Merge pull request #2854 from prometheus/promql-rune

Check for invalid utf-8 in lexer strings.

promql/lex.go

@@ -713,6 +713,8 @@ Loop:
 		switch l.next() {
 		case '\\':
 			lexEscape(l)
+		case utf8.RuneError:
+			return l.errorf("invalid UTF-8 rune")
 		case eof, '\n':
 			return l.errorf("unterminated quoted string")
 		case l.stringOpen:
@@ -728,6 +730,8 @@ func lexRawString(l *lexer) stateFn {
 Loop:
 	for {
 		switch l.next() {
+		case utf8.RuneError:
+			return l.errorf("invalid UTF-8 rune")
 		case eof:
 			return l.errorf("unterminated raw string")
 		case l.stringOpen:

promql/lex_test.go

@@ -396,6 +396,13 @@ var tests = []struct {
 	}, {
 		input: `]`, fail: true,
 	},
+	// Test encoding issues.
+	{
+		input: "\"\xff\"", fail: true,
+	},
+	{
+		input: "`\xff`", fail: true,
+	},
 	// Test series description.
 	{
 		input: `{} _ 1 x .3`,

promql/parse_test.go

@@ -904,6 +904,10 @@ var testExpr = []struct {
 		// TODO(fabxc): willingly lexing wrong tokens allows for more precrise error
 		// messages from the parser - consider if this is an option.
 		errMsg: "unexpected character inside braces: '>'",
+	}, {
+		input:  "some_metric{a=\"\xff\"}",
+		fail:   true,
+		errMsg: "parse error at char 15: invalid UTF-8 rune",
 	}, {
 		input:  `foo{gibberish}`,
 		fail:   true,
@@ -1373,6 +1377,10 @@ var testExpr = []struct {
 		input:  "rate(some_metric)",
 		fail:   true,
 		errMsg: "expected type range vector in call to function \"rate\", got instant vector",
+	}, {
+		input:  "label_replace(a, `b`, `c\xff`, `d`, `.*`)",
+		fail:   true,
+		errMsg: "parse error at char 23: invalid UTF-8 rune",
 	},
 	// Fuzzing regression tests.
 	{