|
@@ 5496-5507 (lines=12) @@
|
| 5493 |
|
$buf .= $cc1 . $cc2; |
| 5494 |
|
} |
| 5495 |
|
|
| 5496 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 5497 |
|
|
| 5498 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5499 |
|
$buf .= $c1 . $c2 . $c3; |
| 5500 |
|
$i += 2; |
| 5501 |
|
} else { // not valid UTF8 - convert it |
| 5502 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 5503 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 5504 |
|
$buf .= $cc1 . $cc2; |
| 5505 |
|
} |
| 5506 |
|
|
| 5507 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 5508 |
|
|
| 5509 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5510 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 5509-5516 (lines=8) @@
|
| 5506 |
|
|
| 5507 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 5508 |
|
|
| 5509 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5510 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 5511 |
|
$i += 3; |
| 5512 |
|
} else { // not valid UTF8 - convert it |
| 5513 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 5514 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 5515 |
|
$buf .= $cc1 . $cc2; |
| 5516 |
|
} |
| 5517 |
|
|
| 5518 |
|
} else { // doesn't look like UTF8, but should be converted |
| 5519 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |