|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @author: Denis Beliaev |
|
4
|
|
|
*/ |
|
5
|
|
|
|
|
6
|
|
|
namespace DenisBeliaev\SitemapParser; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class Link |
|
10
|
|
|
* @package DenisBeliaev\SitemapParser |
|
11
|
|
|
*/ |
|
12
|
|
|
class Link |
|
13
|
|
|
{ |
|
14
|
|
|
static function normalize($link, $base) |
|
|
|
|
|
|
15
|
|
|
{ |
|
16
|
|
|
if (substr($link, 0, 2) == '<a' && substr($link, -4, 4) == '</a>') { |
|
17
|
|
|
preg_match('~href=[\'"](.*?)[\'"]~i', $link, $matches); |
|
18
|
|
|
$link = $matches[1] ?? ''; |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
$link = urldecode($link); |
|
22
|
|
|
$link = explode('#', $link)[0]; |
|
23
|
|
|
|
|
24
|
|
|
if (empty($link)) { |
|
25
|
|
|
return ''; |
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
if (preg_match('~^mailto:|tel:.*~i', $link)) { |
|
29
|
|
|
return ''; |
|
30
|
|
|
} elseif (preg_match('~^https?://[^/].*~i', $link)) { |
|
31
|
|
|
return $link; |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
$isProtocolRelative = preg_match('~^//[^/].*~', $link); |
|
35
|
|
|
|
|
36
|
|
|
$isSiteRelative = preg_match('~^/[^/].*~', $link); |
|
37
|
|
|
$isMainPage = $link == '/'; |
|
38
|
|
|
|
|
39
|
|
|
$components = parse_url($base); |
|
40
|
|
|
if (empty($components['scheme']) || empty($components['host'])) { |
|
41
|
|
|
throw new \InvalidArgumentException('Wrong base value'); |
|
42
|
|
|
} |
|
43
|
|
|
$scheme = $components['scheme']; |
|
44
|
|
|
$host = $components['host']; |
|
45
|
|
|
|
|
46
|
|
|
if ($isMainPage) { |
|
47
|
|
|
$link = $scheme . '://' . $host; |
|
48
|
|
|
} elseif ($isProtocolRelative) { |
|
49
|
|
|
$link = $scheme . ':' . $link; |
|
50
|
|
|
} elseif ($isSiteRelative) { |
|
51
|
|
|
$link = $scheme . '://' . $host . $link; |
|
52
|
|
|
} else { |
|
53
|
|
|
$link = $base . '/' . $link; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
return $link; |
|
57
|
|
|
} |
|
58
|
|
|
} |
cimmwolf /
sitemap-parser
Adding explicit visibility (
private,protected, orpublic) is generally recommend to communicate to other developers how, and from where this method is intended to be used.