Function TestParseStringLoneSurrogateMatchesEncodingJSON

escape_test.go:294–337

Verifies: SYS-REQ-014 [differential] — for a corpus of lone-surrogate inputs, ParseString must byte-match encoding/json.Unmarshal (the oracle exercised by the FuzzJSONStructureAware checkParseStringGate finder).

(t *testing.T)

// inputs, ParseString must byte-match encoding/json.Unmarshal (the oracle
// exercised by the FuzzJSONStructureAware checkParseStringGate finder).
func TestParseStringLoneSurrogateMatchesEncodingJSON(t *testing.T) {
	corpus := []string{
		`\uDB29`,
		`\uDB29A7FA`,
		`\uDB29A7FA71 a`,
		`\uD800`,
		`\uD83D`,
		`\uDBFF`,
		`\uDC00`,
		`\uDFFF`,
		`\uDE03`,
		`\uD800\u0041`,
		`\uD800\uD800`,
		`\uD800\uDBFF`,
		`\uD800\uE000`,
		`\uD800\uFFFF`,
		`\uDB29\uDC00`,
		`\uD83D\uDE03 more`,
		`\uD800\uDC00`,
		`\u0000\uD800\u0000`,
	}
	for _, esc := range corpus {
		// Reference oracle: encoding/json unescapes the quoted string.
		var want string
		jsonInput := `"` + esc + `"`
		if err := json.Unmarshal([]byte(jsonInput), &want); err != nil {
			// encoding/json rejects malformed escapes (e.g. truncated "\u"). The
			// corpus above is chosen to all be accepted; if one is rejected we
			// want to know rather than silently skip.
			t.Errorf("encoding/json rejected corpus input %q: %v", jsonInput, err)
			continue
		}

		got, err := ParseString([]byte(esc))
		if err != nil {
			t.Errorf("ParseString(%q) returned error %v; encoding/json accepted it as %q", esc, err, want)
			continue
		}
		if got != want {
			t.Errorf("ParseString(%q) divergence:\n  got  = %q (% x)\n  want = %q (% x)",
				esc, got, []byte(got), want, []byte(want))
		}
	}
}