voku /
portable-utf8
| @@ 3326-3372 (lines=47) @@ | ||
| 3323 | * |
|
| 3324 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3325 | */ |
|
| 3326 | public static function is_utf16($str) |
|
| 3327 | { |
|
| 3328 | if (self::is_binary($str)) { |
|
| 3329 | self::checkForSupport(); |
|
| 3330 | ||
| 3331 | $maybeUTF16LE = 0; |
|
| 3332 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3333 | if ($test !== false && strlen($test) > 1) { |
|
| 3334 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3335 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3336 | if ($test3 === $test) { |
|
| 3337 | $strChars = self::count_chars($str); |
|
| 3338 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3339 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3340 | $maybeUTF16LE++; |
|
| 3341 | } |
|
| 3342 | } |
|
| 3343 | } |
|
| 3344 | } |
|
| 3345 | ||
| 3346 | $maybeUTF16BE = 0; |
|
| 3347 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3348 | if ($test !== false && strlen($test) > 1) { |
|
| 3349 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3350 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3351 | if ($test3 === $test) { |
|
| 3352 | $strChars = self::count_chars($str); |
|
| 3353 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3354 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3355 | $maybeUTF16BE++; |
|
| 3356 | } |
|
| 3357 | } |
|
| 3358 | } |
|
| 3359 | } |
|
| 3360 | ||
| 3361 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 3362 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3363 | return 1; |
|
| 3364 | } else { |
|
| 3365 | return 2; |
|
| 3366 | } |
|
| 3367 | } |
|
| 3368 | ||
| 3369 | } |
|
| 3370 | ||
| 3371 | return false; |
|
| 3372 | } |
|
| 3373 | ||
| 3374 | /** |
|
| 3375 | * Check if the string is UTF-32. |
|
| @@ 3381-3427 (lines=47) @@ | ||
| 3378 | * |
|
| 3379 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3380 | */ |
|
| 3381 | public static function is_utf32($str) |
|
| 3382 | { |
|
| 3383 | if (self::is_binary($str)) { |
|
| 3384 | self::checkForSupport(); |
|
| 3385 | ||
| 3386 | $maybeUTF32LE = 0; |
|
| 3387 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3388 | if ($test !== false && strlen($test) > 1) { |
|
| 3389 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3390 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3391 | if ($test3 === $test) { |
|
| 3392 | $strChars = self::count_chars($str); |
|
| 3393 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3394 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3395 | $maybeUTF32LE++; |
|
| 3396 | } |
|
| 3397 | } |
|
| 3398 | } |
|
| 3399 | } |
|
| 3400 | ||
| 3401 | $maybeUTF32BE = 0; |
|
| 3402 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3403 | if ($test !== false && strlen($test) > 1) { |
|
| 3404 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3405 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3406 | if ($test3 === $test) { |
|
| 3407 | $strChars = self::count_chars($str); |
|
| 3408 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3409 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3410 | $maybeUTF32BE++; |
|
| 3411 | } |
|
| 3412 | } |
|
| 3413 | } |
|
| 3414 | } |
|
| 3415 | ||
| 3416 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 3417 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3418 | return 1; |
|
| 3419 | } else { |
|
| 3420 | return 2; |
|
| 3421 | } |
|
| 3422 | } |
|
| 3423 | ||
| 3424 | } |
|
| 3425 | ||
| 3426 | return false; |
|
| 3427 | } |
|
| 3428 | ||
| 3429 | /** |
|
| 3430 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
