|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Coyote\Services\Parser\Parsers; |
|
4
|
|
|
|
|
5
|
|
|
use Collective\Html\HtmlBuilder; |
|
6
|
|
|
use TRegx\SafeRegex\Exception\BacktrackLimitPregException; |
|
7
|
|
|
use TRegx\SafeRegex\preg; |
|
8
|
|
|
|
|
9
|
|
|
class UrlFormatter |
|
10
|
|
|
{ |
|
11
|
|
|
// http://daringfireball.net/2010/07/improved_regex_for_matching_urls |
|
12
|
|
|
private const REGEXP_URL = '#(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:\'".,<>?«»“”‘’]))#u'; |
|
13
|
|
|
|
|
14
|
|
|
private const TITLE_LEN = 64; |
|
15
|
|
|
|
|
16
|
|
|
/** @var string */ |
|
17
|
|
|
private $host; |
|
18
|
|
|
/** @var HtmlBuilder */ |
|
19
|
|
|
private $html; |
|
20
|
|
|
|
|
21
|
|
|
public function __construct(string $host, HtmlBuilder $html) |
|
22
|
|
|
{ |
|
23
|
|
|
$this->host = $host; |
|
24
|
|
|
$this->html = $html; |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
public function parse(string $text): string |
|
28
|
|
|
{ |
|
29
|
|
|
try { |
|
30
|
|
|
return preg::replace_callback(self::REGEXP_URL, function (array $match): string { |
|
31
|
|
|
$url = $match[0]; |
|
32
|
|
|
if (pattern('^[\w]+?://', 'i')->fails($url)) { |
|
33
|
|
|
return $this->processLink('http://' . $url, $url); |
|
34
|
|
|
} |
|
35
|
|
|
return $this->processLink($url, $url); |
|
36
|
|
|
}, $text); |
|
37
|
|
|
} catch (BacktrackLimitPregException $exception) { |
|
38
|
|
|
// regexp posiada buga, nie parsuje poprawnie URL-i jezeli zawiera on ( |
|
39
|
|
|
// poki co nie mam rozwiazania na ten problem, dlatego zwracamy nieprzeprasowany tekst |
|
40
|
|
|
// w przypadku bledu |
|
41
|
|
|
// Dokłądniej mówiąc to nie parsuje żadnego urla w poście, jeśli przynajmniej jeden z nich ma ( |
|
42
|
|
|
return $text; |
|
43
|
|
|
} |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
private function processLink(string $url, string $title): string |
|
47
|
|
|
{ |
|
48
|
|
|
$quoted = htmlspecialchars($title, ENT_QUOTES, 'UTF-8', false); // doesn't app('html') take care of this? |
|
49
|
|
|
return $this->html->link($url, $this->truncate($quoted)); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
private function truncate(string $text): string |
|
53
|
|
|
{ |
|
54
|
|
|
if (mb_strlen($text) < self::TITLE_LEN) { |
|
55
|
|
|
return $text; |
|
56
|
|
|
} |
|
57
|
|
|
if ($this->host === parse_url($text, PHP_URL_HOST)) { |
|
58
|
|
|
return $text; |
|
59
|
|
|
} |
|
60
|
|
|
return $this->truncateToLengthWith($text, self::TITLE_LEN, '[...]'); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
private function truncateToLengthWith(string $text, int $length, string $substitute): string |
|
64
|
|
|
{ |
|
65
|
|
|
$padding = ($length - mb_strlen($substitute)) / 2; |
|
66
|
|
|
return mb_substr($text, 0, $padding) . $substitute . mb_substr($text, -$padding); |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
|
adam-boduch /
coyote
