|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace MazenTouati\NoEmoji; |
|
6
|
|
|
|
|
7
|
|
|
use MazenTouati\Simple2wayConfig\S2WConfig; |
|
8
|
|
|
use MazenTouati\NoEmoji\Entities\File; |
|
9
|
|
|
use MazenTouati\NoEmoji\Entities\Group; |
|
10
|
|
|
use MazenTouati\NoEmoji\Entities\GroupsFactory; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* Scrapes the Unicode references file to extract and organize Unicodes |
|
14
|
|
|
* |
|
15
|
|
|
* @author Mazen Touati <[email protected]> |
|
16
|
|
|
*/ |
|
17
|
|
|
class Scrapper |
|
18
|
|
|
{ |
|
19
|
|
|
/** |
|
20
|
|
|
* The class instance |
|
21
|
|
|
* |
|
22
|
|
|
* @var Scrapper |
|
23
|
|
|
*/ |
|
24
|
|
|
private static $_instance = null; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Configuration object |
|
28
|
|
|
* |
|
29
|
|
|
* @var S2WConfig |
|
30
|
|
|
*/ |
|
31
|
|
|
public $config; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* Input file |
|
35
|
|
|
* |
|
36
|
|
|
* @var File |
|
37
|
|
|
*/ |
|
38
|
|
|
private $_file; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Emoji unicodes |
|
42
|
|
|
* |
|
43
|
|
|
* @var array |
|
44
|
|
|
*/ |
|
45
|
|
|
private $_unicodes = []; |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Retrieves the singleton instance |
|
50
|
|
|
* |
|
51
|
|
|
* @return Scrapper |
|
52
|
|
|
*/ |
|
53
|
|
|
public static function getInstance(): Scrapper |
|
54
|
|
|
{ |
|
55
|
|
|
if (self::$_instance == null) { |
|
56
|
|
|
self::$_instance = new Scrapper(); |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
return self::$_instance; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Initializes the Scrapper |
|
64
|
|
|
* |
|
65
|
|
|
* @param S2WConfig $c Configuration object |
|
66
|
|
|
* |
|
67
|
|
|
* @return Scrapper |
|
68
|
|
|
*/ |
|
69
|
|
|
public static function factory(S2WConfig $c): Scrapper |
|
70
|
|
|
{ |
|
71
|
|
|
$instance = self::getInstance(); |
|
72
|
|
|
$instance->config = $c; |
|
73
|
|
|
|
|
74
|
|
|
$instance->_file = new File($c); |
|
75
|
|
|
$instance->_file->prepareFileForScrapping(); |
|
76
|
|
|
|
|
77
|
|
|
return $instance; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* Starts the scrapping |
|
82
|
|
|
* |
|
83
|
|
|
* @return Scrapper |
|
84
|
|
|
*/ |
|
85
|
|
|
public function run(): Scrapper |
|
86
|
|
|
{ |
|
87
|
|
|
$this->_unicodes = (new GroupsFactory()) |
|
88
|
|
|
->extractGroups($this->_file->content) |
|
89
|
|
|
->extractUnicodesFromGroups() |
|
90
|
|
|
->flattenUnicodes() |
|
91
|
|
|
->getUnicodes(); |
|
92
|
|
|
|
|
93
|
|
|
return $this; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Exports the unicodes data into JSON file |
|
98
|
|
|
* |
|
99
|
|
|
* @return boolean|array Returns false on error or the result array |
|
100
|
|
|
*/ |
|
101
|
|
|
public function export() |
|
102
|
|
|
{ |
|
103
|
|
|
$base = $this->config->get('storage.base'); |
|
104
|
|
|
$output = $this->config->get('storage.output.json'); |
|
105
|
|
|
$fileTitle = $this->config->get('storage.output.jsonFileTitle', 'Undefined Title'); |
|
106
|
|
|
return File::toJSON($base . $output, $fileTitle, $this->_unicodes); |
|
107
|
|
|
} |
|
108
|
|
|
} |
|
109
|
|
|
|