|
@@ 6309-6320 (lines=12) @@
|
| 6306 |
|
$buf .= $cc1 . $cc2; |
| 6307 |
|
} |
| 6308 |
|
|
| 6309 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6310 |
|
|
| 6311 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6312 |
|
$buf .= $c1 . $c2 . $c3; |
| 6313 |
|
$i += 2; |
| 6314 |
|
} else { // not valid UTF8 - convert it |
| 6315 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6316 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6317 |
|
$buf .= $cc1 . $cc2; |
| 6318 |
|
} |
| 6319 |
|
|
| 6320 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6321 |
|
|
| 6322 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6323 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6322-6329 (lines=8) @@
|
| 6319 |
|
|
| 6320 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6321 |
|
|
| 6322 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6323 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6324 |
|
$i += 3; |
| 6325 |
|
} else { // not valid UTF8 - convert it |
| 6326 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6327 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6328 |
|
$buf .= $cc1 . $cc2; |
| 6329 |
|
} |
| 6330 |
|
|
| 6331 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6332 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |