|
@@ 6120-6131 (lines=12) @@
|
| 6117 |
|
$buf .= $cc1 . $cc2; |
| 6118 |
|
} |
| 6119 |
|
|
| 6120 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6121 |
|
|
| 6122 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6123 |
|
$buf .= $c1 . $c2 . $c3; |
| 6124 |
|
$i += 2; |
| 6125 |
|
} else { // not valid UTF8 - convert it |
| 6126 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6127 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6128 |
|
$buf .= $cc1 . $cc2; |
| 6129 |
|
} |
| 6130 |
|
|
| 6131 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6132 |
|
|
| 6133 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6134 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6133-6140 (lines=8) @@
|
| 6130 |
|
|
| 6131 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6132 |
|
|
| 6133 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6134 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6135 |
|
$i += 3; |
| 6136 |
|
} else { // not valid UTF8 - convert it |
| 6137 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6138 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6139 |
|
$buf .= $cc1 . $cc2; |
| 6140 |
|
} |
| 6141 |
|
|
| 6142 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6143 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |