|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Coyote\Services\Parser\Parsers; |
|
4
|
|
|
|
|
5
|
|
|
use Collective\Html\HtmlBuilder; |
|
6
|
|
|
use Coyote\Services\Parser\Parsers\Parentheses\ParenthesesParser; |
|
7
|
|
|
use TRegx\CleanRegex\Match\Details\Detail; |
|
|
|
|
|
|
8
|
|
|
use TRegx\CleanRegex\Pattern; |
|
9
|
|
|
|
|
10
|
|
|
class UrlFormatter |
|
11
|
|
|
{ |
|
12
|
|
|
// http://daringfireball.net/2010/07/improved_regex_for_matching_urls |
|
13
|
|
|
// modified - removed part with parsing of parentheses, because that's |
|
14
|
|
|
// done by ParenthesesParser |
|
15
|
|
|
private const REGEXP_URL = '\b(?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)[^\s<>]+[^\s`!(\[\]{};:\'".,<>?«»“”‘’]'; |
|
16
|
|
|
private const TITLE_LEN = 64; |
|
17
|
|
|
|
|
18
|
|
|
/** @var string */ |
|
19
|
|
|
private $host; |
|
20
|
|
|
/** @var HtmlBuilder */ |
|
21
|
|
|
private $html; |
|
22
|
|
|
/** @var ParenthesesParser */ |
|
23
|
|
|
private $parser; |
|
24
|
|
|
/** @var Pattern */ |
|
25
|
|
|
private $urlPattern; |
|
26
|
|
|
|
|
27
|
|
|
public function __construct(string $host, HtmlBuilder $html, ParenthesesParser $parser) |
|
28
|
|
|
{ |
|
29
|
|
|
$this->host = $host; |
|
30
|
|
|
$this->html = $html; |
|
31
|
|
|
$this->parser = $parser; |
|
32
|
|
|
$this->urlPattern = Pattern::of(self::REGEXP_URL, 'i'); |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
public function parse(string $text): string |
|
36
|
|
|
{ |
|
37
|
|
|
return $this->urlPattern->replace($text)->callback(fn(Detail $match) => $this->processLink($match->text())); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
private function processLink(string $url): string |
|
41
|
|
|
{ |
|
42
|
|
|
return join(array_map([$this, 'buildLink'], $this->parser->parse($url))); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
private function buildLink(string $url): string |
|
46
|
|
|
{ |
|
47
|
|
|
if ($this->urlPattern->test($url)) { |
|
48
|
|
|
return $this->html->link($this->prependSchema($url), $this->truncate($url)); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
return $url; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
private function prependSchema(string $url): string |
|
55
|
|
|
{ |
|
56
|
|
|
if (pattern('^[\w]+?://', 'i')->fails($url)) { |
|
57
|
|
|
return "http://$url"; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
return $url; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
private function truncate(string $text): string |
|
64
|
|
|
{ |
|
65
|
|
|
if (mb_strlen($text) < self::TITLE_LEN) { |
|
66
|
|
|
return $text; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
if ($this->host === parse_url($text, PHP_URL_HOST)) { |
|
70
|
|
|
return $text; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return $this->truncateToLengthWith($text, self::TITLE_LEN, '[...]'); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
private function truncateToLengthWith(string $text, int $length, string $substitute): string |
|
77
|
|
|
{ |
|
78
|
|
|
$padding = ($length - mb_strlen($substitute)) / 2; |
|
79
|
|
|
|
|
80
|
|
|
return mb_substr($text, 0, $padding) . $substitute . mb_substr($text, -$padding); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|