|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
namespace MazenTouati\NoEmoji\Entities; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* A utility class that holds and interacts with groups |
|
8
|
|
|
* |
|
9
|
|
|
* @author Mazen Touati <[email protected]> |
|
10
|
|
|
*/ |
|
11
|
|
|
class GroupsFactory |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* The RegEx pattern used to extract the groups from the file |
|
15
|
|
|
* |
|
16
|
|
|
* @var string |
|
17
|
|
|
*/ |
|
18
|
|
|
public $pattern = '/# group: (.*)\n+([^@]+)/m'; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Holds the Group's instances |
|
22
|
|
|
* |
|
23
|
|
|
* @var array |
|
24
|
|
|
*/ |
|
25
|
|
|
public $groups = []; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Holds the Unicodes extracted from the groups |
|
29
|
|
|
* |
|
30
|
|
|
* @var array |
|
31
|
|
|
*/ |
|
32
|
|
|
public $emojiUnicodes = []; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* Unicodes getter |
|
36
|
|
|
* |
|
37
|
|
|
* @return array |
|
38
|
|
|
*/ |
|
39
|
|
|
public function getUnicodes() |
|
40
|
|
|
{ |
|
41
|
|
|
return $this->emojiUnicodes; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Divides the whole file into smaller groups. |
|
46
|
|
|
* |
|
47
|
|
|
* @param string $source The textual source to perform the scrapping on |
|
48
|
|
|
* @return GroupsFactory |
|
49
|
|
|
*/ |
|
50
|
|
|
public function extractGroups($source) |
|
51
|
|
|
{ |
|
52
|
|
|
preg_match_all($this->pattern, $source, $matches); |
|
53
|
|
|
foreach ($matches[1] as $i => $groupName) { |
|
54
|
|
|
$this->groups[] = new Group($groupName, $matches[2][$i]); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
return $this; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* Extracts the Unicodes from the plain text of each group |
|
62
|
|
|
* @return GroupsFactory |
|
63
|
|
|
*/ |
|
64
|
|
|
public function extractUnicodesFromGroups() |
|
65
|
|
|
{ |
|
66
|
|
|
foreach ($this->groups as $group) { |
|
67
|
|
|
$this->emojiUnicodes[$group->name] = $group->extractUnicodes()->getUnicodes(); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
return $this; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* Flattens the Unicode array from being group based to bit based |
|
75
|
|
|
* |
|
76
|
|
|
* @return GroupsFactory |
|
77
|
|
|
*/ |
|
78
|
|
|
public function flattenUnicodes() |
|
79
|
|
|
{ |
|
80
|
|
|
$flattenedUnicodes = [ |
|
81
|
|
|
Group::SINGLE_UNIT_20_BIT => [], |
|
82
|
|
|
Group::MULTI_UNIT_20_BIT => [], |
|
83
|
|
|
Group::SINGLE_UNIT_16_BIT => [], |
|
84
|
|
|
Group::MULTI_UNIT_16_BIT => [], |
|
85
|
|
|
]; |
|
86
|
|
|
|
|
87
|
|
|
$bitKeys = array_keys($flattenedUnicodes); |
|
88
|
|
|
|
|
89
|
|
|
foreach ($bitKeys as $bits) { |
|
90
|
|
|
foreach ($this->groups as $group) { |
|
91
|
|
|
$unicodes = $group->getUnicodes($bits); |
|
92
|
|
|
$flattenedUnicodes[$bits] = array_merge($flattenedUnicodes[$bits], $unicodes); |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
$this->emojiUnicodes = $flattenedUnicodes; |
|
97
|
|
|
$this->mergeMultiWithSingle(Group::MULTI_UNIT_20_BIT, Group::SINGLE_UNIT_20_BIT); |
|
98
|
|
|
|
|
99
|
|
|
return $this; |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* Transforms single unit values from the multi unit array to the single unit array |
|
104
|
|
|
* |
|
105
|
|
|
* @param string $multi Multi unit values array's key |
|
106
|
|
|
* @param string $single Single unit values array's key |
|
107
|
|
|
*/ |
|
108
|
|
|
public function mergeMultiWithSingle(string $multi, string $single): void |
|
109
|
|
|
{ |
|
110
|
|
|
$digits = (int)$single / 4; |
|
111
|
|
|
|
|
112
|
|
|
foreach ($this->emojiUnicodes[$multi] as $k => $v) { |
|
113
|
|
|
// If the following value is single unit then transform it |
|
114
|
|
|
if (!isset($v[$digits])) { |
|
115
|
|
|
unset($this->emojiUnicodes[$multi][$k]); |
|
116
|
|
|
$this->emojiUnicodes[$single][] = $v; |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
// Remove the values holder if all values are transformed to the single unit holder |
|
121
|
|
|
if (empty($this->emojiUnicodes[$multi])) { |
|
122
|
|
|
unset($this->emojiUnicodes[$multi]); |
|
123
|
|
|
} |
|
124
|
|
|
// Otherwise re-base the holder |
|
125
|
|
|
else { |
|
126
|
|
|
$this->emojiUnicodes[$multi] = array_values($this->emojiUnicodes[$multi]); |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
mazentouati /
no-emoji
