|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpMyAdmin\SqlParser; |
|
6
|
|
|
|
|
7
|
|
|
use ArrayAccess; |
|
8
|
|
|
use Exception; |
|
9
|
|
|
use Stringable; |
|
10
|
|
|
|
|
11
|
|
|
use function mb_check_encoding; |
|
12
|
|
|
use function mb_strlen; |
|
13
|
|
|
use function mb_substr; |
|
14
|
|
|
use function ord; |
|
15
|
|
|
use function strlen; |
|
16
|
|
|
use function substr; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Implementation for UTF-8 strings. |
|
20
|
|
|
* |
|
21
|
|
|
* The subscript operator in PHP, when used with string will return a byte and not a character. Because in UTF-8 |
|
22
|
|
|
* strings a character may occupy more than one byte, the subscript operator may return an invalid character. |
|
23
|
|
|
* |
|
24
|
|
|
* Because the lexer relies on the subscript operator this class had to be implemented. |
|
25
|
|
|
* |
|
26
|
|
|
* Implements array-like access for UTF-8 strings. |
|
27
|
|
|
* |
|
28
|
|
|
* In this library, this class should be used to parse UTF-8 queries. |
|
29
|
|
|
* |
|
30
|
|
|
* @implements ArrayAccess<int, string> |
|
31
|
|
|
*/ |
|
32
|
|
|
class UtfString implements ArrayAccess, Stringable |
|
33
|
|
|
{ |
|
34
|
|
|
/** |
|
35
|
|
|
* The raw, multi-byte string. |
|
36
|
|
|
* |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
public $str = ''; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* The index of current byte. |
|
43
|
|
|
* |
|
44
|
|
|
* For ASCII strings, the byte index is equal to the character index. |
|
45
|
|
|
* |
|
46
|
|
|
* @var int |
|
47
|
|
|
*/ |
|
48
|
|
|
public $byteIdx = 0; |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* The index of current character. |
|
52
|
|
|
* |
|
53
|
|
|
* For non-ASCII strings, some characters occupy more than one byte and |
|
54
|
|
|
* the character index will have a lower value than the byte index. |
|
55
|
|
|
* |
|
56
|
|
|
* @var int |
|
57
|
|
|
*/ |
|
58
|
|
|
public $charIdx = 0; |
|
59
|
|
|
|
|
60
|
|
|
/** |
|
61
|
|
|
* The length of the string (in bytes). |
|
62
|
|
|
* |
|
63
|
|
|
* @var int |
|
64
|
|
|
*/ |
|
65
|
|
|
public $byteLen = 0; |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* The length of the string (in characters). |
|
69
|
|
|
* |
|
70
|
|
|
* @var int |
|
71
|
|
|
*/ |
|
72
|
|
|
public $charLen = 0; |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* @param string $str the string |
|
76
|
|
|
*/ |
|
77
|
26 |
|
public function __construct($str) |
|
78
|
|
|
{ |
|
79
|
26 |
|
$this->str = $str; |
|
80
|
26 |
|
$this->byteLen = mb_strlen($str, '8bit'); |
|
81
|
26 |
|
if (! mb_check_encoding($str, 'UTF-8')) { |
|
82
|
2 |
|
$this->charLen = 0; |
|
83
|
|
|
} else { |
|
84
|
24 |
|
$this->charLen = mb_strlen($str, 'UTF-8'); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* Checks if the given offset exists. |
|
90
|
|
|
* |
|
91
|
|
|
* @param int $offset the offset to be checked |
|
92
|
|
|
*/ |
|
93
|
2 |
|
public function offsetExists($offset): bool |
|
94
|
|
|
{ |
|
95
|
2 |
|
return ($offset >= 0) && ($offset < $this->charLen); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Gets the character at given offset. |
|
100
|
|
|
* |
|
101
|
|
|
* @param int $offset the offset to be returned |
|
102
|
|
|
*/ |
|
103
|
20 |
|
public function offsetGet($offset): string|null |
|
104
|
|
|
{ |
|
105
|
20 |
|
if (($offset < 0) || ($offset >= $this->charLen)) { |
|
106
|
4 |
|
return null; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
18 |
|
$delta = $offset - $this->charIdx; |
|
110
|
|
|
|
|
111
|
18 |
|
if ($delta > 0) { |
|
112
|
|
|
// Fast forwarding. |
|
113
|
18 |
|
while ($delta-- > 0) { |
|
114
|
18 |
|
$this->byteIdx += strlen(mb_substr(substr($this->str, $this->byteIdx, 4), 0, 1)); |
|
115
|
18 |
|
++$this->charIdx; |
|
116
|
|
|
} |
|
117
|
16 |
|
} elseif ($delta < 0) { |
|
118
|
|
|
// Rewinding. |
|
119
|
16 |
|
while ($delta++ < 0) { |
|
120
|
|
|
do { |
|
121
|
16 |
|
$byte = ord($this->str[--$this->byteIdx]); |
|
122
|
16 |
|
} while (($byte >= 128) && ($byte < 192)); |
|
123
|
|
|
|
|
124
|
16 |
|
--$this->charIdx; |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
18 |
|
return mb_substr(substr($this->str, $this->byteIdx, 4), 0, 1); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
/** |
|
132
|
|
|
* Sets the value of a character. |
|
133
|
|
|
* |
|
134
|
|
|
* @param int $offset the offset to be set |
|
135
|
|
|
* @param string $value the value to be set |
|
136
|
|
|
* |
|
137
|
|
|
* @throws Exception not implemented. |
|
138
|
|
|
*/ |
|
139
|
2 |
|
public function offsetSet($offset, $value): void |
|
140
|
|
|
{ |
|
141
|
2 |
|
throw new Exception('Not implemented.'); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* Unsets an index. |
|
146
|
|
|
* |
|
147
|
|
|
* @param int $offset the value to be unset |
|
148
|
|
|
* |
|
149
|
|
|
* @throws Exception not implemented. |
|
150
|
|
|
*/ |
|
151
|
2 |
|
public function offsetUnset($offset): void |
|
152
|
|
|
{ |
|
153
|
2 |
|
throw new Exception('Not implemented.'); |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
/** |
|
157
|
|
|
* Returns the length in characters of the string. |
|
158
|
|
|
*/ |
|
159
|
10 |
|
public function length(): int |
|
160
|
|
|
{ |
|
161
|
10 |
|
return $this->charLen; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* Returns the contained string. |
|
166
|
|
|
*/ |
|
167
|
2 |
|
public function __toString(): string |
|
168
|
|
|
{ |
|
169
|
2 |
|
return $this->str; |
|
170
|
|
|
} |
|
171
|
|
|
} |
|
172
|
|
|
|
phpmyadmin /
sql-parser
