|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types = 1); |
|
3
|
|
|
/** |
|
4
|
|
|
* This file is part of GitterBot package. |
|
5
|
|
|
* |
|
6
|
|
|
* @author Serafim <[email protected]> |
|
7
|
|
|
* @date 28.03.2016 23:19 |
|
8
|
|
|
* |
|
9
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
10
|
|
|
* file that was distributed with this source code. |
|
11
|
|
|
*/ |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
namespace Domains\Analyser; |
|
15
|
|
|
|
|
16
|
|
|
use Core\Lazy\Fetch; |
|
17
|
|
|
use Domains\Message\Message; |
|
18
|
|
|
use Domains\Message\Url; |
|
19
|
|
|
use Illuminate\Database\Eloquent\Model; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Class UrlFinder |
|
23
|
|
|
* @package Domains\Analyser |
|
24
|
|
|
*/ |
|
25
|
|
|
class UrlFinder implements Analyser |
|
26
|
|
|
{ |
|
27
|
|
|
/** |
|
28
|
|
|
* @var Model |
|
29
|
|
|
*/ |
|
30
|
|
|
private $entity; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var string |
|
34
|
|
|
*/ |
|
35
|
|
|
private $table; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* UrlAnalyser constructor. |
|
39
|
|
|
*/ |
|
40
|
|
|
public function __construct() |
|
41
|
|
|
{ |
|
42
|
|
|
$this->entity = new Url; |
|
43
|
|
|
$this->table = $this->entity->getTable(); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @return $this|Analyser |
|
48
|
|
|
* @throws \Exception |
|
49
|
|
|
*/ |
|
50
|
|
|
public function clear() : Analyser |
|
51
|
|
|
{ |
|
52
|
|
|
$this->entity->delete(); |
|
53
|
|
|
return $this; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param \Closure|null $progress |
|
58
|
|
|
* @return $this|Analyser |
|
59
|
|
|
*/ |
|
60
|
|
|
public function analyse(\Closure $progress = null) : Analyser |
|
61
|
|
|
{ |
|
62
|
|
|
// Add hidden urls |
|
63
|
|
|
$response = new Fetch(Message::query()); |
|
64
|
|
|
|
|
65
|
|
|
/** @var Message $message */ |
|
66
|
|
|
foreach ($response as $i => $message) { |
|
67
|
|
|
$matches = $this->getUrls($message->text); |
|
68
|
|
|
|
|
69
|
|
|
if (count($matches)) { |
|
70
|
|
|
$urls = []; |
|
71
|
|
|
foreach ($matches as $url) { |
|
72
|
|
|
$urls[] = new Url(['url' => $url]); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
$message->urls()->saveMany($urls); |
|
76
|
|
|
|
|
77
|
|
|
if ($progress !== null) { |
|
78
|
|
|
$progress($message, $matches, $i++); |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
return $this; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
/** |
|
87
|
|
|
* @param string $text |
|
88
|
|
|
* @return mixed |
|
89
|
|
|
*/ |
|
90
|
|
|
private function getUrls(string $text) |
|
91
|
|
|
{ |
|
92
|
|
|
// TODO Add support for non-latin domains |
|
93
|
|
|
// Current RFC 1738 |
|
94
|
|
|
$pattern = static::getPattern(); |
|
95
|
|
|
preg_match_all($pattern, $text . ' ', $matches, PREG_PATTERN_ORDER); |
|
96
|
|
|
|
|
97
|
|
|
return $matches[1]; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* @return string |
|
102
|
|
|
*/ |
|
103
|
|
|
public static function getPattern() |
|
104
|
|
|
{ |
|
105
|
|
|
return '/([a-z]{2,5}:\/\/[a-z]+\.[a-z]{2,}[\w\/\?=%#\-&:\$\.\+\!\*]+)(?:\s|\n)/iu'; |
|
106
|
|
|
} |
|
107
|
|
|
} |
|
108
|
|
|
|
LaravelRUS /
GitterBot
