|
| 1 | +<?php |
| 2 | + |
| 3 | +final class wstring implements ArrayAccess |
| 4 | +{ |
| 5 | + /** @var array */ |
| 6 | + protected $codes; |
| 7 | + /** @var array */ |
| 8 | + protected $chars; |
| 9 | + /** @var string|null */ |
| 10 | + protected $string; |
| 11 | + /** @var int */ |
| 12 | + protected $length; |
| 13 | + |
| 14 | + public function __construct(array $codes, array $chars) |
| 15 | + { |
| 16 | + $this->codes = $codes; |
| 17 | + $this->chars = $chars; |
| 18 | + $this->length = count($codes); |
| 19 | + } |
| 20 | + |
| 21 | + public function offsetExists($offset) |
| 22 | + { |
| 23 | + return isset($this->codes[$offset]); |
| 24 | + } |
| 25 | + |
| 26 | + public function offsetGet($offset) |
| 27 | + { |
| 28 | + return $this->chars[$offset]; |
| 29 | + } |
| 30 | + |
| 31 | + public function offsetSet($offset, $value) |
| 32 | + { |
| 33 | + throw new Exception("Invalid operation"); |
| 34 | + } |
| 35 | + |
| 36 | + public function offsetUnset($offset) |
| 37 | + { |
| 38 | + throw new Exception("Invalid operation"); |
| 39 | + } |
| 40 | + |
| 41 | + /** |
| 42 | + * @return int |
| 43 | + */ |
| 44 | + public function length() |
| 45 | + { |
| 46 | + return $this->length; |
| 47 | + } |
| 48 | + |
| 49 | + /** |
| 50 | + * @return bool |
| 51 | + */ |
| 52 | + public function isEmpty() |
| 53 | + { |
| 54 | + return $this->length === 0; |
| 55 | + } |
| 56 | + |
| 57 | + /** |
| 58 | + * @param int $start |
| 59 | + * @param int|null $length |
| 60 | + * @return wstring |
| 61 | + */ |
| 62 | + public function substr($start, $length = null) |
| 63 | + { |
| 64 | + return $this->substring($start, $length); |
| 65 | + } |
| 66 | + |
| 67 | + |
| 68 | + /** |
| 69 | + * @param int $start |
| 70 | + * @param int|null $length |
| 71 | + * @return wstring |
| 72 | + */ |
| 73 | + public function substring($start, $length = null) |
| 74 | + { |
| 75 | + $cp = array_slice($this->codes, $start, $length); |
| 76 | + $ch = array_slice($this->chars, $start, $length); |
| 77 | + |
| 78 | + return new self($cp, $ch); |
| 79 | + } |
| 80 | + |
| 81 | + /** |
| 82 | + * @param string|wstring $text |
| 83 | + * @param int $offset |
| 84 | + * @return int|false |
| 85 | + * @throws Exception |
| 86 | + */ |
| 87 | + public function indexOf($text, $offset = 0) |
| 88 | + { |
| 89 | + $text = wstring($text); |
| 90 | + |
| 91 | + if($offset < 0){ |
| 92 | + $offset = 0; |
| 93 | + } |
| 94 | + |
| 95 | + $cp1 = $this->codes; |
| 96 | + $cp2 = $text->codes; |
| 97 | + |
| 98 | + for($i = $offset, $l = $this->length - $text->length; $i <= $l; $i++){ |
| 99 | + $match = true; |
| 100 | + |
| 101 | + for($j = 0, $f = $text->length; $j < $f; $j++){ |
| 102 | + if($cp1[$i + $j] != $cp2[$j]){ |
| 103 | + $match = false; |
| 104 | + break; |
| 105 | + } |
| 106 | + } |
| 107 | + |
| 108 | + if($match){ |
| 109 | + return $i; |
| 110 | + } |
| 111 | + } |
| 112 | + |
| 113 | + return false; |
| 114 | + } |
| 115 | + |
| 116 | + /** |
| 117 | + * @param string|wstring $character_mask |
| 118 | + * @return wstring |
| 119 | + */ |
| 120 | + public function trim($character_mask = " \t\n\r\0\x0B") |
| 121 | + { |
| 122 | + return $this->doTrim($character_mask); |
| 123 | + } |
| 124 | + |
| 125 | + /** |
| 126 | + * @param string|wstring $character_mask |
| 127 | + * @return wstring |
| 128 | + */ |
| 129 | + public function ltrim($character_mask = " \t\n\r\0\x0B") |
| 130 | + { |
| 131 | + return $this->doTrim($character_mask, true, false); |
| 132 | + } |
| 133 | + |
| 134 | + /** |
| 135 | + * @param string|wstring $character_mask |
| 136 | + * @return wstring |
| 137 | + */ |
| 138 | + public function rtrim($character_mask = " \t\n\r\0\x0B") |
| 139 | + { |
| 140 | + return $this->doTrim($character_mask, false, true); |
| 141 | + } |
| 142 | + |
| 143 | + /** |
| 144 | + * @param string|wstring $subject |
| 145 | + * @param string|wstring $replace |
| 146 | + * @param int $offset |
| 147 | + * @return wstring |
| 148 | + * @throws Exception |
| 149 | + */ |
| 150 | + public function replace($subject, $replace, $offset = 0) |
| 151 | + { |
| 152 | + $subject = wstring($subject); |
| 153 | + $replace = wstring($replace); |
| 154 | + |
| 155 | + if(false === $pos = $this->indexOf($subject, $offset)){ |
| 156 | + return clone $this; |
| 157 | + } |
| 158 | + |
| 159 | + $cp1 = array_slice($this->codes, 0, $pos); |
| 160 | + $cp2 = array_slice($this->codes, $pos + $subject->length); |
| 161 | + $ch1 = array_slice($this->chars, 0, $pos); |
| 162 | + $ch2 = array_slice($this->chars, $pos + $subject->length); |
| 163 | + |
| 164 | + $cp = array_merge($cp1, $replace->codes, $cp2); |
| 165 | + $ch = array_merge($ch1, $replace->chars, $ch2); |
| 166 | + |
| 167 | + return new self($cp, $ch); |
| 168 | + } |
| 169 | + |
| 170 | + /** |
| 171 | + * @return string |
| 172 | + */ |
| 173 | + public function __toString() |
| 174 | + { |
| 175 | + if($this->string === null){ |
| 176 | + $this->string = implode('', $this->chars); |
| 177 | + } |
| 178 | + return $this->string; |
| 179 | + } |
| 180 | + |
| 181 | + /** |
| 182 | + * @param $character_mask |
| 183 | + * @param bool $left |
| 184 | + * @param bool $right |
| 185 | + * @return wstring |
| 186 | + * @throws Exception |
| 187 | + */ |
| 188 | + protected function doTrim($character_mask, $left = true, $right = true) |
| 189 | + { |
| 190 | + $character_mask = wstring($character_mask); |
| 191 | + |
| 192 | + $cm = $character_mask->codes; |
| 193 | + $cp = $this->codes; |
| 194 | + $l = count($cm); |
| 195 | + $start = 0; |
| 196 | + $end = $this->length; |
| 197 | + |
| 198 | + if($left){ |
| 199 | + for ($i = 0; $i < $this->length; $i++) { |
| 200 | + if (!in_array($cp[$i], $cm)) { |
| 201 | + break; |
| 202 | + } |
| 203 | + } |
| 204 | + $start = $i; |
| 205 | + } |
| 206 | + |
| 207 | + if($right){ |
| 208 | + for ($i = $this->length - 1; $i > $start; $i--) { |
| 209 | + if (!in_array($cp[$i], $cm)) { |
| 210 | + break; |
| 211 | + } |
| 212 | + } |
| 213 | + $end = $i + 1; |
| 214 | + } |
| 215 | + |
| 216 | + $cp = array_slice($cp, $start, $end - $start); |
| 217 | + $ch = array_slice($this->chars, $start, $end - $start); |
| 218 | + |
| 219 | + return new self($cp, $ch); |
| 220 | + } |
| 221 | +} |
| 222 | + |
| 223 | +/** |
| 224 | + * @param $string |
| 225 | + * @return wstring |
| 226 | + * @throws Exception |
| 227 | + */ |
| 228 | +function wstring($string) |
| 229 | +{ |
| 230 | + if($string instanceof wstring){ |
| 231 | + return $string; |
| 232 | + } |
| 233 | + |
| 234 | + $codes = $chars = array(); |
| 235 | + |
| 236 | + if(false === $text = json_encode((string) $string)) { |
| 237 | + throw new Exception("Invalid UTF-8 string"); |
| 238 | + } |
| 239 | + |
| 240 | + for($i = 1, $l = strlen($text) - 1; $i < $l; $i++) { |
| 241 | + $c = $text[$i]; |
| 242 | + |
| 243 | + if($c === '\\'){ |
| 244 | + if(isset($text[$i + 1])){ |
| 245 | + |
| 246 | + if($text[$i + 1] === 'u'){ |
| 247 | + |
| 248 | + $codes[] = $cp = hexdec(substr($text, $i, 6)); |
| 249 | + |
| 250 | + if ($cp < 0x80) { |
| 251 | + $chars[] = chr($cp); |
| 252 | + } elseif ($cp <= 0x7FF) { |
| 253 | + $chars[] = chr(($cp >> 6) + 0xC0) . chr(($cp & 0x3F) + 0x80); |
| 254 | + } elseif ($cp <= 0xFFFF) { |
| 255 | + $chars[] = chr(($cp >> 12) + 0xE0) . chr((($cp >> 6) & 0x3F) + 0x80) . chr(($cp & 0x3F) + 0x80); |
| 256 | + } elseif ($cp <= 0x10FFFF) { |
| 257 | + $chars[] = chr(($cp >> 18) + 0xF0) . chr((($cp >> 12) & 0x3F) + 0x80) |
| 258 | + . chr((($cp >> 6) & 0x3F) + 0x80) . chr(($cp & 0x3F) + 0x80); |
| 259 | + } else { |
| 260 | + throw new Exception("Invalid UTF-8"); |
| 261 | + } |
| 262 | + |
| 263 | + $i += 5; |
| 264 | + continue; |
| 265 | + |
| 266 | + } else{ |
| 267 | + |
| 268 | + switch ($text[$i + 1]){ |
| 269 | + case '\\': |
| 270 | + $c = "\\"; |
| 271 | + break; |
| 272 | + case '\'': |
| 273 | + $c = "'"; |
| 274 | + break; |
| 275 | + case '"': |
| 276 | + $c = '"'; |
| 277 | + break; |
| 278 | + case 'n': |
| 279 | + $c = "\n"; |
| 280 | + break; |
| 281 | + case 'r': |
| 282 | + $c = "\r"; |
| 283 | + break; |
| 284 | + case 't': |
| 285 | + $c = "\t"; |
| 286 | + break; |
| 287 | + case 'b': |
| 288 | + $c = "\b"; |
| 289 | + break; |
| 290 | + case 'f': |
| 291 | + $c = "\f"; |
| 292 | + break; |
| 293 | + case 'v': |
| 294 | + $c = "\v"; |
| 295 | + break; |
| 296 | + case '0': |
| 297 | + $c = "\0"; |
| 298 | + break; |
| 299 | + default: |
| 300 | + $c = $text[$i + 1]; |
| 301 | + } |
| 302 | + |
| 303 | + $codes[] = ord($c); |
| 304 | + $chars[] = $c; |
| 305 | + $i++; |
| 306 | + continue; |
| 307 | + } |
| 308 | + } |
| 309 | + } |
| 310 | + |
| 311 | + $codes[] = ord($c); |
| 312 | + $chars[] = $c; |
| 313 | + } |
| 314 | + |
| 315 | + return new wstring($codes, $chars, null); |
| 316 | +} |
0 commit comments