fix(vt): use Unicode 18 width tables instead of libc wcwidth

dvtm.c

@@ -379,13 +379,21 @@ drawbar(void) {
 	wchar_t wbuf[sizeof bar.text];
 	size_t numchars = mbstowcs(wbuf, bar.text, sizeof bar.text);
 
-	if (numchars != (size_t)-1 && (width = wcswidth(wbuf, maxwidth)) != -1) {
+	width = 0;
+	for (size_t i = 0; numchars != (size_t)-1 && i < numchars && width != -1; i++) {
+		int w = vt_wcwidth(wbuf[i]);
+		width = w < 0 ? -1 : width + w;
+		if (width > maxwidth)
+			break;
+	}
+
+	if (numchars != (size_t)-1 && width != -1) {
 		int pos;
 		for (pos = 0; pos + width < maxwidth; pos++)
 			addch(' ');
 
 		for (size_t i = 0; i < numchars; i++) {
-			pos += wcwidth(wbuf[i]);
+			pos += vt_wcwidth(wbuf[i]);
 			if (pos > maxwidth)
 				break;
 			addnwstr(wbuf+i, 1);

test/test-vt.c

@@ -164,6 +164,25 @@ static void test_device_attributes(void)
 	}
 }
 
+static void test_wcwidth(void)
+{
+	term = "WCWIDTH";
+	CHECK(vt_wcwidth(L'a') == 1);
+	CHECK(vt_wcwidth(L'\0') == 0);
+	CHECK(vt_wcwidth(L'\n') == -1);
+	CHECK(vt_wcwidth(0x4E2D) == 2);
+	CHECK(vt_wcwidth(0x1F600) == 2);
+	CHECK(vt_wcwidth(0x1FAE9) == 2);
+	CHECK(vt_wcwidth(0x0301) == 0);
+	CHECK(vt_wcwidth(0x200D) == 0);
+	CHECK(vt_wcwidth(0x00E9) == 1);
+
+	Vt *t = vt_create(3, 10, 0);
+	feed(t, "\xf0\x9f\xab\xa9x");
+	CHECK(t->buffer->curs_col == 3);
+	vt_destroy(t);
+}
+
 int main(void)
 {
 	setlocale(LC_CTYPE, "C.UTF-8");
@@ -191,6 +210,7 @@ int main(void)
 	setenv("TERMINFO", dir, 1);
 
 	test_device_attributes();
+	test_wcwidth();
 	test_term("xterm", COLOR_RED, COLOR_RED, COLOR_BLACK, COLOR_BLACK);
 	test_term("xterm-256color", 196, 196, 16, 16);
 	if (direct) {

tools/gen-wcwidth.py

@@ -0,0 +1,23 @@
+#!/usr/bin/env python3
+"""Generate wcwidth_table.h from the Unicode tables of the wcwidth package.
+
+usage: pip install wcwidth && tools/gen-wcwidth.py > wcwidth_table.h
+"""
+import wcwidth.table_wide as wide
+import wcwidth.table_zero as zero
+
+version = max(wide.WIDE_EASTASIAN, key=lambda v: tuple(map(int, v.split('.'))))
+
+
+print('typedef struct { unsigned int lo, hi; } WidthRange;\n')
+
+
+def emit(name, ranges):
+    print(f'static const WidthRange {name}[] = {{')
+    for lo, hi in ranges:
+        print(f'\t{{ 0x{lo:04X}, 0x{hi:04X} }},')
+    print('};\n')
+
+
+emit('wcwidth_zero', [r for r in zero.ZERO_WIDTH[version] if r[1] >= 0x80])
+emit('wcwidth_wide', wide.WIDE_EASTASIAN[version])

vt.c

@@ -41,6 +41,7 @@
 #endif
 
 #include "vt.h"
+#include "wcwidth_table.h"
 
 #ifdef _AIX
 # include "forkpty-aix.c"
@@ -1132,6 +1133,35 @@ static void interpret_csi(Vt *t)
 	}
 }
 
+static bool in_ranges(const WidthRange *r, size_t n, unsigned int c)
+{
+	size_t lo = 0, hi = n;
+	while (lo < hi) {
+		size_t mid = lo + (hi - lo) / 2;
+		if (c < r[mid].lo)
+			hi = mid;
+		else if (c > r[mid].hi)
+			lo = mid + 1;
+		else
+			return true;
+	}
+	return false;
+}
+
+int vt_wcwidth(wchar_t wc)
+{
+	unsigned int c = wc;
+	if (c < 0x20 || (c >= 0x7f && c < 0xa0))
+		return c ? -1 : 0;
+	if (c < 0x300)
+		return 1;
+	if (in_ranges(wcwidth_zero, sizeof wcwidth_zero / sizeof *wcwidth_zero, c))
+		return 0;
+	if (in_ranges(wcwidth_wide, sizeof wcwidth_wide / sizeof *wcwidth_wide, c))
+		return 2;
+	return 1;
+}
+
 /* Interpret an 'index' (IND) sequence */
 static void interpret_csi_ind(Vt *t)
 {
@@ -1402,7 +1432,7 @@ static void put_wc(Vt *t, wchar_t wc)
 					wc = gc;
 			}
 			width = 1;
-		} else if ((width = wcwidth(wc)) < 1) {
+		} else if ((width = vt_wcwidth(wc)) < 1) {
 			width = 1;
 		}
 		Buffer *b = t->buffer;
@@ -1576,7 +1606,7 @@ void vt_draw(Vt *t, WINDOW *win, int srow, int scol)
 				size_t len = wcrtomb(buf, cell->text, NULL);
 				if (len != (size_t)-1) {
 					waddnstr(win, buf, len);
-					if (wcwidth(cell->text) > 1)
+					if (vt_wcwidth(cell->text) > 1)
 						j++;
 				}
 			} else {

vt.h

@@ -21,6 +21,7 @@
 #include <curses.h>
 #include <stdbool.h>
 #include <sys/types.h>
+#include <wchar.h>
 
 #ifndef NCURSES_MOUSE_VERSION
 #define mmask_t unsigned long
@@ -56,6 +57,7 @@ void vt_dirty(Vt*);
 void vt_draw(Vt*, WINDOW *win, int startrow, int startcol);
 int vt_color_get(Vt*, int fg, int bg);
 int vt_color_reserve(int fg, int bg);
+int vt_wcwidth(wchar_t wc);
 
 void vt_scroll(Vt*, int rows);
 void vt_noscroll(Vt*);