Fix incorrect handling of Korean Hangul Jamo characters. GitHub issue

4546 from Roy Jung.

screen-write.c

@@ -2038,17 +2038,27 @@ screen_write_combine(struct screen_write_ctx *ctx, const struct grid_cell *gc)
 		return (zero_width);
 
 	/*
-	 * Check if we need to combine characters. This could be zero width
+	 * Check if we need to combine characters. This could be a Korean
-	 * (set above), a modifier character (with an existing Unicode
+	 * Hangul Jamo character, zero width (set above), a modifier character
-	 * character) or a previous ZWJ.
+	 * (with an existing Unicode character) or a previous ZWJ.
 	 */
 	if (!zero_width) {
-		if (utf8_is_modifier(ud)) {
+		switch (hanguljamo_check_state(&last.data, ud)) {
-			if (last.data.size < 2)
+		case HANGULJAMO_STATE_NOT_COMPOSABLE:
-				return (0);
+			return (1);
-			force_wide = 1;
+		case HANGULJAMO_STATE_CHOSEONG:
-		} else if (!utf8_has_zwj(&last.data))
 			return (0);
+		case HANGULJAMO_STATE_COMPOSABLE:
+			break;
+		case HANGULJAMO_STATE_NOT_HANGULJAMO:
+			if (utf8_is_modifier(ud)) {
+				if (last.data.size < 2)
+					return (0);
+				force_wide = 1;
+			} else if (!utf8_has_zwj(&last.data))
+				return (0);
+			break;
+		}
 	}
 
 	/* Check if this combined character would be too long. */

tmux.h

@@ -677,6 +677,14 @@ enum utf8_state {
 	UTF8_ERROR
 };
 
+/* State for processing of Korean characters. */
+enum hanguljamo_state {
+	HANGULJAMO_STATE_NOT_HANGULJAMO,
+	HANGULJAMO_STATE_CHOSEONG,
+	HANGULJAMO_STATE_COMPOSABLE,
+	HANGULJAMO_STATE_NOT_COMPOSABLE
+};
+
 /* Colour flags. */
 #define COLOUR_FLAG_256 0x01000000
 #define COLOUR_FLAG_RGB 0x02000000
@@ -3465,6 +3473,8 @@ int		 utf8_has_zwj(const struct utf8_data *);
 int		 utf8_is_zwj(const struct utf8_data *);
 int		 utf8_is_vs(const struct utf8_data *);
 int		 utf8_is_modifier(const struct utf8_data *);
+enum hanguljamo_state hanguljamo_check_state(const struct utf8_data *,
+		    const struct utf8_data *);
 
 /* procname.c */
 char   *get_proc_name(int, char *);

utf8-combined.c

@@ -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);
+}