|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace UCD\Unicode\Codepoint; |
|
4
|
|
|
|
|
5
|
|
|
use UCD\Unicode\Codepoint; |
|
6
|
|
|
use UCD\Unicode\Codepoint\Range\RangeRegexBuilder; |
|
7
|
|
|
use UCD\Unicode\Collection\TraversableBackedCollection; |
|
8
|
|
|
use UCD\Unicode\TransformationFormat; |
|
9
|
|
|
|
|
10
|
|
|
class Collection extends TraversableBackedCollection |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @return \Traversable|int[] |
|
14
|
|
|
*/ |
|
15
|
|
|
public function flatten() |
|
16
|
|
|
{ |
|
17
|
|
|
/** @var Codepoint $codepoint */ |
|
18
|
|
|
foreach ($this as $codepoint) { |
|
19
|
|
|
yield $codepoint->getValue(); |
|
20
|
|
|
} |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @param Codepoint $codepoint |
|
25
|
|
|
* @return bool |
|
26
|
|
|
*/ |
|
27
|
|
|
public function has(Codepoint $codepoint) |
|
28
|
|
|
{ |
|
29
|
|
|
foreach ($this as $check) { |
|
30
|
|
|
if ($codepoint->equals($check)) { |
|
31
|
|
|
return true; |
|
32
|
|
|
} |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
return false; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @return Range[]|Range\Collection |
|
40
|
|
|
*/ |
|
41
|
|
|
public function aggregate() |
|
42
|
|
|
{ |
|
43
|
|
|
$aggregator = new Aggregator(); |
|
44
|
|
|
|
|
45
|
|
|
$this->traverseWith(function (Codepoint $codepoint) use ($aggregator) { |
|
46
|
|
|
$aggregator->addCodepoint($codepoint); |
|
47
|
|
|
}); |
|
48
|
|
|
|
|
49
|
|
|
return $aggregator->getAggregated(); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @return string |
|
54
|
|
|
*/ |
|
55
|
|
|
public function toRegexCharacterClass() |
|
56
|
|
|
{ |
|
57
|
|
|
$builder = new RegexBuilder(); |
|
58
|
|
|
|
|
59
|
|
|
$this->traverseWith(function (Codepoint $codepoint) use ($builder) { |
|
60
|
|
|
$builder->addCodepoint($codepoint); |
|
61
|
|
|
}); |
|
62
|
|
|
|
|
63
|
|
|
return $builder->getCharacterClass(); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @param string $string |
|
68
|
|
|
* @return static |
|
69
|
|
|
*/ |
|
70
|
|
|
public static function fromUTF8($string) |
|
71
|
|
|
{ |
|
72
|
|
|
return self::fromEncodedString( |
|
73
|
|
|
$string, |
|
74
|
|
|
TransformationFormat::ofType(TransformationFormat::EIGHT) |
|
75
|
|
|
); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* @param string $string |
|
80
|
|
|
* @return static |
|
81
|
|
|
*/ |
|
82
|
|
|
public static function fromUTF16($string) |
|
83
|
|
|
{ |
|
84
|
|
|
return self::fromEncodedString( |
|
85
|
|
|
$string, |
|
86
|
|
|
TransformationFormat::ofType(TransformationFormat::SIXTEEN) |
|
87
|
|
|
); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* @param string $string |
|
92
|
|
|
* @return static |
|
93
|
|
|
*/ |
|
94
|
|
|
public static function fromUTF32($string) |
|
95
|
|
|
{ |
|
96
|
|
|
return self::fromEncodedString( |
|
97
|
|
|
$string, |
|
98
|
|
|
TransformationFormat::ofType(TransformationFormat::THIRTY_TWO) |
|
99
|
|
|
); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* @param string $string |
|
104
|
|
|
* @param TransformationFormat $encoding |
|
105
|
|
|
* @return static |
|
106
|
|
|
*/ |
|
107
|
|
|
public static function fromEncodedString($string, TransformationFormat $encoding) |
|
108
|
|
|
{ |
|
109
|
|
|
$characters = TransformationFormat\StringUtility::split($string, $encoding); |
|
110
|
|
|
|
|
111
|
|
|
$mapper = function ($character) use ($encoding) { |
|
112
|
|
|
return Codepoint::fromEncodedCharacter($character, $encoding); |
|
113
|
|
|
}; |
|
114
|
|
|
|
|
115
|
|
|
return static::fromArray( |
|
116
|
|
|
array_map($mapper, $characters) |
|
117
|
|
|
); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* @return string |
|
122
|
|
|
*/ |
|
123
|
|
|
public function toUTF8() |
|
124
|
|
|
{ |
|
125
|
|
|
return $this->toEncodedString( |
|
126
|
|
|
TransformationFormat::ofType(TransformationFormat::EIGHT) |
|
127
|
|
|
); |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* @return string |
|
132
|
|
|
*/ |
|
133
|
|
|
public function toUTF16() |
|
134
|
|
|
{ |
|
135
|
|
|
return $this->toEncodedString( |
|
136
|
|
|
TransformationFormat::ofType(TransformationFormat::SIXTEEN_BIG_ENDIAN) |
|
137
|
|
|
); |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* @return string |
|
142
|
|
|
*/ |
|
143
|
|
|
public function toUTF32() |
|
144
|
|
|
{ |
|
145
|
|
|
return $this->toEncodedString( |
|
146
|
|
|
TransformationFormat::ofType(TransformationFormat::THIRTY_TWO_BIG_ENDIAN) |
|
147
|
|
|
); |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* @param TransformationFormat $encoding |
|
152
|
|
|
* @return string |
|
153
|
|
|
*/ |
|
154
|
|
|
public function toEncodedString(TransformationFormat $encoding) |
|
155
|
|
|
{ |
|
156
|
|
|
$characters = ''; |
|
157
|
|
|
|
|
158
|
|
|
$this->traverseWith(function (Codepoint $codepoint) use ($encoding, &$characters) { |
|
159
|
|
|
$characters .= $codepoint->toEncodedCharacter($encoding); |
|
160
|
|
|
}); |
|
161
|
|
|
|
|
162
|
|
|
return $characters; |
|
163
|
|
|
} |
|
164
|
|
|
} |
nick-jones /
php-ucd
