|
@@ 2011-2022 (lines=12) @@
|
| 2008 |
|
$buf .= $cc1 . $cc2; |
| 2009 |
|
} |
| 2010 |
|
|
| 2011 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 2012 |
|
|
| 2013 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 2014 |
|
$buf .= $c1 . $c2 . $c3; |
| 2015 |
|
$i += 2; |
| 2016 |
|
} else { // not valid UTF8 - convert it |
| 2017 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 2018 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 2019 |
|
$buf .= $cc1 . $cc2; |
| 2020 |
|
} |
| 2021 |
|
|
| 2022 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 2023 |
|
|
| 2024 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 2025 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 2024-2031 (lines=8) @@
|
| 2021 |
|
|
| 2022 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 2023 |
|
|
| 2024 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 2025 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 2026 |
|
$i += 3; |
| 2027 |
|
} else { // not valid UTF8 - convert it |
| 2028 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 2029 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 2030 |
|
$buf .= $cc1 . $cc2; |
| 2031 |
|
} |
| 2032 |
|
|
| 2033 |
|
} else { // doesn't look like UTF8, but should be converted |
| 2034 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |