expr: Add support for UTF-8 character constants

4 files changed, 11 insertions, 14 deletions

diff --git a/expr.c b/expr.c
index 27acbcf..6ff7dd4 100644
--- a/expr.c
+++ b/expr.c
@@ -596,11 +596,11 @@ primaryexpr(struct scope *s)
 		break;
 	case TCHARCONST:
 		src = tok.lit;
-		t = &typeint;
 		switch (*src) {
 		case 'L': ++src; t = targ->typewchar; break;
-		case 'u': ++src; t = &typeushort;     break;
-		case 'U': ++src; t = &typeuint;       break;
+		case 'u': ++src; t = *src == '8' ? ++src, &typeuchar : &typeushort; break;
+		case 'U': ++src; t = &typeuint; break;
+		default: t = &typeint;
 		}
 		assert(*src == '\'');
 		++src;
diff --git a/scan.c b/scan.c
index cd85d9d..bd0435c 100644
--- a/scan.c
+++ b/scan.c
@@ -388,18 +388,11 @@ again:
 	case 'u':
 		s->usebuf = true;
 		nextchar(s);
-		switch (s->chr) {
-		case '\'':
-			return charconst(s);
-		case '8':
-			if (s->buf.str[0] != 'u')
-				break;
+		if (s->buf.str[0] == 'u' && s->chr == '8')
 			nextchar(s);
-			if (s->chr != '"')
-				break;
-			/* fallthrough */
-		case '"':
-			return stringlit(s);
+		switch (s->chr) {
+		case '\'': return charconst(s);
+		case '"': return stringlit(s);
 		}
 		return ident(s);
 	case EOF:
diff --git a/test/char-const-u8.c b/test/char-const-u8.c
new file mode 100644
index 0000000..9f20452
--- /dev/null
+++ b/test/char-const-u8.c
@@ -0,0 +1,3 @@
+unsigned char u8 = u8'a';
+_Static_assert(__builtin_types_compatible_p(__typeof__(u8'b'), unsigned char),
+	"UTF-8 character constant has incorrect type");
diff --git a/test/char-const-u8.qbe b/test/char-const-u8.qbe
new file mode 100644
index 0000000..2d199a4
--- /dev/null
+++ b/test/char-const-u8.qbe
@@ -0,0 +1 @@
+export data $u8 = align 1 { b 97, }