PHPOffice /
PhpSpreadsheet
| Total Complexity | 70 |
| Total Lines | 716 |
| Duplicated Lines | 0 % |
| Coverage | 93.4% |
| Changes | 0 | ||
Complex classes like StringHelper often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use StringHelper, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 7 | class StringHelper |
||
| 8 | { |
||
| 9 | /** Constants */ |
||
| 10 | /** Regular Expressions */ |
||
| 11 | // Fraction |
||
| 12 | const STRING_REGEXP_FRACTION = '(-?)(\d+)\s+(\d+\/\d+)'; |
||
| 13 | |||
| 14 | /** |
||
| 15 | * Control characters array. |
||
| 16 | * |
||
| 17 | * @var string[] |
||
| 18 | */ |
||
| 19 | private static $controlCharacters = []; |
||
| 20 | |||
| 21 | /** |
||
| 22 | * SYLK Characters array. |
||
| 23 | * |
||
| 24 | * @var array |
||
| 25 | */ |
||
| 26 | private static $SYLKCharacters = []; |
||
| 27 | |||
| 28 | /** |
||
| 29 | * Decimal separator. |
||
| 30 | * |
||
| 31 | * @var string |
||
| 32 | */ |
||
| 33 | private static $decimalSeparator; |
||
| 34 | |||
| 35 | /** |
||
| 36 | * Thousands separator. |
||
| 37 | * |
||
| 38 | * @var string |
||
| 39 | */ |
||
| 40 | private static $thousandsSeparator; |
||
| 41 | |||
| 42 | /** |
||
| 43 | * Currency code. |
||
| 44 | * |
||
| 45 | * @var string |
||
| 46 | */ |
||
| 47 | private static $currencyCode; |
||
| 48 | |||
| 49 | /** |
||
| 50 | * Is iconv extension avalable? |
||
| 51 | * |
||
| 52 | * @var bool |
||
| 53 | */ |
||
| 54 | private static $isIconvEnabled; |
||
| 55 | |||
| 56 | /** |
||
| 57 | * iconv options. |
||
| 58 | * |
||
| 59 | * @var string |
||
| 60 | */ |
||
| 61 | private static $iconvOptions = '//IGNORE//TRANSLIT'; |
||
| 62 | |||
| 63 | /** |
||
| 64 | * Build control characters array. |
||
| 65 | */ |
||
| 66 | 56 | private static function buildControlCharacters() |
|
| 67 | { |
||
| 68 | 56 | for ($i = 0; $i <= 31; ++$i) { |
|
| 69 | 56 | if ($i != 9 && $i != 10 && $i != 13) { |
|
| 70 | 56 | $find = '_x' . sprintf('%04s', strtoupper(dechex($i))) . '_'; |
|
| 71 | 56 | $replace = chr($i); |
|
| 72 | 56 | self::$controlCharacters[$find] = $replace; |
|
| 73 | } |
||
| 74 | } |
||
| 75 | 56 | } |
|
| 76 | |||
| 77 | /** |
||
| 78 | * Build SYLK characters array. |
||
| 79 | */ |
||
| 80 | 56 | private static function buildSYLKCharacters() |
|
| 81 | { |
||
| 82 | self::$SYLKCharacters = [ |
||
| 83 | 56 | "\x1B 0" => chr(0), |
|
| 84 | 56 | "\x1B 1" => chr(1), |
|
| 85 | 56 | "\x1B 2" => chr(2), |
|
| 86 | 56 | "\x1B 3" => chr(3), |
|
| 87 | 56 | "\x1B 4" => chr(4), |
|
| 88 | 56 | "\x1B 5" => chr(5), |
|
| 89 | 56 | "\x1B 6" => chr(6), |
|
| 90 | 56 | "\x1B 7" => chr(7), |
|
| 91 | 56 | "\x1B 8" => chr(8), |
|
| 92 | 56 | "\x1B 9" => chr(9), |
|
| 93 | 56 | "\x1B :" => chr(10), |
|
| 94 | 56 | "\x1B ;" => chr(11), |
|
| 95 | 56 | "\x1B <" => chr(12), |
|
| 96 | 56 | "\x1B =" => chr(13), |
|
| 97 | 56 | "\x1B >" => chr(14), |
|
| 98 | 56 | "\x1B ?" => chr(15), |
|
| 99 | 56 | "\x1B!0" => chr(16), |
|
| 100 | 56 | "\x1B!1" => chr(17), |
|
| 101 | 56 | "\x1B!2" => chr(18), |
|
| 102 | 56 | "\x1B!3" => chr(19), |
|
| 103 | 56 | "\x1B!4" => chr(20), |
|
| 104 | 56 | "\x1B!5" => chr(21), |
|
| 105 | 56 | "\x1B!6" => chr(22), |
|
| 106 | 56 | "\x1B!7" => chr(23), |
|
| 107 | 56 | "\x1B!8" => chr(24), |
|
| 108 | 56 | "\x1B!9" => chr(25), |
|
| 109 | 56 | "\x1B!:" => chr(26), |
|
| 110 | 56 | "\x1B!;" => chr(27), |
|
| 111 | 56 | "\x1B!<" => chr(28), |
|
| 112 | 56 | "\x1B!=" => chr(29), |
|
| 113 | 56 | "\x1B!>" => chr(30), |
|
| 114 | 56 | "\x1B!?" => chr(31), |
|
| 115 | 56 | "\x1B'?" => chr(127), |
|
| 116 | 56 | "\x1B(0" => '€', // 128 in CP1252 |
|
| 117 | 56 | "\x1B(2" => '‚', // 130 in CP1252 |
|
| 118 | 56 | "\x1B(3" => 'ƒ', // 131 in CP1252 |
|
| 119 | 56 | "\x1B(4" => '„', // 132 in CP1252 |
|
| 120 | 56 | "\x1B(5" => '…', // 133 in CP1252 |
|
| 121 | 56 | "\x1B(6" => '†', // 134 in CP1252 |
|
| 122 | 56 | "\x1B(7" => '‡', // 135 in CP1252 |
|
| 123 | 56 | "\x1B(8" => 'ˆ', // 136 in CP1252 |
|
| 124 | 56 | "\x1B(9" => '‰', // 137 in CP1252 |
|
| 125 | 56 | "\x1B(:" => 'Š', // 138 in CP1252 |
|
| 126 | 56 | "\x1B(;" => '‹', // 139 in CP1252 |
|
| 127 | 56 | "\x1BNj" => 'Œ', // 140 in CP1252 |
|
| 128 | 56 | "\x1B(>" => 'Ž', // 142 in CP1252 |
|
| 129 | 56 | "\x1B)1" => '‘', // 145 in CP1252 |
|
| 130 | 56 | "\x1B)2" => '’', // 146 in CP1252 |
|
| 131 | 56 | "\x1B)3" => '“', // 147 in CP1252 |
|
| 132 | 56 | "\x1B)4" => '”', // 148 in CP1252 |
|
| 133 | 56 | "\x1B)5" => '•', // 149 in CP1252 |
|
| 134 | 56 | "\x1B)6" => '–', // 150 in CP1252 |
|
| 135 | 56 | "\x1B)7" => '—', // 151 in CP1252 |
|
| 136 | 56 | "\x1B)8" => '˜', // 152 in CP1252 |
|
| 137 | 56 | "\x1B)9" => '™', // 153 in CP1252 |
|
| 138 | 56 | "\x1B):" => 'š', // 154 in CP1252 |
|
| 139 | 56 | "\x1B);" => '›', // 155 in CP1252 |
|
| 140 | 56 | "\x1BNz" => 'œ', // 156 in CP1252 |
|
| 141 | 56 | "\x1B)>" => 'ž', // 158 in CP1252 |
|
| 142 | 56 | "\x1B)?" => 'Ÿ', // 159 in CP1252 |
|
| 143 | 56 | "\x1B*0" => ' ', // 160 in CP1252 |
|
| 144 | 56 | "\x1BN!" => '¡', // 161 in CP1252 |
|
| 145 | 56 | "\x1BN\"" => '¢', // 162 in CP1252 |
|
| 146 | 56 | "\x1BN#" => '£', // 163 in CP1252 |
|
| 147 | 56 | "\x1BN(" => '¤', // 164 in CP1252 |
|
| 148 | 56 | "\x1BN%" => '¥', // 165 in CP1252 |
|
| 149 | 56 | "\x1B*6" => '¦', // 166 in CP1252 |
|
| 150 | 56 | "\x1BN'" => '§', // 167 in CP1252 |
|
| 151 | 56 | "\x1BNH " => '¨', // 168 in CP1252 |
|
| 152 | 56 | "\x1BNS" => '©', // 169 in CP1252 |
|
| 153 | 56 | "\x1BNc" => 'ª', // 170 in CP1252 |
|
| 154 | 56 | "\x1BN+" => '«', // 171 in CP1252 |
|
| 155 | 56 | "\x1B*<" => '¬', // 172 in CP1252 |
|
| 156 | 56 | "\x1B*=" => '', // 173 in CP1252 |
|
| 157 | 56 | "\x1BNR" => '®', // 174 in CP1252 |
|
| 158 | 56 | "\x1B*?" => '¯', // 175 in CP1252 |
|
| 159 | 56 | "\x1BN0" => '°', // 176 in CP1252 |
|
| 160 | 56 | "\x1BN1" => '±', // 177 in CP1252 |
|
| 161 | 56 | "\x1BN2" => '²', // 178 in CP1252 |
|
| 162 | 56 | "\x1BN3" => '³', // 179 in CP1252 |
|
| 163 | 56 | "\x1BNB " => '´', // 180 in CP1252 |
|
| 164 | 56 | "\x1BN5" => 'µ', // 181 in CP1252 |
|
| 165 | 56 | "\x1BN6" => '¶', // 182 in CP1252 |
|
| 166 | 56 | "\x1BN7" => '·', // 183 in CP1252 |
|
| 167 | 56 | "\x1B+8" => '¸', // 184 in CP1252 |
|
| 168 | 56 | "\x1BNQ" => '¹', // 185 in CP1252 |
|
| 169 | 56 | "\x1BNk" => 'º', // 186 in CP1252 |
|
| 170 | 56 | "\x1BN;" => '»', // 187 in CP1252 |
|
| 171 | 56 | "\x1BN<" => '¼', // 188 in CP1252 |
|
| 172 | 56 | "\x1BN=" => '½', // 189 in CP1252 |
|
| 173 | 56 | "\x1BN>" => '¾', // 190 in CP1252 |
|
| 174 | 56 | "\x1BN?" => '¿', // 191 in CP1252 |
|
| 175 | 56 | "\x1BNAA" => 'À', // 192 in CP1252 |
|
| 176 | 56 | "\x1BNBA" => 'Á', // 193 in CP1252 |
|
| 177 | 56 | "\x1BNCA" => 'Â', // 194 in CP1252 |
|
| 178 | 56 | "\x1BNDA" => 'Ã', // 195 in CP1252 |
|
| 179 | 56 | "\x1BNHA" => 'Ä', // 196 in CP1252 |
|
| 180 | 56 | "\x1BNJA" => 'Å', // 197 in CP1252 |
|
| 181 | 56 | "\x1BNa" => 'Æ', // 198 in CP1252 |
|
| 182 | 56 | "\x1BNKC" => 'Ç', // 199 in CP1252 |
|
| 183 | 56 | "\x1BNAE" => 'È', // 200 in CP1252 |
|
| 184 | 56 | "\x1BNBE" => 'É', // 201 in CP1252 |
|
| 185 | 56 | "\x1BNCE" => 'Ê', // 202 in CP1252 |
|
| 186 | 56 | "\x1BNHE" => 'Ë', // 203 in CP1252 |
|
| 187 | 56 | "\x1BNAI" => 'Ì', // 204 in CP1252 |
|
| 188 | 56 | "\x1BNBI" => 'Í', // 205 in CP1252 |
|
| 189 | 56 | "\x1BNCI" => 'Î', // 206 in CP1252 |
|
| 190 | 56 | "\x1BNHI" => 'Ï', // 207 in CP1252 |
|
| 191 | 56 | "\x1BNb" => 'Ð', // 208 in CP1252 |
|
| 192 | 56 | "\x1BNDN" => 'Ñ', // 209 in CP1252 |
|
| 193 | 56 | "\x1BNAO" => 'Ò', // 210 in CP1252 |
|
| 194 | 56 | "\x1BNBO" => 'Ó', // 211 in CP1252 |
|
| 195 | 56 | "\x1BNCO" => 'Ô', // 212 in CP1252 |
|
| 196 | 56 | "\x1BNDO" => 'Õ', // 213 in CP1252 |
|
| 197 | 56 | "\x1BNHO" => 'Ö', // 214 in CP1252 |
|
| 198 | 56 | "\x1B-7" => '×', // 215 in CP1252 |
|
| 199 | 56 | "\x1BNi" => 'Ø', // 216 in CP1252 |
|
| 200 | 56 | "\x1BNAU" => 'Ù', // 217 in CP1252 |
|
| 201 | 56 | "\x1BNBU" => 'Ú', // 218 in CP1252 |
|
| 202 | 56 | "\x1BNCU" => 'Û', // 219 in CP1252 |
|
| 203 | 56 | "\x1BNHU" => 'Ü', // 220 in CP1252 |
|
| 204 | 56 | "\x1B-=" => 'Ý', // 221 in CP1252 |
|
| 205 | 56 | "\x1BNl" => 'Þ', // 222 in CP1252 |
|
| 206 | 56 | "\x1BN{" => 'ß', // 223 in CP1252 |
|
| 207 | 56 | "\x1BNAa" => 'à', // 224 in CP1252 |
|
| 208 | 56 | "\x1BNBa" => 'á', // 225 in CP1252 |
|
| 209 | 56 | "\x1BNCa" => 'â', // 226 in CP1252 |
|
| 210 | 56 | "\x1BNDa" => 'ã', // 227 in CP1252 |
|
| 211 | 56 | "\x1BNHa" => 'ä', // 228 in CP1252 |
|
| 212 | 56 | "\x1BNJa" => 'å', // 229 in CP1252 |
|
| 213 | 56 | "\x1BNq" => 'æ', // 230 in CP1252 |
|
| 214 | 56 | "\x1BNKc" => 'ç', // 231 in CP1252 |
|
| 215 | 56 | "\x1BNAe" => 'è', // 232 in CP1252 |
|
| 216 | 56 | "\x1BNBe" => 'é', // 233 in CP1252 |
|
| 217 | 56 | "\x1BNCe" => 'ê', // 234 in CP1252 |
|
| 218 | 56 | "\x1BNHe" => 'ë', // 235 in CP1252 |
|
| 219 | 56 | "\x1BNAi" => 'ì', // 236 in CP1252 |
|
| 220 | 56 | "\x1BNBi" => 'í', // 237 in CP1252 |
|
| 221 | 56 | "\x1BNCi" => 'î', // 238 in CP1252 |
|
| 222 | 56 | "\x1BNHi" => 'ï', // 239 in CP1252 |
|
| 223 | 56 | "\x1BNs" => 'ð', // 240 in CP1252 |
|
| 224 | 56 | "\x1BNDn" => 'ñ', // 241 in CP1252 |
|
| 225 | 56 | "\x1BNAo" => 'ò', // 242 in CP1252 |
|
| 226 | 56 | "\x1BNBo" => 'ó', // 243 in CP1252 |
|
| 227 | 56 | "\x1BNCo" => 'ô', // 244 in CP1252 |
|
| 228 | 56 | "\x1BNDo" => 'õ', // 245 in CP1252 |
|
| 229 | 56 | "\x1BNHo" => 'ö', // 246 in CP1252 |
|
| 230 | 56 | "\x1B/7" => '÷', // 247 in CP1252 |
|
| 231 | 56 | "\x1BNy" => 'ø', // 248 in CP1252 |
|
| 232 | 56 | "\x1BNAu" => 'ù', // 249 in CP1252 |
|
| 233 | 56 | "\x1BNBu" => 'ú', // 250 in CP1252 |
|
| 234 | 56 | "\x1BNCu" => 'û', // 251 in CP1252 |
|
| 235 | 56 | "\x1BNHu" => 'ü', // 252 in CP1252 |
|
| 236 | 56 | "\x1B/=" => 'ý', // 253 in CP1252 |
|
| 237 | 56 | "\x1BN|" => 'þ', // 254 in CP1252 |
|
| 238 | 56 | "\x1BNHy" => 'ÿ', // 255 in CP1252 |
|
| 239 | ]; |
||
| 240 | 56 | } |
|
| 241 | |||
| 242 | /** |
||
| 243 | * Get whether iconv extension is available. |
||
| 244 | * |
||
| 245 | * @return bool |
||
| 246 | */ |
||
| 247 | 156 | public static function getIsIconvEnabled() |
|
| 248 | { |
||
| 249 | 156 | if (isset(self::$isIconvEnabled)) { |
|
| 250 | 156 | return self::$isIconvEnabled; |
|
| 251 | } |
||
| 252 | |||
| 253 | // Assume no problems with iconv |
||
| 254 | 96 | self::$isIconvEnabled = true; |
|
| 255 | |||
| 256 | // Fail if iconv doesn't exist |
||
| 257 | 96 | if (!function_exists('iconv')) { |
|
| 258 | self::$isIconvEnabled = false; |
||
| 259 | 96 | } elseif (!@iconv('UTF-8', 'UTF-16LE', 'x')) { |
|
| 260 | // Sometimes iconv is not working, and e.g. iconv('UTF-8', 'UTF-16LE', 'x') just returns false, |
||
| 261 | self::$isIconvEnabled = false; |
||
| 262 | 96 | } elseif (defined('PHP_OS') && @stristr(PHP_OS, 'AIX') && defined('ICONV_IMPL') && (@strcasecmp(ICONV_IMPL, 'unknown') == 0) && defined('ICONV_VERSION') && (@strcasecmp(ICONV_VERSION, 'unknown') == 0)) { |
|
| 263 | // CUSTOM: IBM AIX iconv() does not work |
||
| 264 | self::$isIconvEnabled = false; |
||
| 265 | } |
||
| 266 | |||
| 267 | // Deactivate iconv default options if they fail (as seen on IMB i) |
||
| 268 | 96 | if (self::$isIconvEnabled && !@iconv('UTF-8', 'UTF-16LE' . self::$iconvOptions, 'x')) { |
|
| 269 | self::$iconvOptions = ''; |
||
| 270 | } |
||
| 271 | |||
| 272 | 96 | return self::$isIconvEnabled; |
|
| 273 | } |
||
| 274 | |||
| 275 | 66 | private static function buildCharacterSets() |
|
| 276 | { |
||
| 277 | 66 | if (empty(self::$controlCharacters)) { |
|
| 278 | 56 | self::buildControlCharacters(); |
|
| 279 | } |
||
| 280 | |||
| 281 | 66 | if (empty(self::$SYLKCharacters)) { |
|
| 282 | 56 | self::buildSYLKCharacters(); |
|
| 283 | } |
||
| 284 | 66 | } |
|
| 285 | |||
| 286 | /** |
||
| 287 | * Convert from OpenXML escaped control character to PHP control character. |
||
| 288 | * |
||
| 289 | * Excel 2007 team: |
||
| 290 | * ---------------- |
||
| 291 | * That's correct, control characters are stored directly in the shared-strings table. |
||
| 292 | * We do encode characters that cannot be represented in XML using the following escape sequence: |
||
| 293 | * _xHHHH_ where H represents a hexadecimal character in the character's value... |
||
| 294 | * So you could end up with something like _x0008_ in a string (either in a cell value (<v>) |
||
| 295 | * element or in the shared string <t> element. |
||
| 296 | * |
||
| 297 | * @param string $value Value to unescape |
||
| 298 | * |
||
| 299 | * @return string |
||
| 300 | */ |
||
| 301 | 20 | public static function controlCharacterOOXML2PHP($value) |
|
| 302 | { |
||
| 303 | 20 | self::buildCharacterSets(); |
|
| 304 | |||
| 305 | 20 | return str_replace(array_keys(self::$controlCharacters), array_values(self::$controlCharacters), $value); |
|
| 306 | } |
||
| 307 | |||
| 308 | /** |
||
| 309 | * Convert from PHP control character to OpenXML escaped control character. |
||
| 310 | * |
||
| 311 | * Excel 2007 team: |
||
| 312 | * ---------------- |
||
| 313 | * That's correct, control characters are stored directly in the shared-strings table. |
||
| 314 | * We do encode characters that cannot be represented in XML using the following escape sequence: |
||
| 315 | * _xHHHH_ where H represents a hexadecimal character in the character's value... |
||
| 316 | * So you could end up with something like _x0008_ in a string (either in a cell value (<v>) |
||
| 317 | * element or in the shared string <t> element. |
||
| 318 | * |
||
| 319 | * @param string $value Value to escape |
||
| 320 | * |
||
| 321 | * @return string |
||
| 322 | */ |
||
| 323 | 61 | public static function controlCharacterPHP2OOXML($value) |
|
| 324 | { |
||
| 325 | 61 | self::buildCharacterSets(); |
|
| 326 | |||
| 327 | 61 | return str_replace(array_values(self::$controlCharacters), array_keys(self::$controlCharacters), $value); |
|
| 328 | } |
||
| 329 | |||
| 330 | /** |
||
| 331 | * Try to sanitize UTF8, stripping invalid byte sequences. Not perfect. Does not surrogate characters. |
||
| 332 | * |
||
| 333 | * @param string $value |
||
| 334 | * |
||
| 335 | * @return string |
||
| 336 | */ |
||
| 337 | 125 | public static function sanitizeUTF8($value) |
|
| 338 | { |
||
| 339 | 125 | if (self::getIsIconvEnabled()) { |
|
| 340 | 125 | $value = @iconv('UTF-8', 'UTF-8', $value); |
|
| 341 | |||
| 342 | 125 | return $value; |
|
| 343 | } |
||
| 344 | |||
| 345 | $value = mb_convert_encoding($value, 'UTF-8', 'UTF-8'); |
||
| 346 | |||
| 347 | return $value; |
||
| 348 | } |
||
| 349 | |||
| 350 | /** |
||
| 351 | * Check if a string contains UTF8 data. |
||
| 352 | * |
||
| 353 | * @param string $value |
||
| 354 | * |
||
| 355 | * @return bool |
||
| 356 | */ |
||
| 357 | public static function isUTF8($value) |
||
| 358 | { |
||
| 359 | return $value === '' || preg_match('/^./su', $value) === 1; |
||
| 360 | } |
||
| 361 | |||
| 362 | /** |
||
| 363 | * Formats a numeric value as a string for output in various output writers forcing |
||
| 364 | * point as decimal separator in case locale is other than English. |
||
| 365 | * |
||
| 366 | * @param mixed $value |
||
| 367 | * |
||
| 368 | * @return string |
||
| 369 | */ |
||
| 370 | 94 | public static function formatNumber($value) |
|
| 371 | { |
||
| 372 | 94 | if (is_float($value)) { |
|
| 373 | 94 | return str_replace(',', '.', $value); |
|
| 374 | } |
||
| 375 | |||
| 376 | 83 | return (string) $value; |
|
| 377 | } |
||
| 378 | |||
| 379 | /** |
||
| 380 | * Converts a UTF-8 string into BIFF8 Unicode string data (8-bit string length) |
||
| 381 | * Writes the string using uncompressed notation, no rich text, no Asian phonetics |
||
| 382 | * If mbstring extension is not available, ASCII is assumed, and compressed notation is used |
||
| 383 | * although this will give wrong results for non-ASCII strings |
||
| 384 | * see OpenOffice.org's Documentation of the Microsoft Excel File Format, sect. 2.5.3. |
||
| 385 | * |
||
| 386 | * @param string $value UTF-8 encoded string |
||
| 387 | * @param mixed[] $arrcRuns Details of rich text runs in $value |
||
| 388 | * |
||
| 389 | * @return string |
||
| 390 | */ |
||
| 391 | 43 | public static function UTF8toBIFF8UnicodeShort($value, $arrcRuns = []) |
|
| 392 | { |
||
| 393 | // character count |
||
| 394 | 43 | $ln = self::countCharacters($value, 'UTF-8'); |
|
| 395 | // option flags |
||
| 396 | 43 | if (empty($arrcRuns)) { |
|
| 397 | 43 | $data = pack('CC', $ln, 0x0001); |
|
| 398 | // characters |
||
| 399 | 43 | $data .= self::convertEncoding($value, 'UTF-16LE', 'UTF-8'); |
|
| 400 | } else { |
||
| 401 | 9 | $data = pack('vC', $ln, 0x09); |
|
| 402 | 9 | $data .= pack('v', count($arrcRuns)); |
|
| 403 | // characters |
||
| 404 | 9 | $data .= self::convertEncoding($value, 'UTF-16LE', 'UTF-8'); |
|
| 405 | 9 | foreach ($arrcRuns as $cRun) { |
|
| 406 | 9 | $data .= pack('v', $cRun['strlen']); |
|
| 407 | 9 | $data .= pack('v', $cRun['fontidx']); |
|
| 408 | } |
||
| 409 | } |
||
| 410 | |||
| 411 | 43 | return $data; |
|
| 412 | } |
||
| 413 | |||
| 414 | /** |
||
| 415 | * Converts a UTF-8 string into BIFF8 Unicode string data (16-bit string length) |
||
| 416 | * Writes the string using uncompressed notation, no rich text, no Asian phonetics |
||
| 417 | * If mbstring extension is not available, ASCII is assumed, and compressed notation is used |
||
| 418 | * although this will give wrong results for non-ASCII strings |
||
| 419 | * see OpenOffice.org's Documentation of the Microsoft Excel File Format, sect. 2.5.3. |
||
| 420 | * |
||
| 421 | * @param string $value UTF-8 encoded string |
||
| 422 | * |
||
| 423 | * @return string |
||
| 424 | */ |
||
| 425 | 43 | public static function UTF8toBIFF8UnicodeLong($value) |
|
| 426 | { |
||
| 427 | // character count |
||
| 428 | 43 | $ln = self::countCharacters($value, 'UTF-8'); |
|
| 429 | |||
| 430 | // characters |
||
| 431 | 43 | $chars = self::convertEncoding($value, 'UTF-16LE', 'UTF-8'); |
|
| 432 | |||
| 433 | 43 | $data = pack('vC', $ln, 0x0001) . $chars; |
|
| 434 | |||
| 435 | 43 | return $data; |
|
| 436 | } |
||
| 437 | |||
| 438 | /** |
||
| 439 | * Convert string from one encoding to another. |
||
| 440 | * |
||
| 441 | * @param string $value |
||
| 442 | * @param string $to Encoding to convert to, e.g. 'UTF-8' |
||
| 443 | * @param string $from Encoding to convert from, e.g. 'UTF-16LE' |
||
| 444 | * |
||
| 445 | * @return string |
||
| 446 | */ |
||
| 447 | 61 | public static function convertEncoding($value, $to, $from) |
|
| 448 | { |
||
| 449 | 61 | if (self::getIsIconvEnabled()) { |
|
| 450 | 61 | $result = iconv($from, $to . self::$iconvOptions, $value); |
|
| 451 | 61 | if (false !== $result) { |
|
| 452 | 61 | return $result; |
|
| 453 | } |
||
| 454 | } |
||
| 455 | |||
| 456 | return mb_convert_encoding($value, $to, $from); |
||
| 457 | } |
||
| 458 | |||
| 459 | /** |
||
| 460 | * Get character count. |
||
| 461 | * |
||
| 462 | * @param string $value |
||
| 463 | * @param string $enc Encoding |
||
| 464 | * |
||
| 465 | * @return int Character count |
||
| 466 | */ |
||
| 467 | 238 | public static function countCharacters($value, $enc = 'UTF-8') |
|
| 468 | { |
||
| 469 | 238 | return mb_strlen($value, $enc); |
|
| 470 | } |
||
| 471 | |||
| 472 | /** |
||
| 473 | * Get a substring of a UTF-8 encoded string. |
||
| 474 | * |
||
| 475 | * @param string $pValue UTF-8 encoded string |
||
| 476 | * @param int $pStart Start offset |
||
| 477 | * @param int $pLength Maximum number of characters in substring |
||
| 478 | * |
||
| 479 | * @return string |
||
| 480 | */ |
||
| 481 | 214 | public static function substring($pValue, $pStart, $pLength = 0) |
|
| 482 | { |
||
| 483 | 214 | return mb_substr($pValue, $pStart, $pLength, 'UTF-8'); |
|
| 484 | } |
||
| 485 | |||
| 486 | /** |
||
| 487 | * Convert a UTF-8 encoded string to upper case. |
||
| 488 | * |
||
| 489 | * @param string $pValue UTF-8 encoded string |
||
| 490 | * |
||
| 491 | * @return string |
||
| 492 | */ |
||
| 493 | 4 | public static function strToUpper($pValue) |
|
| 494 | { |
||
| 495 | 4 | return mb_convert_case($pValue, MB_CASE_UPPER, 'UTF-8'); |
|
| 496 | } |
||
| 497 | |||
| 498 | /** |
||
| 499 | * Convert a UTF-8 encoded string to lower case. |
||
| 500 | * |
||
| 501 | * @param string $pValue UTF-8 encoded string |
||
| 502 | * |
||
| 503 | * @return string |
||
| 504 | */ |
||
| 505 | 4 | public static function strToLower($pValue) |
|
| 506 | { |
||
| 507 | 4 | return mb_convert_case($pValue, MB_CASE_LOWER, 'UTF-8'); |
|
| 508 | } |
||
| 509 | |||
| 510 | /** |
||
| 511 | * Convert a UTF-8 encoded string to title/proper case |
||
| 512 | * (uppercase every first character in each word, lower case all other characters). |
||
| 513 | * |
||
| 514 | * @param string $pValue UTF-8 encoded string |
||
| 515 | * |
||
| 516 | * @return string |
||
| 517 | */ |
||
| 518 | 3 | public static function strToTitle($pValue) |
|
| 519 | { |
||
| 520 | 3 | return mb_convert_case($pValue, MB_CASE_TITLE, 'UTF-8'); |
|
| 521 | } |
||
| 522 | |||
| 523 | 21 | public static function mbIsUpper($char) |
|
| 524 | { |
||
| 525 | 21 | return mb_strtolower($char, 'UTF-8') != $char; |
|
| 526 | } |
||
| 527 | |||
| 528 | 21 | public static function mbStrSplit($string) |
|
| 529 | { |
||
| 530 | // Split at all position not after the start: ^ |
||
| 531 | // and not before the end: $ |
||
| 532 | 21 | return preg_split('/(?<!^)(?!$)/u', $string); |
|
| 533 | } |
||
| 534 | |||
| 535 | /** |
||
| 536 | * Reverse the case of a string, so that all uppercase characters become lowercase |
||
| 537 | * and all lowercase characters become uppercase. |
||
| 538 | * |
||
| 539 | * @param string $pValue UTF-8 encoded string |
||
| 540 | * |
||
| 541 | * @return string |
||
| 542 | */ |
||
| 543 | 21 | public static function strCaseReverse($pValue) |
|
| 544 | { |
||
| 545 | 21 | $characters = self::mbStrSplit($pValue); |
|
| 546 | 21 | foreach ($characters as &$character) { |
|
| 547 | 21 | if (self::mbIsUpper($character)) { |
|
| 548 | 14 | $character = mb_strtolower($character, 'UTF-8'); |
|
| 549 | } else { |
||
| 550 | 21 | $character = mb_strtoupper($character, 'UTF-8'); |
|
| 551 | } |
||
| 552 | } |
||
| 553 | |||
| 554 | 21 | return implode('', $characters); |
|
|
1 ignored issue
–
show
|
|||
| 555 | } |
||
| 556 | |||
| 557 | /** |
||
| 558 | * Identify whether a string contains a fractional numeric value, |
||
| 559 | * and convert it to a numeric if it is. |
||
| 560 | * |
||
| 561 | * @param string &$operand string value to test |
||
| 562 | * |
||
| 563 | * @return bool |
||
| 564 | */ |
||
| 565 | 2 | public static function convertToNumberIfFraction(&$operand) |
|
| 566 | { |
||
| 567 | 2 | if (preg_match('/^' . self::STRING_REGEXP_FRACTION . '$/i', $operand, $match)) { |
|
| 568 | $sign = ($match[1] == '-') ? '-' : '+'; |
||
| 569 | $fractionFormula = '=' . $sign . $match[2] . $sign . $match[3]; |
||
| 570 | $operand = Calculation::getInstance()->_calculateFormulaValue($fractionFormula); |
||
| 571 | |||
| 572 | return true; |
||
| 573 | } |
||
| 574 | |||
| 575 | 2 | return false; |
|
| 576 | } |
||
| 577 | |||
| 578 | // function convertToNumberIfFraction() |
||
| 579 | |||
| 580 | /** |
||
| 581 | * Get the decimal separator. If it has not yet been set explicitly, try to obtain number |
||
| 582 | * formatting information from locale. |
||
| 583 | * |
||
| 584 | * @return string |
||
| 585 | */ |
||
| 586 | 45 | public static function getDecimalSeparator() |
|
| 587 | { |
||
| 588 | 45 | if (!isset(self::$decimalSeparator)) { |
|
| 589 | 10 | $localeconv = localeconv(); |
|
| 590 | 10 | self::$decimalSeparator = ($localeconv['decimal_point'] != '') |
|
| 591 | 10 | ? $localeconv['decimal_point'] : $localeconv['mon_decimal_point']; |
|
| 592 | |||
| 593 | 10 | if (self::$decimalSeparator == '') { |
|
| 594 | // Default to . |
||
| 595 | self::$decimalSeparator = '.'; |
||
| 596 | } |
||
| 597 | } |
||
| 598 | |||
| 599 | 45 | return self::$decimalSeparator; |
|
| 600 | } |
||
| 601 | |||
| 602 | /** |
||
| 603 | * Set the decimal separator. Only used by NumberFormat::toFormattedString() |
||
| 604 | * to format output by \PhpOffice\PhpSpreadsheet\Writer\Html and \PhpOffice\PhpSpreadsheet\Writer\Pdf. |
||
| 605 | * |
||
| 606 | * @param string $pValue Character for decimal separator |
||
| 607 | */ |
||
| 608 | 245 | public static function setDecimalSeparator($pValue) |
|
| 611 | 245 | } |
|
| 612 | |||
| 613 | /** |
||
| 614 | * Get the thousands separator. If it has not yet been set explicitly, try to obtain number |
||
| 615 | * formatting information from locale. |
||
| 616 | * |
||
| 617 | * @return string |
||
| 618 | */ |
||
| 619 | 53 | public static function getThousandsSeparator() |
|
| 620 | { |
||
| 621 | 53 | if (!isset(self::$thousandsSeparator)) { |
|
| 622 | 10 | $localeconv = localeconv(); |
|
| 623 | 10 | self::$thousandsSeparator = ($localeconv['thousands_sep'] != '') |
|
| 624 | 10 | ? $localeconv['thousands_sep'] : $localeconv['mon_thousands_sep']; |
|
| 625 | |||
| 626 | 10 | if (self::$thousandsSeparator == '') { |
|
| 627 | // Default to . |
||
| 628 | self::$thousandsSeparator = ','; |
||
| 629 | } |
||
| 630 | } |
||
| 631 | |||
| 632 | 53 | return self::$thousandsSeparator; |
|
| 633 | } |
||
| 634 | |||
| 635 | /** |
||
| 636 | * Set the thousands separator. Only used by NumberFormat::toFormattedString() |
||
| 637 | * to format output by \PhpOffice\PhpSpreadsheet\Writer\Html and \PhpOffice\PhpSpreadsheet\Writer\Pdf. |
||
| 638 | * |
||
| 639 | * @param string $pValue Character for thousands separator |
||
| 640 | */ |
||
| 641 | 245 | public static function setThousandsSeparator($pValue) |
|
| 642 | { |
||
| 643 | 245 | self::$thousandsSeparator = $pValue; |
|
| 644 | 245 | } |
|
| 645 | |||
| 646 | /** |
||
| 647 | * Get the currency code. If it has not yet been set explicitly, try to obtain the |
||
| 648 | * symbol information from locale. |
||
| 649 | * |
||
| 650 | * @return string |
||
| 651 | */ |
||
| 652 | 19 | public static function getCurrencyCode() |
|
| 653 | { |
||
| 654 | 19 | if (!empty(self::$currencyCode)) { |
|
| 655 | 18 | return self::$currencyCode; |
|
| 656 | } |
||
| 657 | 3 | self::$currencyCode = '$'; |
|
| 658 | 3 | $localeconv = localeconv(); |
|
| 659 | 3 | if (!empty($localeconv['currency_symbol'])) { |
|
| 660 | 3 | self::$currencyCode = $localeconv['currency_symbol']; |
|
| 661 | |||
| 662 | 3 | return self::$currencyCode; |
|
| 663 | } |
||
| 664 | if (!empty($localeconv['int_curr_symbol'])) { |
||
| 665 | self::$currencyCode = $localeconv['int_curr_symbol']; |
||
| 666 | |||
| 667 | return self::$currencyCode; |
||
| 668 | } |
||
| 669 | |||
| 670 | return self::$currencyCode; |
||
| 671 | } |
||
| 672 | |||
| 673 | /** |
||
| 674 | * Set the currency code. Only used by NumberFormat::toFormattedString() |
||
| 675 | * to format output by \PhpOffice\PhpSpreadsheet\Writer\Html and \PhpOffice\PhpSpreadsheet\Writer\Pdf. |
||
| 676 | * |
||
| 677 | * @param string $pValue Character for currency code |
||
| 678 | */ |
||
| 679 | 197 | public static function setCurrencyCode($pValue) |
|
| 682 | 197 | } |
|
| 683 | |||
| 684 | /** |
||
| 685 | * Convert SYLK encoded string to UTF-8. |
||
| 686 | * |
||
| 687 | * @param string $pValue |
||
| 688 | * |
||
| 689 | * @return string UTF-8 encoded string |
||
| 690 | */ |
||
| 691 | 2 | public static function SYLKtoUTF8($pValue) |
|
| 692 | { |
||
| 693 | 2 | self::buildCharacterSets(); |
|
| 694 | |||
| 695 | // If there is no escape character in the string there is nothing to do |
||
| 696 | 2 | if (strpos($pValue, '') === false) { |
|
| 697 | 1 | return $pValue; |
|
| 698 | } |
||
| 699 | |||
| 700 | 1 | foreach (self::$SYLKCharacters as $k => $v) { |
|
| 701 | 1 | $pValue = str_replace($k, $v, $pValue); |
|
| 702 | } |
||
| 703 | |||
| 704 | 1 | return $pValue; |
|
| 705 | } |
||
| 706 | |||
| 707 | /** |
||
| 708 | * Retrieve any leading numeric part of a string, or return the full string if no leading numeric |
||
| 709 | * (handles basic integer or float, but not exponent or non decimal). |
||
| 710 | * |
||
| 711 | * @param string $value |
||
| 712 | * |
||
| 713 | * @return mixed string or only the leading numeric part of the string |
||
| 714 | */ |
||
| 715 | 84 | public static function testStringAsNumeric($value) |
|
| 723 | } |
||
| 724 | } |
||
| 725 |
Loading history...