|
@@ 6872-6877 (lines=6) @@
|
| 6869 |
|
if ($c2 >= "\x80" && $c2 <= "\xBF") { // yeah, almost sure it's UTF8 already |
| 6870 |
|
$buf .= $c1 . $c2; |
| 6871 |
|
$i++; |
| 6872 |
|
} else { // not valid UTF8 - convert it |
| 6873 |
|
$cc1tmp = ord($c1) / 64; |
| 6874 |
|
$cc1 = self::chr_and_parse_int($cc1tmp) | "\xC0"; |
| 6875 |
|
$cc2 = ($c1 & "\x3F") | "\x80"; |
| 6876 |
|
$buf .= $cc1 . $cc2; |
| 6877 |
|
} |
| 6878 |
|
|
| 6879 |
|
} elseif ($c1 >= "\xE0" && $c1 <= "\xEF") { // looks like 3 bytes UTF8 |
| 6880 |
|
|
|
@@ 6887-6892 (lines=6) @@
|
| 6884 |
|
if ($c2 >= "\x80" && $c2 <= "\xBF" && $c3 >= "\x80" && $c3 <= "\xBF") { // yeah, almost sure it's UTF8 already |
| 6885 |
|
$buf .= $c1 . $c2 . $c3; |
| 6886 |
|
$i += 2; |
| 6887 |
|
} else { // not valid UTF8 - convert it |
| 6888 |
|
$cc1tmp = ord($c1) / 64; |
| 6889 |
|
$cc1 = self::chr_and_parse_int($cc1tmp) | "\xC0"; |
| 6890 |
|
$cc2 = ($c1 & "\x3F") | "\x80"; |
| 6891 |
|
$buf .= $cc1 . $cc2; |
| 6892 |
|
} |
| 6893 |
|
|
| 6894 |
|
} elseif ($c1 >= "\xF0" && $c1 <= "\xF7") { // looks like 4 bytes UTF8 |
| 6895 |
|
|
|
@@ 6903-6908 (lines=6) @@
|
| 6900 |
|
if ($c2 >= "\x80" && $c2 <= "\xBF" && $c3 >= "\x80" && $c3 <= "\xBF" && $c4 >= "\x80" && $c4 <= "\xBF") { // yeah, almost sure it's UTF8 already |
| 6901 |
|
$buf .= $c1 . $c2 . $c3 . $c4; |
| 6902 |
|
$i += 3; |
| 6903 |
|
} else { // not valid UTF8 - convert it |
| 6904 |
|
$cc1tmp = ord($c1) / 64; |
| 6905 |
|
$cc1 = self::chr_and_parse_int($cc1tmp) | "\xC0"; |
| 6906 |
|
$cc2 = ($c1 & "\x3F") | "\x80"; |
| 6907 |
|
$buf .= $cc1 . $cc2; |
| 6908 |
|
} |
| 6909 |
|
|
| 6910 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6911 |
|
$cc1tmp = ord($c1) / 64; |
|
@@ 6910-6915 (lines=6) @@
|
| 6907 |
|
$buf .= $cc1 . $cc2; |
| 6908 |
|
} |
| 6909 |
|
|
| 6910 |
|
} else { // doesn't look like UTF8, but should be converted |
| 6911 |
|
$cc1tmp = ord($c1) / 64; |
| 6912 |
|
$cc1 = self::chr_and_parse_int($cc1tmp) | "\xC0"; |
| 6913 |
|
$cc2 = ($c1 & "\x3F") | "\x80"; |
| 6914 |
|
$buf .= $cc1 . $cc2; |
| 6915 |
|
} |
| 6916 |
|
|
| 6917 |
|
} elseif (($c1 & "\xC0") === "\x80") { // needs conversion |
| 6918 |
|
|
|
@@ 6922-6926 (lines=5) @@
|
| 6919 |
|
$ordC1 = ord($c1); |
| 6920 |
|
if (isset(self::$WIN1252_TO_UTF8[$ordC1])) { // found in Windows-1252 special cases |
| 6921 |
|
$buf .= self::$WIN1252_TO_UTF8[$ordC1]; |
| 6922 |
|
} else { |
| 6923 |
|
$cc1 = self::chr_and_parse_int($ordC1 / 64) | "\xC0"; |
| 6924 |
|
$cc2 = ($c1 & "\x3F") | "\x80"; |
| 6925 |
|
$buf .= $cc1 . $cc2; |
| 6926 |
|
} |
| 6927 |
|
|
| 6928 |
|
} else { // it doesn't need conversion |
| 6929 |
|
$buf .= $c1; |