1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* This file is part of PHPOffice Common |
4
|
|
|
* |
5
|
|
|
* PHPOffice Common is free software distributed under the terms of the GNU Lesser |
6
|
|
|
* General Public License version 3 as published by the Free Software Foundation. |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please read the LICENSE |
9
|
|
|
* file that was distributed with this source code. For the full list of |
10
|
|
|
* contributors, visit https://github.com/PHPOffice/Common/contributors. |
11
|
|
|
* |
12
|
|
|
* @link https://github.com/PHPOffice/Common |
13
|
|
|
* @copyright 2009-2016 PHPOffice Common contributors |
14
|
|
|
* @license http://www.gnu.org/licenses/lgpl.txt LGPL version 3 |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
namespace PhpOffice\Common; |
18
|
|
|
|
19
|
|
|
/** |
20
|
|
|
* Text |
21
|
|
|
*/ |
22
|
|
|
class Text |
23
|
|
|
{ |
24
|
|
|
/** |
25
|
|
|
* Control characters array |
26
|
|
|
* |
27
|
|
|
* @var string[] |
28
|
|
|
*/ |
29
|
|
|
private static $controlCharacters = array(); |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Build control characters array |
33
|
|
|
*/ |
34
|
1 |
|
private static function buildControlCharacters() |
35
|
|
|
{ |
36
|
1 |
|
for ($i = 0; $i <= 19; ++$i) { |
37
|
1 |
|
if ($i != 9 && $i != 10 && $i != 13) { |
38
|
1 |
|
$find = '_x' . sprintf('%04s', strtoupper(dechex($i))) . '_'; |
39
|
1 |
|
$replace = chr($i); |
40
|
1 |
|
self::$controlCharacters[$find] = $replace; |
41
|
1 |
|
} |
42
|
1 |
|
} |
43
|
1 |
|
} |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* Convert from PHP control character to OpenXML escaped control character |
47
|
|
|
* |
48
|
|
|
* Excel 2007 team: |
49
|
|
|
* ---------------- |
50
|
|
|
* That's correct, control characters are stored directly in the shared-strings table. |
51
|
|
|
* We do encode characters that cannot be represented in XML using the following escape sequence: |
52
|
|
|
* _xHHHH_ where H represents a hexadecimal character in the character's value... |
53
|
|
|
* So you could end up with something like _x0008_ in a string (either in a cell value (<v>) |
54
|
|
|
* element or in the shared string <t> element. |
55
|
|
|
* |
56
|
|
|
* @param string $value Value to escape |
57
|
|
|
* @return string |
58
|
|
|
*/ |
59
|
1 |
|
public static function controlCharacterPHP2OOXML($value = '') |
60
|
|
|
{ |
61
|
1 |
|
if (empty(self::$controlCharacters)) { |
62
|
1 |
|
self::buildControlCharacters(); |
63
|
1 |
|
} |
64
|
|
|
|
65
|
1 |
|
return str_replace(array_values(self::$controlCharacters), array_keys(self::$controlCharacters), $value); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* Return a number formatted for being integrated in xml files |
70
|
|
|
* @param float $number |
71
|
|
|
* @param integer $decimals |
72
|
|
|
* @return string |
73
|
|
|
*/ |
74
|
1 |
|
public static function numberFormat($number, $decimals) |
75
|
|
|
{ |
76
|
1 |
|
return number_format($number, $decimals, '.', ''); |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
/** |
80
|
|
|
* @param int $dec |
81
|
|
|
* @link http://stackoverflow.com/a/7153133/2235790 |
82
|
|
|
* @author velcrow |
83
|
|
|
* @return string |
84
|
|
|
*/ |
85
|
1 |
|
public static function chr($dec) |
86
|
|
|
{ |
87
|
1 |
|
if ($dec<=0x7F) { |
88
|
1 |
|
return chr($dec); |
89
|
|
|
} |
90
|
1 |
|
if ($dec<=0x7FF) { |
91
|
1 |
|
return chr(($dec>>6)+192).chr(($dec&63)+128); |
92
|
|
|
} |
93
|
1 |
|
if ($dec<=0xFFFF) { |
94
|
1 |
|
return chr(($dec>>12)+224).chr((($dec>>6)&63)+128).chr(($dec&63)+128); |
95
|
|
|
} |
96
|
1 |
|
if ($dec<=0x1FFFFF) { |
97
|
1 |
|
return chr(($dec>>18)+240).chr((($dec>>12)&63)+128).chr((($dec>>6)&63)+128).chr(($dec&63)+128); |
98
|
|
|
} |
99
|
1 |
|
return ''; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* Convert from OpenXML escaped control character to PHP control character |
104
|
|
|
* |
105
|
|
|
* @param string $value Value to unescape |
106
|
|
|
* @return string |
107
|
|
|
*/ |
108
|
1 |
|
public static function controlCharacterOOXML2PHP($value = '') |
109
|
|
|
{ |
110
|
1 |
|
if (empty(self::$controlCharacters)) { |
111
|
|
|
self::buildControlCharacters(); |
112
|
|
|
} |
113
|
|
|
|
114
|
1 |
|
return str_replace(array_keys(self::$controlCharacters), array_values(self::$controlCharacters), $value); |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Check if a string contains UTF-8 data |
119
|
|
|
* |
120
|
|
|
* @param string $value |
121
|
|
|
* @return boolean |
122
|
|
|
*/ |
123
|
1 |
|
public static function isUTF8($value = '') |
124
|
|
|
{ |
125
|
1 |
|
return is_string($value) && ($value === '' || preg_match('/^./su', $value) == 1); |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
/** |
129
|
|
|
* Return UTF8 encoded value |
130
|
|
|
* |
131
|
|
|
* @param string $value |
132
|
|
|
* @return string |
133
|
|
|
*/ |
134
|
|
|
public static function toUTF8($value = '') |
135
|
|
|
{ |
136
|
|
|
if (!is_null($value) && !self::isUTF8($value)) { |
137
|
|
|
$value = utf8_encode($value); |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
return $value; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* Returns unicode from UTF8 text |
145
|
|
|
* |
146
|
|
|
* The function is splitted to reduce cyclomatic complexity |
147
|
|
|
* |
148
|
|
|
* @param string $text UTF8 text |
149
|
|
|
* @return string Unicode text |
150
|
|
|
* @since 0.11.0 |
151
|
|
|
*/ |
152
|
1 |
|
public static function toUnicode($text) |
153
|
|
|
{ |
154
|
1 |
|
return self::unicodeToEntities(self::utf8ToUnicode($text)); |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* Returns unicode array from UTF8 text |
159
|
|
|
* |
160
|
|
|
* @param string $text UTF8 text |
161
|
|
|
* @return array |
162
|
|
|
* @since 0.11.0 |
163
|
|
|
* @link http://www.randomchaos.com/documents/?source=php_and_unicode |
164
|
|
|
*/ |
165
|
1 |
|
public static function utf8ToUnicode($text) |
166
|
|
|
{ |
167
|
1 |
|
$unicode = array(); |
168
|
1 |
|
$values = array(); |
169
|
1 |
|
$lookingFor = 1; |
170
|
|
|
|
171
|
|
|
// Gets unicode for each character |
172
|
1 |
|
for ($i = 0; $i < strlen($text); $i++) { |
173
|
1 |
|
$thisValue = ord($text[$i]); |
174
|
1 |
|
if ($thisValue < 128) { |
175
|
1 |
|
$unicode[] = $thisValue; |
176
|
1 |
|
} else { |
177
|
1 |
|
if (count($values) == 0) { |
178
|
1 |
|
$lookingFor = $thisValue < 224 ? 2 : 3; |
179
|
1 |
|
} |
180
|
1 |
|
$values[] = $thisValue; |
181
|
1 |
|
if (count($values) == $lookingFor) { |
182
|
1 |
|
if ($lookingFor == 3) { |
183
|
1 |
|
$number = (($values[0] % 16) * 4096) + (($values[1] % 64) * 64) + ($values[2] % 64); |
184
|
1 |
|
} else { |
185
|
1 |
|
$number = (($values[0] % 32) * 64) + ($values[1] % 64); |
186
|
|
|
} |
187
|
1 |
|
$unicode[] = $number; |
188
|
1 |
|
$values = array(); |
189
|
1 |
|
$lookingFor = 1; |
190
|
1 |
|
} |
191
|
|
|
} |
192
|
1 |
|
} |
193
|
|
|
|
194
|
1 |
|
return $unicode; |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* Returns entites from unicode array |
199
|
|
|
* |
200
|
|
|
* @param array $unicode |
201
|
|
|
* @return string |
202
|
|
|
* @since 0.11.0 |
203
|
|
|
* @link http://www.randomchaos.com/documents/?source=php_and_unicode |
204
|
|
|
*/ |
205
|
1 |
|
private static function unicodeToEntities($unicode) |
206
|
|
|
{ |
207
|
1 |
|
$entities = ''; |
208
|
|
|
|
209
|
1 |
|
foreach ($unicode as $value) { |
210
|
1 |
|
if ($value != 65279) { |
211
|
1 |
|
$entities .= $value > 127 ? '\uc0{\u' . $value . '}' : chr($value); |
212
|
1 |
|
} |
213
|
1 |
|
} |
214
|
|
|
|
215
|
1 |
|
return $entities; |
216
|
|
|
} |
217
|
|
|
|
218
|
|
|
/** |
219
|
|
|
* Return name without underscore for < 0.10.0 variable name compatibility |
220
|
|
|
* |
221
|
|
|
* @param string $value |
222
|
|
|
* @return string |
223
|
|
|
*/ |
224
|
1 |
|
public static function removeUnderscorePrefix($value) |
225
|
|
|
{ |
226
|
1 |
|
if (!is_null($value)) { |
227
|
1 |
|
if (substr($value, 0, 1) == '_') { |
228
|
1 |
|
$value = substr($value, 1); |
229
|
1 |
|
} |
230
|
1 |
|
} |
231
|
|
|
|
232
|
1 |
|
return $value; |
233
|
|
|
} |
234
|
|
|
} |
235
|
|
|
|