fix escaped { is not handled in format string passed to format()

Author: rhysd

expr_sema.go

@@ -28,6 +28,55 @@ func ordinal(i int) string {
 	return fmt.Sprintf("%d%s", i, suffix)
 }
 
+// parseFormatFuncSpecifiers parses the format string passed to `format()` calls.
+// https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/evaluate-expressions-in-workflows-and-actions#format
+func parseFormatFuncSpecifiers(f string, n int) map[int]struct{} {
+	ret := make(map[int]struct{}, n)
+
+	type state int
+	const (
+		init  state = iota // Initial state
+		brace              // {
+		digit              // 0..9
+	)
+
+	var cur state
+	var start int
+	for i, r := range f {
+		switch cur {
+		case init:
+			switch r {
+			case '{':
+				cur = brace
+				start = i + 1 // `+ 1` because `i` points char '{'
+			}
+		case brace:
+			switch {
+			case '0' <= r && r <= '9':
+				cur = digit
+			default:
+				cur = init
+			}
+		case digit:
+			switch {
+			case '0' <= r && r <= '9':
+				// Do nothing
+			case r == '{':
+				cur = brace
+				start = i + 1
+			case r == '}':
+				i, _ := strconv.Atoi(f[start:i])
+				ret[i] = struct{}{}
+				cur = init
+			default:
+				cur = init
+			}
+		}
+	}
+
+	return ret
+}
+
 // Functions
 
 // FuncSignature is a signature of function, which holds return and arguments types.
@@ -798,16 +847,10 @@ func (sema *ExprSemanticsChecker) checkBuiltinFunctionCall(n *FuncCallNode, _ *F
 		}
 		l := len(n.Args) - 1 // -1 means removing first format string argument
 
-		// Find all placeholders in format string
-		holders := make(map[int]struct{}, l)
-		for _, m := range reFormatPlaceholder.FindAllString(lit.Value, -1) {
-			i, _ := strconv.Atoi(m[1 : len(m)-1])
-			holders[i] = struct{}{}
-		}
+		holders := parseFormatFuncSpecifiers(lit.Value, l)
 
 		for i := 0; i < l; i++ {
-			_, ok := holders[i]
-			if !ok {
+			if _, ok := holders[i]; !ok {
 				sema.errorf(n, "format string %q does not contain placeholder {%d}. remove argument which is unused in the format string", lit.Value, i)
 				continue
 			}

expr_sema_test.go

@@ -719,6 +719,16 @@ func TestExprSemanticsCheckOK(t *testing.T) {
 			input:    "!!('foo' || 10) && 20",
 			expected: NumberType{},
 		},
+		{
+			what:     "escaped braces in format string",
+			input:    "format('hello {{1}} {0}', 42)",
+			expected: StringType{},
+		},
+		{
+			what:     "format specifier is escaped",
+			input:    "format('hello {{{0}', 'world')", // First {{ is escaped. {0} is not escaped
+			expected: StringType{},
+		},
 	}
 
 	allSPFuncs := []string{}
@@ -1023,6 +1033,20 @@ func TestExprSemanticsCheckError(t *testing.T) {
 				`format string "{0}" does not contain placeholder {1}`,
 			},
 		},
+		{
+			what:  "format specifier is escaped",
+			input: "format('hello {{0}}', 'world')",
+			expected: []string{
+				"does not contain placeholder {0}",
+			},
+		},
+		{
+			what:  "format specifier is still escaped",
+			input: "format('hello {{{{0}}', 'world')", // First {{ is escaped. {{0}} is still escaped
+			expected: []string{
+				"does not contain placeholder {0}",
+			},
+		},
 		{
 			what:  "undefined matrix value",
 			input: "matrix.bar",
@@ -1660,3 +1684,95 @@ func TestBuiltinGlobalVariableTypesValidation(t *testing.T) {
 		testObjectPropertiesAreInLowerCase(t, ty)
 	}
 }
+
+func TestParseFormatSpecifiers(t *testing.T) {
+	tests := []struct {
+		what string
+		in   string
+		want []int // Specifiers in the `in` string
+	}{
+		{
+			what: "empty input",
+			in:   "",
+		},
+		{
+			what: "no specifier",
+			in:   "hello, world!",
+		},
+		{
+			what: "single specifier",
+			in:   "Hello{0}specifier",
+			want: []int{0},
+		},
+		{
+			what: "mutliple specifiers",
+			in:   "{0} {1}{2}x{3}}{4}!",
+			want: []int{0, 1, 2, 3, 4},
+		},
+		{
+			what: "many specifiers",
+			in:   "{0}{1}{2}{3}{4}{5}{6}{7}{8}{9}{10}!",
+			want: []int{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10},
+		},
+		{
+			what: "unordered",
+			in:   "{2}foo{4} {0}{3}",
+			want: []int{2, 4, 0, 3},
+		},
+		{
+			what: "uncontiguous",
+			in:   "{0} {2}foo{5} {1}",
+			want: []int{0, 2, 5, 1},
+		},
+		{
+			what: "unclosed",
+			in:   "{12foo",
+		},
+		{
+			what: "not digit",
+			in:   "{hello}",
+		},
+		{
+			what: "space in digits",
+			in:   "{1 2}",
+		},
+		{
+			what: "empty",
+			in:   "{}",
+		},
+		{
+			what: "specifier inside specifier",
+			in:   "{1{0}2}",
+			want: []int{0},
+		},
+		{
+			what: "escaped",
+			in:   "{{hello{{0}{{{{1}world}}",
+		},
+		{
+			what: "after escaped",
+			in:   "{{{{{0}",
+			want: []int{0},
+		},
+		{
+			what: "kuma-",
+			in:   "{・{ᴥ}・}",
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.what, func(t *testing.T) {
+			want := map[int]struct{}{}
+			for _, i := range tc.want {
+				want[i] = struct{}{}
+			}
+			have := parseFormatFuncSpecifiers(tc.in, len(tc.want))
+
+			if !cmp.Equal(want, have) {
+				t.Fatal(cmp.Diff(want, have))
+			}
+		})
+	}
+}
+
+// vim: nofoldenable