voku /
portable-utf8
| @@ 3441-3487 (lines=47) @@ | ||
| 3438 | * |
|
| 3439 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3440 | */ |
|
| 3441 | public static function is_utf16($str) |
|
| 3442 | { |
|
| 3443 | if (self::is_binary($str)) { |
|
| 3444 | self::checkForSupport(); |
|
| 3445 | ||
| 3446 | $maybeUTF16LE = 0; |
|
| 3447 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3448 | if ($test !== false && strlen($test) > 1) { |
|
| 3449 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3450 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3451 | if ($test3 === $test) { |
|
| 3452 | $strChars = self::count_chars($str); |
|
| 3453 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3454 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3455 | $maybeUTF16LE++; |
|
| 3456 | } |
|
| 3457 | } |
|
| 3458 | } |
|
| 3459 | } |
|
| 3460 | ||
| 3461 | $maybeUTF16BE = 0; |
|
| 3462 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3463 | if ($test !== false && strlen($test) > 1) { |
|
| 3464 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3465 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3466 | if ($test3 === $test) { |
|
| 3467 | $strChars = self::count_chars($str); |
|
| 3468 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3469 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3470 | $maybeUTF16BE++; |
|
| 3471 | } |
|
| 3472 | } |
|
| 3473 | } |
|
| 3474 | } |
|
| 3475 | ||
| 3476 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 3477 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3478 | return 1; |
|
| 3479 | } else { |
|
| 3480 | return 2; |
|
| 3481 | } |
|
| 3482 | } |
|
| 3483 | ||
| 3484 | } |
|
| 3485 | ||
| 3486 | return false; |
|
| 3487 | } |
|
| 3488 | ||
| 3489 | /** |
|
| 3490 | * Check if the string is UTF-32. |
|
| @@ 3496-3542 (lines=47) @@ | ||
| 3493 | * |
|
| 3494 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3495 | */ |
|
| 3496 | public static function is_utf32($str) |
|
| 3497 | { |
|
| 3498 | if (self::is_binary($str)) { |
|
| 3499 | self::checkForSupport(); |
|
| 3500 | ||
| 3501 | $maybeUTF32LE = 0; |
|
| 3502 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3503 | if ($test !== false && strlen($test) > 1) { |
|
| 3504 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3505 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3506 | if ($test3 === $test) { |
|
| 3507 | $strChars = self::count_chars($str); |
|
| 3508 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3509 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3510 | $maybeUTF32LE++; |
|
| 3511 | } |
|
| 3512 | } |
|
| 3513 | } |
|
| 3514 | } |
|
| 3515 | ||
| 3516 | $maybeUTF32BE = 0; |
|
| 3517 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3518 | if ($test !== false && strlen($test) > 1) { |
|
| 3519 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3520 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3521 | if ($test3 === $test) { |
|
| 3522 | $strChars = self::count_chars($str); |
|
| 3523 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3524 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3525 | $maybeUTF32BE++; |
|
| 3526 | } |
|
| 3527 | } |
|
| 3528 | } |
|
| 3529 | } |
|
| 3530 | ||
| 3531 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 3532 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3533 | return 1; |
|
| 3534 | } else { |
|
| 3535 | return 2; |
|
| 3536 | } |
|
| 3537 | } |
|
| 3538 | ||
| 3539 | } |
|
| 3540 | ||
| 3541 | return false; |
|
| 3542 | } |
|
| 3543 | ||
| 3544 | /** |
|
| 3545 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
