|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
namespace MazenTouati\NoEmoji\Entities; |
|
5
|
|
|
|
|
6
|
|
|
/** |
|
7
|
|
|
* A group is a self-contained object that stores and manipulates the data of a single unicodes group |
|
8
|
|
|
* |
|
9
|
|
|
* @author Mazen Touati <[email protected]> |
|
10
|
|
|
*/ |
|
11
|
|
|
class Group |
|
12
|
|
|
{ |
|
13
|
|
|
/** |
|
14
|
|
|
* The array key for unicodes with single unit of 16 bit |
|
15
|
|
|
* |
|
16
|
|
|
* @var string |
|
17
|
|
|
*/ |
|
18
|
|
|
const SINGLE_UNIT_16_BIT = '16bit'; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* The array key for unicodes with single unit of 20 bit |
|
22
|
|
|
* |
|
23
|
|
|
* @var string |
|
24
|
|
|
*/ |
|
25
|
|
|
const SINGLE_UNIT_20_BIT = '20bit'; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* The array key for unicodes with multi unit of 16 bit |
|
29
|
|
|
* |
|
30
|
|
|
* @var string |
|
31
|
|
|
*/ |
|
32
|
|
|
const MULTI_UNIT_16_BIT = '16bit-MultiUnitValues'; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* The array key for unicodes with multi unit of 16 bit |
|
36
|
|
|
* |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
const MULTI_UNIT_20_BIT = '20bit-MultiUnitValues'; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Group's name or title |
|
43
|
|
|
* |
|
44
|
|
|
* @var string |
|
45
|
|
|
*/ |
|
46
|
|
|
public $name = ''; |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* The content of the group as a plain text |
|
50
|
|
|
* |
|
51
|
|
|
* @var string |
|
52
|
|
|
*/ |
|
53
|
|
|
private $_content = ''; |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* The unicodes that are extracted from the group |
|
57
|
|
|
* |
|
58
|
|
|
* @var array |
|
59
|
|
|
*/ |
|
60
|
|
|
private $_unicodes = []; |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Set of pattern to extract the different unicodes from the group |
|
64
|
|
|
* |
|
65
|
|
|
* @var array |
|
66
|
|
|
*/ |
|
67
|
|
|
private $_patterns = [ |
|
68
|
|
|
self::SINGLE_UNIT_20_BIT => '/^([0-9A-F]{5}(?: [0-9A-F]+)*)\s*;/m', |
|
69
|
|
|
self::SINGLE_UNIT_16_BIT => '/^([0-9A-F]{4}(?: [0-9A-F]+)*)\s*;/m', |
|
70
|
|
|
]; |
|
71
|
|
|
|
|
72
|
|
|
public function __construct(string $name, string $content) |
|
73
|
|
|
{ |
|
74
|
|
|
$this->name = $name; |
|
75
|
|
|
$this->_content = $content; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* Unicodes getter |
|
80
|
|
|
* |
|
81
|
|
|
* @param string $bits When not null it the method will return the data of the selected bits |
|
82
|
|
|
* |
|
83
|
|
|
* @return array The group's unicodes |
|
84
|
|
|
*/ |
|
85
|
|
|
public function getUnicodes($bits = null) |
|
86
|
|
|
{ |
|
87
|
|
|
if ($bits === null) { |
|
88
|
|
|
return $this->_unicodes; |
|
89
|
|
|
} |
|
90
|
|
|
return isset($this->_unicodes[$bits]) ? $this->_unicodes[$bits] : []; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Extracts the unicodes from the group's plain text content |
|
95
|
|
|
* |
|
96
|
|
|
* @return Group |
|
97
|
|
|
*/ |
|
98
|
|
|
public function extractUnicodes() |
|
99
|
|
|
{ |
|
100
|
|
|
$unicodes = []; |
|
101
|
|
|
|
|
102
|
|
|
foreach ($this->_patterns as $bits => $pattern) { |
|
103
|
|
|
preg_match_all($pattern, $this->_content, $matches); |
|
104
|
|
|
|
|
105
|
|
|
$extractedUnicodes = $matches[1]; |
|
106
|
|
|
|
|
107
|
|
|
$multiUnitUnicodes = $this->popMultiUnitValues($extractedUnicodes, $bits); |
|
108
|
|
|
|
|
109
|
|
|
if (!empty($extractedUnicodes)) { |
|
110
|
|
|
$unicodes[$bits] = $extractedUnicodes; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
if (!empty($multiUnitUnicodes)) { |
|
114
|
|
|
// For the 20 bit multi unit values we gonna transform the multi unit to single unit |
|
115
|
|
|
if ($bits == self::SINGLE_UNIT_20_BIT) { |
|
116
|
|
|
$unicodes[self::MULTI_UNIT_20_BIT] = $this->flattenMultiUnitValues($multiUnitUnicodes); |
|
117
|
|
|
} |
|
118
|
|
|
// The 16 bit multi unit values use an initial unit that is used by ASCII characters so we have to avoid flattening them |
|
119
|
|
|
else { |
|
120
|
|
|
$unicodes[self::MULTI_UNIT_16_BIT] = $multiUnitUnicodes; |
|
121
|
|
|
} |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
$this->_unicodes = $unicodes; |
|
126
|
|
|
return $this; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* Removes the unicodes with multi units from the input array and return an array with the removed values and discard the modifiers |
|
131
|
|
|
* |
|
132
|
|
|
* @param array $data Unicodes' array |
|
133
|
|
|
* @param string $bits The target bit's key |
|
134
|
|
|
* |
|
135
|
|
|
* @return array The removed multi unit values |
|
136
|
|
|
*/ |
|
137
|
|
|
public function popMultiUnitValues(array &$data, string $bits): array |
|
138
|
|
|
{ |
|
139
|
|
|
$multiUnitValues = []; |
|
140
|
|
|
$digits = (int)$bits / 4; |
|
141
|
|
|
|
|
142
|
|
|
// Extract the multi unit values |
|
143
|
|
|
// -- |
|
144
|
|
|
foreach ($data as $k => $v) { |
|
145
|
|
|
// it's like if count($v) >= $digits ? |
|
146
|
|
|
if (isset($v[$digits])) { |
|
147
|
|
|
$multiUnitValues[] = $v; |
|
148
|
|
|
unset($data[$k]); |
|
149
|
|
|
} |
|
150
|
|
|
} |
|
151
|
|
|
// Re-base source array |
|
152
|
|
|
$data = array_values($data); |
|
153
|
|
|
|
|
154
|
|
|
// Discard the modifiers |
|
155
|
|
|
// -- |
|
156
|
|
|
$joinedData = implode(',', $data); |
|
157
|
|
|
foreach ($multiUnitValues as $k => $v) { |
|
158
|
|
|
$firstUnit = substr($v, 0, $digits); |
|
159
|
|
|
// If the first unit is already extracted then this multi unit value represents a modifier => discard |
|
160
|
|
|
if (strpos($joinedData, $firstUnit) !== false) { |
|
161
|
|
|
unset($multiUnitValues[$k]); |
|
162
|
|
|
} |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
return array_values($multiUnitValues); |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
|
|
/** |
|
169
|
|
|
* Returns only the first unit of the given multi unit values |
|
170
|
|
|
* |
|
171
|
|
|
* @param array $data Input multi unit values |
|
172
|
|
|
* |
|
173
|
|
|
* @return array Output single unit values |
|
174
|
|
|
*/ |
|
175
|
|
|
public function flattenMultiUnitValues($data): array |
|
176
|
|
|
{ |
|
177
|
|
|
$new = []; |
|
178
|
|
|
foreach ($data as $v) { |
|
179
|
|
|
$units = explode(' ', $v); |
|
180
|
|
|
// Ignore already existent units |
|
181
|
|
|
if (in_array($units[0], $new)) { |
|
182
|
|
|
continue; |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
$new[] = $units[0]; |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
|
|
return $new; |
|
189
|
|
|
} |
|
190
|
|
|
} |
|
191
|
|
|
|
mazentouati /
no-emoji
