| 1 | <?php |
|---|
| 2 | |
|---|
| 3 | /** |
|---|
| 4 | * @version $Id: ord.php,v 1.4 2006/09/11 15:22:54 harryf Exp $ |
|---|
| 5 | * @package utf8 |
|---|
| 6 | * @subpackage strings |
|---|
| 7 | */ |
|---|
| 8 | |
|---|
| 9 | /** |
|---|
| 10 | * UTF-8 aware alternative to ord |
|---|
| 11 | * Returns the unicode ordinal for a character |
|---|
| 12 | * @param string UTF-8 encoded character |
|---|
| 13 | * @return int unicode ordinal for the character |
|---|
| 14 | * @see http://www.php.net/ord |
|---|
| 15 | * @see http://www.php.net/manual/en/function.ord.php#46267 |
|---|
| 16 | */ |
|---|
| 17 | function utf8_ord($chr) |
|---|
| 18 | { |
|---|
| 19 | $ord0 = ord($chr); |
|---|
| 20 | |
|---|
| 21 | if ($ord0 >= 0 && $ord0 <= 127) |
|---|
| 22 | return $ord0; |
|---|
| 23 | |
|---|
| 24 | if (!isset($chr{1})) |
|---|
| 25 | { |
|---|
| 26 | trigger_error('Short sequence - at least 2 bytes expected, only 1 seen'); |
|---|
| 27 | return false; |
|---|
| 28 | } |
|---|
| 29 | |
|---|
| 30 | $ord1 = ord($chr{1}); |
|---|
| 31 | if ($ord0 >= 192 && $ord0 <= 223) |
|---|
| 32 | return ($ord0 - 192) * 64 + ($ord1 - 128); |
|---|
| 33 | |
|---|
| 34 | if (!isset($chr{2})) |
|---|
| 35 | { |
|---|
| 36 | trigger_error('Short sequence - at least 3 bytes expected, only 2 seen'); |
|---|
| 37 | return false; |
|---|
| 38 | } |
|---|
| 39 | |
|---|
| 40 | $ord2 = ord($chr{2}); |
|---|
| 41 | if ($ord0 >= 224 && $ord0 <= 239) |
|---|
| 42 | return ($ord0-224)*4096 + ($ord1-128)*64 + ($ord2-128); |
|---|
| 43 | |
|---|
| 44 | if (!isset($chr{3})) |
|---|
| 45 | { |
|---|
| 46 | trigger_error('Short sequence - at least 4 bytes expected, only 3 seen'); |
|---|
| 47 | return false; |
|---|
| 48 | } |
|---|
| 49 | |
|---|
| 50 | $ord3 = ord($chr{3}); |
|---|
| 51 | if ($ord0>=240 && $ord0<=247) |
|---|
| 52 | return ($ord0-240)*262144 + ($ord1-128)*4096 + ($ord2-128)*64 + ($ord3-128); |
|---|
| 53 | |
|---|
| 54 | if (!isset($chr{4})) |
|---|
| 55 | { |
|---|
| 56 | trigger_error('Short sequence - at least 5 bytes expected, only 4 seen'); |
|---|
| 57 | return false; |
|---|
| 58 | } |
|---|
| 59 | |
|---|
| 60 | $ord4 = ord($chr{4}); |
|---|
| 61 | if ($ord0>=248 && $ord0<=251) |
|---|
| 62 | return ($ord0-248)*16777216 + ($ord1-128)*262144 + ($ord2-128)*4096 + ($ord3-128)*64 + ($ord4-128); |
|---|
| 63 | |
|---|
| 64 | if (!isset($chr{5})) |
|---|
| 65 | { |
|---|
| 66 | trigger_error('Short sequence - at least 6 bytes expected, only 5 seen'); |
|---|
| 67 | return false; |
|---|
| 68 | } |
|---|
| 69 | |
|---|
| 70 | if ($ord0>=252 && $ord0<=253) |
|---|
| 71 | return ($ord0-252) * 1073741824 + ($ord1-128)*16777216 + ($ord2-128)*262144 + ($ord3-128)*4096 + ($ord4-128)*64 + (ord($c{5})-128); |
|---|
| 72 | |
|---|
| 73 | if ($ord0 >= 254 && $ord0 <= 255) |
|---|
| 74 | { |
|---|
| 75 | trigger_error('Invalid UTF-8 with surrogate ordinal '.$ord0); |
|---|
| 76 | return false; |
|---|
| 77 | } |
|---|
| 78 | } |
|---|
