1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of Bom. |
5
|
|
|
* (c) Fabrice de Stefanis / https://github.com/fab2s/Bom |
6
|
|
|
* This source file is licensed under the MIT license which you will |
7
|
|
|
* find in the LICENSE file or at https://opensource.org/licenses/MIT |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
namespace fab2s\Bom; |
11
|
|
|
|
12
|
|
|
/** |
13
|
|
|
* class Bom |
14
|
|
|
*/ |
15
|
|
|
class Bom |
16
|
|
|
{ |
17
|
|
|
/** |
18
|
|
|
* UTF8 BOM & encoding |
19
|
|
|
*/ |
20
|
|
|
const BOM_UTF8 = "\xEF\xBB\xBF"; |
21
|
|
|
const ENC_UTF8 = 'UTF-8'; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* UTF16_BE BOM & encoding |
25
|
|
|
*/ |
26
|
|
|
const BOM_UTF16_BE = "\xFE\xFF"; |
27
|
|
|
const ENC_UTF16_BE = 'UTF-16BE'; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* UTF16_LE BOM & encoding |
31
|
|
|
*/ |
32
|
|
|
const BOM_UTF16_LE = "\xFF\xFE"; |
33
|
|
|
const ENC_UTF16_LE = 'UTF-16LE'; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* UTF32_BE BOM & encoding |
37
|
|
|
*/ |
38
|
|
|
const BOM_UTF32_BE = "\x00\x00\xFE\xFF"; |
39
|
|
|
const ENC_UTF32_BE = 'UTF-32BE'; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* UTF32_LE BOM & encoding |
43
|
|
|
*/ |
44
|
|
|
const BOM_UTF32_LE = "\xFF\xFE\x00\x00"; |
45
|
|
|
const ENC_UTF32_LE = 'UTF-32LE'; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* UTF8 | UTF16_BE | UTF32_LE | UTF16_LE | UTF32_BE |
49
|
|
|
*/ |
50
|
|
|
const BOM_REGEX = '\xEF\xBB\xBF|\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE|\x00\x00\xFE\xFF'; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @var string[] |
54
|
|
|
*/ |
55
|
|
|
protected static $boms = [ |
56
|
|
|
self::ENC_UTF8 => self::BOM_UTF8, |
57
|
|
|
self::ENC_UTF16_BE => self::BOM_UTF16_BE, |
58
|
|
|
self::ENC_UTF16_LE => self::BOM_UTF16_LE, |
59
|
|
|
self::ENC_UTF32_BE => self::BOM_UTF32_BE, |
60
|
|
|
self::ENC_UTF32_LE => self::BOM_UTF32_LE, |
61
|
|
|
]; |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* @var string[] |
65
|
|
|
*/ |
66
|
|
|
protected static $smob = [ |
67
|
|
|
self::BOM_UTF8 => self::ENC_UTF8, |
68
|
|
|
self::BOM_UTF16_BE => self::ENC_UTF16_BE, |
69
|
|
|
self::BOM_UTF16_LE => self::ENC_UTF16_LE, |
70
|
|
|
self::BOM_UTF32_BE => self::ENC_UTF32_BE, |
71
|
|
|
self::BOM_UTF32_LE => self::ENC_UTF32_LE, |
72
|
|
|
]; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @param string $string |
76
|
|
|
* |
77
|
|
|
* @return string|null |
78
|
|
|
*/ |
79
|
|
|
public static function extract(string $string): ? string |
80
|
|
|
{ |
81
|
|
|
return preg_match('`^(' . static::BOM_REGEX . ')`', $string, $match) ? $match[1] : null; |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @param string $string |
86
|
|
|
* |
87
|
|
|
* @return string |
88
|
|
|
*/ |
89
|
|
|
public static function drop(string $string): string |
90
|
|
|
{ |
91
|
|
|
return preg_replace('`^(' . static::BOM_REGEX . ')`', '', $string); |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* @param string $bom |
96
|
|
|
* |
97
|
|
|
* @return string|null |
98
|
|
|
*/ |
99
|
|
|
public static function getBomEncoding(string $bom): ? string |
100
|
|
|
{ |
101
|
|
|
return static::$smob[$bom] ?? null; |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
/** |
105
|
|
|
* @param string $encoding |
106
|
|
|
* |
107
|
|
|
* @return null|string |
108
|
|
|
*/ |
109
|
|
|
public static function getEncodingBom(string $encoding): ? string |
110
|
|
|
{ |
111
|
|
|
return static::$boms[$encoding] ?? null; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* @return array |
116
|
|
|
*/ |
117
|
|
|
public static function getBoms(): array |
118
|
|
|
{ |
119
|
|
|
return static::$boms; |
120
|
|
|
} |
121
|
|
|
} |
122
|
|
|
|