fix: correct error message for invalid bytes in comments

Author: WillLillis

lib/std/zig/AstGen.zig

@@ -13987,10 +13987,10 @@ fn lowerAstErrors(astgen: *AstGen) !void {
     const tok = parse_err.token + @intFromBool(parse_err.token_is_prev);
     const tok_start = token_starts[tok];
 
-    if (token_tags[tok] == .invalid and tree.source[tok_start] == '\\') {
+    if (token_tags[tok] == .invalid and (tree.source[tok_start] == '\\' or tree.source[tok_start] == '/')) {
+        const tok_len: u32 = @intCast(tree.tokenSlice(tok).len);
+        const tok_end = tok_start + tok_len;
         const bad_off = blk: {
-            const tok_len: u32 = @intCast(tree.tokenSlice(tok).len);
-            const tok_end = tok_start + tok_len;
             var idx = tok_start;
             while (idx < tok_end) : (idx += 1) {
                 switch (tree.source[idx]) {
@@ -14005,10 +14005,20 @@ fn lowerAstErrors(astgen: *AstGen) !void {
         try notes.append(gpa, try astgen.errNoteTokOff(tok, bad_off, "invalid byte: '{'}'", .{
             std.zig.fmtEscapes(tree.source[byte_abs..][0..1]),
         }));
+        const expected_tag = switch (tree.source[tok_start]) {
+            '\\' => std.zig.Token.Tag.multiline_string_literal_line,
+            '/' => blk: {
+                if (std.mem.indexOf(u8, tree.source[tok_start..tok_end], "///")) |idx| {
+                    if (idx == 0) break :blk std.zig.Token.Tag.doc_comment;
+                }
+                break :blk std.zig.Token.Tag.comment;
+            },
+            else => unreachable,
+        };
         const err: Ast.Error = .{
             .tag = Ast.Error.Tag.expected_token,
             .token = tok,
-            .extra = .{ .expected_tag = std.zig.Token.Tag.multiline_string_literal_line },
+            .extra = .{ .expected_tag = expected_tag },
         };
         msg.clearRetainingCapacity();
         try tree.renderError(err, msg.writer(gpa));

lib/std/zig/tokenizer.zig

@@ -137,6 +137,7 @@ pub const Token = struct {
         angle_bracket_angle_bracket_right_equal,
         tilde,
         number_literal,
+        comment,
         doc_comment,
         container_doc_comment,
         keyword_addrspace,
@@ -199,6 +200,7 @@ pub const Token = struct {
                 .eof,
                 .builtin,
                 .number_literal,
+                .comment,
                 .doc_comment,
                 .container_doc_comment,
                 => null,
@@ -327,6 +329,7 @@ pub const Token = struct {
                 .eof => "EOF",
                 .builtin => "a builtin function",
                 .number_literal => "a number literal",
+                .comment => "a comment",
                 .doc_comment, .container_doc_comment => "a document comment",
                 else => unreachable,
             };

test/cases/compile_errors/tab_inside_comment.zig

@@ -0,0 +1,9 @@
+// Some		comment
+export fn entry() void {}
+
+// error
+// backend=stage2
+// target=native
+//
+// :1:1: error: expected 'a comment', found invalid bytes
+// :1:8: note: invalid byte: '\t'

test/cases/compile_errors/tab_inside_doc_comment.zig

@@ -0,0 +1,9 @@
+/// Some doc		comment
+export fn entry() void {}
+
+// error
+// backend=stage2
+// target=native
+//
+// :1:1: error: expected 'a document comment', found invalid bytes
+// :1:13: note: invalid byte: '\t'

test/compile_errors.zig

@@ -225,7 +225,8 @@ pub fn addCases(ctx: *Cases, b: *std.Build) !void {
         const case = ctx.obj("invalid byte in comment", b.graph.host);
 
         case.addError("//\x01Q", &[_][]const u8{
-            ":1:1: error: expected type expression, found 'invalid token'",
+            ":1:1: error: expected 'a comment', found invalid bytes",
+            ":1:3: note: invalid byte: '\\x01'"
         });
     }