|
@@ 6607-6618 (lines=12) @@
|
| 6604 |
|
$buf .= $cc1 . $cc2; |
| 6605 |
|
} |
| 6606 |
|
|
| 6607 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6608 |
|
|
| 6609 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6610 |
|
$buf .= $c1 . $c2 . $c3; |
| 6611 |
|
$i += 2; |
| 6612 |
|
} else { // not valid UTF8 - convert it |
| 6613 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6614 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6615 |
|
$buf .= $cc1 . $cc2; |
| 6616 |
|
} |
| 6617 |
|
|
| 6618 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6619 |
|
|
| 6620 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6621 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6620-6627 (lines=8) @@
|
| 6617 |
|
|
| 6618 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6619 |
|
|
| 6620 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6621 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6622 |
|
$i += 3; |
| 6623 |
|
} else { // not valid UTF8 - convert it |
| 6624 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6625 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6626 |
|
$buf .= $cc1 . $cc2; |
| 6627 |
|
} |
| 6628 |
|
|
| 6629 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6630 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |