|
@@ 6344-6355 (lines=12) @@
|
| 6341 |
|
$buf .= $cc1 . $cc2; |
| 6342 |
|
} |
| 6343 |
|
|
| 6344 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6345 |
|
|
| 6346 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6347 |
|
$buf .= $c1 . $c2 . $c3; |
| 6348 |
|
$i += 2; |
| 6349 |
|
} else { // not valid UTF8 - convert it |
| 6350 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6351 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6352 |
|
$buf .= $cc1 . $cc2; |
| 6353 |
|
} |
| 6354 |
|
|
| 6355 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6356 |
|
|
| 6357 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6358 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6357-6364 (lines=8) @@
|
| 6354 |
|
|
| 6355 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6356 |
|
|
| 6357 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6358 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6359 |
|
$i += 3; |
| 6360 |
|
} else { // not valid UTF8 - convert it |
| 6361 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6362 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6363 |
|
$buf .= $cc1 . $cc2; |
| 6364 |
|
} |
| 6365 |
|
|
| 6366 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6367 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |