diff options
| author | Carl Drougge <bearded@longhaired.org> | 2022-02-23 16:23:22 +0200 |
|---|---|---|
| committer | Alexander Naumov <alexander_naumov@opensuse.org> | 2022-02-23 16:23:22 +0200 |
| commit | 916d0523137b6ce9d56c2ec6a602a3dc471779a4 (patch) | |
| tree | 499523fe32788389476dce4e1c96a4b5d211c1f5 | |
| parent | 1bfe555cd334feccdf55a0fd2891f743b2ffc7a0 (diff) | |
| download | screen-916d0523137b6ce9d56c2ec6a602a3dc471779a4.tar.gz | |
[PATCH] encoding: Replace out of range unicode with U+FFFD
Also correct the mask for four byte UTF-8.
bug #62097
| -rw-r--r-- | src/encoding.c | 25 |
1 files changed, 11 insertions, 14 deletions
diff --git a/src/encoding.c b/src/encoding.c index 9cebddd..976978a 100644 --- a/src/encoding.c +++ b/src/encoding.c @@ -675,14 +675,14 @@ int c; AddUtf8(combchars[c - 0xd800]->c1); c = combchars[c - 0xd800]->c2; } + + /* replace out of range values with U+FFFD "replacement character" */ + if (c < 0 || c > 0x10ffff) + c = 0xfffd; + if (c >= 0x10000) { - if (c >= 0x200000) - { - AddChar((c & 0x3000000) >> 12 ^ 0xf8); - c = (c & 0xffffff) ^ ((0xf0 ^ 0x80) << 18); - } - AddChar((c & 0x1fc0000) >> 18 ^ 0xf0); + AddChar((c & 0x1c0000) >> 18 ^ 0xf0); c = (c & 0x3ffff) ^ ((0xe0 ^ 0x80) << 12); } if (c >= 0x800) @@ -719,17 +719,14 @@ char *p; int c; { int l = 1; + /* replace out of range values with U+FFFD "replacement character" */ + if (c < 0 || c > 0x10ffff) + c = 0xfffd; + if (c >= 0x10000) { - if (c >= 0x200000) - { - if (p) - *p++ = (c & 0x3000000) >> 12 ^ 0xf8; - l++; - c = (c & 0xffffff) ^ ((0xf0 ^ 0x80) << 18); - } if (p) - *p++ = (c & 0x1fc0000) >> 18 ^ 0xf0; + *p++ = (c & 0x1c0000) >> 18 ^ 0xf0; l++; c = (c & 0x3ffff) ^ ((0xe0 ^ 0x80) << 12); } |