|
@@ 6155-6166 (lines=12) @@
|
| 6152 |
|
$buf .= $cc1 . $cc2; |
| 6153 |
|
} |
| 6154 |
|
|
| 6155 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 6156 |
|
|
| 6157 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6158 |
|
$buf .= $c1 . $c2 . $c3; |
| 6159 |
|
$i += 2; |
| 6160 |
|
} else { // not valid UTF8 - convert it |
| 6161 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6162 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6163 |
|
$buf .= $cc1 . $cc2; |
| 6164 |
|
} |
| 6165 |
|
|
| 6166 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6167 |
|
|
| 6168 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6169 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 6168-6175 (lines=8) @@
|
| 6165 |
|
|
| 6166 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 6167 |
|
|
| 6168 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 6169 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6170 |
|
$i += 3; |
| 6171 |
|
} else { // not valid UTF8 - convert it |
| 6172 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 6173 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 6174 |
|
$buf .= $cc1 . $cc2; |
| 6175 |
|
} |
| 6176 |
|
|
| 6177 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6178 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |