circlecloud/tmux
1
0
Fork 0
mirror of https://github.com/tmux/tmux.git synced 2025-09-02 13:37:12 +00:00
Code Issues Projects Releases Wiki Activity

Fix incorrect handling of Korean Hangul Jamo characters. GitHub issue

Browse Source 
4546 from Roy Jung.
This commit is contained in:
nicm
2025-07-22 07:42:52 +00:00
parent 6549aaf393
commit 206f492401
3 changed files with 161 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

133
utf8-combined.c
View File

@ -24,6 +24,28 @@
#include "tmux.h"
enum hanguljamo_subclass {
HANGULJAMO_SUBCLASS_NOT_HANGULJAMO,
HANGULJAMO_SUBCLASS_CHOSEONG, // U+1100 - U+1112
HANGULJAMO_SUBCLASS_OLD_CHOSEONG, // U+1113 - U+115E
HANGULJAMO_SUBCLASS_CHOSEONG_FILLER, // U+115F
HANGULJAMO_SUBCLASS_JUNGSEONG_FILLER, // U+1160
HANGULJAMO_SUBCLASS_JUNGSEONG, // U+1161 - U+1175
HANGULJAMO_SUBCLASS_OLD_JUNGSEONG, // U+1176 - U+11A7
HANGULJAMO_SUBCLASS_JONGSEONG, // U+11A8 - U+11C2
HANGULJAMO_SUBCLASS_OLD_JONGSEONG, // U+11C3 - U+11FF
HANGULJAMO_SUBCLASS_EXTENDED_OLD_CHOSEONG, // U+A960 - U+A97C
HANGULJAMO_SUBCLASS_EXTENDED_OLD_JUNGSEONG, // U+D7B0 - U+D7C6
HANGULJAMO_SUBCLASS_EXTENDED_OLD_JONGSEONG // U+D7CB - U+D7FB
};
enum hanguljamo_class {
HANGULJAMO_CLASS_NOT_HANGULJAMO,
HANGULJAMO_CLASS_CHOSEONG,
HANGULJAMO_CLASS_JUNGSEONG,
HANGULJAMO_CLASS_JONGSEONG
};
/* Has this got a zero width joiner at the end? */
int
utf8_has_zwj(const struct utf8_data *ud)
@ -95,3 +117,114 @@ utf8_is_modifier(const struct utf8_data *ud)
}
return (0);
}
static enum hanguljamo_subclass
hanguljamo_get_subclass(const u_char *s)
{
switch (s[0]) {
case 0xE1:
switch (s[1]) {
case 0x84:
if (s[2] >= 0x80 && s[2] <= 0x92)
return (HANGULJAMO_SUBCLASS_CHOSEONG);
if (s[2] >= 0x93 && s[2] <= 0xBF)
return (HANGULJAMO_SUBCLASS_OLD_CHOSEONG);
break;
case 0x85:
if (s[2] == 0x9F)
return (HANGULJAMO_SUBCLASS_CHOSEONG_FILLER);
if (s[2] == 0xA0)
return (HANGULJAMO_SUBCLASS_JUNGSEONG_FILLER);
if (s[2] >= 0x80 && s[2] <= 0x9E)
return (HANGULJAMO_SUBCLASS_OLD_CHOSEONG);
if (s[2] >= 0xA1 && s[2] <= 0xB5)
return (HANGULJAMO_SUBCLASS_JUNGSEONG);
if (s[2] >= 0xB6 && s[2] <= 0xBF)
return (HANGULJAMO_SUBCLASS_OLD_JUNGSEONG);
break;
case 0x86:
if (s[2] >= 0x80 && s[2] <= 0xA7)
return (HANGULJAMO_SUBCLASS_OLD_JUNGSEONG);
if (s[2] >= 0xA8 && s[2] <= 0xBF)
return (HANGULJAMO_SUBCLASS_JONGSEONG);
break;
case 0x87:
if (s[2] >= 0x80 && s[2] <= 0x82)
return (HANGULJAMO_SUBCLASS_JONGSEONG);
if (s[2] >= 0x83 && s[2] <= 0xBF)
return (HANGULJAMO_SUBCLASS_OLD_JONGSEONG);
break;
}
break;
case 0xEA:
if (s[1] == 0xA5 && s[2] >= 0xA0 && s[2] <= 0xBC)
return (HANGULJAMO_SUBCLASS_EXTENDED_OLD_CHOSEONG);
break;
case 0xED:
if (s[1] == 0x9E && s[2] >= 0xB0 && s[2] <= 0xBF)
return (HANGULJAMO_SUBCLASS_EXTENDED_OLD_JUNGSEONG);
if (s[1] != 0x9F)
break;
if (s[2] >= 0x80 && s[2] <= 0x86)
return (HANGULJAMO_SUBCLASS_EXTENDED_OLD_JUNGSEONG);
if (s[2] >= 0x8B && s[2] <= 0xBB)
return (HANGULJAMO_SUBCLASS_EXTENDED_OLD_JONGSEONG);
break;
}
return (HANGULJAMO_SUBCLASS_NOT_HANGULJAMO);
}
static enum hanguljamo_class
hanguljamo_get_class(const u_char *s)
{
switch (hanguljamo_get_subclass(s)) {
case HANGULJAMO_SUBCLASS_CHOSEONG:
case HANGULJAMO_SUBCLASS_CHOSEONG_FILLER:
case HANGULJAMO_SUBCLASS_OLD_CHOSEONG:
case HANGULJAMO_SUBCLASS_EXTENDED_OLD_CHOSEONG:
return (HANGULJAMO_CLASS_CHOSEONG);
case HANGULJAMO_SUBCLASS_JUNGSEONG:
case HANGULJAMO_SUBCLASS_JUNGSEONG_FILLER:
case HANGULJAMO_SUBCLASS_OLD_JUNGSEONG:
case HANGULJAMO_SUBCLASS_EXTENDED_OLD_JUNGSEONG:
return (HANGULJAMO_CLASS_JUNGSEONG);
case HANGULJAMO_SUBCLASS_JONGSEONG:
case HANGULJAMO_SUBCLASS_OLD_JONGSEONG:
case HANGULJAMO_SUBCLASS_EXTENDED_OLD_JONGSEONG:
return (HANGULJAMO_CLASS_JONGSEONG);
case HANGULJAMO_SUBCLASS_NOT_HANGULJAMO:
return (HANGULJAMO_CLASS_NOT_HANGULJAMO);
}
return (HANGULJAMO_CLASS_NOT_HANGULJAMO);
}
enum hanguljamo_state
hanguljamo_check_state(const struct utf8_data *p_ud, const struct utf8_data *ud)
{
const u_char *s;
if (ud->size != 3)
return (HANGULJAMO_STATE_NOT_HANGULJAMO);
switch (hanguljamo_get_class(ud->data)) {
case HANGULJAMO_CLASS_CHOSEONG:
return (HANGULJAMO_STATE_CHOSEONG);
case HANGULJAMO_CLASS_JUNGSEONG:
if (p_ud->size < 3)
return (HANGULJAMO_STATE_NOT_COMPOSABLE);
s = p_ud->data + p_ud->size - 3;
if (hanguljamo_get_class(s) == HANGULJAMO_CLASS_CHOSEONG)
return (HANGULJAMO_STATE_COMPOSABLE);
return (HANGULJAMO_STATE_NOT_COMPOSABLE);
case HANGULJAMO_CLASS_JONGSEONG:
if (p_ud->size < 3)
return (HANGULJAMO_STATE_NOT_COMPOSABLE);
s = p_ud->data + p_ud->size - 3;
if (hanguljamo_get_class(s) == HANGULJAMO_CLASS_JUNGSEONG)
return (HANGULJAMO_STATE_COMPOSABLE);
return (HANGULJAMO_STATE_NOT_COMPOSABLE);
case HANGULJAMO_CLASS_NOT_HANGULJAMO:
return (HANGULJAMO_STATE_NOT_HANGULJAMO);
}
return (HANGULJAMO_STATE_NOT_HANGULJAMO);
}