1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Jclyons52\PagePreview; |
4
|
|
|
|
5
|
|
|
class Url |
6
|
|
|
{ |
7
|
|
|
public $original; |
8
|
|
|
|
9
|
|
|
public $components; |
10
|
|
|
|
11
|
54 |
|
public function __construct($url) |
12
|
|
|
{ |
13
|
54 |
|
$this->original = $url; |
14
|
54 |
|
$this->components = $this->parse($url); |
15
|
51 |
|
} |
16
|
|
|
|
17
|
6 |
|
public static function findFirst($text) |
18
|
|
|
{ |
19
|
6 |
|
if ($urls = self::extract($text)) { |
20
|
3 |
|
return new self($urls[0]); |
21
|
|
|
} |
22
|
|
|
|
23
|
3 |
|
return null; |
24
|
|
|
} |
25
|
|
|
|
26
|
6 |
|
public static function findAll($text) |
27
|
|
|
{ |
28
|
6 |
|
if ($urls = self::extract($text)) { |
29
|
3 |
|
return array_map(function ($url) { |
30
|
3 |
|
return new self($url); |
31
|
3 |
|
}, $urls); |
32
|
|
|
} |
33
|
3 |
|
return null; |
34
|
|
|
} |
35
|
|
|
|
36
|
9 |
|
public static function extract($text) |
37
|
|
|
{ |
38
|
9 |
|
if (preg_match_all('!https?://\S+!', $text, $matches) === 0) { |
39
|
3 |
|
return null; |
40
|
|
|
} |
41
|
|
|
|
42
|
6 |
|
return $matches[0]; |
43
|
|
|
} |
44
|
|
|
|
45
|
54 |
|
public function parse($url) |
46
|
|
|
{ |
47
|
54 |
|
$urlComponents = parse_url($url); |
48
|
|
|
|
49
|
54 |
|
if (filter_var($url, FILTER_VALIDATE_URL) === false) { |
50
|
3 |
|
throw new \Exception("url {$url} is invalid"); |
51
|
|
|
} |
52
|
|
|
|
53
|
51 |
|
return $urlComponents; |
54
|
|
|
} |
55
|
|
|
|
56
|
36 |
|
public function formatRelativeToAbsolute($url) |
57
|
|
|
{ |
58
|
36 |
|
if (substr($url, 0, 5) === "data:") { |
59
|
30 |
|
return $url; |
60
|
|
|
} |
61
|
|
|
|
62
|
36 |
|
$path = array_key_exists('path', $this->components) ? $this->components['path'] : ''; |
63
|
|
|
|
64
|
36 |
|
if (filter_var($url, FILTER_VALIDATE_URL) !== false) { |
65
|
36 |
|
return $url; |
66
|
|
|
} |
67
|
36 |
|
if (substr($url, 0, 1) === '/') { |
68
|
36 |
|
return 'http://' . $this->components['host'] . $url; |
69
|
|
|
} |
70
|
|
|
|
71
|
36 |
|
$host = trim($this->components['host'], '/') . '/'; |
72
|
36 |
|
$path = trim($path, '/') . '/'; |
73
|
|
|
|
74
|
36 |
|
return 'http://' . $host . $path . $url; |
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
|