1
|
|
|
<?php |
2
|
|
|
namespace Xls; |
3
|
|
|
|
4
|
|
|
class StringUtils |
5
|
|
|
{ |
6
|
|
|
const STRING_REGEXP_FRACTION = '(-?)(\d+)\s+(\d+\/\d+)'; |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* Get whether mbstring extension is available |
10
|
|
|
* |
11
|
|
|
* @return boolean |
12
|
|
|
*/ |
13
|
|
|
public static function isMbstringEnabled() |
14
|
|
|
{ |
15
|
|
|
return function_exists('mb_convert_encoding'); |
16
|
|
|
} |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Get whether iconv extension is available |
20
|
|
|
* |
21
|
|
|
* @return boolean |
22
|
|
|
*/ |
23
|
|
|
public static function isIconvEnabled() |
24
|
|
|
{ |
25
|
|
|
return function_exists('iconv'); |
26
|
|
|
} |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* @return bool |
30
|
|
|
*/ |
31
|
|
|
public static function isMbstringOrIconvEnabled() |
32
|
|
|
{ |
33
|
|
|
return self::isMbstringEnabled() || self::isIconvEnabled(); |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Converts a UTF-8 string into BIFF8 Unicode string data (8-bit string length) |
38
|
|
|
* Writes the string using uncompressed notation, no rich text, no Asian phonetics |
39
|
|
|
* If mbstring extension is not available, ASCII is assumed, and compressed notation is used |
40
|
|
|
* although this will give wrong results for non-ASCII strings |
41
|
|
|
* see OpenOffice.org's Documentation of the Microsoft Excel File Format, sect. 2.5.3 |
42
|
|
|
* |
43
|
|
|
* @param string $value UTF-8 encoded string |
44
|
|
|
* @return string |
45
|
|
|
*/ |
46
|
|
|
public static function toBiff8UnicodeShort($value) |
47
|
|
|
{ |
48
|
|
|
return self::toBiff8Unicode($value, 8); |
49
|
|
|
} |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* Converts a UTF-8 string into BIFF8 Unicode string data (16-bit string length) |
53
|
|
|
* Writes the string using uncompressed notation, no rich text, no Asian phonetics |
54
|
|
|
* If mbstring extension is not available, ASCII is assumed, and compressed notation is used |
55
|
|
|
* although this will give wrong results for non-ASCII strings |
56
|
|
|
* see OpenOffice.org's Documentation of the Microsoft Excel File Format, sect. 2.5.3 |
57
|
|
|
* |
58
|
|
|
* @param string $value UTF-8 encoded string |
59
|
|
|
* @return string |
60
|
|
|
*/ |
61
|
|
|
public static function toBiff8UnicodeLong($value) |
62
|
|
|
{ |
63
|
|
|
return self::toBiff8Unicode($value, 16); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @param string $value |
68
|
|
|
* @param int $lengthSize 8 or 16 |
69
|
|
|
* |
70
|
|
|
* @return string |
71
|
|
|
*/ |
72
|
|
|
public static function toBiff8Unicode($value, $lengthSize) |
73
|
|
|
{ |
74
|
|
|
$ln = self::countCharacters($value); |
75
|
|
|
$lengthFormat = ($lengthSize == 8) ? 'C' : 'v'; |
76
|
|
|
$data = pack($lengthFormat, $ln); |
77
|
|
|
|
78
|
|
|
$opt = (self::isMbstringOrIconvEnabled()) ? 0x01 : 0x00; |
79
|
|
|
$data .= pack('C', $opt); |
80
|
|
|
|
81
|
|
|
$data .= self::toUtf16Le($value); |
82
|
|
|
|
83
|
|
|
return $data; |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* @param $value |
88
|
|
|
* |
89
|
|
|
* @return string |
90
|
|
|
*/ |
91
|
|
|
public static function toBiff8UnicodeLongWoLenInfo($value) |
92
|
|
|
{ |
93
|
|
|
return substr(self::toBiff8UnicodeLong($value), 2); |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* Convert string from one encoding to another. First try mbstring, then iconv, finally strlen |
98
|
|
|
* |
99
|
|
|
* @param string $value |
100
|
|
|
* @param string $from Encoding to convert from, e.g. 'UTF-16LE' |
101
|
|
|
* @param string $to Encoding to convert to, e.g. 'UTF-8' |
102
|
|
|
* @return string |
103
|
|
|
*/ |
104
|
|
|
public static function convertEncoding($value, $from, $to) |
105
|
|
|
{ |
106
|
|
|
if (self::isMbstringEnabled()) { |
107
|
|
|
return mb_convert_encoding($value, $to, $from); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
if (self::isIconvEnabled()) { |
111
|
|
|
return iconv($from, $to, $value); |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
return $value; |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Get character count. First try mbstring, then iconv, finally strlen |
119
|
|
|
* |
120
|
|
|
* @param string $value |
121
|
|
|
* @param string $enc Encoding |
122
|
|
|
* @return int Character count |
123
|
|
|
*/ |
124
|
|
|
public static function countCharacters($value, $enc = 'UTF-8') |
125
|
|
|
{ |
126
|
|
|
if (self::isMbstringEnabled()) { |
127
|
|
|
return mb_strlen($value, $enc); |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
if (self::isIconvEnabled()) { |
131
|
|
|
return iconv_strlen($value, $enc); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
// else strlen |
135
|
|
|
return strlen($value); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* Get a substring of a UTF-8 encoded string. First try mbstring, then iconv, finally strlen |
140
|
|
|
* |
141
|
|
|
* @param string $str UTF-8 encoded string |
142
|
|
|
* @param int $start Start offset |
143
|
|
|
* @param int $length Maximum number of characters in substring |
144
|
|
|
* |
145
|
|
|
* @return string |
146
|
|
|
*/ |
147
|
|
|
public static function substr($str = '', $start = 0, $length = 0) |
148
|
|
|
{ |
149
|
|
|
if (self::isMbstringEnabled()) { |
150
|
|
|
return mb_substr($str, $start, $length, 'UTF-8'); |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
if (self::isIconvEnabled()) { |
154
|
|
|
return iconv_substr($str, $start, $length, 'UTF-8'); |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
// else substr |
158
|
|
|
return substr($str, $start, $length); |
159
|
|
|
} |
160
|
|
|
|
161
|
|
|
/** |
162
|
|
|
* @param $value |
163
|
|
|
* |
164
|
|
|
* @return string |
165
|
|
|
*/ |
166
|
|
|
public static function toUtf16Le($value) |
167
|
|
|
{ |
168
|
|
|
return self::convertEncoding($value, 'UTF-8', 'UTF-16LE'); |
169
|
|
|
} |
170
|
|
|
|
171
|
|
|
/** |
172
|
|
|
* @param $str |
173
|
|
|
* |
174
|
|
|
* @return string |
175
|
|
|
*/ |
176
|
|
|
public static function toNullTerminatedWchar($str) |
177
|
|
|
{ |
178
|
|
|
$str = join("\0", preg_split("''", $str, -1, PREG_SPLIT_NO_EMPTY)); |
179
|
|
|
$str = $str . "\0\0\0"; |
180
|
|
|
|
181
|
|
|
return $str; |
182
|
|
|
} |
183
|
|
|
} |
184
|
|
|
|