|
@@ 5383-5394 (lines=12) @@
|
| 5380 |
|
$buf .= $cc1 . $cc2; |
| 5381 |
|
} |
| 5382 |
|
|
| 5383 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 5384 |
|
|
| 5385 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5386 |
|
$buf .= $c1 . $c2 . $c3; |
| 5387 |
|
$i += 2; |
| 5388 |
|
} else { // not valid UTF8 - convert it |
| 5389 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 5390 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 5391 |
|
$buf .= $cc1 . $cc2; |
| 5392 |
|
} |
| 5393 |
|
|
| 5394 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 5395 |
|
|
| 5396 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5397 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 5396-5403 (lines=8) @@
|
| 5393 |
|
|
| 5394 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 5395 |
|
|
| 5396 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 5397 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 5398 |
|
$i += 3; |
| 5399 |
|
} else { // not valid UTF8 - convert it |
| 5400 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 5401 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 5402 |
|
$buf .= $cc1 . $cc2; |
| 5403 |
|
} |
| 5404 |
|
|
| 5405 |
|
} else { // doesn't look like UTF8, but should be converted |
| 5406 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |