|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* |
|
4
|
|
|
* This file is part of Aura for PHP. |
|
5
|
|
|
* |
|
6
|
|
|
* @license http://opensource.org/licenses/bsd-license.php BSD |
|
7
|
|
|
* |
|
8
|
|
|
*/ |
|
9
|
|
|
namespace Aura\Filter\Rule; |
|
10
|
|
|
|
|
11
|
|
|
use Aura\Filter\Exception; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* |
|
15
|
|
|
* Abstract rule for string-length filters; supports the `iconv` and `mbstring` |
|
16
|
|
|
* extensions. |
|
17
|
|
|
* |
|
18
|
|
|
* @package Aura.Filter |
|
19
|
|
|
* |
|
20
|
|
|
*/ |
|
21
|
|
|
abstract class AbstractStrlen |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* |
|
25
|
|
|
* Is the `mbstring` extension loaded? |
|
26
|
|
|
* |
|
27
|
|
|
* @return bool |
|
28
|
|
|
* |
|
29
|
|
|
*/ |
|
30
|
75 |
|
protected function mbstring() |
|
31
|
|
|
{ |
|
32
|
75 |
|
return extension_loaded('mbstring'); |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* |
|
37
|
|
|
* Is the `iconv` extension loaded? |
|
38
|
|
|
* |
|
39
|
|
|
* @return bool |
|
40
|
|
|
* |
|
41
|
|
|
*/ |
|
42
|
119 |
|
protected function iconv() |
|
43
|
|
|
{ |
|
44
|
119 |
|
return extension_loaded('iconv'); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* |
|
49
|
|
|
* Proxy to `iconv_strlen()` or `mb_strlen()` when available; fall back to |
|
50
|
|
|
* `utf8_decode()` and `strlen()` otherwise. |
|
51
|
|
|
* |
|
52
|
|
|
* @param string $str Return the number of characters in this string. |
|
53
|
|
|
* |
|
54
|
|
|
* @return int |
|
55
|
|
|
* |
|
56
|
|
|
*/ |
|
57
|
121 |
|
protected function strlen($str) |
|
58
|
|
|
{ |
|
59
|
121 |
|
if ($this->iconv()) { |
|
60
|
117 |
|
return $this->strlenIconv($str); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
4 |
|
if ($this->mbstring()) { |
|
64
|
2 |
|
return mb_strlen($str, 'UTF-8'); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
2 |
|
return strlen(utf8_decode($str)); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* |
|
72
|
|
|
* Wrapper for `iconv_substr()` to throw an exception on malformed UTF-8. |
|
73
|
|
|
* |
|
74
|
|
|
* @param string $str The string to work with. |
|
75
|
|
|
* |
|
76
|
|
|
* @param int $start Start at this position. |
|
77
|
|
|
* |
|
78
|
|
|
* @param int $length End after this many characters. |
|
79
|
|
|
* |
|
80
|
|
|
* @return string |
|
81
|
|
|
* |
|
82
|
|
|
* @throws Exception\MalformedUtf8 |
|
83
|
|
|
* |
|
84
|
|
|
*/ |
|
85
|
49 |
|
protected function substrIconv($str,$start,$length) |
|
86
|
|
|
{ |
|
87
|
49 |
|
$level = error_reporting(0); |
|
88
|
49 |
|
$substr = iconv_substr($str,$start,$length, 'UTF-8'); |
|
89
|
49 |
|
error_reporting($level); |
|
90
|
|
|
|
|
91
|
49 |
|
if ($substr !== false) { |
|
92
|
48 |
|
return $substr; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
1 |
|
throw new Exception\MalformedUtf8(); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* |
|
100
|
|
|
* Wrapper for `iconv_strlen()` to throw an exception on malformed UTF-8. |
|
101
|
|
|
* |
|
102
|
|
|
* @param string $str Return the number of characters in this string. |
|
103
|
|
|
* |
|
104
|
|
|
* @return int |
|
105
|
|
|
* |
|
106
|
|
|
* @throws Exception\MalformedUtf8 |
|
107
|
|
|
* |
|
108
|
|
|
*/ |
|
109
|
117 |
|
protected function strlenIconv($str) |
|
110
|
|
|
{ |
|
111
|
117 |
|
$level = error_reporting(0); |
|
112
|
117 |
|
$strlen = iconv_strlen($str, 'UTF-8'); |
|
113
|
117 |
|
error_reporting($level); |
|
114
|
|
|
|
|
115
|
117 |
|
if ($strlen !== false) { |
|
116
|
116 |
|
return $strlen; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
1 |
|
throw new Exception\MalformedUtf8(); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* |
|
124
|
|
|
* Proxy to `iconv_substr()` or `mb_substr()` when the `mbstring` available; |
|
125
|
|
|
* polyfill via `preg_split()` and `array_slice()` otherwise. |
|
126
|
|
|
* |
|
127
|
|
|
* @param string $str The string to work with. |
|
128
|
|
|
* |
|
129
|
|
|
* @param int $start Start at this position. |
|
130
|
|
|
* |
|
131
|
|
|
* @param int $length End after this many characters. |
|
132
|
|
|
* |
|
133
|
|
|
* @return string |
|
134
|
|
|
* |
|
135
|
|
|
*/ |
|
136
|
53 |
|
protected function substr($str, $start, $length = null) |
|
137
|
|
|
{ |
|
138
|
53 |
|
if ($this->iconv()) { |
|
139
|
49 |
|
return $this->substrIconv($str, $start, $length); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
4 |
|
if ($this->mbstring()) { |
|
143
|
2 |
|
return mb_substr($str, $start, $length, 'UTF-8'); |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
2 |
|
$split = preg_split("//u", $str, -1, PREG_SPLIT_NO_EMPTY); |
|
147
|
2 |
|
return implode('', array_slice($split, $start, $length)); |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* |
|
152
|
|
|
* Userland UTF-8-aware implementation of `str_pad()`. |
|
153
|
|
|
* |
|
154
|
|
|
* @param string $input The input string. |
|
155
|
|
|
* |
|
156
|
|
|
* @param int $pad_length If the value of pad_length is negative, less than, |
|
157
|
|
|
* or equal to the length of the input string, no padding takes place. |
|
158
|
|
|
* |
|
159
|
|
|
* @param string $pad_str Pad with this string. The pad_string may be |
|
160
|
|
|
* truncated if the required number of padding characters can't be evenly |
|
161
|
|
|
* divided by the pad_string's length. |
|
162
|
|
|
* |
|
163
|
|
|
* @param int $pad_type Optional argument pad_type can be STR_PAD_RIGHT, |
|
164
|
|
|
* STR_PAD_LEFT, or STR_PAD_BOTH. If pad_type is not specified it is |
|
165
|
|
|
* assumed to be STR_PAD_RIGHT. |
|
166
|
|
|
* |
|
167
|
|
|
* @return string |
|
168
|
|
|
* |
|
169
|
|
|
*/ |
|
170
|
9 |
|
protected function strpad($input, $pad_length, $pad_str = ' ', $pad_type = STR_PAD_RIGHT) |
|
171
|
|
|
{ |
|
172
|
9 |
|
$input_len = $this->strlen($input); |
|
173
|
9 |
|
if ($pad_length <= $input_len) { |
|
174
|
3 |
|
return $input; |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
9 |
|
$pad_str_len = $this->strlen($pad_str); |
|
178
|
9 |
|
$pad_len = $pad_length - $input_len; |
|
179
|
|
|
|
|
180
|
9 |
|
if ($pad_type == STR_PAD_LEFT) { |
|
181
|
3 |
|
$repeat_times = ceil($pad_len / $pad_str_len); |
|
182
|
3 |
|
$prefix = str_repeat($pad_str, $repeat_times); |
|
183
|
3 |
|
return $this->substr($prefix, 0, floor($pad_len)) . $input; |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
9 |
|
if ($pad_type == STR_PAD_BOTH) { |
|
187
|
3 |
|
$pad_len /= 2; |
|
188
|
3 |
|
$pad_amount_left = floor($pad_len); |
|
189
|
3 |
|
$pad_amount_right = ceil($pad_len); |
|
190
|
3 |
|
$repeat_times_left = ceil($pad_amount_left / $pad_str_len); |
|
191
|
3 |
|
$repeat_times_right = ceil($pad_amount_right / $pad_str_len); |
|
192
|
|
|
|
|
193
|
3 |
|
$prefix = str_repeat($pad_str, $repeat_times_left); |
|
194
|
3 |
|
$padding_left = $this->substr($prefix, 0, $pad_amount_left); |
|
195
|
|
|
|
|
196
|
3 |
|
$suffix = str_repeat($pad_str, $repeat_times_right); |
|
197
|
3 |
|
$padding_right = $this->substr($suffix, 0, $pad_amount_right); |
|
198
|
|
|
|
|
199
|
3 |
|
return $padding_left . $input . $padding_right; |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
// STR_PAD_RIGHT |
|
203
|
9 |
|
$repeat_times = ceil($pad_len / $pad_str_len); |
|
204
|
9 |
|
$input .= str_repeat($pad_str, $repeat_times); |
|
205
|
9 |
|
return $this->substr($input, 0, $pad_length); |
|
206
|
|
|
} |
|
207
|
|
|
} |
|
208
|
|
|
|