edo

commit 1abe3bee69f88fe171dc46eb86f9c81049b69632
parent 420caa07a66b197ee51017a96eba8ae98f099c11
Author: Claudio Alessi <smoppy@gmail.com>
Date:   Sat, 17 Jan 2026 19:00:53 +0100

Improve handling of non-spacing characters.

UTF-8 code points are rendered using their hex values with the exception
of combining code points, identified by the utf8comp library which is
now a dependency.

Diffstat:
M
config.mk
 | 
2
+-
M
tui.c
 | 
62
++++++++++++++++++++++++++++++++++++++------------------------
M
utf8.c
 | 
19
++++++++++++++-----
M
utf8.h
 | 
1
+

4 files changed, 54 insertions(+), 30 deletions(-)
diff --git a/config.mk b/config.mk
@@ -8,7 +8,7 @@ MANPREFIX = ${PREFIX}/share/man
 # flags
 CPPFLAGS = -D_DEFAULT_SOURCE -DVERSION=\"${VERSION}\"
 CFLAGS   = -std=c99 -g -pedantic -Wall -O0 ${CPPFLAGS}
-LDFLAGS  = -lgrapheme
+LDFLAGS  = -lgrapheme -lutf8proc
 #CFLAGS  = -std=c99 -pedantic -Wall -Wno-deprecated-declarations -Os ${CPPFLAGS}
 
 # compiler and linker
diff --git a/tui.c b/tui.c
@@ -124,6 +124,18 @@ tui_frame_flush(void) {
 	ab_flush(&frame);
 }
 
+/* XXX move to utils? */
+int
+hexlen(unsigned int n) {
+	int len = 0;
+
+	do {
+		len++;
+		n >>= 4;
+	} while(n > 0);
+	return len;
+}
+
 int
 tui_text_width(char *s, int len, int x) {
 	int tabstop = 8;
@@ -140,22 +152,24 @@ tui_text_width(char *s, int len, int x) {
 
 		wc = -1;
 		if(compat_mode) {
-			if(cp == 0x200D) wc = 6;
-			else if(IS_RIS(cp)) wc = 2;
+			if(IS_RIS(cp)) wc = 2;
 		} else {
 			/* force 2 cells width for emoji followed by VS16 */
 			int nxi = i + step;
 			if(nxi < len) {
 				unsigned int nxcp;
 				utf8_decode(s + nxi, len - nxi, &nxcp);
-
 				if(nxcp == 0xFE0F && ((cp >= 0x203C && cp <= 0x3299) || cp >= 0x1F000))
 					wc = 2;
 			}
 		}
 
-		if(wc == -1) wc = wcwidth(cp);
+		if(wc < 0) wc = wcwidth(cp);
+		assert(wc != -1);
+
 		if(wc > 0) w += wc;
+		else if(!utf8_is_combining(cp)) w += hexlen(cp) + 2; /* 2 for < and > */
+		//else w += hexlen(cp) + 2; /* 2 for < and > */
 	}
 	return w;
 }
@@ -200,35 +214,39 @@ tui_draw_line_compat(UI *ui, int x, int y, Cell *cells, int count) {
 			int step = utf8_decode(txt + o, cells[i].len - o, &cp);
 			int cw = cells[i].width;
 			int neederase = cells[i].len > 1 && (cells[i].width == 1 || IS_RIS(cp));
-			char tag[] = "<200d>";
-			int tagw = 6;
-			int offset = 0;
 
 			switch(cp) {
-			case 0x200D:
-				/* TODO: */
-				if(cells[i].flags & CELL_TRUNC_L) {
-					offset = tagw - cells[i].width;
-					if(offset < 0) offset = 0;
-				}
-				int len_to_print = cells[i].width;
-				assert(offset + len_to_print <= tagw);
-
-				if(len_to_print > 0) ab_write(&frame, tag + offset, len_to_print);
-				break;
 			case '\t':
 				for(int t = 0; t < cells[i].width; t++)
 					ab_write(&frame, " ", 1);
 				break;
 			default:
-				if(cells[i].flags & CELL_TRUNC_L) {
+				cw = wcwidth(cp);
+				if(cw < 0) break;
+
+				if(!cw) {
 					if(!cells[i].width) break;
+
+					char tag[16];
+					snprintf(tag, sizeof tag, "<%0x>", cp);
+
+					if(cells[i].flags & CELL_TRUNC_L) {
+						cw = tui_text_width(txt + o, cells[i].len, 0);
+						o = cw - cells[i].width;
+						if(o < 0) o = 0;
+					}
+					{ const char t[] = ESC"[48;5;233m"; ab_write(&frame, t, sizeof t - 1); }
+					cw = ab_printf(&frame, "%.*s", cells[i].width, tag + o);
+					{ const char t[] = ESC"[0m"; ab_write(&frame, t, sizeof t - 1); }
+					break;
+				}
+
+				if(cells[i].flags & CELL_TRUNC_L) {
 					ab_write(&frame, "<", 1);
 					cw = 1;
 					break;
 				}
 				if(cells[i].flags & CELL_TRUNC_R) {
-					if(!cells[i].width) break;
 					ab_write(&frame, ">", 1);
 					cw = 1;
 					break;
@@ -239,10 +257,6 @@ tui_draw_line_compat(UI *ui, int x, int y, Cell *cells, int count) {
 					ab_write(&frame, t, sizeof t - 1);
 				}
 
-				/* don't expect negative values here */
-				cw = wcwidth(cp);
-				if(cw < 0) cw = 0;
-
 				ab_write(&frame, txt + o, step);
 
 				/* to preserve coherence between terminals always split RIS
diff --git a/utf8.c b/utf8.c
@@ -1,10 +1,9 @@
 #define _XOPEN_SOURCE
 #include <wchar.h>
 #include <grapheme.h>
+#include <utf8proc.h>
 
-int utf8_len(char *buf, int len);
-int utf8_decode(char *buf, int len, unsigned int *cp);
-size_t utf8_len_compat(char *buf, int len);
+#include "utf8.h"
 
 size_t
 utf8_len_compat(char *buf, int len) {
@@ -12,10 +11,12 @@ utf8_len_compat(char *buf, int len) {
 	int step, i;
 
 	i = step = utf8_decode(buf, len, &cp);
-	if(wcwidth(cp) < 0) return step;
+	//if(wcwidth(cp) >= 0) return step;
+	if(!utf8_is_combining(cp) && wcwidth(cp) < 0) return step;
 	while(i < len) {
 		step = utf8_decode(buf + i, len - i, &cp);
-		if(cp == 0x200D || wcwidth(cp)) break;
+		//if(wcwidth(cp) >= 0) break;
+		if(!utf8_is_combining(cp) || wcwidth(cp)) break;
 		i += step;
 	}
 	return i;
@@ -30,3 +31,11 @@ int
 utf8_decode(char *buf, int len, unsigned int *cp) {
 	return grapheme_decode_utf8(buf, len, cp);
 }
+
+int
+utf8_is_combining(unsigned int cp) {
+	const utf8proc_property_t *prop = utf8proc_get_property(cp);
+
+	return (prop->category == UTF8PROC_CATEGORY_MN
+			|| prop->category == UTF8PROC_CATEGORY_ME);
+}
diff --git a/utf8.h b/utf8.h
@@ -9,3 +9,4 @@
 int utf8_len(char *buf, int len);
 size_t utf8_len_compat(char *buf, int len);
 int utf8_decode(char *buf, int len, unsigned int *cp);
+int utf8_is_combining(unsigned int cp);