|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of the Purl package, a project by Jonathan H. Wage. |
|
5
|
|
|
* |
|
6
|
|
|
* (c) 2013 Jonathan H. Wage |
|
7
|
|
|
* |
|
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
9
|
|
|
* file that was distributed with this source code. |
|
10
|
|
|
*/ |
|
11
|
|
|
|
|
12
|
|
|
namespace Purl; |
|
13
|
|
|
|
|
14
|
|
|
use Pdp\Parser as PslParser; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Parser class. |
|
18
|
|
|
* |
|
19
|
|
|
* @author Jonathan H. Wage <[email protected]> |
|
20
|
|
|
*/ |
|
21
|
|
|
class Parser implements ParserInterface |
|
22
|
|
|
{ |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var PslParser Public Suffix List parser |
|
26
|
|
|
*/ |
|
27
|
|
|
private static $pslParser = null; |
|
28
|
|
|
|
|
29
|
|
|
private static $defaultParts = [ |
|
30
|
|
|
'scheme' => null, |
|
31
|
|
|
'host' => null, |
|
32
|
|
|
'port' => null, |
|
33
|
|
|
'user' => null, |
|
34
|
|
|
'pass' => null, |
|
35
|
|
|
'path' => null, |
|
36
|
|
|
'query' => null, |
|
37
|
|
|
'fragment' => null, |
|
38
|
|
|
'publicSuffix' => null, |
|
39
|
|
|
'registerableDomain' => null, |
|
40
|
|
|
'subdomain' => null, |
|
41
|
|
|
'canonical' => null, |
|
42
|
|
|
'resource' => null, |
|
43
|
|
|
]; |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* Public constructor |
|
47
|
|
|
* |
|
48
|
|
|
* @param PslParser $pslParser Public Suffix List parser |
|
49
|
|
|
*/ |
|
50
|
61 |
|
public function __construct(PslParser $pslParser) |
|
51
|
|
|
{ |
|
52
|
61 |
|
if (self::$pslParser === null) { |
|
53
|
1 |
|
self::$pslParser = $pslParser; |
|
54
|
|
|
} |
|
55
|
61 |
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @inheritDoc |
|
59
|
|
|
*/ |
|
60
|
61 |
|
public function parseUrl($url) |
|
61
|
|
|
{ |
|
62
|
61 |
|
$url = (string)$url; |
|
63
|
|
|
|
|
64
|
61 |
|
$parsedUrl = $this->doParseUrl($url); |
|
65
|
|
|
|
|
66
|
60 |
|
if ($parsedUrl === false) { |
|
67
|
|
|
throw new \InvalidArgumentException(sprintf('Invalid url %s', $url)); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
60 |
|
$parsedUrl = array_merge(self::$defaultParts, $parsedUrl); |
|
71
|
|
|
|
|
72
|
60 |
|
if (isset($parsedUrl['host'])) { |
|
73
|
59 |
|
$parsedUrl['publicSuffix'] = self::$pslParser->getPublicSuffix($parsedUrl['host']); |
|
74
|
59 |
|
$parsedUrl['registerableDomain'] = self::$pslParser->getRegistrableDomain($parsedUrl['host']); |
|
75
|
59 |
|
$parsedUrl['subdomain'] = self::$pslParser->getSubdomain($parsedUrl['host']); |
|
76
|
59 |
|
$parsedUrl['canonical'] = implode('.', array_reverse(explode('.', $parsedUrl['host']))) . ($parsedUrl['path'] ?? '') . (isset($parsedUrl['query']) ? '?' . $parsedUrl['query'] : ''); |
|
77
|
|
|
|
|
78
|
59 |
|
$parsedUrl['resource'] = $parsedUrl['path'] ?? ''; |
|
79
|
|
|
|
|
80
|
59 |
|
if (isset($parsedUrl['query'])) { |
|
81
|
6 |
|
$parsedUrl['resource'] .= '?' . $parsedUrl['query']; |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
60 |
|
return $parsedUrl; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @param string $url |
|
90
|
|
|
* |
|
91
|
|
|
* @return array $parsedUrl |
|
92
|
|
|
*/ |
|
93
|
61 |
|
protected function doParseUrl(string $url): array |
|
94
|
|
|
{ |
|
95
|
|
|
// If there's a single leading forward slash, use parse_url() |
|
96
|
|
|
// Expected matches: |
|
97
|
|
|
// |
|
98
|
|
|
// "/one/two" YES |
|
99
|
|
|
// "/" YES PLEASE |
|
100
|
|
|
// "//" NO |
|
101
|
|
|
// "//one/two" NO |
|
102
|
|
|
// "" HELL NO |
|
103
|
61 |
|
if (preg_match('#^[\/]([^\/]|$)#', $url) === 1) { |
|
104
|
2 |
|
return parse_url($url); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
// Otherwise use the PSL parser |
|
108
|
60 |
|
return self::$pslParser->parseUrl($url)->toArray(); |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|