|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Class Hanzi |
|
4
|
|
|
* |
|
5
|
|
|
* @created 19.11.2020 |
|
6
|
|
|
* @author smiley <[email protected]> |
|
7
|
|
|
* @copyright 2020 smiley |
|
8
|
|
|
* @license MIT |
|
9
|
|
|
*/ |
|
10
|
|
|
|
|
11
|
|
|
namespace chillerlan\QRCode\Data; |
|
12
|
|
|
|
|
13
|
|
|
use chillerlan\QRCode\Common\{BitBuffer, Mode}; |
|
14
|
|
|
|
|
15
|
|
|
use function chr, implode, is_string, mb_convert_encoding, mb_detect_encoding, |
|
16
|
|
|
mb_detect_order, mb_internal_encoding, mb_strlen, ord, sprintf, strlen; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Hanzi (simplified Chinese) mode, GBT18284-2000: double-byte characters from the GB2312/GB18030 character set |
|
20
|
|
|
* |
|
21
|
|
|
* Please note that this is not part of the QR Code specification and may not be supported by all readers (ZXing-based ones do). |
|
22
|
|
|
* |
|
23
|
|
|
* @see https://en.wikipedia.org/wiki/GB_2312 |
|
24
|
|
|
* @see http://www.herongyang.com/GB2312/Introduction-of-GB2312.html |
|
25
|
|
|
* @see https://en.wikipedia.org/wiki/GBK_(character_encoding)#Encoding |
|
26
|
|
|
* @see https://gist.github.com/codemasher/91da33c44bfb48a81a6c1426bb8e4338 |
|
27
|
|
|
* @see https://github.com/zxing/zxing/blob/dfb06fa33b17a9e68321be151c22846c7b78048f/core/src/main/java/com/google/zxing/qrcode/decoder/DecodedBitStreamParser.java#L172-L209 |
|
28
|
|
|
* @see https://www.chinesestandard.net/PDF/English.aspx/GBT18284-2000 |
|
29
|
|
|
*/ |
|
30
|
|
|
final class Hanzi extends QRDataModeAbstract{ |
|
31
|
|
|
|
|
32
|
|
|
// GB2312, GB18030 |
|
33
|
|
|
public const ENCODING = 'GB18030'; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* @inheritDoc |
|
37
|
|
|
*/ |
|
38
|
|
|
protected static int $datamode = Mode::HANZI; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @inheritDoc |
|
42
|
|
|
*/ |
|
43
|
|
|
protected function getCharCount():int{ |
|
44
|
|
|
return mb_strlen($this->data, self::ENCODING); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @inheritDoc |
|
49
|
|
|
*/ |
|
50
|
|
|
public function getLengthInBits():int{ |
|
51
|
|
|
return $this->getCharCount() * 13; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @inheritDoc |
|
56
|
|
|
*/ |
|
57
|
|
|
public static function convertEncoding(string $string):string{ |
|
58
|
|
|
mb_detect_order([mb_internal_encoding(), 'UTF-8', 'GB2312', 'GB18030', 'CP936', 'EUC-CN', 'HZ']); |
|
59
|
|
|
|
|
60
|
|
|
$detected = mb_detect_encoding($string, null, true); |
|
61
|
|
|
|
|
62
|
|
|
if($detected === false){ |
|
63
|
|
|
throw new QRCodeDataException('mb_detect_encoding error'); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
if($detected === self::ENCODING){ |
|
67
|
|
|
return $string; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
$string = mb_convert_encoding($string, self::ENCODING, $detected); |
|
71
|
|
|
|
|
72
|
|
|
if(!is_string($string)){ |
|
73
|
|
|
throw new QRCodeDataException('mb_convert_encoding error'); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return $string; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* checks if a string qualifies as Hanzi/GB2312 |
|
81
|
|
|
*/ |
|
82
|
|
|
public static function validateString(string $string):bool{ |
|
83
|
|
|
$string = self::convertEncoding($string); |
|
84
|
|
|
$len = strlen($string); |
|
85
|
|
|
|
|
86
|
|
|
if($len < 2 || $len % 2 !== 0){ |
|
87
|
|
|
return false; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
for($i = 0; $i < $len; $i += 2){ |
|
91
|
|
|
$byte1 = ord($string[$i]); |
|
92
|
|
|
$byte2 = ord($string[$i + 1]); |
|
93
|
|
|
|
|
94
|
|
|
// byte 1 unused ranges |
|
95
|
|
|
if($byte1 < 0xa1 || ($byte1 > 0xa9 && $byte1 < 0xb0) || $byte1 > 0xf7){ |
|
96
|
|
|
return false; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
// byte 2 unused ranges |
|
100
|
|
|
if($byte2 < 0xa1 || $byte2 > 0xfe){ |
|
101
|
|
|
return false; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
return true; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* @inheritDoc |
|
111
|
|
|
* |
|
112
|
|
|
* @throws \chillerlan\QRCode\Data\QRCodeDataException on an illegal character occurence |
|
113
|
|
|
*/ |
|
114
|
|
|
public function write(BitBuffer $bitBuffer, int $versionNumber):void{ |
|
115
|
|
|
|
|
116
|
|
|
$bitBuffer |
|
117
|
|
|
->put($this::$datamode, 4) |
|
118
|
|
|
->put($this->getCharCount(), $this::getLengthBits($versionNumber)) |
|
119
|
|
|
; |
|
120
|
|
|
|
|
121
|
|
|
$len = strlen($this->data); |
|
122
|
|
|
|
|
123
|
|
|
for($i = 0; $i + 1 < $len; $i += 2){ |
|
124
|
|
|
$c = ((0xff & ord($this->data[$i])) << 8) | (0xff & ord($this->data[$i + 1])); |
|
125
|
|
|
|
|
126
|
|
|
if($c >= 0xa1a1 && $c <= 0xaafe){ |
|
127
|
|
|
$c -= 0x0a1a1; |
|
128
|
|
|
} |
|
129
|
|
|
elseif($c >= 0xb0a1 && $c <= 0xfafe){ |
|
130
|
|
|
$c -= 0x0a6a1; |
|
131
|
|
|
} |
|
132
|
|
|
else{ |
|
133
|
|
|
throw new QRCodeDataException(sprintf('illegal char at %d [%d]', $i + 1, $c)); |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
$bitBuffer->put(((($c >> 8) & 0xff) * 0x060) + ($c & 0xff), 13); |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
if($i < $len){ |
|
140
|
|
|
throw new QRCodeDataException(sprintf('illegal char at %d', $i + 1)); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* See specification GBT 18284-2000 |
|
147
|
|
|
* |
|
148
|
|
|
* @throws \chillerlan\QRCode\Data\QRCodeDataException |
|
149
|
|
|
*/ |
|
150
|
|
|
public static function decodeSegment(BitBuffer $bitBuffer, int $versionNumber):string{ |
|
151
|
|
|
$length = $bitBuffer->read(self::getLengthBits($versionNumber)); |
|
152
|
|
|
|
|
153
|
|
|
if($bitBuffer->available() < $length * 13){ |
|
154
|
|
|
throw new QRCodeDataException('not enough bits available'); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
// Each character will require 2 bytes. Read the characters as 2-byte pairs and decode as GB2312 afterwards |
|
158
|
|
|
$buffer = []; |
|
159
|
|
|
$offset = 0; |
|
160
|
|
|
|
|
161
|
|
|
while($length > 0){ |
|
162
|
|
|
// Each 13 bits encodes a 2-byte character |
|
163
|
|
|
$twoBytes = $bitBuffer->read(13); |
|
164
|
|
|
$assembledTwoBytes = (($twoBytes / 0x060) << 8) | ($twoBytes % 0x060); |
|
165
|
|
|
|
|
166
|
|
|
$assembledTwoBytes += ($assembledTwoBytes < 0x00a00) // 0x003BF |
|
167
|
|
|
? 0x0a1a1 // In the 0xA1A1 to 0xAAFE range |
|
168
|
|
|
: 0x0a6a1; // In the 0xB0A1 to 0xFAFE range |
|
169
|
|
|
|
|
170
|
|
|
$buffer[$offset] = chr(0xff & ($assembledTwoBytes >> 8)); |
|
171
|
|
|
$buffer[$offset + 1] = chr(0xff & $assembledTwoBytes); |
|
172
|
|
|
$offset += 2; |
|
173
|
|
|
$length--; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
return mb_convert_encoding(implode($buffer), mb_internal_encoding(), self::ENCODING); |
|
|
|
|
|
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
} |
|
180
|
|
|
|
chillerlan /
php-qrcode
Loading history...