|
@@ 6009-6020 (lines=12) @@
|
| 6006 |
|
$buf .= $cc1 . $cc2; |
| 6007 |
|
} |
| 6008 |
|
|
| 6009 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6010 |
|
|
| 6011 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6012 |
|
$buf .= $c1 . $c2 . $c3; |
| 6013 |
|
$i += 2; |
| 6014 |
|
} else { // not valid UTF8 - convert it |
| 6015 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6016 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6017 |
|
$buf .= $cc1 . $cc2; |
| 6018 |
|
} |
| 6019 |
|
|
| 6020 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6021 |
|
|
| 6022 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6023 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6022-6029 (lines=8) @@
|
| 6019 |
|
|
| 6020 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6021 |
|
|
| 6022 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6023 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6024 |
|
$i += 3; |
| 6025 |
|
} else { // not valid UTF8 - convert it |
| 6026 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6027 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6028 |
|
$buf .= $cc1 . $cc2; |
| 6029 |
|
} |
| 6030 |
|
|
| 6031 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6032 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |