| 1 | <?php |
| 2 | /** |
| 3 | * takes a string of unicode entities and converts it to a utf-8 encoded string |
| 4 | * each unicode entitiy has the form &#nnn(nn); n={0..9} and can be displayed by utf-8 supporting |
| 5 | * browsers. Ascii will not be modified. |
| 6 | * @param $source string of unicode entities [STRING] |
| 7 | * @return a utf-8 encoded string [STRING] |
| 8 | * @access public |
| 9 | */ |
| 10 | function charset_encode_utf_8 ($source) { |
| 11 | $utf8Str = ''; |
| 12 | $entityArray = explode ("&#", $source); |
| 13 | $size = count ($entityArray); |
| 14 | for ($i = 0; $i < $size; $i++) { |
| 15 | $subStr = $entityArray[$i]; |
| 16 | $nonEntity = strstr ($subStr, ';'); |
| 17 | if ($nonEntity !== false) { |
| 18 | $unicode = intval (substr ($subStr, 0, (strpos ($subStr, ';') + 1))); |
| 19 | // determine how many chars are needed to reprsent this unicode char |
| 20 | if ($unicode < 128) { |
| 21 | $utf8Substring = chr ($unicode); |
| 22 | } |
| 23 | else if ($unicode >= 128 && $unicode < 2048) { |
| 24 | $binVal = str_pad (decbin ($unicode), 11, "0", STR_PAD_LEFT); |
| 25 | $binPart1 = substr ($binVal, 0, 5); |
| 26 | $binPart2 = substr ($binVal, 5); |
| 27 | |
| 28 | $char1 = chr (192 + bindec ($binPart1)); |
| 29 | $char2 = chr (128 + bindec ($binPart2)); |
| 30 | $utf8Substring = $char1 . $char2; |
| 31 | } |
| 32 | else if ($unicode >= 2048 && $unicode < 65536) { |
| 33 | $binVal = str_pad (decbin ($unicode), 16, "0", STR_PAD_LEFT); |
| 34 | $binPart1 = substr ($binVal, 0, 4); |
| 35 | $binPart2 = substr ($binVal, 4, 6); |
| 36 | $binPart3 = substr ($binVal, 10); |
| 37 | |
| 38 | $char1 = chr (224 + bindec ($binPart1)); |
| 39 | $char2 = chr (128 + bindec ($binPart2)); |
| 40 | $char3 = chr (128 + bindec ($binPart3)); |
| 41 | $utf8Substring = $char1 . $char2 . $char3; |
| 42 | } |
| 43 | else { |
| 44 | $binVal = str_pad (decbin ($unicode), 21, "0", STR_PAD_LEFT); |
| 45 | $binPart1 = substr ($binVal, 0, 3); |
| 46 | $binPart2 = substr ($binVal, 3, 6); |
| 47 | $binPart3 = substr ($binVal, 9, 6); |
| 48 | $binPart4 = substr ($binVal, 15); |
| 49 | |
| 50 | $char1 = chr (240 + bindec ($binPart1)); |
| 51 | $char2 = chr (128 + bindec ($binPart2)); |
| 52 | $char3 = chr (128 + bindec ($binPart3)); |
| 53 | $char4 = chr (128 + bindec ($binPart4)); |
| 54 | $utf8Substring = $char1 . $char2 . $char3 . $char4; |
| 55 | } |
| 56 | |
| 57 | if (strlen ($nonEntity) > 1) |
| 58 | $nonEntity = substr ($nonEntity, 1); // chop the first char (';') |
| 59 | else |
| 60 | $nonEntity = ''; |
| 61 | |
| 62 | $utf8Str .= $utf8Substring . $nonEntity; |
| 63 | } |
| 64 | else { |
| 65 | $utf8Str .= $subStr; |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | return $utf8Str; |
| 70 | } |
| 71 | ?> |