|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* File containing the {@see AppUtils\ConvertHelper_URLFinder} class. |
|
4
|
|
|
* |
|
5
|
|
|
* @package Application Utils |
|
6
|
|
|
* @subpackage ConvertHelper |
|
7
|
|
|
* @see AppUtils\ConvertHelper_URLFinder |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
declare(strict_types=1); |
|
11
|
|
|
|
|
12
|
|
|
namespace AppUtils; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Can find any URLs in a string, be it plain text or HTML, XML. |
|
16
|
|
|
* |
|
17
|
|
|
* @package Application Utils |
|
18
|
|
|
* @subpackage ConvertHelper |
|
19
|
|
|
* @author Sebastian Mordziol <[email protected]> |
|
20
|
|
|
* |
|
21
|
|
|
* @see ConvertHelper::createURLFinder() |
|
22
|
|
|
*/ |
|
23
|
|
|
class ConvertHelper_URLFinder |
|
24
|
|
|
{ |
|
25
|
|
|
/** |
|
26
|
|
|
* @var string |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $subject; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @var boolean |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $sorting = false; |
|
34
|
|
|
|
|
35
|
|
|
protected $schemes = array( |
|
36
|
|
|
'http', |
|
37
|
|
|
'https', |
|
38
|
|
|
'ftp', |
|
39
|
|
|
'ftps', |
|
40
|
|
|
'mailto', |
|
41
|
|
|
'svn', |
|
42
|
|
|
'ssl', |
|
43
|
|
|
'tel', |
|
44
|
|
|
); |
|
45
|
|
|
|
|
46
|
|
|
public function __construct(string $subject) |
|
47
|
|
|
{ |
|
48
|
|
|
$this->subject = $subject; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* Whether to enable sorting the URLs alphabetically (disabled by default). |
|
53
|
|
|
* |
|
54
|
|
|
* @param bool $enabled |
|
55
|
|
|
* @return ConvertHelper_URLFinder |
|
56
|
|
|
*/ |
|
57
|
|
|
public function enableSorting(bool $enabled=true) : ConvertHelper_URLFinder |
|
58
|
|
|
{ |
|
59
|
|
|
$this->sorting = $enabled; |
|
60
|
|
|
|
|
61
|
|
|
return $this; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* Prepares the subject string by adding a newline before all URL schemes, |
|
66
|
|
|
* to make it possible to parse even lists of links separated by commas or |
|
67
|
|
|
* the like (http://domain.com,http://domain2.com). |
|
68
|
|
|
*/ |
|
69
|
|
|
protected function prepareSubject() : void |
|
70
|
|
|
{ |
|
71
|
|
|
$replaces = array(); |
|
72
|
|
|
|
|
73
|
|
|
foreach($this->schemes as $scheme) |
|
74
|
|
|
{ |
|
75
|
|
|
$replaces[$scheme.':'] = PHP_EOL.$scheme.':'; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
$this->subject = str_replace(array_keys($replaces), array_values($replaces), $this->subject); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* Fetches all URLs that can be found in the subject string. |
|
83
|
|
|
* |
|
84
|
|
|
* @return string[] |
|
85
|
|
|
* |
|
86
|
|
|
* @see https://gist.github.com/gruber/249502 |
|
87
|
|
|
*/ |
|
88
|
|
|
public function getURLs() : array |
|
89
|
|
|
{ |
|
90
|
|
|
$this->prepareSubject(); |
|
91
|
|
|
|
|
92
|
|
|
$matches = array(); |
|
93
|
|
|
preg_match_all('#(?i)\b((?:[a-z][\w-]+:(?:/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:\'".,<>?«»“”‘’]))#i', $this->subject, $matches, PREG_PATTERN_ORDER); |
|
94
|
|
|
|
|
95
|
|
|
$result = array(); |
|
96
|
|
|
|
|
97
|
|
|
if(is_array($matches)) |
|
98
|
|
|
{ |
|
99
|
|
|
foreach($matches[0] as $match) |
|
100
|
|
|
{ |
|
101
|
|
|
if(strstr($match, '://') && !in_array($match, $result)) |
|
102
|
|
|
{ |
|
103
|
|
|
$result[] = $match; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
if($this->sorting) |
|
109
|
|
|
{ |
|
110
|
|
|
usort($result, function(string $a, string $b) { |
|
111
|
|
|
return strnatcasecmp($a, $b); |
|
112
|
|
|
}); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
return $result; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Retrieves all URLs as URLInfo instances. |
|
120
|
|
|
* |
|
121
|
|
|
* @return URLInfo[] |
|
122
|
|
|
*/ |
|
123
|
|
|
public function getInfos() |
|
124
|
|
|
{ |
|
125
|
|
|
$urls = $this->getURLs(); |
|
126
|
|
|
|
|
127
|
|
|
$result = array(); |
|
128
|
|
|
|
|
129
|
|
|
foreach($urls as $url) |
|
130
|
|
|
{ |
|
131
|
|
|
$result[] = parseURL($url); |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
return $result; |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
|
Mistralys /
application-utils
