|
@@ 6404-6415 (lines=12) @@
|
| 6401 |
|
$buf .= $cc1 . $cc2; |
| 6402 |
|
} |
| 6403 |
|
|
| 6404 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6405 |
|
|
| 6406 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6407 |
|
$buf .= $c1 . $c2 . $c3; |
| 6408 |
|
$i += 2; |
| 6409 |
|
} else { // not valid UTF8 - convert it |
| 6410 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6411 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6412 |
|
$buf .= $cc1 . $cc2; |
| 6413 |
|
} |
| 6414 |
|
|
| 6415 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6416 |
|
|
| 6417 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6418 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6417-6424 (lines=8) @@
|
| 6414 |
|
|
| 6415 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6416 |
|
|
| 6417 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6418 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6419 |
|
$i += 3; |
| 6420 |
|
} else { // not valid UTF8 - convert it |
| 6421 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6422 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6423 |
|
$buf .= $cc1 . $cc2; |
| 6424 |
|
} |
| 6425 |
|
|
| 6426 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6427 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |