Restore code to handle wcwidth failure so that unknown codepoints still

do the most likely right thing. GitHub issue 3427, patch based on an
diff from Jesse Luehrs in GitHub issue 3003.
This commit is contained in:
nicm 2023-01-08 22:15:30 +00:00
parent cb51942669
commit 7ced0a03d2

11
utf8.c
View File

@ -226,9 +226,16 @@ utf8_width(struct utf8_data *ud, int *width)
case 0: case 0:
return (UTF8_ERROR); return (UTF8_ERROR);
} }
log_debug("UTF-8 %.*s is %08X", (int)ud->size, ud->data, (u_int)wc);
*width = wcwidth(wc); *width = wcwidth(wc);
log_debug("UTF-8 %.*s %#x, wcwidth() %d", (int)ud->size, ud->data, log_debug("wcwidth(%08X) returned %d", (u_int)wc, *width);
(u_int)wc, *width); if (*width < 0) {
/*
* C1 control characters are nonprintable, so they are always
* zero width.
*/
*width = (wc >= 0x80 && wc <= 0x9f) ? 0 : 1;
}
if (*width >= 0 && *width <= 0xff) if (*width >= 0 && *width <= 0xff)
return (UTF8_DONE); return (UTF8_DONE);
return (UTF8_ERROR); return (UTF8_ERROR);