encoding/json/v2: fix UnmarshalDecode regression with EOF

When EOF is encountered within jsontext.Decoder stream without
starting to parse any token, UnmarshalDecode should report EOF,
rather than converting it into ErrUnexpectedEOF.

This fixes a regression introduced by https://go.dev/cl/689919.

This change only affects code compiled under goexperiment.jsonv2.

Fixes golang#74835

Change-Id: I7e8e57ab11b462c422c538503ed8c6b91ead53bd
Reviewed-on: https://go-review.googlesource.com/c/go/+/692175
LUCI-TryBot-Result: Go LUCI <[email protected]>
Reviewed-by: Sean Liao <[email protected]>
Reviewed-by: David Chase <[email protected]>
Reviewed-by: Jake Bailey <[email protected]>
Reviewed-by: Dmitri Shuralyov <[email protected]>
Auto-Submit: Joseph Tsai <[email protected]>

Author: dsnet

src/encoding/json/v2/arshal.go

@@ -470,7 +470,7 @@ func unmarshalDecode(in *jsontext.Decoder, out any, uo *jsonopts.Struct, last bo
 	// was validated before attempting to unmarshal it.
 	if uo.Flags.Get(jsonflags.ReportErrorsWithLegacySemantics) {
 		if err := export.Decoder(in).CheckNextValue(last); err != nil {
-			if err == io.EOF {
+			if err == io.EOF && last {
 				offset := in.InputOffset() + int64(len(in.UnreadBuffer()))
 				return &jsontext.SyntacticError{ByteOffset: offset, Err: io.ErrUnexpectedEOF}
 			}
@@ -487,7 +487,7 @@ func unmarshalDecode(in *jsontext.Decoder, out any, uo *jsonopts.Struct, last bo
 		if !uo.Flags.Get(jsonflags.AllowDuplicateNames) {
 			export.Decoder(in).Tokens.InvalidateDisabledNamespaces()
 		}
-		if err == io.EOF {
+		if err == io.EOF && last {
 			offset := in.InputOffset() + int64(len(in.UnreadBuffer()))
 			return &jsontext.SyntacticError{ByteOffset: offset, Err: io.ErrUnexpectedEOF}
 		}

src/encoding/json/v2/arshal_test.go

@@ -9413,6 +9413,51 @@ func TestUnmarshalDecodeOptions(t *testing.T) {
 	}
 }
 
+func TestUnmarshalDecodeStream(t *testing.T) {
+	tests := []struct {
+		in   string
+		want []any
+		err  error
+	}{
+		{in: ``, err: io.EOF},
+		{in: `{`, err: &jsontext.SyntacticError{ByteOffset: len64(`{`), Err: io.ErrUnexpectedEOF}},
+		{in: `{"`, err: &jsontext.SyntacticError{ByteOffset: len64(`{"`), Err: io.ErrUnexpectedEOF}},
+		{in: `{"k"`, err: &jsontext.SyntacticError{ByteOffset: len64(`{"k"`), JSONPointer: "/k", Err: io.ErrUnexpectedEOF}},
+		{in: `{"k":`, err: &jsontext.SyntacticError{ByteOffset: len64(`{"k":`), JSONPointer: "/k", Err: io.ErrUnexpectedEOF}},
+		{in: `{"k",`, err: &jsontext.SyntacticError{ByteOffset: len64(`{"k"`), JSONPointer: "/k", Err: jsonwire.NewInvalidCharacterError(",", "after object name (expecting ':')")}},
+		{in: `{"k"}`, err: &jsontext.SyntacticError{ByteOffset: len64(`{"k"`), JSONPointer: "/k", Err: jsonwire.NewInvalidCharacterError("}", "after object name (expecting ':')")}},
+		{in: `[`, err: &jsontext.SyntacticError{ByteOffset: len64(`[`), Err: io.ErrUnexpectedEOF}},
+		{in: `[0`, err: &jsontext.SyntacticError{ByteOffset: len64(`[0`), Err: io.ErrUnexpectedEOF}},
+		{in: ` [0`, err: &jsontext.SyntacticError{ByteOffset: len64(` [0`), Err: io.ErrUnexpectedEOF}},
+		{in: `[0.`, err: &jsontext.SyntacticError{ByteOffset: len64(`[`), JSONPointer: "/0", Err: io.ErrUnexpectedEOF}},
+		{in: `[0. `, err: &jsontext.SyntacticError{ByteOffset: len64(`[0.`), JSONPointer: "/0", Err: jsonwire.NewInvalidCharacterError(" ", "in number (expecting digit)")}},
+		{in: `[0,`, err: &jsontext.SyntacticError{ByteOffset: len64(`[0,`), Err: io.ErrUnexpectedEOF}},
+		{in: `[0:`, err: &jsontext.SyntacticError{ByteOffset: len64(`[0`), Err: jsonwire.NewInvalidCharacterError(":", "after array element (expecting ',' or ']')")}},
+		{in: `n`, err: &jsontext.SyntacticError{ByteOffset: len64(`n`), Err: io.ErrUnexpectedEOF}},
+		{in: `nul`, err: &jsontext.SyntacticError{ByteOffset: len64(`nul`), Err: io.ErrUnexpectedEOF}},
+		{in: `fal `, err: &jsontext.SyntacticError{ByteOffset: len64(`fal`), Err: jsonwire.NewInvalidCharacterError(" ", "in literal false (expecting 's')")}},
+		{in: `false`, want: []any{false}, err: io.EOF},
+		{in: `false0.0[]null`, want: []any{false, 0.0, []any{}, nil}, err: io.EOF},
+	}
+	for _, tt := range tests {
+		d := jsontext.NewDecoder(strings.NewReader(tt.in))
+		var got []any
+		for {
+			var v any
+			if err := UnmarshalDecode(d, &v); err != nil {
+				if !reflect.DeepEqual(err, tt.err) {
+					t.Errorf("`%s`: UnmarshalDecode error = %v, want %v", tt.in, err, tt.err)
+				}
+				break
+			}
+			got = append(got, v)
+		}
+		if !reflect.DeepEqual(got, tt.want) {
+			t.Errorf("`%s`: UnmarshalDecode = %v, want %v", tt.in, got, tt.want)
+		}
+	}
+}
+
 // BenchmarkUnmarshalDecodeOptions is a minimal decode operation to measure
 // the overhead options setup before the unmarshal operation.
 func BenchmarkUnmarshalDecodeOptions(b *testing.B) {