
|
{ |
77
|
|
|
return ( |
78
|
|
|
filter_var($url, FILTER_VALIDATE_URL) && |
79
|
|
|
($parsed = parse_url($url)) !== false && |
80
|
|
|
$this->urlValidateHost($parsed['host']) && |
81
|
|
|
$this->urlValidateScheme($parsed['scheme']) |
82
|
|
|
); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Validate host name |
87
|
|
|
* |
88
|
|
|
* @link http://stackoverflow.com/questions/1755144/how-to-validate-domain-name-in-php |
89
|
|
|
* |
90
|
|
|
* @param string $host |
91
|
|
|
* @return bool |
92
|
|
|
*/ |
93
|
|
|
protected static function urlValidateHost($host) |
94
|
|
|
{ |
95
|
|
|
return ( |
96
|
|
|
mb_ereg_match('^([a-z\d](-*[a-z\d])*)(\.([a-z\d](-*[a-z\d])*))*$', $host) && //valid chars check |
97
|
|
|
mb_ereg_match('^.{1,253}$', $host) && //overall length check |
98
|
|
|
mb_ereg_match('^[^\.]{1,63}(\.[^\.]{1,63})*$', $host) && //length of each label |
99
|
|
|
!filter_var($host, FILTER_VALIDATE_IP) //is not an IP address |
100
|
|
|
); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Validate URL scheme |
105
|
|
|
* |
106
|
|
|
* @param string $scheme |
107
|
|
|
* @return bool |
108
|
|
|
*/ |
109
|
|
|
protected static function urlValidateScheme($scheme) |
110
|
|
|
{ |
111
|
|
|
return in_array($scheme, [ |
112
|
|
|
'http', |
113
|
|
|
'https', |
114
|
|
|
'ftp', |
115
|
|
|
'sftp', |
116
|
|
|
] |
117
|
|
|
); |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* Base URL |
122
|
|
|
* |
123
|
|
|
* @param string $url |
124
|
|
|
* @return string |
125
|
|
|
* @throws ClientException |
126
|
|
|
*/ |
127
|
|
|
protected function urlBase($url) |
128
|
|
|
{ |
129
|
|
|
if ($this->urlValidate($url) === false) { |
130
|
|
|
throw new ClientException('Invalid URL'); |
131
|
|
|
} |
132
|
|
|
$parts = [ |
133
|
|
|
'scheme' => parse_url($url, PHP_URL_SCHEME), |
134
|
|
|
'host' => parse_url($url, PHP_URL_HOST), |
135
|
|
|
]; |
136
|
|
|
$parts['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp'); |
137
|
|
|
return $parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port']; |
138
|
|
|
} |
139
|
|
|
} |
140
|
|
|
|