|
@@ 1919-1930 (lines=12) @@
|
| 1916 |
|
$buf .= $cc1 . $cc2; |
| 1917 |
|
} |
| 1918 |
|
|
| 1919 |
|
} elseif ($c1 >= "\xe0" & $c1 <= "\xef") { // looks like 3 bytes UTF8 |
| 1920 |
|
|
| 1921 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 1922 |
|
$buf .= $c1 . $c2 . $c3; |
| 1923 |
|
$i += 2; |
| 1924 |
|
} else { // not valid UTF8 - convert it |
| 1925 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 1926 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 1927 |
|
$buf .= $cc1 . $cc2; |
| 1928 |
|
} |
| 1929 |
|
|
| 1930 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 1931 |
|
|
| 1932 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 1933 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
|
@@ 1932-1939 (lines=8) @@
|
| 1929 |
|
|
| 1930 |
|
} elseif ($c1 >= "\xf0" & $c1 <= "\xf7") { // looks like 4 bytes UTF8 |
| 1931 |
|
|
| 1932 |
|
if ($c2 >= "\x80" && $c2 <= "\xbf" && $c3 >= "\x80" && $c3 <= "\xbf" && $c4 >= "\x80" && $c4 <= "\xbf") { // yeah, almost sure it's UTF8 already |
| 1933 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 1934 |
|
$i += 3; |
| 1935 |
|
} else { // not valid UTF8 - convert it |
| 1936 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |
| 1937 |
|
$cc2 = ($c1 & "\x3f") | "\x80"; |
| 1938 |
|
$buf .= $cc1 . $cc2; |
| 1939 |
|
} |
| 1940 |
|
|
| 1941 |
|
} else { // doesn't look like UTF8, but should be converted |
| 1942 |
|
$cc1 = (chr(ord($c1) / 64) | "\xc0"); |