ww source: reject malformed UTF-8

internal/wwpackage/package.ww

@@ -699,23 +699,34 @@ fn pkgident(c: u8) bool = {
 	return c >= '0' && c <= '9';
 };
 
-fn pkgnulerrors(path: str, src: str) bool = {
+// Validate the selected physical source before package-clause classification.
+// Go's source reader consumes malformed UTF-8 one byte at a time; raw NUL is
+// the independent ASCII source error. Reporting both in byte order preserves
+// the decoder's diagnostic precedence and raw-byte columns.
+fn pkgsourceerrors(path: str, src: str) bool = {
 	let i: i32 = 0;
 	let ln: i32 = 1;
 	let cl: i32 = 1;
 	let found: bool = false;
 	for (i < src.len) {
-		if (src[i] == 0u8) {
+		let c: u8 = src[i];
+		let width: i32 = 1;
+		if (c >= 128u8) { width = pkgutf8width(src, i); };
+		if (c == 0u8) {
 			pkgfailsource(path, ln, cl, "invalid NUL character");
 			found = true;
+		} else { if (c >= 128u8 && width == 1) {
+			pkgfailsource(path, ln, cl, "invalid UTF-8 encoding");
+			found = true;
 		};
-		if (src[i] == '\n') {
+		};
+		if (c == '\n') {
 			ln += 1;
 			cl = 1;
 		} else {
-			cl += 1;
+			cl += width;
 		};
-		i += 1;
+		i += width;
 	};
 	return found;
 };
@@ -2299,7 +2310,7 @@ export fn packagecommand(args: []str) int = {
 			pkgfailpath(ds.paths[i], "invalid or missing package clause");
 			return pkgteststatusfail(explicitstatus, compileonly);
 		};
-		if (pkgnulerrors(ds.paths[i], body)) {
+		if (pkgsourceerrors(ds.paths[i], body)) {
 			return pkgteststatusfail(explicitstatus, compileonly);
 		};
 		if (!pkgclause(body, &pn)) {